Every lesson rests on published research. Here is each source and what it does and doesn’t support. Lessons are checked against their sources; none has been reviewed by a subject expert yet, and Friendly College hasn’t been tested in a learner study.
What it supports: Supports retrieval practice as a learning principle. Findings vary by study conditions and do not validate this app or establish broader life outcomes.
Open the published source (opens in a new tab)What it supports: Concerns verbal recall and the relationship between practice spacing and retention intervals. It does not prescribe one universal schedule or lesson duration.
Open the published source (opens in a new tab)What it supports: Supports linking identifiable situations with goal-directed responses across studied contexts. Feasibility, goals, resources, and context still matter.
Open the published source (opens in a new tab)What it supports: Supports evaluating progress monitoring as a goal-management practice. It does not establish that app engagement, self-reported check-ins, or this lesson's format cause lasting change.
Open the published source (opens in a new tab)What it supports: Illustrates tracking the timing of inflows and outflows. United States benefit examples and legal context do not transfer automatically to other countries.
Open the published source (opens in a new tab)What it supports: Describes job-related questions about actual past behaviour or hypothetical situations. It does not establish that this practice exercise improves hiring outcomes.
Open the published source (opens in a new tab)What it supports: Self-reported automaticity varied across participants practising chosen behaviours. This is not a randomized test of this app or proof of a universal habit deadline.
Open the published source (opens in a new tab)What it supports: Examined perceived understanding and other immediate conversation ratings. It did not establish lasting relationship improvement, clinical benefit, or safety in coercive relationships.
Open the published source (opens in a new tab)What it supports: Recognises diverse forms of physical activity and the role of context and access. It does not determine a safe individualized routine from a goal picker.
Open the published source (opens in a new tab)What it supports: Tested a structured scientific decision approach with Italian startups. Findings do not establish profitability of a particular idea or this short lesson.
Open the published source (opens in a new tab)What it supports: Four randomized trials with 759 firms found effects on idea termination and a nonlinear effect on radical pivots. The abstract does not report revenue effects.
Open the published source (opens in a new tab)What it supports: Incubation effects depend on task and conditions. Laboratory task results do not guarantee creative quality or prescribe one ideal break duration.
Open the published source (opens in a new tab)What it supports: Supports the 2024 global estimate of adults who were overweight, the statement that eating patterns are shaped and often constrained by environmental, commercial and societal conditions, the aim of making healthier options available, affordable and desirable, and limiting sugar-sweetened drinks as a preventive action. It is population-level guidance, not evidence that rearranging one person’s surroundings changes health outcomes.
Open the published source (opens in a new tab)What it supports: Pooled 299 effect sizes from field experiments on seven kinds of healthy-eating nudge. Behaviourally oriented nudges, such as convenience enhancements, had larger average effects than cognitive ones such as labels, and effects were larger for reducing unhealthy eating than for increasing healthy eating. Settings were organisational, such as restaurants, cafeterias and grocery stores, so self-arranged home or desk changes were not directly tested, and effects varied by setting and country.
Open the published source (opens in a new tab)What it supports: Supports the 2024 global estimates of adults aged 30–79 with hypertension and those unaware of it, that most people feel no symptoms, that detection requires measurement by a health professional, that diagnosis uses readings on two different days, that professional evaluation matters even with home devices, and when to seek care immediately. It is public-health guidance, not an individual diagnostic tool.
Open the published source (opens in a new tab)What it supports: Cited only for the published overview, which highlights the urgent need to scale up detection, treatment and control of hypertension, especially at the primary health care level. The full report text was not reviewed for this draft, and the report sets policy priorities for countries rather than testing individual check-up planning.
Open the published source (opens in a new tab)What it supports: Supports the global estimate that 1 in 6 people are affected by loneliness, WHO’s description of loneliness, and its suggestion of simple everyday steps such as reaching out to a friend. It summarises a global report and does not test any individual technique.
Open the published source (opens in a new tab)What it supports: Supports the finding that, among randomised studies, interventions addressing maladaptive social cognition were the most successful, and that single-group and non-randomised designs showed larger effects than randomised ones. It did not test single invitations or self-guided exercises.
Open the published source (opens in a new tab)What it supports: Supports the global figures of 1.2 billion tobacco users and more than 7 million deaths a year, that most smokers aware of the dangers want to quit, and that counselling and medication can more than double the chance of quitting. It gives no individual treatment advice.
Open the published source (opens in a new tab)What it supports: Supports high-quality evidence that face-to-face individual counselling from a trained counsellor increased quitting compared with minimal contact, with a smaller, moderate-quality benefit when added to medication. It did not evaluate a self-guided trigger-planning exercise.
Open the published source (opens in a new tab)What it supports: Supports the global scale figure of more than 1 billion people living with mental health conditions, with anxiety and depressive disorders the most common. It summarises two WHO reports and says nothing about grounding or any self-help technique.
Open the published source (opens in a new tab)What it supports: Describes the grounding skill (notice thoughts and feelings, slow down and connect with the body, refocus on the world around you) and suggests brief, planned practice; WHO states the guide is informed by evidence and extensive field testing. The guide is not itself a trial, and this lesson teaches only the first of its five skills.
Open the published source (opens in a new tab)What it supports: Supports the scale figure: in 2024 survey data from low- and middle-income economies, only 56 percent of adults could easily access extra money for an emergency within 30 days, unchanged since 2021. Financial health questions were not asked in high-income economies, and the report does not test savings techniques.
Open the published source (opens in a new tab)What it supports: Clients of one Philippine bank offered a commitment savings account had higher average savings balances after 12 months than a control group (an intent-to-treat estimate significant at the 90 percent level); 28.4 percent of those offered took it up. It does not test home buffers, automatic saving or this lesson, and the authors note that reduced access to money does not necessarily improve welfare.
Open the published source (opens in a new tab)What it supports: Supplies the global estimate of about 2.6 million deaths caused by alcohol in 2019 and states that even low levels of drinking carry some risk. It does not set individual drinking limits or evaluate this exercise.
Open the published source (opens in a new tab)What it supports: Moderate-quality evidence that brief conversations with health workers, typically including feedback, advice and a personal plan, reduced weekly drinking at one year among primary care patients not seeking help for alcohol. Effects were modest, with little change in binge frequency or drinking days; digital interventions and self-guided counting were not assessed.
Open the published source (opens in a new tab)What it supports: Advises having at least a few drink-free days every week, and warns that people with physical withdrawal symptoms should get medical advice before stopping because stopping too quickly without proper help can be dangerous. United Kingdom guidance, not a trial or individual medical advice.
Open the published source (opens in a new tab)What it supports: Supplies the 2024 global jobs-gap estimate of about 402 million people who want to work but do not have a job. These are modelled global estimates; they do not describe any one country, sector or person’s prospects.
Open the published source (opens in a new tab)What it supports: Across 47 evaluated interventions, participants had higher odds of employment than controls, and interventions promoted employment only when they combined skill development with motivation enhancement. It evaluated organised programmes, not a self-written weekly plan, and cannot guarantee results in any labour market.
Open the published source (opens in a new tab)What it supports: Supports the cited prevalence estimate, the definition of procrastination and the strongest correlates (task aversiveness, task delay, low self-efficacy, impulsiveness, distractibility), plus the review's theoretical suggestion that proximal goals may shorten delays. Findings are mainly correlational, most included studies used young university students, and the 15%–20% adult figure is cited from earlier sources rather than a worldwide survey.
Open the published source (opens in a new tab)What it supports: Supports the statement that psychological treatments for procrastination showed small average benefits (12 trials, g = 0.34) with significant between-study variation, and that interventions in this field often draw on goal setting, sub-goals and implementation intentions. It did not isolate the effect of a tiny first step, outcomes were self-reported, and the authors note some risk of bias.
Open the published source (opens in a new tab)What it supports: Supports the estimate that 30.5% of adults reported sleeping less than 7 hours on average in a 24-hour period in 2024. Data are self-reported and nationally representative of the United States civilian noninstitutionalized population only, so they are not a global figure.
Open the published source (opens in a new tab)What it supports: Supports the panel's conclusions that consistency of sleep onset and offset timing is important for health, safety and performance, and that weekend or non-work-day catch-up sleep may be beneficial when workday sleep is insufficient. It is an expert consensus based on 63 publications, not a trial of a fixed-wake-time routine; the National Sleep Foundation summary describes consistent timing as associated with improved outcomes.
Open the published source (opens in a new tab)What it supports: Supports the study size, that 45 percent of interventions significantly raised weekly gym visits by 9 to 27 percent during the four weeks, that the top performer gave a bonus of 125 points (about US$0.09) for returning after a missed workout (an estimated 27 percent increase over the placebo control, or 16 percent over the basic planning, reminders and points programme), that all 24 effective treatments included planning, reminders and incentives, that only 8 percent had significant effects in the four weeks after the programme, that forecasts failed to predict the best interventions, and that top effects are probably overestimated. It covers gym visits among members of one U.S. gym chain who enrolled in the programme, and it did not test written comeback plans.
Open the published source (opens in a new tab)What it supports: Supports the sample of 2,250 adults, mind wandering in 46.9 percent of samples, lower happiness when minds wandered, no gain in happiness from pleasant topics and lower happiness for neutral or unpleasant ones, and time-lag analyses that strongly suggested wandering was generally a cause rather than merely a consequence of unhappiness. It is correlational, about 74 percent of the sample lived in the United States, each person gave an average of 7.9 samples, and it did not test any attention practice.
Open the published source (opens in a new tab)What it supports: Supports an overall small association between unprompted thought (mind wandering) and worse affective well-being, whose size and direction changed with thought content, intentionality and measurement method, including questionnaires versus experience sampling, so it does not on its own confirm experience-sampling findings such as the wandering-mind study's. It reports associations, not causes, and did not test noticing or returning attention.
Open the published source (opens in a new tab)What it supports: Supports the survey’s size (more than 150,000 adults in more than 140 economies, interviewed in 2014), its four concepts, its definition of financial literacy as at least three of four correct, the 33 percent figure, and its note that many short-term credit users do not fully understand how fast compounding can inflate amounts owed. It measures knowledge at one point in time, not borrowing decisions, and tests no lesson or technique.
Open the published source (opens in a new tab)What it supports: Supports the finding that, across 76 randomised experiments with more than 160,000 people, financial education programmes on average had positive causal effects on financial knowledge and later financial behaviours, similar in size to educational interventions in other fields. It reports average effects across varied programmes and does not test this lesson or these borrowing sums.
Open the published source (opens in a new tab)What it supports: Supports the survey finding that giving to charity in the past month was associated with higher subjective well-being across 136 countries (more than 200,000 respondents, 2006 to 2008; positive in 120 countries), and the experimental findings that recalling or making a purchase for others raised happiness or positive affect compared with spending on oneself. The survey is correlational; only it used nationally representative samples, and the experiments were smaller, often with students.
Open the published source (opens in a new tab)What it supports: Supports the statements that the preregistered evidence appears somewhat mixed, that the authors judge the emotional benefits of prosocial spending robust and replicable in large samples, and that benefits are particularly likely when people have some choice about whether or how to give and understand how their generosity makes a difference. It is a review by researchers behind the original work, it concerns spending money rather than giving time, and only the abstract was inspected.
Open the published source (opens in a new tab)What it supports: Supports the 14 potentially modifiable risk factors named in the 2024 update, the estimate that nearly half of dementias could theoretically be prevented by eliminating these 14 risk factors (population attributable fractions calculated from worldwide risk and prevalence literature), that it is never too early or too late to reduce risk, and the Commission’s calls for accessible hearing aids and accessible vision screening and treatment. The Commission notes that some associations may be only partly causal and that the evidence is still disproportionately from high-income countries. Wording was checked against the published report, whose headline sentence differs slightly from the author accepted manuscript. It gives population estimates, not individual risk predictions, and did not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the estimate that 57 million people were living with dementia worldwide in 2021 and WHO’s list of actions that can reduce the risk of cognitive decline and dementia, including using hearing aids where needed. It is public-health guidance, not an individual risk assessment, and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the pooled analysis of 218 randomised trials (14,170 participants who met clinical cut-offs for major depression); moderate reductions in depression compared with active controls for walking or jogging, yoga and strength training; effects proportional to the intensity prescribed, with light activity such as walking still showing clinically meaningful effects; lower dropout for yoga and strength training than active controls; and confidence (CINeMA) rated low for walking or jogging and very low for other treatments, with only one study at low risk of bias. Results are averages across trials of prescribed programmes; it did not test self-guided plans or this lesson.
Open the published source (opens in a new tab)What it supports: Supports the estimate that 5.2% of adults worldwide experienced depression in 2023, WHO’s self-care suggestions (keep doing activities you used to enjoy, stay connected to friends and family, exercise regularly even if it is just a short walk), that there are effective treatments for depression, and the advice to seek care for symptoms. It is public-health guidance, not a diagnosis or individual treatment plan, and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports that, among 47 couples from the Harvard Study’s college group, being in a satisfying relationship went with a stronger positive daily link between time with one’s partner and mood, that in less satisfying marriages more time with a partner went with more negative mood, and that marital satisfaction buffered links between health stressors and mood. Participants were all white and socioeconomically advantaged, the men were aged 80 to 88, and the authors note the design cannot show with certainty that daily contact shapes mood.
Open the published source (opens in a new tab)What it supports: Supports the description of the study: two groups of men followed over the last 80 years, 268 Harvard graduates from the classes of 1939 to 1944 and 456 men who grew up in the inner-city neighbourhoods of Boston, and a newer study of their children. It is a descriptive web page, not a report of findings.
Open the published source (opens in a new tab)What it supports: Supports that, across 148 studies with an average follow-up of 7.5 years, participants with stronger social relationships had a 50 percent increased likelihood of survival (odds ratio 1.50). Studies were mainly from North America and Europe, most did not measure relationship quality, and the authors note that causality is not easily established.
Open the published source (opens in a new tab)What it supports: Supports the study size, that the top relationship-specific predictors of relationship quality were perceived-partner commitment, appreciation, sexual satisfaction, perceived-partner satisfaction and conflict, that relationship-specific variables predicted up to 45 percent of variance at baseline against 21 percent for individual differences, that individual differences and partner reports added nothing beyond a person’s own relationship-specific reports, and that change in relationship quality was largely unpredictable. Appreciation here means how appreciative people felt towards their partner. Data were self-reports from Western countries, and the findings are predictive, not causal.
Open the published source (opens in a new tab)What it supports: Supports that people who felt more appreciated by their partners reported being more appreciative of them, that more appreciative people reported more responsiveness and commitment and were more likely to remain in their relationships over time, and that outside observers rated appreciative partners as more responsive and committed in laboratory interactions. It did not experimentally test instructions to express appreciation.
Open the published source (opens in a new tab)What it supports: Supports the 10-year figures for South Africa Equity funds to 30 June 2026: 96.48 percent underperformed the S&P South Africa 50 and 95.07 percent the S&P South Africa Composite Capped (Report 1a), from 142 funds at the start of the period (Report 2), with funds that were merged or liquidated counted as underperformers, and fund returns net of fees but excluding loads and entry fees. It reports past results for one fund category against indices; the lesson does not use it to explain why funds underperformed, and it does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the explanation that fees and expenses reduce the amount of money in a portfolio earning a return, and that over time even small ongoing fees have a big impact, illustrated with a 100,000-dollar investment growing 4 percent a year for 20 years under fees of 0.25, 0.50 and 1.00 percent. It is U.S. educational guidance with an illustrative example, not a study of fund returns, and it does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the 2026 edition’s figures: over the 10 years ended 31 December 2025, the average dollar invested in U.S. mutual funds and exchange-traded funds earned 8.7 percent a year, about 1.2 percentage points a year less than the funds’ 9.9 percent aggregate annual total return, a gap explained by the timing and magnitude of investors’ purchases and sales; gaps tended to be smaller for less volatile funds (0.4 points a year for the least-volatile quintile) and larger for more volatile ones (more than 2.0 points for the most-volatile quintile). The author notes that impulsive buying high or selling low can hold back returns, but that routine practices such as investing part of every paycheck can also open a gap, so it is not proof of individual investors’ fallibility. It covers U.S. funds only and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the statement that, since 1900, equities and bonds have on several occasions lost more than 70 percent in real terms, in a yearbook covering 35 markets. The public summary gives no recovery times, so the lesson does not say how quickly markets recovered. It describes history and states that historical performance is no guarantee of future performance; it does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the sample (66,465 households with accounts at a large discount broker, 1991 to 1996, about 42 percent of them in the western United States), the net annualised returns of the highest-turnover fifth (11.4 percent), the average household (16.4 percent) and a value-weighted market index (17.9 percent), the small differences in gross returns across turnover groups, the authors’ conclusion that the cost and frequency of trading, not portfolio selections, explain the poor performance, and the comparison of each household with its own beginning-of-year portfolio. It is observational, covers one broker’s customers in the 1990s, and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports that, using the same discount-broker data as Barber and Odean (2000), with over 35,000 households from February 1991 to January 1997, men traded 45 percent more than women, trading reduced men’s net returns by 2.65 percentage points a year against 1.72 for women, and both men and women lowered their returns by trading compared with holding their beginning-of-year portfolios. It is observational, covers one broker’s customers in the 1990s, and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the sample (all 19,646 individuals who first day traded mini-Ibovespa futures in Brazil from 2013 to 2015, from data provided by the Brazilian securities regulator), that 1,551 of them day traded on more than 300 days, that 97 percent of those 1,551 lost money net of exchange and brokerage fees, that only 17 of them (1.1 percent) earned more than the Brazilian minimum wage, and that the authors found no evidence of learning. It is an unpublished working paper (the authors list it as a working paper; a 2020 FGV version reports the same 97 percent figure); its figures exclude income taxes and costs such as courses and trading platforms; it covers one contract in one country and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports that, among day traders in Taiwan from 1992 to 2006 (about 450,000 individuals in an average year), approximately 20 percent earned profits net of fees in a typical year, but less than 1 percent could predictably and reliably earn positive abnormal returns, measured against a market index, net of fees. It also found that a small group of top-ranked traders kept outperforming the next year. It covers one market and period, net returns assume a discounted commission plus the transaction tax, and it does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports WHO’s recommendation that adults do muscle-strengthening activities at moderate or greater intensity that involve all major muscle groups on 2 or more days a week (a strong recommendation, moderate-certainty evidence); the older-adult advice on varied multicomponent activity emphasising functional balance and strength on 3 or more days a week; the good-practice advice to start with small amounts and gradually increase; and that people who develop new symptoms when increasing activity should consult a healthcare provider. It is population guidance, found no evidence of a dose-response with higher volumes of muscle-strengthening, and is not an individual exercise prescription or a test of this lesson.
Open the published source (opens in a new tab)What it supports: Supports that 16 prospective cohort studies (minimum follow-up 2 years; adults aged 18 or over without severe health conditions; 3,809 to 479,856 participants per study; most in the USA) were pooled; that muscle-strengthening activities were associated with a 10–17% lower risk of all-cause mortality, cardiovascular disease, total cancer, diabetes and lung cancer; that the largest risk reduction for all-cause mortality, cardiovascular disease and total cancer was seen at about 30–60 minutes a week, with the effect of higher volumes unclear (J-shaped associations); and the authors’ view that the current recommendation of at least 2 days a week could be reasonable. The studies are observational and mostly self-reported, so they show associations, not causation; the authors note possible publication bias, residual confounding and limited generalisability.
Open the published source (opens in a new tab)What it supports: Supports that randomised trials of healthy adults (at least 6 weeks; athletes, military personnel, people with or at risk of comorbidities and unsupervised home-based training excluded) compared 12 prescriptions differing in load, sets and weekly frequency; that all prescriptions were superior to non-exercise control for strength and hypertrophy; that higher-load (over 80% of one-repetition maximum) prescriptions maximised strength gains while all prescriptions comparably promoted hypertrophy; and the authors’ view that healthy adults can adopt a prescription of their choice and preference. It did not study progression over time, self-kept logs or this lesson, and its results are averages across trials.
Open the published source (opens in a new tab)What it supports: Supports that the position stand synthesised 137 systematic reviews (more than 30,000 participants) of healthy adults; that progression is not necessary to achieve beneficial outcomes, but increasing the stimulus (such as load, volume, training frequency, exercise selection or duration, even if only slightly) is likely a requirement for continued longer-term progress; that where inexperience and safety concerns mean loads are necessarily low, progression is a requirement for a meaningful benefit; that training to momentary muscular failure is not necessary for benefits; and its conclusion that healthy adults should perform progressive resistance training. It is expert guidance for healthy adults, does not address training logs, and is not an individual prescription or a test of this lesson.
Open the published source (opens in a new tab)What it supports: Supports the pooled analysis of 347 studies of positive psychology interventions, defined as interventions aiming to enhance well-being through pathways consistent with positive psychology theory, with over 72,000 participants from clinical and non-clinical child and adult populations in 41 countries; interventions averaged ten sessions over six weeks; at post-test, a small to medium effect on well-being (g = 0.39); gains maintained at three-month follow-up; and larger benefits for people in non-western countries with clinical problems in longer programmes with several exercises. It pooled many kinds of multi-session programmes and did not test the three good things exercise on its own or this lesson.
Open the published source (opens in a new tab)What it supports: Supports the design (577 adults recruited through a website, randomly assigned to five exercises or an early-memories placebo; 411 completed all five follow-ups), the three good things instructions (write down three things that went well each day and their causes, every night for one week), the finding that this group was happier and less depressed than at baseline from the one-month follow-up through three and six months, that all groups improved at the immediate posttest while the placebo group returned to baseline within a week, and that continuing the exercise beyond the week mediated long-term benefits. A single trial with a convenience sample that was largely well educated, White and financially comfortable, mildly depressed on average and motivated to become happier; outcomes were self-reported. It did not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the sample of 19,806 people from the Monitor of Engagement with the Natural Environment survey of England (2014/15–2015/16), weighted to be nationally representative; that, compared with no nature contact in the last week, the likelihood of reporting good health or high well-being became significantly greater with contact of at least 120 minutes; that positive associations peaked between 200 and 300 minutes with no further gain; that the pattern was consistent across key groups including older adults and those with long-term health issues; that it did not matter how the 120 minutes was achieved; and the authors’ view that the threshold is tentative. The data are cross-sectional and self-reported, so reverse causation cannot be ruled out; time in one’s own garden and routine shopping trips were excluded; one country.
Open the published source (opens in a new tab)What it supports: Supports that, among 50 studies of structured outdoor activities in green or blue space for community-based adults (16 randomised trials), meta-analysis of the randomised trials found improved depressive mood and positive affect and reduced negative affect compared with controls, with green exercise, gardening and nature-based therapy among the activities. Few trials fed each analysis, heterogeneity was high for depressive mood, risk of bias for random sequence generation was generally unclear, most studies measured only short-term effects, and the studies came from Europe, Asia, North America and Australia. It did not test self-planned visits or this lesson.
Open the published source (opens in a new tab)What it supports: Supports the 195-country, 1990–2017 scope; the estimate that 11 million deaths in 2017 (22% of all deaths among adults) were attributable to dietary risks; high sodium, low whole grains and low fruit as the leading dietary risks globally; the authors’ note that the leading risks for mortality were mostly diets low in whole grains, fruit, nuts and seeds, and vegetables, and that promoting these foods might have a greater effect than policies only targeting sugar and fat; that intake of healthy foods was below the optimal level in almost all regions; and that a diet low in fruits was the leading dietary risk in southern sub-Saharan Africa (a GBD region that includes South Africa, per IHME’s GHDx country list). These are population attributable estimates; the authors note dietary data were from mixed sources and not available for all countries, and that evidence for diet–disease links was mostly observational. It did not test any individual change or this lesson.
Open the published source (opens in a new tab)What it supports: Supports WHO’s description of a healthy diet as including a variety of fruits, vegetables, pulses (such as lentils, chickpeas, beans and dried peas) and whole grains (such as unprocessed maize, millet, oats, wheat and brown rice), and its practical tips for reducing salt: limiting salt and high-sodium condiments such as bouillon when cooking, not having salt or high-sodium sauces on the table, limiting salty snacks and choosing lower-sodium products. It is general public-health guidance, not individual dietary advice, and does not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the trial’s size (7,447 adults aged 55 to 80 in Spain at high cardiovascular risk without cardiovascular disease), the three diet groups, the median follow-up of 4.8 years, the hazard ratios of 0.69 (olive oil) and 0.72 (nuts) against a reduced-fat control diet for the composite of heart attack, stroke or cardiovascular death (a relative difference of about 30%), the foods the Mediterranean groups were encouraged to eat, and the authors’ account of the protocol deviations (425 household members assigned without randomisation, one site assigning 467 people by clinic, and inconsistent use of randomisation tables at another site), their withdrawal of the 2013 report and the similar results of the reanalysis. The authors note that participants were at high risk, that most already ate a diet similar to the Mediterranean one, and that the main between-group differences were the extra olive oil and nuts. It did not test single food swaps or this lesson.
Open the published source (opens in a new tab)What it supports: Supports the wider, more cautious picture: for primary prevention, low-quality evidence of little or no effect of the PREDIMED intervention compared with a low-fat diet on cardiovascular or total mortality over 4.8 years, moderate-quality evidence of fewer strokes, the note that PREDIMED was retracted and re-analysed after concerns about randomisation at two of 11 sites, and the conclusion that some uncertainty remains. Searches ran to September 2018, and it did not test single food swaps or this lesson.
Open the published source (opens in a new tab)What it supports: Supports Ford’s 1913 turnover of 370 percent, the hiring of 50,448 men to keep an average workforce of 13,623, daily absenteeism of about 10 percent (roughly 1,300 to 1,400 replacements a day), the previous minimum of about $2.34 for a nine-hour day, the $5 rate for an eight-hour day announced in January 1914, turnover of 54 percent in 1914 and 16 percent in 1915, absenteeism falling from 10 to 2.5 percent, and the authors’ conclusion that productivity and profits rose. The authors note the case may be atypical. Full text read in the NBER working-paper version (w2101); the journal record was checked on RePEc. Pages follow the RePEc journal record (S57–S86, issue 4, October 1987). No DOI was confirmed on RePEc, NBER or by search, so none is recorded.
Open the published source (opens in a new tab)What it supports: Supports experiments with the moving assembly line through 1913 and into 1914 cutting Model T build time from 12½ hours to 93 minutes, the 370 percent turnover, the 5 January 1914 announcement doubling pay to five dollars a day, thousands of applicants coming to Detroit and being turned away with fire hoses in icy January weather, and the conditions workers had to meet to qualify, and the Sociological Department that monitored employees’ habits beyond the workplace. The article gives the old wage as $2.30 with a $2.70 bonus for qualifying workers; the lesson’s $2.34 figure is cited to Raff and Summers. The museum’s $5 Day expert set was also read; it gives $2.34 for a nine-hour day but states turnover as 380 percent, so the lesson uses the 370 percent figure given by both this article and Raff and Summers.
Open the published source (opens in a new tab)What it supports: Supports a significant negative relationship between total turnover rates and organisational performance (rho = −.15), a similar link for voluntary turnover (−.15) and reduction-in-force turnover (−.17), and almost none for involuntary turnover (−.01). The abstract does not say what the N of 309,245 counts (the studies are organisation-level), so the lesson does not call it people. The data are mainly correlational and do not test pay rises or this lesson’s exercise.
Open the published source (opens in a new tab)What it supports: Supports that relative pay was modestly linked to lower collective turnover (mean r = −.13) and that turnover was negatively related to performance outcomes including customer satisfaction and production efficiency (each about −.22). Links are mainly correlational and do not show that a pay rise causes lower turnover.
Open the published source (opens in a new tab)What it supports: Supports the book’s central argument as summarised by its publisher: that the belief that rewards like money are the best motivator is mistaken, and that high performance and satisfaction come from the desire to direct our own lives, get better at meaningful work and serve a larger purpose, drawing on decades of research and offering techniques for applying it. Bestseller status: the publisher (Penguin Random House, Riverhead imprint, first published 29 December 2009) describes it as a New York Times bestseller, and the author’s official site states that it spent 159 weeks on the New York Times main and extended bestseller lists. The book itself was not read for this lesson and is not quoted.
Open the published source (opens in a new tab)What it supports: Supports that engagement-, completion- and performance-contingent rewards, and expected and tangible rewards, reduced free-choice intrinsic motivation (d = −0.40, −0.36 and −0.28 for the three contingencies), while positive feedback increased free-choice behaviour (d = 0.33). It covers experiments, not workplace pay systems.
Open the published source (opens in a new tab)What it supports: Supports a combined sample of 212,468 (the abstract does not define the unit), that intrinsic motivation was a medium to strong predictor of performance (rho = .21 to .45) whether or not incentives were present, that it predicted more unique variance in quality while incentives better predicted quantity, and the authors’ conclusion that incentives and intrinsic motivation are not necessarily antagonistic and are best considered together.
Open the published source (opens in a new tab)What it supports: Supports that Nightingale arrived at Scutari on 4 November 1854 with 38 nurses, that soldiers lay on bare floors among vermin, that cholera and typhus were widespread and injured soldiers were 7 times more likely to die of disease in hospital than on the battlefield, that she returned in August 1856, and that in May 1857 her request for a formal investigation was granted, leading to the Royal Commission on the health of the army. Its hospital mortality percentages are not used, because other sources dispute them. It is a secondary biography (last updated October 2003), not a primary archive.
Open the published source (opens in a new tab)What it supports: Supports that after her return to England in 1856 Nightingale and William Farr worked together to organise data from the Crimea, that most deaths came not from battle but from preventable causes (mainly zymotic, that is contagious, diseases and inadequate hospital sanitation), that she pressed for a Royal Commission on army mortality, and that her diagrams were designed to shock the viewer and compare deaths before and after the Sanitary Commission. Page range and DOI are omitted because the records conflict: the journal’s provisional PDF prints pages 1–16 and DOI 10.2436/20.8080.02.77, the author’s copy prints pages 1–18 and the same DOI, and the UPCommons repository record (peer reviewed, issued 15 June 2021) gives pages 3–18 and DOI 10.2436/20.8080.02.106; neither DOI could be resolved on the day. It dates the Sanitary Commission to around April 1855; McDonald and other accounts give its arrival as March 1855, which the lesson uses. It is a history of graphics, not a mortality study.
Open the published source (opens in a new tab)What it supports: Supports Nightingale’s own figure that 10,053 men, or 60 percent per annum, died of disease alone in seven months (on an average strength of 28,939), that the British government sent a Sanitary Commission that cleaned the hospitals’ sewers, drains and water supply, that she marked its arrival in March 1855 on her charts, that she never attributed the fall in deaths to nursing, that she worked closely with Dr William Farr and his staff at the General Register Office, and that Crimean mortality data came from several conflicting sources. The newsletter is not a peer-reviewed journal, so the peer-reviewed Friendly and Andrews article supports the key facts too. Friendly and Andrews say the commission was formed around April 1855; the lesson follows McDonald’s March 1855 arrival date. Lesson facts that rest on this source alone: Nightingale’s 10,053 figure, the Sanitary Commission’s work, that she marked its arrival on her charts and never credited nursing. The article gives 3 conflicting data sources on one page and 6 on another.
Open the published source (opens in a new tab)What it supports: Supports that the review included 140 randomised trials of audit and feedback (a summary of clinical performance over a period), that after excluding high-risk-of-bias studies the weighted median adjusted gain in compliance with desired practice was 4.3 percentage points (interquartile range 0.5 to 16), that effects on patient outcomes were small or unclear, and that feedback may work better when baseline performance is low, the source is a supervisor or colleague, it is given more than once, in verbal and written form, with explicit targets and an action plan. It covers health professionals in general, not nurses alone, and did not test this lesson.
Open the published source (opens in a new tab)What it supports: Supports the book’s central argument as described by its publisher and reviewers: skilled professionals still make serious errors as their work grows more complex, longer training and better technology have not prevented them, and short checklists, drawn from aviation and applied in hospitals and in fields such as construction and investing, can prevent them; the publisher description reports a brief surgical checklist cutting fatalities. Publishers Weekly describes Gawande as a Harvard Medical School professor and New Yorker writer. Bestseller status: a reproduction of the New York Times Hardcover Nonfiction Bestsellers list dated 24 January 2010 (NextReads/LibraryAware newsletter, copyright New York Times Syndication Sales Corp.) shows it at No. 3, and EarlyWord (18 January 2010) reported it debuted at No. 3 that week; Publishers Weekly’s hardcover nonfiction list of 18 January 2010 had it at No. 9 in its first week. Publishers Weekly’s review says Gawande gets carried away and treats mundane checklists as breakthroughs; the lesson does not use this criticism. The book is an argument, not a study, and is described here in the lesson’s own words.
Open the published source (opens in a new tab)What it supports: Supports that 8 hospitals in 8 cities (including Toronto and Ifakara) introduced a 19-item surgical safety checklist, that data came from 3,733 patients before and 3,955 after, that the death rate fell from 1.5 to 0.8 percent and inpatient complications from 11.0 to 7.0 percent, and that Gawande was a co-author. It is not randomised, so other changes over time may have contributed.
Open the published source (opens in a new tab)What it supports: Supports that in 3-month periods before and after checklist adoption, 101 Ontario hospitals performed 109,341 and 106,370 procedures, that adjusted operative mortality went from 0.71 to 0.65 percent (odds ratio 0.91, not significant) and complications from 3.86 to 3.82 percent (not significant), and that mandatory-style adoption was not associated with significant reductions. It measured early effects of a policy, not how well each team used the checklist.
Open the published source (opens in a new tab)What it supports: Supports that Kahneman was drafted in 1954 as a second lieutenant, served a year as a platoon leader, moved to the Psychology branch of the Israel Defense Forces, helped assess officer candidates whose assessors felt confident although monthly statistics showed their predictions barely matched later performance (which he called the illusion of validity), and at 21 was asked to set up an interviewing system for screening combat recruits starting in 1955; that the earlier interviewers gave global ratings with poor validity and were proud of their skill; that someone gave him Meehl's 1954 book, which had a big effect on him; that his structured interview used questions about civilian life to rate six dimensions, including masculine pride and sense of obligation; that interviewers felt like unthinking robots and that he faced a near-mutiny; that he let them add a global evaluation after the six ratings; that validity rose to close to .30 from about .10, the global evaluation being by far the most predictive rating; and that the system remained in use with little modification for many decades. The page does not give the year of his first degree (Britannica does). It is Kahneman's recollection written decades later, not an independent record of the army data; a search for an independent scholarly account found only later retellings that rest on Kahneman's own writing.
Open the published source (opens in a new tab)What it supports: Supports that Kahneman was born on 5 March 1934 in Tel Aviv, studied psychology at the Hebrew University of Jerusalem (B.A., 1954) and the University of California, Berkeley (Ph.D., 1961), shared the 2002 Nobel Prize for Economics, published the best-selling Thinking, Fast and Slow in 2011, and died on 27 March 2024. It does not describe his army service or the interview system.
Open the published source (opens in a new tab)What it supports: Supports that mechanical prediction was on average about 10 percent more accurate than clinical prediction, substantially more accurate in roughly a third to a half of the studies (33% to 47%, depending on the analysis), while clinical prediction was substantially more accurate in only about 6 to 16 percent, and that the advantage held across tasks, judge types, experience and data types, with clinicians doing relatively worse when interview data were included. It pools studies of varying quality and did not test this lesson's exercise.
Open the published source (opens in a new tab)What it supports: Supports that earlier range-restriction corrections overstated validities, that revised mean validities fall by about 0.10 to 0.20, that structured interviews rank first among selection procedures, and that the revised operational validity estimates are 0.42 for structured interviews and 0.19 for unstructured interviews (Table 3). The estimates depend on the authors' correction choices and pooled studies; they do not test army interviews or this lesson's exercise.
Open the published source (opens in a new tab)What it supports: Library catalogue record (Farrar, Straus and Giroux, New York, 2011, 1st edition, 499 pages) supports that the book describes a fast, intuitive, emotional mode of thinking and a slower, more deliberate, logical one, examines the strengths and biases of fast thinking, discusses when intuitions can be trusted, and offers techniques to reduce common errors in business and personal decisions. Bestseller status: the New York Times paperback nonfiction list for the week ending 24 March 2018, as republished by WRAL, shows it at number 4 after 164 weeks on the list; Britannica also calls it best-selling. The book is a popular synthesis, not a new study.
Open the published source (opens in a new tab)What it supports: Supports that the authors compare the heuristics-and-biases and naturalistic decision-making views of intuition and conclude that whether intuitive judgements are skilled depends on how predictable the environment is and whether the person has had an adequate chance to learn its regularities, and that subjective confidence is not a reliable indicator of accuracy. It is a conceptual review, not a new experiment.
Open the published source (opens in a new tab)What it supports: Supports that a replication of Bargh, Chen and Burrows (1996) with 120 participants (against 60 in the original experiments 2a and 2b combined) and infrared timing found no difference in walking speed between age-primed and unprimed participants, and that in a second experiment with 50 participants, in which experimenters' expectations were manipulated, primed participants walked more slowly only when tested by experimenters who expected them to. It tests one priming experiment, not the whole book.
Open the published source (opens in a new tab)What it supports: Supports that chapter 4 of Thinking, Fast and Slow described the elderly-walking priming study, in which students primed with old-age words walked to an elevator more slowly, that independent replications have since failed, and that Kahneman later conceded he had placed too much faith in underpowered studies and that behavioural priming effects cannot be as large and robust as his chapter suggested. It reports Kahneman's concession second-hand rather than reproducing his original statement, and it is not a peer-reviewed study.
Open the published source (opens in a new tab)What it supports: Supports that Darwin went to Edinburgh in 1825 to study medicine and disliked it, entered Christ's College, Cambridge, was taught field geology by Adam Sedgwick, served as the Beagle's naturalist on a voyage of about five years, read Malthus in September 1838, received Wallace's essay on 18 June 1858, that the two men's papers were read at the Linnean Society on 1 July 1858, and that the Origin appeared in 1859. The page carries no publication date; the year recorded is that of its last update shown on the page (31 October 2025), checked 2 November 2026. It does not itself give 1837 for Darwin's London notebooks; that comes from M004B. It does not test any principle about success.
Open the published source (opens in a new tab)What it supports: Supports that in 1837, living in London and just off the Beagle, Darwin began tackling the origin of new species in pocket notebooks recording observations, readings and conversations (including with his barber, dog breeders, sheep farmers and gardeners); that his first theory, drawing on Erasmus Darwin, worked through a process akin to sexual reproduction; that in September 1838 he abandoned it after reading Malthus's Essay on the Principle of Population (first published 1798); that Malthus argued limited resources restrict population growth; and that Darwin applied this to the whole living world, seeing that the best-adapted individuals would survive to reproduce and that gradual inherited change could produce new species. Published 7 July 2022. Not a test of any principle.
Open the published source (opens in a new tab)What it supports: Supports that in February 1858, ill with intermittent fever at Ternate in the Moluccas, Wallace thought about how species arise, recalled Malthus's Essay on Population read about a dozen years earlier, reasoned that checks on population would weed out the less fit, wrote the idea out and posted it to Darwin, and that the two papers were read at the Linnean Society on 1 July 1858. It is Wallace's recollection 45 years later. His location is disputed: the essay is headed Ternate, but George Beccaloni (Wallace Fund) argues from Wallace's field journal and species notebook that he was at Dodinga on Gilolo (Halmahera) for most of February 1858, so the lesson says only 'in the Moluccas'. The 1858 dates are corroborated on the 1858 dates by van Wyhe (M004A) and by the Wallace Letters project page.
Open the published source (opens in a new tab)What it supports: Supports that across 17.9 million papers, with hits defined as the top 5 percent by citations within 8 years, papers combining high median conventionality with high tail novelty had a hit rate of 9.11 per 100 (described as about twice as likely to be highly cited), against 5.82 for high conventionality and low novelty, 5.33 for low conventionality and high novelty, and 2.05 for low on both. Correlational, uses citations as the measure of impact, and does not study individual careers or education.
Open the published source (opens in a new tab)What it supports: The book argues that broad experience and later specialisation serve people better than early specialisation in complex, unpredictable fields, partly because breadth helps people connect ideas across domains, while early specialisation suits stable settings such as chess. Bestseller status: Penguin Random House describes it as David Epstein's '#1 New York Times bestseller Range' on its page for the young readers' edition. Publishers Weekly records Riverhead, 352 pages, ISBN 978-0-7352-1448-4, and its review called the case studies fascinating but said the quick movement between them can create the impression of evidence in search of a thesis; that review is not cited in the lesson text because it is not a package source with a confirmed date. The book is described, not quoted, and is not itself a study.
Open the published source (opens in a new tab)What it supports: Supports that world-class adult athletes, compared with national-class peers, had more multisport practice in childhood and adolescence, started their main sport later, accumulated less main-sport practice and progressed more slowly at first; that higher-performing junior athletes showed the opposite pattern (earlier start, more main-sport practice, less other-sport practice, faster early progress); and that Nobel laureates had multidisciplinary study or work experience and slower early progress than national-level award winners. Correlational, mostly about sport, and does not test Range itself.
Open the published source (opens in a new tab)What it supports: Supports that highly novel papers (those making new journal combinations) had 57 percent higher odds of being top 1 percent cited over a 15-year window, were less likely to be top cited in short windows (delayed recognition), had about 18 percent higher citation dispersion and were more likely to land in both the top and bottom 10 percent, and appeared in journals with lower impact factors. Correlational; some results are not robust in the arts, humanities and social sciences; it does not study individual careers.
Open the published source (opens in a new tab)What it supports: Supports that Thompson, Ritchie and colleagues had worked on Multics; that Multics's failure to deliver a usable system promptly became obvious and Bell Labs slowly withdrew in 1968–1969; that the group did not want to lose its pleasant working niche and wanted a system around which a fellowship could form; that throughout 1969 they lobbied for a medium-scale machine such as a PDP-10 or Sigma 7, proposals were never clearly answered and the final one was rejected; Ritchie's own retrospective judgement that they asked the Labs to spend too much money on too few people with too vague a plan; that Thompson found a little-used PDP-7 that the group did not own and that was already obsolete; that he implemented a file system, processes, a few utilities and a simple shell; that the name Unix was suggested around 1970 as a pun on Multics; and that much in Unix was inspired by Multics ideas. It is a participant's account, not an independent history. The author's online rendition gives the reprint as 63 No. 6 Part 2, October 1984, pp. 1577–93, but the journal's own running header (as reprinted) reads Vol. 63, No. 8, October 1984, and the Waterloo Computer Museum record lists the UNIX System special issue as Volume 63, Number 8, Part 2, 1984, so the issue is recorded as 8, Part 2. Pages follow the author's note (17 pages, consistent with the reprint's running page numbers). No DOI could be confirmed on an inspected page, so none is recorded.
Open the published source (opens in a new tab)What it supports: Supports that the earliest version (circa 1969–70) ran on DEC PDP-7 and PDP-9 computers, that Thompson found a little-used PDP-7, that by 1974 UNIX could run on hardware costing as little as $40,000, that less than two man-years were spent on the main system software, that about 40 installations had been put into service since PDP-11 UNIX became operational in February 1971, and that the authors attributed its success largely to its not being designed to meet any predefined objectives, and that fairly severe size constraints encouraged economy and elegance of design. It is the creators' own assessment, not an independent evaluation.
Open the published source (opens in a new tab)What it supports: Supports that across more than 65 million papers, patents and software products from 1954 to 2014, smaller teams tended to disrupt science and technology with new ideas while larger teams tended to develop existing ones; that large-team work built on recent, popular developments and drew attention quickly, while small-team work searched deeper into the past and its impact emerged more slowly, if at all; that most of the effect occurred within individuals moving between team sizes; and that the authors call for supporting a diversity of team sizes. It is observational, uses a citation-based disruption measure, and did not study Unix.
Open the published source (opens in a new tab)What it supports: Supports that the book (Simon & Schuster hardcover, October 2014, ISBN 978-1-4767-0869-0) is a history of the people who created the computer and the internet, and that its publisher describes it as a saga of collaborative genius in which the innovators' ability to collaborate and master teamwork made them even more creative. Bestseller status: number 8 on The New York Times hardcover nonfiction list, as reported in Waltham Public Library's post 'This Week's Best Seller Lists — November 9, 2014' (its NYT links are dated 2014-11-16), and number 13 on the NPR hardcover nonfiction bestseller list for the week of 18 December 2014. The lesson describes its argument in its own words and does not quote it. It is not peer-reviewed evidence.
Open the published source (opens in a new tab)What it supports: Supports that across 19.9 million papers (Web of Science) and 2.1 million U.S. patents, teamwork increased in 170 of 171 science and engineering subfields, team papers were cited more than solo papers, and in science and engineering a team paper was 6.3 times more likely than a solo paper to receive at least 1,000 citations. It measures citations, not quality, did not examine team size and disruption, and did not study software teams.
Open the published source (opens in a new tab)What it supports: Supports target beat 144 of 80 residential square blocks (the report also says “80-by-10 block area”; the journal abstract says 10 by 8, so the lesson says 80 square blocks), the police and academic team that included Sherman, its 1991 homicide rate of 177 per 100,000 (about 20 times the national average), 8 homicides, 24 drive-by shootings and 349 violent felonies in 1991, federally funded overtime patrols from 7 July 1992 working 7 p.m. to 1 a.m. for 29 weeks, 4,512 officer-hours, guns found through searches on arrest, frisks and plain view (traffic stops being the most productive route), guns seized rising 65 percent (46 to 76), gun crimes falling 49 percent (169 to 86), drive-by shootings falling from 7 to 1, the comparison beat’s 4 percent rise in gun crimes (184 to 192, not significant), no measurable displacement, residents becoming less fearful, and the authors’ cautions about replication and police–community relations.
Open the published source (opens in a new tab)What it supports: Supports the test of stricter enforcement of concealed-weapon laws through intensive patrol in a 10-by-8-block area with a homicide rate 20 times the national average, the 65 percent rise in guns seized, the 49 percent fall in gun crimes, no significant displacement to surrounding beats, and no significant change in a comparison beat several miles away. Checked through the NCJRS abstract record, which gives the DOI.
Open the published source (opens in a new tab)What it supports: Supports that 62 of 78 tests reported noteworthy crime and disorder reductions, that the pooled effect was small but statistically significant and smaller in randomised designs, and that displacement was not inevitable, with a diffusion of crime prevention benefits to surrounding areas associated with hot spots policing. It does not establish community views of the policing.
Open the published source (opens in a new tab)What it supports: Supports, from the abstract, strong support for a law of crime concentration: across cities compared with a common microgeographic unit, crime concentrated within a narrow range of percentages of places, and that concentration stayed within a narrow band over time. The abstract does not name the unit as the street segment; Gill and colleagues’ later paper describes Weisburd (2015) as using street segments and a bandwidth of about 4 percent of segments for 50 percent of crime, but the full text was not read, so the lesson says only “a common small unit of place” and uses no city percentages.
Open the published source (opens in a new tab)What it supports: Supports the book’s argument, as described by its publisher, that the tools we use to make sense of strangers are flawed and lead to misunderstanding and conflict, using cases including Sandra Bland; the paperback page lists it as a best book of the year for the Financial Times, Bloomberg, Chicago Tribune and Detroit Free Press. Bestseller status: Hachette Book Group reported that it debuted at #1 on both the New York Times Print Hardcover Nonfiction and Combined Print & E-Book Nonfiction lists (week of 16–20 September 2019); first published 10 September 2019 by Little, Brown and Company. George Mason University’s news page confirms the book draws on Weisburd’s hot spots and crime-concentration research. The book itself was not read for this lesson and is not quoted.
Open the published source (opens in a new tab)What it supports: Supports that people discriminating lies from truths in real time without special aids or training averaged 54 percent correct, classifying 47 percent of lies as deceptive and 61 percent of truths as nondeceptive, and that people tend to regard their interaction partners as honest. It does not test Gladwell’s policing argument.
Open the published source (opens in a new tab)What it supports: Supports the review’s account of the book: the truth-default idea drawn from Tim Levine, coupling of crime to place, the opening with Sandra Bland’s 2015 traffic stop in Texas that ended in her arrest and death in custody three days later, and Gladwell’s use of the Kansas City gun experiment as the backdrop for aggressive proactive stops. Also supports Weisburd’s view that Bland was stopped in an area that was not a hot spot, that the book is an important read for police, his view that Gladwell identified a real shift in officer behaviour, and his criticisms that the Kansas City Gun Experiment takes a more central role than it should (it was not a test of preventive patrol), that the book misses the centrality of the Minneapolis Hot Spots Experiment (the key study overturning the Kansas City Preventive Patrol finding), and that the story perhaps oversimplifies a phenomenon with complex causes. Weisburd hedges these points and says they do not mean Gladwell is wrong. Weisburd’s own research features in the book. The author copy (George Mason CINA) shows the journal, DOI and a 2021 copyright line but no volume, issue or pages; the CINA record dates it January/February 2021. Volume, issue and pages could not be confirmed and are not recorded.
Open the published source (opens in a new tab)What it supports: Supports that Andersen audited Enron’s publicly filed financial statements and provided internal audit and consulting services; that Enron’s 1990s growth came with aggressive accounting; that its performance suffered from 2000 and worsened in 2001; that Skilling resigned on 14 August 2001 and Watkins warned Lay within days that Enron could implode in accounting scandals; that the SEC opened an informal investigation; that on 10 October a supervising partner urged a training meeting of 89 employees, 10 from the Enron team, to comply with the document retention policy; that on 12 October in-house counsel Nancy Temple suggested reminding the team of the policy; that substantial destruction of paper and electronic documents followed despite some managers’ reservations; that subpoenas were served on 8 November and shredding stopped on 9 November; that Andersen was indicted in March 2002; and that on 31 May 2005 the Court unanimously reversed because the jury instructions failed to convey the elements of the offence. Decision date and vote also checked on the official docket. It does not describe the firm’s collapse.
Open the published source (opens in a new tab)What it supports: Supports that Andersen’s reputation collapsed in 2002 amid accounting scandals and a criminal conviction, that about 95 percent of clients did not change auditors until after Andersen was indicted for misconduct tied to its Enron audit, and that more visible clients left sooner. It does not test auditor bias.
Open the published source (opens in a new tab)What it supports: Supports that the study covered all 284 S&P 1500 clients of Andersen (Vanderbilt news release), that after Andersen admitted shredding a significant number of documents, its other clients experienced a statistically significant negative market reaction, larger for clients of its Houston office, and that the authors could not show that the level of non-audit fees affected the reaction. It measures share prices over days, not audit quality directly.
Open the published source (opens in a new tab)What it supports: Supports that in Experiment 1, 130 auditors (139 recruited, 9 withdrawn) judged five accounting vignettes and those in the role of auditor hired by the firm approved its accounting 29 percent of the time against 21 percent for those working for an outside investor (p = .037); and that in Experiments 2 and 3, agents for a seller gave higher private valuations than agents for a buyer even when paid for accuracy, and reported less bias than they showed. Short role-play tasks, not real audits.
Open the published source (opens in a new tab)What it supports: Supports, from the author-repository abstract, that higher client-specific non-audit fees are associated with lower financial reporting quality and that the underlying studies are not homogeneous. The repository working version (42 studies) reports an overall correlation of 0.02, calls the association quite low, and finds highly significant heterogeneity (Q = 220.26). Associations, not causes; the publisher page shows no abstract.
Open the published source (opens in a new tab)What it supports: Supports, from the publishers’ descriptions, that the book recounts Enron’s rise and collapse, that McLean, then at Fortune, wrote an article in 2001 asking how Enron made its money, that within a year Enron was bankrupt in what was then the largest US bankruptcy, that it profiles Lay, Skilling and Fastow among others, that it presents the story as one of greed, arrogance and deceit in which top insiders made millions while small investors lost everything (Penguin UK description), and that it is a picture of what can go wrong in business. Bestseller status: Penguin Random House’s own product page carries a Best Seller label (a publisher statement, not a list position or sales figure). The encyclopaedia entry cites a New York Times Best Sellers list dated 16 November 2003, but the checker could not open that NYT page, so the lesson does not claim a New York Times listing. First published 13 October 2003. Journalism, not a study; described in the lesson’s own words.
Open the published source (opens in a new tab)What it supports: Supports that about one-quarter of the misstatements met the legal standards of intent, that the rest began with an optimistic bias that was not necessarily intentional but could later lead to intentional misstatement, and that overconfident executives are more prone to this bias. Associations in US enforcement cases, not a test of the book.
Open the published source (opens in a new tab)What it supports: Supports that the study covered alleged corporate frauds at US companies with more than 750 million dollars in assets from 1996 to 2004, including Enron, that no single type of actor accounted for more than 20 percent of detected cases (216 cases, per the authors’ Harvard Law School Forum summary), and that employees detected more frauds than auditors (in the NBER working version, 19 against 14 percent). The journal DOI was not confirmed on a page inspected, so none is recorded. Detection, not causes of fraud.
Open the published source (opens in a new tab)What it supports: Supports that First Index Investment Trust launched on 31 August 1976 and was later renamed Vanguard 500 Index Fund, an investible version of the S&P 500 index; that Bogle hoped to raise 50 to 150 million U.S. dollars and the offering raised a little more than 11 million, which he called an abject failure; that critics called it Bogle's Folly and a 1976 Financial Analysts Journal article called index funds a cop-out and a fad; that for about its first six months it was sold through brokers with sales loads, with Vanguard going no-load in early 1977; that annual operating expenses at inception were 0.43 percent of average net assets; and that Bogle's 1951 Princeton thesis highlighted the role of costs. It is Vanguard's own account. Its fundraising target (50 to 150 million) differs from the 250 million in Bogle's 2017 speech, so the lesson states only the amount raised and that it fell far below Bogle's hopes, which both sources support.
Open the published source (opens in a new tab)What it supports: Supports that Bogle was fired as chief executive of Wellington Management Company on 23 January 1974 by his new partners; that on 24 September 1974 the Wellington funds' board chose the option under which the funds handled their own administration, under the Vanguard name Bogle had chosen; that his March 1951 Princeton thesis concluded mutual funds could make no claim to superiority over market indexes; that the index fund's offering on 31 August 1976 was led by Wall Street's four largest retail brokers and produced 11.3 million U.S. dollars against a plan of 250 million; that the fund struggled to gain traction, met its first index fund competitor in 1984 and began growing in the early 1990s; and that the Vanguard 500 index funds held 581 billion U.S. dollars at the time of the speech. It is Bogle's own recollection.
Open the published source (opens in a new tab)What it supports: Supports that common factors in stock returns and investment expenses almost completely explain persistence in equity mutual funds' mean and risk-adjusted returns, that apparent hot hands are mostly explained by one-year momentum which funds do not earn from following momentum themselves, that the only significant unexplained persistence is strong underperformance by the worst funds, and that the results do not support skilled or informed managers. The sample size was not visible on the pages of the article itself that could be opened; Choi and Zhao's NBER working paper reports that Carhart's sample contained 1,892 unique funds over 1962 to 1993, and finds that persistence does not continue in 1994 to 2018. The lesson does not state the sample size.
Open the published source (opens in a new tab)What it supports: Supports that the sample covered January 1984 to September 2006, with 3,156 funds in the group above 5 million dollars in assets, excluding index funds; that the aggregate portfolio of active funds is close to the market portfolio; that its value-weighted net three-factor alpha was about minus 0.81 percent a year, roughly the size of expense ratios, while gross alpha was close to zero (0.13 percent); that few funds produce benchmark-adjusted returns that cover their costs; and that, before expenses, there is evidence of some inferior and superior true performance that is hard to separate from luck. It covers U.S. funds in that period only.
Open the published source (opens in a new tab)What it supports: The publisher's description of the 50th anniversary edition (hardcover 3 January 2023, paperback 2 January 2024; the ISBN in the record is the paperback) supports that the book argues that a person who saves steadily and invests in a diversified set of index funds can achieve above-average results, and that this edition critiques cryptocurrencies, NFTs and meme stocks. Bestseller evidence: the publisher's description reports 2 million copies in print, and Princeton University's Julis-Rabinowitz Center (28 November 2022) reports the book, first published in 1973 by Malkiel as a Princeton economics professor, has sold close to two million copies and is in its 13th edition, advising that the average investor should not waste time picking stocks but own a broad slice of the market. No New York Times list entry was inspected.
Open the published source (opens in a new tab)What it supports: Supports that the paper derives limits on price volatility implied by the efficient markets model, compares them with historical data, and found stock prices far more volatile than later changes in dividends could justify (Yale's economics department summarises the finding as prices far more volatile than any rational forecast of future dividends could justify); Yale's economics department reports it was named one of the top 20 articles in the journal's first 100 years. The exact excess-volatility multiples were not confirmed from the pages inspected. It does not test investor or fund returns. No DOI was found for the journal article: RePEc lists none, and the DOI 10.3386/w0456 belongs to the 1980 NBER working-paper version, so no DOI is recorded.
Open the published source (opens in a new tab)What it supports: Supports that strategies buying past winners and selling past losers generated significant positive returns over 3- to 12-month holding periods, that the profits are not due to systematic risk or delayed reactions to common factors, and that part of the abnormal returns in the first year dissipates over the following two years. It does not measure fund or investor returns after trading costs.
Open the published source (opens in a new tab)What it supports: Supports that at 9:00 a.m. on 27 October 1962 Radio Moscow broadcast Khrushchev's message offering to remove missiles from Cuba if the U.S. removed its Jupiter missiles from Turkey, after his 26 October letter had asked only for a no-invasion pledge; that Major Rudolf Anderson's U-2 was shot down over Cuba and he was killed; that Chairman of the Joint Chiefs of Staff General Maxwell D. Taylor urged a strike on the SAM site and Secretary of Defense Robert McNamara said an invasion had become inevitable; that the administration answered the earlier, conciliatory letter and set aside the harsher one, publicly offering a no-invasion pledge for verified withdrawal; that Robert Kennedy secretly told Ambassador Dobrynin the Jupiters would be removed within six months; and that the Soviet submarine B-59's nuclear torpedo launch was blocked by Vasili Arkhipov. First published 2012; the page inspected is the 2026 republication.
Open the published source (opens in a new tab)What it supports: Supports that at 9:00 a.m. Washington time on 28 October 1962 Radio Moscow broadcast Khrushchev's message that the weapons would be dismantled and returned, accepting Kennedy's proposals of the previous night; that Kennedy had privately promised to remove the Jupiters from Turkey and the few officials who knew kept it confidential; and that the details of the Jupiter pledge were confirmed in 1989. First published 2012; the page inspected is the 2026 republication.
Open the published source (opens in a new tab)What it supports: Supports that on 27 October 1962 Kennedy's advisers offered him two choices, an air strike or accepting Soviet nuclear missiles in Cuba, and that he rejected both; that he publicly pledged not to invade Cuba if the missiles were withdrawn; that during the dinner break he sent Robert Kennedy to tell the Soviet ambassador that the U.S. would withdraw its missiles from Turkey within six months; and that this was kept so secret that most ExComm members did not know. The lesson does not use the article's present-day policy recommendations. The URL is a university course copy of the article (a scan of the Foreign Affairs pages), inspected because the publisher page was not reachable; the citation (vol. 91, no. 4, July/August 2012, pp. 11-16) is independently confirmed by the Sciences Po Lyon journal index. The wording 'attack or accept' is Allison's.
Open the published source (opens in a new tab)What it supports: Supports that lose-lose agreements, a faulty belief about the other party's interests combined with an agreement that fails to use compatible interests, are widespread across 32 experiments, occur even when parties recognise their interests are compatible, and reduce prosperity and satisfaction for both parties; it also examines judgements of others' interests and the role of practice. The abstract inspected gives no participant count or percentages. Experiments, not real diplomacy. Pages: the Kellogg faculty record prints 369–409, but the DOI itself encodes first page 396 (APA DOIs take the form volume.issue.first page) and the eCampusOntario reference list gives 120(3), 396–409, so 396–409 is kept and 369 is treated as a transposition. The abstract says lose-lose outcomes happen 'sometimes' even when parties recognise compatible interests, and the reduced prosperity and satisfaction is the authors' description of the effect, not a pooled estimate.
Open the published source (opens in a new tab)What it supports: Supports that negotiators with a prosocial rather than egoistic motive were less contentious, engaged in more problem solving and achieved higher joint outcomes, but only when resistance to yielding was high or unknown, consistent with Dual Concern Theory; results were moderated by study features such as class exercises versus participant pools and publication status. It does not test Getting to Yes directly.
Open the published source (opens in a new tab)What it supports: The book argues for principled negotiation instead of positional bargaining: separate the people from the problem, focus on interests rather than positions, invent options for mutual gain, and insist on objective criteria. Bestseller status: the Penguin Random House Australia page says it has sold over 2 million copies in over 20 languages and calls it the world's bestselling guide to negotiation (its author bio gives a higher figure of over 8 million); The Texas A&M catalogue records the first edition as Boston: Houghton Mifflin, 1981; the Wake County and Berkeley Law records of the 2011 3rd edition note first publication in 1981 and list chapters including focusing on interests and inventing options for mutual gain. Fisher directed the Harvard Negotiation Project and Ury is at the Program on Negotiation at Harvard Law School (publisher bios). The publisher's figure is attributed, not endorsed: the same page's author bio gives over 8 million. Criticism of the book is cited to White (1984; M009H), not to the book. Described, not quoted; not itself a study.
Open the published source (opens in a new tab)What it supports: Supports that across 3 experiments whichever party made the first offer obtained a better outcome and first offers strongly predicted final settlement prices, and that focusing on the opponent's alternatives, the opponent's reservation price or one's own target eliminated that advantage. Distributive price negotiations between participants, not diplomacy, and not a test of Getting to Yes itself.
Open the published source (opens in a new tab)What it supports: Supports only what the repository abstract states: that White found Getting to Yes 'frequently naive, occasionally self-righteous, but often helpful', credited it with a persuasive criticism of traditional negotiating behaviour, and said the authors seem to deny the existence of a significant part of the negotiation process and to oversimplify or explain away its most troublesome problems. The full text PDF could not be retrieved, so the review is not used for any claim about anecdotes or distributive bargaining. A review, not a study.
Open the published source (opens in a new tab)What it supports: Supports that early department-store sales of Tupperware were unsuccessful because Americans were still sceptical of plastics and the Tupper seal required demonstration; that Wise had sold Stanley Home Products, moved to Miami in 1949 and ran a direct-selling business there; that she and Thomas Damigella in Massachusetts became among the fastest movers of the product, drawing Tupper's attention; that in 1951 Tupper recruited her to develop the hostess party plan and named her vice president; that in Miami she and her mother ran a direct-selling business, Patio Parties, that sold Tupperware through a home party plan; that the board forced her out in January 1958 after clashes with Tupper; and that Tupper sold the company to Dart/Rexall shortly afterwards. It gives no sales figures. The finding aid shows no separate publication date; it was processed by Mimi Minnick in July 1995 (the date used here, as the biographical note dates from that processing), and Series 9 (2021 addenda) was processed in 2024, so the online version read is later than 1995. It does not itself say the party plan let the seal be demonstrated.
Open the published source (opens in a new tab)What it supports: Supports that in 1946 Tupper obtained polyethylene pellets from DuPont, which doubted raw polyethylene could be moulded; that the patent for the airtight container with its burping seal was granted in 1949; that Wise had been a Stanley Home Products representative; that it was Wise's suggestion that Tupperware be removed from stores and sold only through the party plan; that she was vice president of Tupperware Home Parties from 1951 to 1958; and that parties were promoted as a way for women to sell to friends and neighbours, with hosts receiving gifts based on the value sold. It is a short article, not a scholarly history.
Open the published source (opens in a new tab)What it supports: Supports that Tupperware sat unpurchased on hardware and department store shelves and catalogue sales were nearly nonexistent, partly from consumer distrust of plastic and lack of instructions on sealing; that in April 1951 Tupper took the product off store shelves and hired Wise to build a direct-selling system, Tupperware Home Parties, Inc.; that Wise established a campus-like headquarters in Kissimmee, Florida; that parties included games, recipes and giveaways; and that hostesses taught guests about the bowls and their seals, the product having come with no sealing instructions. It gives no sales figures.
Open the published source (opens in a new tab)What it supports: Supports that referred customers had an approximately 18 percent lower churn hazard (33-month survival 82.0 percent versus 79.2 percent), had higher contribution margins that faded over about 29 months, and had an average value at least 16 percent higher than comparable non-referred customers; and that the value difference was absent for customers over 55 and for low-margin customers. It is one bank in one country, lifetime values are partly predicted, and it did not study home parties. Volume 75, issue 1 (January 2011), pages 46–59 are confirmed by the article PDF and the Deakin repository record; neither shows a DOI, and the DOI landing page could not be opened, so none is recorded.
Open the published source (opens in a new tab)What it supports: Supports that electronic word of mouth is positively correlated with sales, with an average correlation of .091, and that effects vary by platform, product and metric (volume matters more than valence). It covers online word of mouth, not face-to-face selling or home parties.
Open the published source (opens in a new tab)What it supports: Supports that the book (Simon & Schuster, New York, 2013, ISBN 9781451686579) argues that people are influenced more by peers than by advertising and that six principles, with chapters on social currency, triggers, emotion, public visibility, practical value and stories, drive things to become popular; and that it offers practical techniques for making messages people share. Bestseller status: the library catalogue's publisher summary calls it 'The New York Times bestseller', and the author's site describes it as a New York Times and Wall Street Journal bestseller; an independent copy of a New York Times or Wall Street Journal list entry could not be inspected, so the lesson attributes the claim to the publisher rather than stating it as fact. The lesson describes the argument in its own words and does not quote the book. It is not peer-reviewed evidence.
Open the published source (opens in a new tab)What it supports: Supports that, among nearly 7,000 New York Times articles from August to November 2008, positive content was more viral than negative content, content evoking high-arousal emotions (awe, anger, anxiety) was more likely to make the most-emailed list, and sad content was less likely; small experiments (about 45 to 49 people each) supported a causal role for arousal. It is the book author's own study, of one newspaper, and measures sharing, not sales.
Open the published source (opens in a new tab)What it supports: Supports that across about a billion Twitter diffusion events (about 622 million pieces of content), only 0.025 percent of diffusion trees reached at least 100 nodes, structural virality was typically low, and popularity was largely driven by the size of the largest broadcast rather than multi-generation person-to-person spread. It covers one online platform and did not study offline word of mouth or sales.
Open the published source (opens in a new tab)What it supports: Supports the date and WBBM location of the first televised U.S. presidential debate; Nixon as vice president and Kennedy as a little-known senator; Nixon’s use of Lazy Shave, which worsened his pallor on screen; Kennedy’s touch-up by staff (the HISTORY.com article says only that sources later claimed this, so the lesson says “reportedly”); and that Kennedy later said he could not have won without the debates. The Smithsonian page shows no date or byline as read; the author and the 2010 year come from the History News Network reprint, so 2010 is the reprint date (14 September 2010). The original cannot be later than that, but its exact date was not seen. The article gives a 65 million audience, which the lesson does not use. It does not prove the debate decided the election.
Open the published source (opens in a new tab)What it supports: Supports that Nixon spent 3 weeks in hospital with an inflamed knee and hit it on a car door on arrival; that Kennedy spent much of the previous week practising with mock opponents; that Kennedy declined makeup and Nixon then did too; a nearly 70 million audience; that Nixon arrived for the second debate rested and healthy in a dark suit; that analysts generally gave him the second and third debates; that Gallup showed a virtual tie before and after the first debate; that studies do not clearly support the radio versus television story; and the 112,827 popular vote margin.
Open the published source (opens in a new tab)What it supports: Supports the August car door injury and infection, about 2 weeks in hospital, about 20 pounds underweight, recent flu, CBS’s top makeup artist from New York, both candidates declining, Kennedy’s tan (the touch-up only as something sources later claimed), Nixon’s light grey suit about 70 million viewers and that Nixon improved in the later debates. It repeats the claim that radio listeners scored the debate differently, which the lesson does not adopt because Newcott (2020) and Druckman (2003) describe that evidence as weak.
Open the published source (opens in a new tab)What it supports: Supports random assignment to the televised or audio version of the first debate; 81 percent of viewers against 60 percent of listeners saying Kennedy won; viewers weighing perceived integrity significantly more; and issue agreement predicting listeners’ judgements but not viewers’, although the gap between the two issue coefficients was not itself significant. Some models use fewer cases (148 in the evaluation regression). Participants were mostly young University of Minnesota summer students. It cannot recreate the 1960 electorate and may not apply to candidates who look alike. Its results rest on a new experiment, not 1960 data.
Open the published source (opens in a new tab)What it supports: Supports that judgements of competence from faces alone predicted U.S. congressional election outcomes above chance. Princeton’s news release supplies the figures of 843 student participants, 71.6 percent of 95 Senate races and 66.8 percent of 600 House races, with familiar candidates excluded. It shows a link, not that faces cause votes. The paper’s full text was not read and could not be found, so the lesson’s figures rest on the university’s release; the abstract gives 68.8 percent for 2004 Senate races alone.
Open the published source (opens in a new tab)What it supports: The publisher’s page (on sale 22 December 2015) describes presence as a state of no longer worrying about the impression you make, reached through small adjustments in body language, behaviour and mindset, ties the book to her TED talk on power poses, and labels the book a New York Times, Wall Street Journal, USA Today and Publishers Weekly bestseller. Cuddy’s own site reports more than 500,000 copies sold and 35 languages, with no independent support. That site also carries the values and feelings, posture and self-talk wording, which the lesson paraphrases. No list page was inspected, so the lesson attributes the label. Only Wikipedia calls power posing the book’s centrepiece, so the lesson does not; the book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports that high-power poses raised self-reported feelings of power (d about 0.34) but did not significantly change testosterone, cortisol, risk taking or willingness to compete. The experimenter was blind to condition. It is a conceptual replication, not an exact one, of a 2010 paper with 42 participants. Volume, issue and pages are confirmed by PubMed.
Open the published source (opens in a new tab)What it supports: Supports that the studies cited by Carney and colleagues in 2015 lacked evidential value for behavioural and physiological effects, and that the authors did not conclude the effect is exactly zero or rule out benefits in future well-powered research. It covers only 24 of those 33 studies, and feelings of power were analysed separately (not significantly right-skewed, on 7 results).
Open the published source (opens in a new tab)What it supports: Supports only that the authors replied to Simmons and Simonsohn with a wider p-curve analysis and claim clear evidential value, as the title states. The full text was behind a login and its methods and results were not read. PubMed confirms the title, authors, volume, issue, pages and DOI, but gives no abstract.
Open the published source (opens in a new tab)What it supports: The authors’ own summary on the SPSP blog (March 2023) and the University of Bamberg’s release support modest effects of body position on confidence, feelings of power and mood; a behaviour effect that nearly disappeared after correcting for publication bias; and almost no effect on hormones, blood pressure or heart rate. Most studies lacked a neutral posture. The blog says more than 10,000 participants and the release nearly 10,000. The paper’s full text was not read, and the Bamberg release gives the full author list.
Open the published source (opens in a new tab)What it supports: Supports that the lead author of the 2010 paper no longer believes power-pose effects are real, and her stated reasons (a tiny sample and flexible analysis). The statement itself shows no date or signature and is hosted in her Berkeley faculty directory; Fortune dates it to 25 September 2016 and a blog reported it on 26 September 2016. It is an opinion from one author, not a new study.
Open the published source (opens in a new tab)What it supports: Supports that in the early 1990s economists widely believed higher minimum wages reduce employment, while the evidence, mainly negative correlations open to reverse causation, was not fully convincing; the 4.25 to 5.05 dollar rise; Pennsylvania as the comparison; no measurable job loss; later research finding small negative effects; and Card’s half share of the 2021 prize. It states the assumption of similar trends on both sides of the border.
Open the published source (opens in a new tab)What it supports: Supports the coffee counter origin, the newspaper report, Card’s minimum-wage jobs as a youth, Krueger’s wish for real-world experiments like those in medical journals, the study’s criticism and later acceptance, and Krueger’s death in 2019 at age 58. It comes from the authors’ own university and is not an independent history. The coffee counter story has this single source, so the lesson attributes it, and no date is given for the morning.
Open the published source (opens in a new tab)What it supports: Supports the 1 April 1992 rise from 4.25 to 5.05 dollars, the recession after the rise was enacted, the phase-in attempt, which the governor vetoed after letting 5.05 dollars take effect, the survey dates, chains and counts (410, then 399), mainly telephone interviews, visits to non-respondents, and a relative gain of about 2.76 full-time-equivalent jobs per store. The authors could not reject a small negative effect in some specifications, tested store openings only through separate McDonald’s data across states, and studied one industry in one episode.
Open the published source (opens in a new tab)What it supports: Supports that the number of low-wage jobs stayed essentially unchanged over 5 years after the rises, from 1979 to 2016, and that the minimums reached about 37 to 59 percent of the median wage, with higher levels an open question. Tradable sector employment may have fallen, though imprecisely estimated. It covers the U.S. only.
Open the published source (opens in a new tab)What it supports: Supports the use of payroll records from 230 restaurants and a 4.6 percent fall in New Jersey employment relative to Pennsylvania. These are the working paper’s figures, and the lesson says so. The published 2000 comment (doi 10.1257/aer.90.5.1362) has no abstract on RePEc and its full text was not read; a press quotation on a University of California, Irvine page reports it as a 3.9 to 4.0 percent fall, which was not confirmed from the paper.
Open the published source (opens in a new tab)What it supports: The summary of its working paper version (NBER 6386) supports that Bureau of Labor Statistics records show New Jersey fast-food employment growing at a similar or slightly faster rate than in eastern Pennsylvania, and that the other sample’s discrepancies trace to a small group of restaurants owned by one franchisee. The published reply itself was not read.
Open the published source (opens in a new tab)What it supports: Bestseller evidence: Levitt’s University of Chicago pages report more than 2 years on the New York Times Best Seller list and more than 5 million copies sold worldwide (his own claim); an EBSCO reference entry reports more than 2 years on the list and over 4 million copies; the publisher’s description, seen on a bookseller page for the 2020 revised edition, calls it a legendary bestseller and frames it as about incentives and the hidden side of the modern world. A university library catalogue confirms William Morrow, New York, 2005. No list page was inspected, so the lesson attributes the list claims to Levitt and the Minneapolis review. The book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports the book’s argument about incentives and information, its estate agent and abortion and crime cases, its number 2 place on the New York Times list when the review was written, the reviewer’s view that the abortion evidence is compelling but not conclusive, and critics’ view that it is too clever and lacks lasting substance, which the lesson no longer uses. It is one reviewer’s opinion, written before the 2008 papers.
Open the published source (opens in a new tab)What it supports: Supports a coding mistake in the final regressions of Donohue and Levitt (2001), much weaker results after correcting it and using a per capita specification, and cross-state tests that are not robust to differing state trends. The RePEc abstract and the 2005 Boston Fed working paper were read; the journal text was not. The working paper’s estimates fall to essentially zero per capita, and it claims little evidence for a selection effect, not a proven zero effect.
Open the published source (opens in a new tab)What it supports: Supports that the authors accepted their final table omitted state-year interactions, that correcting it kept the sign and significance while shrinking the size, and that a more precise abortion measure still indicates an effect on crime. Levitt is an author, so this is not an independent view. The abstract was read; the full text was not.
Open the published source (opens in a new tab)What it supports: Supports that homes owned by estate agents sold for 3.7 percent more and stayed on the market 9.5 days longer after controlling for observable features, read by the authors as an incentive and information effect. The 98,000 suburban Cook County, Illinois sample comes from the 2005 working paper. The authors say unobserved quality and extra effort may bias the estimate. Levitt is an author, so this is not an independent test of the book, and the journal text was not read.
Open the published source (opens in a new tab)What it supports: Abstract and journal record only (full text paywalled). Supports that Morris’s group dealt with possible selection and confounding, compared conductors with drivers and postmen with less active government workers, then turned to leisure-time activity in civil servants because work intensity varied too little. PubMed classes it as biography and historical article. Its later claim that studies solidified cause and effect comes from admirers of Morris and is not adopted by this lesson as settled.
Open the published source (opens in a new tab)What it supports: Supports the numbers in the story: about 31,000 male transport workers aged 35 to 65 examined in 1949–1950; conductors climbing about 500 to 750 steps per working day and drivers sitting over 90 percent of a shift; early case fatality of about 30 versus 50 percent; postmen compared with sedentary staff; considerable scepticism; the uniform study; the 10-year sudden-death comparison; 16,882 civil servants (analysis concentrating on executive-grade men aged 40 to 64), the 33 percent relative risk, given in the lesson as about a third of the risk, and no protection from lighter activity. Its printed annual incidence figures (2.7 for conductors, 1.9 for drivers) contradict its own text, appear transposed, and are deliberately not used. It gives the Lancet volume as 265, where PubMed gives 262; the lesson uses 262. The December 2007 issue date, volume, issue and pages were confirmed on the page. Its figures have not been checked against the Lancet originals.
Open the published source (opens in a new tab)What it supports: Bibliographic record confirmed on PubMed: 21 November 1953, volume 262, issue 6795 (no abstract there; full text not read; the DOI page could not be opened). This is part one of a two-part report; M013B cites part two at pages 1111–1120. Cited as the 1953 report; its findings in this lesson are taken from M013B and M013A, and an LSHTM library page confirms the report’s title, year and conductor-driver comparison.
Open the published source (opens in a new tab)What it supports: Bibliographic record confirmed on PubMed: 15 September 1956 (no abstract there; full text not read; the DOI page could not be opened). Sample sizes and findings about uniform sizes are taken from M013B’s account of this paper.
Open the published source (opens in a new tab)What it supports: Supports Morris’s birth in Liverpool in 1910, wartime service with the Royal Army Medical Corps in India, becoming director of the Medical Research Council’s Social Medicine Unit in 1948, and his death in 2009 aged 99. Also states the conductor-driver finding, independently of M013B.
Open the published source (opens in a new tab)What it supports: Abstract read on PubMed. Supports 661,137 adults, 116,686 deaths, median follow-up 14.2 years, self-reported activity, a 20 percent lower mortality risk below the recommended minimum and a 31 percent lower risk at 1 to 2 times it, compared with no leisure activity. Also reports 37 percent at 2 to 3 times the minimum and no evidence of harm at 10 or more times, which the lesson omits. Observational: it cannot show cause.
Open the published source (opens in a new tab)What it supports: Abstract read on PubMed. Supports an 18 percent higher risk of early death in men with high versus low occupational activity (HR 1.18, 95% CI 1.05 to 1.34) after adjustment, and no such link in women (HR 0.90, 95% CI 0.80 to 1.01). Does not test bus work.
Open the published source (opens in a new tab)What it supports: Publisher pages support the prevention-focused argument, that treatment for the main diseases of ageing often comes too late. Only publisher descriptions and one review were read; the book itself was not read. Bestseller evidence: Penguin Random House Canada’s page calls it a number 1 New York Times bestseller; Penguin New Zealand’s page calls it an instant Sunday Times and New York Times bestseller, reports over 1 million copies sold globally citing Nielsen BookScan UK and Circana BookScan US (April 2024), and elsewhere says over 2 million without a source. No actual New York Times list page was inspected; the publisher label is attributed in the lesson. The Penguin Random House US page could not be opened.
Open the published source (opens in a new tab)What it supports: Supports the book’s emphasis on healthspan, its focus on heart disease, cancer, dementia and metabolic disease, and core principles of exercise, nutrition, sleep and emotional health, the lesson’s only source for the book’s stress on exercise; that it has become a centrepiece of conversations about healthspan; and the reviewer’s judgement that the underlying research is treated superficially and the approach is resource-intensive. The reviewer also calls the book thought-provoking. One reviewer’s opinion, not a test. No page numbers or DOI were shown.
Open the published source (opens in a new tab)What it supports: Abstract read on PubMed. Supports 122,007 patients referred for treadmill testing at one academic centre, and that low fitness was linked to mortality about 5 times that of the fittest (adjusted HR 5.04 for low versus elite), comparable to or greater than smoking and diabetes. Patients referred for testing; association only.
Open the published source (opens in a new tab)What it supports: Abstract read on PubMed. Supports that, in trials with about 50,000 participants, exercise did not reduce all-cause mortality or new cardiovascular disease in older adults or people with chronic conditions, and that the authors note shortcomings and say additional large trials would be of value, especially on fractures. It also finds exercise can prevent type 2 diabetes when combined with diet, which the lesson omits. Its conclusions concern those groups. It is one critical review, not the last word.
Open the published source (opens in a new tab)What it supports: Full text read in PubMed Central. Finnish Twin Cohort, 4,190 complete pairs. Supports no difference in mortality within 34 activity-discordant identical pairs (HR 1.00, 95% CI 0.52 to 1.94, wide), and the proposal that genetic pleiotropy might partly explain the link. Discordant non-identical pairs did show a difference (HR 0.58), and individuals with persistent vigorous activity had lower mortality overall (adjusted HR 0.73). The rat experiment is not used in the lesson’s figures.
Open the published source (opens in a new tab)What it supports: Supports Roebuck’s flooded Bo’ness mines; his paying Watt’s debts for two-thirds of the invention by the end of 1765; Watt’s drawing of 10 November 1765 for a cylinder to be cast and bored at Carron; the engine erected in summer 1769 and tests from September; the inconclusive 1769–1770 experiments, Watt blaming poor workmanship from Carron; abandonment in 1770 for land surveying and civil engineering; Roebuck’s 1773 bankruptcy; and Boulton’s purchase of his share. The record, added 3 August 2017, also supports the secluded workshop. It says Watt’s equipment was moved in May 1773; the Friends of Kinneil agree the engine was shipped in 1773 and that Watt himself left in May 1774, so these do not conflict. Britannica dates Roebuck’s bankruptcy to 1772; the lesson follows this record.
Open the published source (opens in a new tab)What it supports: Supports that the bar is said to come from the cylinder-boring mill Wilkinson built at Bersham around 1775; that the engine’s success depended on steam-tight contact between piston and cylinder; that the mill gave unprecedented accuracy; and that without the object Boulton and Watt’s success would have been in doubt. The record notes its information may not be fully accurate, gives no tolerance figures, and its attribution is partly “allegedly”.
Open the published source (opens in a new tab)What it supports: Supports Watt’s experimental engine being moved from Kinneil to Soho in 1773; that it appears (the page’s hedge) to have had a block-tin cylinder of sheet 1/4 inch thick; and a new cast-iron cylinder from Wilkinson at Bersham, ordered 26 January 1775 and delivered 24 April 1775. The page credits Hills, James Watt vol. 1 (2002), pp. 401–402, and Dickinson & Jenkins (1927), which were not read. No second source for the tin or the exact dates was found in museum, encyclopaedia or archive pages, so the lesson hedges the tin, gives only the months and names the wiki. The Friends of Kinneil confirm the engine’s shipment and reassembly at Soho, but say nothing of tin or Wilkinson, and describe it as running smoothly, where this page calls it unsuited to practical work. The lesson does not call it Old Bess.
Open the published source (opens in a new tab)What it supports: Supports the separate condenser developed by 1765, Watt’s move to Birmingham in 1774, first engines working in commercial enterprises in 1776, and 1,164 engines by 1824. It does not mention Wilkinson or Roebuck. It dates the partnership with Boulton to 1776 where Britannica says 1775, so the lesson gives no partnership date. Britannica also confirms the 1774 move and two engines installed in 1776.
Open the published source (opens in a new tab)What it supports: Supports a machine that made the bore more accurate, and Boulton and Watt’s own foundry at Smethwick, set up as the patent neared expiry. It also says he proposed solid casting in 1774, supplied accurately bored cylinders and secretly built pirated engines; the lesson cites M014L for those. It dates the machine to 1775, so the lesson does not date the machine itself. EBSCO’s Research Starter and steamindex.com’s Watt page corroborate his unmatched accuracy, but EBSCO’s date of 1770 for Watt’s finding this is wrong for the story and is not used.
Open the published source (opens in a new tab)What it supports: Early-access journal version read in a PDF copy. Supports growth in textiles where wages were low and mechanical skill high, about 70 percent of variation in 1831 textile employment explained by skill and wages, and that coal, literacy and banks add little. Only 41 observations; skill measured retrospectively from the 1851 census, as men aged 60 and over, a proxy for the 1790s. Page numbers 59–94 come from the RePEc record; the copy read did not show them.
Open the published source (opens in a new tab)What it supports: Supports over 3,200 stolen watches valued in Old Bailey trials, 1685–1810, and real prices falling about 1.3 percent a year, roughly 75 percent a century, before quality adjustment. Trial records may over-represent dear watches. It shows sustained progress in a complex product, not a bottleneck as such.
Open the published source (opens in a new tab)What it supports: Library catalogue confirms authors, publisher and edition. TIME (Rawlings, 9 August 2011) supports 1984, Goldratt being a physicist, the novel form, Alex Rogo, a production manager, having three months to save a factory, the Theory of Constraints and focus on bottlenecks, and listing among the 25 most influential business management books. Bestseller evidence: Routledge’s page for the graphic-novel version says it has been a bestseller since 1984, with over 7 million copies; a 2005 Times Higher Education review reports a cover claim of more than 3 million, and an appendix of ten managers, chosen by Goldratt and interviewed by a Fortune Small Business journalist, who all said it helped; that review is by Winston Fletcher (25 February 2005). No newspaper list page was inspected, so the label is attributed to Routledge.
Open the published source (opens in a new tab)What it supports: Abstract read on the publisher’s page. The Wellington repository record is Balderstone’s related 1999 master’s thesis, not this article. Supports significant operational and financial gains in over 80 successful applications with quantitative data, mostly in manufacturing, and no reports of failure found. The sample is of reported successes, so it can’t show how often the method fails.
Open the published source (opens in a new tab)What it supports: Abstract read on RePEc. Supports that ranking products by throughput per unit of constraint use can fail when output must be whole units, and that the product-mix decision is NP-complete. It tests one heuristic of the theory, not whole plants. No DOI could be confirmed: RePEc lists only the publisher item S0925-5273(09)00136-4, and the DOI resolver could not be opened, so none is recorded.
Open the published source (opens in a new tab)What it supports: Journal record read at LSE and abstract read at an academic index; figures read in the May 2003 working paper version (LSE WP 75/03). The abstract confirms little contribution before 1830, a peak about a century after Watt, and high pressure from about 1850. Supports steam adding 0.01 and 0.02 percentage points a year to British labour productivity growth in 1760–1800 and 1800–1830, and 0.38 at its 1850–1870 peak, the last including railways and the earlier ones engines only, with low-pressure Watt engines limited and high pressure becoming economic later. The numbers come from the working paper; the journal’s own table was not seen.
Open the published source (opens in a new tab)What it supports: Added by the checker to reduce reliance on the wiki. Supports Wilkinson’s 1774 cannon-boring patent, his near monopoly of boring cylinders for Boulton and Watt engines from 1775 to 1795, and that he was accused of pirating the engines. Says nothing of the tin cylinder or the January and April 1775 dates.
Open the published source (opens in a new tab)What it supports: Supports that Collins taught in Chicago public schools, grew disillusioned, used her 5,000-dollar pension to start her own school on the second floor of her home; that visitors described a no-frills school using drills and repeatedly telling pupils success was within reach; and that the school closed in 2008 because families could not or would not pay 5,500 dollars a year. It does not mention substitute teaching and gives no test data.
Open the published source (opens in a new tab)What it supports: Supports 5,000 dollars from her retirement fund, the second-floor home setting, four starting students including her daughter, and that Collins blamed teachers and school systems when children failed. The headline date of Sept. 1, 1975 is the column’s only date; its text gives no opening day, so the lesson says only 1975. Its claim about President George H. W. Bush conflicts with other sources and is not used.
Open the published source (opens in a new tab)What it supports: Supports that by 1975 Collins was tired of pupils spending more time playing at recess than learning, and that she educated low-income Black pupils, many labelled learning disabled, whom many thought could not be educated. Its claims about their later success are not independently verified and are not used.
Open the published source (opens in a new tab)What it supports: By Janis Johnson. The page shows no article date; 2004 is the medal year it carries, so the year here is that and no more. Supports that Collins received the 2004 National Humanities Medal, opened the school in her Chicago home after being disenchanted with the education her children received at private schools and with her experience of public schools, and that students read authors such as Homer and Tolstoy. Its statements about results rest on the school’s own claims and are not used.
Open the published source (opens in a new tab)What it supports: Supports her birth in Monroeville, Alabama, in 1936, Clark College, teaching in Alabama, work as a medical secretary and 14 years as a Chicago substitute; four starting students including her daughter; the start in a college basement and the move to the second floor of her home to be free of red tape, the same reason she said she refused federal funds; enrolment over 200; the 1981 television film; the 1982 accusations that she was not certified and had overstated her success, with parents defending her. The page text shows no byline, date or wire credit and says only that she died on Wednesday, so the year 2015 is taken from her death date in M015A; it may be a reprint whose origin was not identified.
Open the published source (opens in a new tab)What it supports: Supports that self-fulfilling prophecies occur in classrooms but are typically small (typically changing the achievement of about 5 to 10 percent of students, citing Brophy, 1983), may be more likely to dissipate than accumulate, may be larger among stigmatised and low-achieving students with only two studies testing this, and that expectations predict outcomes mainly because they are accurate.
Open the published source (opens in a new tab)What it supports: PubMed abstract supports that the benefits of a short intervention in the first year of U.S. high school depended on teachers’ own growth mindsets (9,167 student records, 223 maths teachers). Effect sizes were not read and are not stated. Dweck is a co-author. PubMed dates it online 22 December 2021; the issue is 2022.
Open the published source (opens in a new tab)What it supports: A description headed only "Description", almost certainly supplied by the publisher but not labelled so, appears on the Harvard Book Store, MIT Press Bookstore and Chaucer’s Books pages for the Ballantine paperback (ISBN 9780345472328). It says beliefs about whether abilities are fixed or can be developed affect success and which paths people take, and that parents, teachers, managers and athletes can use the idea. Bestseller evidence: that description calls this the updated edition of the million-copy bestseller. Penguin Random House’s own page makes no bestseller claim and has no description, and no New York Times list page or sales figure was inspected, so the lesson attributes the label to booksellers’ descriptions. Dweck’s Stanford affiliation comes from M015K.
Open the published source (opens in a new tab)What it supports: Supports the weak mindset and achievement link (r = .10, 129 studies, 365,915 students), the small intervention effect (d = 0.08, 29 studies, 57,155 students), and a larger effect for low-income students (d = 0.34, N = 577), which the authors say to interpret with care.
Open the published source (opens in a new tab)What it supports: PubMed abstract supports a small overall effect, nonsignificant after publication bias correction; a nonsignificant effect in the 6 highest-quality studies (13,571 students); major design and reporting flaws; and larger effects from authors with a financial incentive.
Open the published source (opens in a new tab)What it supports: PubMed Central full text (read through the PubMed tool) supports the sample, the under-one-hour online programme, a 0.10 grade point gain among 6,320 lower-achieving students, stronger effects where peer norms supported challenge, and the idea that needing to try hard can look like lacking ability under a fixed mindset. Dweck is a co-author. Author list abbreviated here; full list in PubMed.
Open the published source (opens in a new tab)What it supports: Supports the start in the basement of Daniel Hale Williams University before the move to her home, and that in 1982 critics charged she had exaggerated test scores and taken federal money, that an independent investigation (not named) found she had received 69,000 dollars through the Comprehensive Employment and Training Act, that she denied knowing it came from Washington, and that parents largely supported her. It reports no finding on the test scores. It also says about 20 children were taught at the home, later than the four at the start. The entry is undated, so the year field records the year of access.
Open the published source (opens in a new tab)What it supports: Introduction and Part I support printed sheets issued in groups of four with authors withheld, comments collected after a week, most readers undergraduates reading English for Honours, about 60 percent returning comments, few reading a poem fewer than four times, the 10 difficulties (including prose sense and stock responses), the claim that failure was widespread even among good readers and reputable scholars, language habits and inadequate technique as causes he points to, educational methods as an aim, and his admission that selection could mislead. The book has 13 poems. The scan as served stops at about 117,000 characters, in the commentary on poem 4, so the appendix naming poets could not be reached and the readers’ comments on later poems were not seen. The Routledge reissue page confirms the 1929 original.
Open the published source (opens in a new tab)What it supports: Supports that Richards graduated in Moral Sciences (philosophy), lectured at Magdalene from 1922 to 1929, helped create the Cambridge English course in the 1920s, which helped make English a recognised academic subject; that Empson wrote Seven Types of Ambiguity (1930) as his undergraduate student; and that his stress on close reading is seen as foundational to modern criticism. Britannica independently confirms the 1922 to 1929 dates and the role of Practical Criticism in close reading.
Open the published source (opens in a new tab)What it supports: Supports the results named in the lesson, which the lesson attributes to Annan: a Longfellow poem rated above one by Hardy, Ella Wheeler Wilcox and Woodbine Willie preferred to Hopkins and Lawrence, almost no agreement on meaning, and a birthday greeting to Meredith read as a cavalier drinking song. It is a secondary account in passing; these specific judgments could not be checked against Richards’s appendix or later protocols. It says a dozen poems where Richards prints 13, so the lesson gives no count.
Open the published source (opens in a new tab)What it supports: Published version (Taylor & Francis pages in the Cambridge repository) supports that practical criticism has become the byword for analytical, decontextualised reading, and reports A. C. Goodson’s argument that Richards set Cambridge English toward close reading at the expense of extensive reading. No volume, issue or pages are shown, and the page does not itself state peer review.
Open the published source (opens in a new tab)What it supports: The author’s own research bulletin supports the design and findings: PhD students in English against undergraduates and advanced high school students, thinking aloud on Marvell’s poem without the poet’s name; experts used prior knowledge to recognise patterns; novices noticed oppositions but dismissed them as confusing and ignored those lines. The sample of 8 and 8, two period poems, and the dismissal of oppositions come from the summary in Lee and Hughes (2012, International Electronic Journal of Elementary Education, 4(3), 489–506). The article itself was not inspected; the publisher page was not reachable and no DOI is recorded. Volume, issue and pages confirmed from the reference list in M016H.
Open the published source (opens in a new tab)What it supports: Author-hosted version supports two 12th-grade classes at one urban U.S. high school, a 4.5-week unit of 18 periods (the abstract says four-week), 19 and 18 consenting students, and interpretive written responses of 95 percent (18 of 19, 58 percent before) against 56 percent (10 of 18, 67 percent before). The author notes that teachers and texts differed. The hosted copy shows a placeholder for volume and issue; those details come from the reference list in M016H.
Open the published source (opens in a new tab)What it supports: The publisher’s description supports the argument that literature has layers beneath plot, with a road as a quest and a shared meal as communion. Bestseller evidence: the Associated Press (Hillel Italie, 20 October 2016) reports sales topping 1 million copies according to HarperCollins, more than 100,000 that year per Nielsen BookScan, a word-of-mouth bestseller used in hundreds of courses, and Foster’s post at the University of Michigan-Flint. Lansing City Pulse reports number 2 on the New York Times education list about 12 years after publication, that sales rose quickly after Advanced Placement English teachers discovered it, and that the book grew from his lectures; the article date and byline were not visible. The list page itself was not inspected. The Apple Books page has no bestseller wording; its description speaks of a third edition though it lists a 2014 release, so the edition wording could not be resolved.
Open the published source (opens in a new tab)What it supports: Supports 18 instructional days, 5 average or below-average high school readers, 5 experienced teachers as experts, higher attention to high-salience passages, literal responses falling from 78 to 21 percent, and rules-of-notice comments rising from 6 to 15 percent against 64 percent for experts. One text; no control group. It tested an emotional-language strategy, not Foster’s book.
Open the published source (opens in a new tab)What it supports: PubMed abstract supports moderate to strong correlations between leisure reading and comprehension and language outcomes. The studies did not measure literary interpretation.
Open the published source (opens in a new tab)What it supports: Supports the invitation from the University of Pennsylvania and the College Settlement, the critics who blamed the Black community after the 1893–96 depression, the title Assistant Instructor from August 1896, Du Bois’s post at Wilberforce, residence with his wife near the College Settlement for over a year, the Seventh Ward’s roughly 9,700 of 40,000 Black residents, a door-to-door canvass with more than 5,000 interviews, the questionnaires’ topics (occupations, health, education, religious, social and family life), the 1899 publication, his emphasis on exclusion, slavery’s legacy and discrimination over a monolithic pathology, and most of his methods lying dormant until the 1920s. The page spells his wife’s surname Gormer; the lesson gives only her first name. The page does not report a household count.
Open the published source (opens in a new tab)What it supports: Supports that Du Bois had no office and taught no students, that he was left out of the 1896–97 catalogue but appeared in 1897–98, and that he took a position at Atlanta University in 1897 after it became clear Penn would not offer the post he hoped for. The page reports this through named scholars and does not record a formal refusal. The archivist quoted attributes the catalogue omission to a post the Trustees never formally created, not necessarily to intent; the lesson states only the omission.
Open the published source (opens in a new tab)What it supports: Supports the appointment from 1 August 1896 to 31 December 1897 to study the Seventh Ward, the house-to-house interviews with some 5,000 people, a questionnaire, and the 1899 book. It does not list the questionnaire’s topics. It gives the title Assistant in Sociology, which differs from M017A. The University Almanac (7 February 2012) gives Assistant Lecturer. The page shows no publication date; its metadata gives a last-modified date of 8 July 2019. The year field follows the site’s convention for undated pages.
Open the published source (opens in a new tab)What it supports: Supports that Penn named Du Bois an honorary professor in February 2012, 49 years after his death, and independently that he was not allowed to teach and had no office. The report states the door-to-door method.
Open the published source (opens in a new tab)What it supports: Supports that, among more than 7 million U.S. families who moved, the outcomes of children moving to a better area, defined by how children already living there fared, converge on those outcomes at about 4 percent per year of childhood exposure. So the 4 percent is a share of the gap between places, not a 4 percent rise in income. The baseline estimate comes from a subsample of about 1.6 million children in families who moved once, and the paper uses sibling comparisons and moves triggered by displacement shocks. It studies movers in the United States and does not test Du Bois’s findings. Published article read, to about 100,000 of 122,000 characters, on an MIT-hosted copy.
Open the published source (opens in a new tab)What it supports: Supports the 28 experiments, 55,842 applications for 26,326 positions, 24 studies since 1989, 36 percent and 24 percent more callbacks for white applicants, and no change against African Americans over 25 years with modest evidence of decline against Latinos. Limited to U.S. hiring at the point of hire. PubMed record 28900012 read.
Open the published source (opens in a new tab)What it supports: Argues that segregation in American metropolitan areas was deliberately created by federal, state and local government, through racial zoning, segregating public housing, subsidies for whites-only suburbs, tax exemptions, and government support for violent resistance, and that this creates an obligation to remedy it. Described from a bookseller page carrying the publisher’s text, not from the book. The publisher’s own page could not be opened. A public library catalogue confirms Liveright, a division of W. W. Norton, New York, 2017. The bookseller page calls it a New York Times Bestseller. The bestseller evidence is M017H.
Open the published source (opens in a new tab)What it supports: Supports, as a news report, that The Color of Law stood at number 6 on the New York Times combined print and e-book nonfiction list in June 2020, and that 9 of the top 10 titles concerned racism in America. The New York Times list page itself could not be opened, and the lesson attributes the position to this report. It does not explain why any reader bought the book.
Open the published source (opens in a new tab)What it supports: Supports that the Home Owners’ Loan Corporation maps led to lower homeownership, house values and rents and more segregation in later decades, with similar results across a city population cutoff, attributed likely to reduced credit and disinvestment. Abstract read; the full paper and effect sizes were not.
Open the published source (opens in a new tab)What it supports: Supports that the Federal Housing Administration did not insure mortgages in low-income urban neighbourhoods where most urban Black residents lived, before the maps existed and with little change after; that the Home Owners’ Loan Corporation lent broadly, including to Black homeowners; and that the maps had little effect on where either programme lent. It does not measure later neighbourhood outcomes, as M017I does. The loan counts, over 16,000 in 3 cities, and the footnote naming Rothstein among those who credit the maps with a huge impact come from NBER Working Paper 29244, the earlier version; the authors oppose that view through their conclusion, not by answering him directly. Volume 141 (May 2024) comes from the University of Arizona record, and the DOI dates from 2022. The journal article’s final text was not read.
Open the published source (opens in a new tab)What it supports: Supports that The Sunday Times asked Trevor-Roper, the British Hitler scholar, to verify the diaries and paid for Commonwealth rights; that a teaser quote ran in The Times on 23 April; that Frank Giles phoned him that day and he admitted last-minute doubts, an account the page takes from Robert Harris’s Selling Hitler (1986), which was not read; that over 1 million copies were being printed and the paper presented the diaries as genuine the next day; and the Hamburg press conference the day after that. The date of 25 April is derived from this page and agrees with M018C’s Monday. The page shows no date. The year 2019 is inferred: the page anchors on Giles’s death on 30 October at 100, and a November 2019 obituary confirms the year.
Open the published source (opens in a new tab)What it supports: Supports that Trevor-Roper was a director of Times Newspapers Ltd., examined some of the books in a Swiss bank, wrote that his doubts gradually dissolved as he turned the pages, and later said some documents were forgeries. Only page 2 of the 11 web pages was read.
Open the published source (opens in a new tab)What it supports: Supports that Trevor-Roper had half a day with the journals at a Swiss bank, wrote in The Times that he was satisfied, and at Hamburg on the Monday retreated and said normal verification had been to some extent sacrificed to the scoop. The date of 23 April comes from M018A.
Open the published source (opens in a new tab)What it supports: Supports the German Federal Archives’ announcement on 6 May 1983, the paper additive not used until 1954, the post-war materials in the seal threads, and the conviction and sentences of four and a half years, which it gives for both Kujau and Heidemann. Its figures for volumes and price differ from other sources and within the entry, so the lesson does not rely on them. The entry is undated; the site footer reads 2019. The year field follows the site’s convention for undated pages.
Open the published source (opens in a new tab)What it supports: Supports the reported origin of the diaries in a 1945 crash near Dresden and hiding in East Germany, the three handwriting tests, Stern’s refusal to let German war experts see them, fearing a leak, and the magazine’s judgement that the scandal ruined Trevor-Roper’s reputation. Its sentence of 42 months differs from M018D and M018F, and the lesson does not use it.
Open the published source (opens in a new tab)What it supports: Supports Stern’s payment of 9.3 million Deutschmarks, Kujau as forger and his sentence of four and a half years, and that the German Federal Archives first revealed the forgery. It does not report the time he served.
Open the published source (opens in a new tab)What it supports: Supports the 8 historians and 8 high-achieving high school students, the Lexington documents, use of the sourcing heuristic, defined as reading a document’s source before its text, at about 98 and 31 percent, and historians ranking the textbook excerpt below an excerpt from Howard Fast’s novel, while one student rated it the most trustworthy of the 8 documents. The paper does not report how many students ranked the textbook high. Tiny, non-random sample; does not test forged documents.
Open the published source (opens in a new tab)What it supports: Supports the 6-month intervention, 236 eleventh graders in 5 San Francisco high schools, the document-based design, and significant treatment effects on historical thinking, factual knowledge, comprehension and transfer. Not randomised; effect sizes not read.
Open the published source (opens in a new tab)What it supports: Argues that American history textbooks should be critiqued and offers a retelling of how the history could be taught, from 12 surveyed textbooks, with six new textbooks in a revised edition. Bestseller evidence: The New Press’s page says over a million copies across editions; an undated synopsis on Blackwell’s says the paperback licensed to Simon and Schuster has sold over 2 million and calls it the best-selling book in The New Press’s history. The figures differ, and the lesson attributes each to where it appears. The American Book Award and the new chapter come from The New Press’s page. No bestseller list was inspected.
Open the published source (opens in a new tab)What it supports: Supports the 6 high school textbooks published from 2002 to 2005 and the heroic, celebratory and messianic portrait of King, omitting struggles, mistakes and ordinary participants. Details confirmed on a hosted PDF of the article; the publisher’s own page was not read. A content analysis, not a measure of what students learn.
Open the published source (opens in a new tab)What it supports: Supports about 2,000 juniors and seniors in all 50 states, a non-random sample, asked to list famous Americans other than presidents, naming King at 67 percent, Parks at 60 and Tubman at 44, that Black Americans were once largely absent from textbooks, and that the authors credit curriculum change and a wider cultural curriculum. It does not evaluate Loewen’s book.
Open the published source (opens in a new tab)What it supports: Supports that he tried vocations including art dealer and clergyman, decided at 27 to become an artist, was largely self-taught, zealously copied prints and studied manuals such as Bargue’s, believed in mastering black and white before colour; that in 1889 to 1890 at Saint-Rémy he made 21 copies after Millet, called them interpretations or translations, working from photographs and prints; and that his decade-long career left nearly 900 paintings and more than 1,100 works on paper. It does not test whether copying caused his development.
Open the published source (opens in a new tab)What it supports: Supports that Van Gogh arrived in 1878 at age 25, worked as an unsuccessful Protestant evangelist in the Borinage, decided in August 1880 to become an artist, and made drawings after prints of masterworks by artists such as Millet. Only the publisher’s page was read, not the catalogue. That page gives no title line or publication date beyond the 2015 exhibition; title, Yale University Press and 2015 are confirmed by the Lehmanns catalogue record (by Sjraar van Heugten and others).
Open the published source (opens in a new tab)What it supports: Supports his December 1878 arrival, the February 1879 six-month trial as a preacher, the end of that post mainly over his weak speaking, his family’s reaction, nearly a year without letters, resuming to Theo in June 1880, the August 1880 letter on becoming an artist, and Theo sending engravings, works by Millet and handbooks. It is a tourism page, and it spells Bargue as Barques.
Open the published source (opens in a new tab)What it supports: Supports the three conditions of Experiment 1, the qualitative difference from the models after copying unfamiliar abstract drawings, the gain from unfamiliar but not familiar styles, and the gain from copying or thorough viewing but not from merely thinking about other styles. Abstract read through PubMed; sample sizes are not in the abstract. It does not test long-term art careers.
Open the published source (opens in a new tab)What it supports: Supports 30 non-art undergraduates, a three-day design and two artist raters in Experiment 1, listed there as under review, at a talk at Hong Kong Shue Yan University. Used only for these details, as the journal abstract omits them and the journal full text was not available; the published paper may differ. An undated Ishibashi and Okada conference paper on eScholarship also reports 30 undergraduates, 3 days and 2 professional artists, but it was not shown to be the same experiment.
Open the published source (opens in a new tab)What it supports: The book argues that creative work grows from what you absorb from others, remixed and reimagined, with curiosity and a conventional daily life. Argument taken from a bookseller’s description, as the book was not read. Bestseller: the Workman imprint page on the Hachette site calls it a New York Times bestseller; Kleon’s site calls him a New York Times bestselling author and the bestselling author of this book; no list page, rank or date was seen, and Wikipedia repeats the label without a reference. Kleon’s site says his books together have sold over 2 million copies, with no figure for this title. Publisher and date (28 February 2012) from the MIT Press Bookstore record.
Open the published source (opens in a new tab)What it supports: Supports that in 2011 he was a 27-year-old copywriter, spoke at Broome Community College in New York, wrote a blog post that crashed his website, sold the book to Workman, and that his editor told him on tour that he was a New York Times bestselling author. It is the author’s own recollection.
Open the published source (opens in a new tab)What it supports: Supports that a meta-analysis of design studies (N = 43) found examples brought more example-related ideas, fewer categories and more novel ideas, that quality was positively correlated with copying, and that benefits grew with fewer and less common examples. The abstract was read in a library record, not the full paper; authors and citation confirmed in the British Library catalogue, which shows no separate issue number. It covers design, not fine art.
Open the published source (opens in a new tab)What it supports: Supports that in August 1880 he was making large drawings after Millet, asked Theo to lend him prints including Millet’s Les Travaux des Champs, and wanted Bargue’s Exercices au Fusain. Read in this web copy only, because vangoghletters.org returned an empty page shell. It does not itself state a decision to become an artist.
Open the published source (opens in a new tab)What it supports: Supports the garbage boxes, the high death rate in the crowded foreign colonies, 12 club members going into the alleys 3 evenings a week, 1,037 reports from Hull-House in August and September, 3 inspectors moved, the death rate still high through the next winter, the bid with 2 businessmen the following spring, the technicality, the appointment at 1,000 dollars a year, rising at 6, following wagons, 9 to 17 wagons, court cases, third to seventh with credit to inspection and other causes, and the alderman abolishing the post. It is her own account and makes no causal test. The chapter gives no year and places the crisis in her fourth summer, which sits loosely with the 1895 date from M020B and M020C. Macmillan, New York, 1910 confirmed in the University of Southern Indiana catalogue record and a Hindman auction record.
Open the published source (opens in a new tab)What it supports: Supports Hull-House in the Nineteenth Ward from 1889, and the 1895 appointment as garbage inspector as her first paid job. Its claim of America’s first settlement house is not used.
Open the published source (opens in a new tab)What it supports: Supports that Addams and Starr opened Hull-House in 1889 on Halsted Street, and that the city appointed her garbage inspector in 1895 after ignoring her reports. The page calls it Chicago’s toughest, poorest neighbourhood, with no measure given; the lesson softens this to one of the poorest.
Open the published source (opens in a new tab)What it supports: Supports 950 participants, assignment to Housing First with assertive community treatment or usual treatment, 71 versus 29 percent of time in stable housing over 2 years, and narrowing of functioning and quality of life differences by the end of year 2. Abstract read through PubMed. Covers people with serious mental illness in Canada only.
Open the published source (opens in a new tab)What it supports: Supports 506 boys, matched pairs split by coin toss, counsellors visiting families for an average of 5 years, 168 men with convictions in each group, worse results on some measures for treated men, and about two-thirds of respondents saying the programme helped. McCord notes many comparisons and calls the side effects subtle. Full text read via a third-party copy; citation and Massachusetts setting confirmed in PubMed (which prints the DOI as 10.1037//0003-066x.33.3.284).
Open the published source (opens in a new tab)What it supports: The book argues that people react to problems instead of preventing them by fixing the systems behind them, partly because of problem blindness. Argument from catalogue and bookseller descriptions, as the book was not read. Bestseller: the Barnes & Noble overview and the Porchlight description both carry the words Wall Street Journal Bestseller, with no date or rank. Two searches found no Wall Street Journal list page naming the book; the 2 Journal lists opened (2018 and 2019) predate it, and the Simon & Schuster Canada page would not load, so the publisher’s own use of the label was not seen. No sales figure was seen. Publication date 3 March 2020 from NetGalley and Porchlight.
Open the published source (opens in a new tab)What it supports: Supports 400 women enrolled, 324 in follow-up, and 0.29 versus 0.54 verified reports of child abuse and neglect. Abstract read through PubMed, which gives no DOI; the DOI comes from the Scilit record. Semirural New York only.
Open the published source (opens in a new tab)What it supports: Supports 823 plus 822 women, teenage first-time mothers at 18 English sites, no additional short-term benefit on the primary outcomes up to 24 months, and that the programme is an adaptation of David Olds’s Nurse Family Partnership. Abstract and full text read through PubMed Central. Some secondary outcomes showed small gains, and it does not cover longer-term effects.
Open the published source (opens in a new tab)What it supports: Supports that some preventive measures save money, that the vast majority reviewed in the health economics literature do not, and that opportunities for efficient investment are roughly equal for prevention and treatment. The record was read through PubMed, which has no abstract, and the wording through quotations in APS Observer, PolitiFact and The Skeptical OB; the article itself could not be opened. It concerns health care, not social services.
Open the published source (opens in a new tab)What it supports: Supports that it stayed unproved for more than 350 years; that Wiles, then at Princeton, retreated to his attic to prove the Taniyama-Shimura conjecture after Ken Ribet’s 1986 work; that he worked alone and in secret for 7 years; that in June 1993 in Cambridge he gave a series of lectures and about 20 minutes into the third concluded with the theorem; that the New York Times put it on its front page the next day; that about 200 pages went to Inventiones Mathematicae and 6 reviewers; that Nick Katz and Luc Illusie checked one section for 2 months and that in late August 1993 Wiles’s explanation did not satisfy them and he saw a serious gap; that by December 1993 rumours spread, peers pressed for a draft and he kept the gap mostly to himself; that Peter Sarnak urged a collaborator and Wiles called Richard Taylor; that small fixes failed and wider searches ran through spring and most of summer; that in September 1994, near admitting defeat, he took one last look and saw that his Iwasawa theory approach, abandoned 3 years earlier, could close the gap; and that they repaired it within weeks. It does not say outright that he refused the draft requests. The 20 minutes, 6 reviewers, Sarnak’s advice and the 3 years appear only here among the sources read. It is not peer reviewed and does not establish the mathematics.
Open the published source (opens in a new tab)What it supports: Supports that in late summer 1986 a friend told Wiles that Ken Ribet had linked the theorem to the Taniyama-Shimura conjecture, that Wiles compared research to entering dark rooms in a mansion and stumbling for months before finding the light switch, that when stuck he would go for a walk, often down by the lake, because it lets the subconscious work, and that the error lay in a crucial part of the argument. It is his own recollection, and it does not test whether walks help. He says 8 years here for the whole work and the interviewer says 7; the lesson uses the 7 years before the announcement, which is 1986 to 1993.
Open the published source (opens in a new tab)What it supports: Supports that at age 10, in 1963, Wiles found a book on the theorem in his local library; that he studied it in private for 7 years at Princeton before announcing a proof in 1993; that a referee found a serious error, fixed with former student Richard Taylor; that the corrected proof appeared in 1995; and that he received the 2016 Abel Prize. The page gives both 7 years (to the 1993 announcement) and 8 years (to the proof), so it is not inconsistent with the lesson’s 7. The page shows no publication date; 2016 comes from the prize it reports, and the Oxford Mathematical Institute’s own notice of the prize is dated 15 March 2016.
Open the published source (opens in a new tab)What it supports: An independent second source that Wiles announced in lectures in Cambridge in June 1993; that Nick Katz found a gap while refereeing; that Wiles worked on it first alone and then with former student Richard Taylor; that by September 1994 they were almost ready to admit defeat before a sudden insight; and that the corrected proof appeared in 1995 as two papers. It names the Annals of Mathematics as the journal where Brown names Inventiones Mathematicae, so the lesson names neither. It leans on Simon Singh’s book for part of this account, and the Katz sentence has no inline citation.
Open the published source (opens in a new tab)What it supports: Supports the statement of the theorem for positive whole numbers; that in December 1993 Wiles said the proof was mostly correct but one final calculation was not yet complete; and that in the week before 31 October 1994 he sent a revised proof to about 20 colleagues, with Richard Taylor as collaborator, and submitted it to the Annals of Mathematics. It reports the verification as still under way.
Open the published source (opens in a new tab)What it supports: Supports that trying problems before instruction gave a moderate overall learning advantage over instruction first (Hedges’ g = 0.36), larger with close adherence to Productive Failure design, and that instruction first did better for younger learners (grades 2 to 5) and domain-general skills. Grade level, intervention time span and (quasi-)experimental design also contributed to the advantage. The abstract also gives a publication-bias-adjusted g of 0.87, not used in the lesson. The total number of participants was not seen. It is not specific to mathematics or to professional research.
Open the published source (opens in a new tab)What it supports: Supports that 54 seventh-grade classes taught by 15 teachers in 5 middle schools of one Florida district were randomly assigned to mostly interleaved or mostly blocked practice; that 1,103 students were in the classes at recruitment and 787 were analysed; that all took the same interleaved review and, about 33 days later, an unannounced test, on which the interleaved group scored 60.7 percent and the blocked group 37.6 percent (d = 0.83). The course was Honors Advanced Grade 7 Mathematics. The authors note that time on task was not measured, that the sample came from one district, that feedback may matter, that benefits may be smaller at shorter delays and that some blocked practice may be needed first. The IES record gives volume 112 and pages 40–52 with no issue number, and the author copy has placeholder numbers, so none is recorded.
Open the published source (opens in a new tab)What it supports: The publisher’s page supports publication on 31 July 2014 by Tarcher; the argument that mathematics takes creative as well as analytical thinking, that working at a problem in a tightly focused way until it is solved is not an effective way to learn, and that stepping away lets a more relaxed part of the brain help; that the author flunked high school mathematics and science, enlisted in the army and returned to study them; and that it is aimed at students struggling with these subjects and people changing careers, as companion to a popular online course. Bestseller evidence: the publisher’s page for Oakley’s later book calls it “the bestselling book A Mind for Numbers”, and its own U.S. page shows a Best Seller badge; its Australian page has none. Shortform, a book-summary site, reports it ranked 14th on the New York Times list of best-selling science books in 2014, and describes focused and diffuse thinking; it links to a New York Times page dated 9 September 2014, which could not be opened. A California State University, Fullerton overview confirms the two modes and does not call the book a bestseller. No New York Times list page was inspected and no sales figure was seen, so the lesson attributes the label. The book’s text was not read.
Open the published source (opens in a new tab)What it supports: Supports a positive overall effect of setting a problem aside (weighted mean 0.29, 95% CI 0.21 to 0.39), with creative divergent-thinking tasks benefiting more (14 studies; weighted mean 0.29, wide interval) than visual and linguistic insight problems (0.26 and 0.22), in the authors’ moderator analysis. The authors’ regression found no sign of publication bias, but they still say effects may be overstated by it. Median 25 people per study. The included experiments used puzzles and creative tasks; the paper reviews no maths courses and not this book.
Open the published source (opens in a new tab)What it supports: Supports the ratings of practice testing and distributed practice as high utility, interleaved practice as moderate, and rereading and highlighting as low. It rates general techniques, not this book, and did not test relaxed and focused thinking.
Open the published source (opens in a new tab)What it supports: Supports that Wiles gave 3 days of lectures from 21 June 1993, closing on 23 June, after 7 years in virtual secrecy, and reports Gina Kolata saying the New York Times story was displayed twice on Page 1. It does not date those two Page 1 appearances, so ‘the next day’ rests on Brown; a University of Waterloo bulletin of 24 June 1993 confirms that the Times reported the news that morning. The page shows no date. Its volume and issue come from the URL. The year field holds the access year because the schema needs one, as with other undated sources; the citation stays n.d.
Open the published source (opens in a new tab)What it supports: Supports that Wiles went to Cambridge for 3 lectures at the Isaac Newton Institute and announced his result on 23 June 1993; that referees then found a flaw; and that by 25 October 1994 he had fixed it with Iwasawa theory, a technique he had earlier tried and rejected. It is a review of Simon Singh’s book, so it is not independent of that book. The closing page reads printed page 514 and carries Stent’s byline; the first page’s heading gives the title, and its OCR was unclear on the reviewer’s name, so the byline at the end is used. Page range not recorded.
Open the published source (opens in a new tab)What it supports: Supports that Codman resigned his full-time Massachusetts General post in August 1911 and opened his private End Result Hospital in November 1911; that 337 patients were discharged from 1911 to 1916, with 123 errors recorded and end results measured for all; that he recorded his own errors; that he paid for annual reports and mailed them to major hospitals, challenging them to publish theirs; that on 6 January 1915 he presented a large cartoon at a Surgical Society meeting he chaired, showing greedy surgeons grasping for gold from wealthy patients, with the patients as an ostrich with its head buried in Beacon Hill, and was ostracised and ordered to step down as chair; and that his practice income rose to 10,000 dollars by 1914 and fell to 5,000 that year (lifetime average 6,000). It is a short profile, not a full biography.
Open the published source (opens in a new tab)What it supports: Supports Codman’s birth in Boston in 1869 and training at Massachusetts General (final-year intern, medical degree 1895, then under Francis Harrington), the date on the lesson’s timeline scene; his proposal to use results rather than seniority for promotion; that the hospital refused the End Result system; that he kept a part-time post until 1914, when he resigned entirely; and the 1915 meeting, ostrich cartoon and requested resignation from the Surgical Section of the Suffolk Medical Society. It does not give income figures.
Open the published source (opens in a new tab)What it supports: Supports the description of the End Result Idea: following patients long enough to determine whether treatment succeeded and taking measures to prevent new failures. Only the PubMed record and abstract were inspected, not the full article.
Open the published source (opens in a new tab)What it supports: Supports that Codman’s end-result system inspired Franklin Martin, a founder of the American College of Surgeons, and that the College’s standardisation programme reviewed 692 hospitals with 100 or more beds, of which 89 met the minimum standards. It gives no year for that review, so the lesson states none. The 692 and 89 are corroborated by a Springer book chapter on the Joint Commission (1918, 100 beds or more), and Moseley (2008) gives 692 and 13 percent in 1918.
Open the published source (opens in a new tab)What it supports: Current version of the review (published 22 June 2026; same abstract as the March 2025 version, pub4). Supports that 292 studies were included; that across 558 dichotomous outcomes from 177 studies the median absolute improvement in desired practice was 2.7 percent (interquartile range 0.0 to 8.6), with a meta-analytic mean of 6.2 percent (95% CI 4.1 to 8.2, moderate certainty); and that exploratory analyses linked larger effects to lower baseline performance, individual-level data, comparison with top peers or a benchmark and action plans, and repeated delivery to lower effects. It reviews feedback to professionals, not published error reports. Abstract inspected through PubMed. It replaces the 2012 version (pub3: 140 studies, median 4.3 percent), which called repeated feedback more effective.
Open the published source (opens in a new tab)What it supports: Supports that 12 studies were included; that low-certainty evidence suggests public release may make little or no difference to long-term utilisation by consumers or providers and may slightly improve some patient outcomes; and that the authors judged the evidence inadequate to directly inform policy and practice. Abstract inspected through PubMed.
Open the published source (opens in a new tab)What it supports: Publisher and bookseller descriptions support that it first went on sale on 10 September 2015; a 10th anniversary edition appeared on 25 September 2025; that it argues failure is a key to innovation and improvement, that openly examining mistakes beats denying them, and that failing to learn is costly, citing preventable medical errors; that its cases come from elite sport, medicine, aviation and business; and that a publisher synopsis on Waterstones’ page says it extends to business, politics, healthcare, aviation and innovation. Preventable medical errors appear in the description on Schweitzer’s page, not on Hachette’s. Bestseller evidence: Hachette UK and Hachette India bill the 10th anniversary edition as an edition of “the Sunday Times No.1 bestseller”. One search and no Sunday Times list page or book-specific sales figure was seen, so the lesson attributes the label to the publisher. The book’s own text was not read.
Open the published source (opens in a new tab)What it supports: Supports that 71 cardiothoracic surgeons and 6,516 analysed procedures in Massachusetts from October 1999 to September 2009 were studied; that surgeons learned more from their own successes than their own failures, with performance worsening after a failure; and that colleagues’ failures helped more than colleagues’ successes (weak statistical support, p about 0.08 to 0.10). One procedure, one state, in-hospital mortality as the outcome. It did not test the book.
Open the published source (opens in a new tab)What it supports: Supports that 43 studies were included; that there was no strong evidence that incident-reporting systems beat other methods; that there was some evidence of changes to clinical settings or processes but little of improved outcomes, and insubstantial evidence of cultural change; and that systems could be more effective when owned by clinical teams. It did not test the book. Abstract inspected through PubMed.
Open the published source (opens in a new tab)What it supports: Supports the 25,000 dollar licence; the buyers’ slight understanding of semiconductors and the belief that transistors could not work at radio frequencies; that Western Electric viewed the transistor as a telecommunications technology and the hearing aid as its only consumer use; that Ibuka wanted a pocket-sized radio; that high failure rates for high-frequency transistors raised costs and put the project in doubt; that the 1955 TR-55 had poor sound by Morita’s admission, was neither cheap nor compact, used less battery power and sold as the only transistor radio in Japan; and that over 12 million transistor radios had sold by 1960. It dates the end of the licence negotiations to 1953. The 25,000 dollar fee is also given by the Semiconductor History Museum of Japan essay (M023I), a company history from the International Directory of Company Histories and the Encyclopedia of World Biography.
Open the published source (opens in a new tab)What it supports: Supports the transistor’s invention at Bell Laboratories in 1947; the company’s formation in 1946; Ibuka as an electrical engineer who in 1952 visited the U.S. and learned Western Electric was preparing to license the patent; Regency’s first transistor radio in 1954; the TR-55 in 1955; and the 1957 Type 63, which sold in the millions. It dates the licence purchase to 1954, so the lesson gives no year. Sony’s audio history page and a Sony Europe company history confirm the TR-55 and TR-63 dates; the company history on Funding Universe confirms Regency’s precedence.
Open the published source (opens in a new tab)What it supports: Supports the study of semiconductor photolithographic alignment equipment from 1962 to 1986; over 100 interviews; four waves of architectural innovation each followed by a new entrant taking the lead; established firms misreading new designs as extensions or copies; the note that RCA built a portable transistor prototype in the mid-1950s yet stayed a follower of Sony; and the limits of one industry with other causes contributing.
Open the published source (opens in a new tab)What it supports: Supports the analysed sample of 64 innovations (from 93) in consumer durables and office products, 1851 to 1998, of which 53 met the radicalness threshold; that about 73 percent of older radical innovations came from non-incumbents and about 74 percent of post-war ones from incumbents (the paper’s figure gives no n; 53 is the likely base); and the authors’ limits, including a focus on relatively successful innovations and two product classes. The authors also find size effects strong and say the incumbency result may partly reflect collinearity with size, which the lesson does not cover.
Open the published source (opens in a new tab)What it supports: Publishers Weekly (M023J) gives Harvard Business School Press, May 1997, and Christensen as a Harvard Business School associate professor. Barnes & Noble and Porchlight pages bill it as a Wall Street Journal and Businessweek bestseller; Porchlight names Harvard Business Review Press for the 2024 edition. The publisher’s own page and any such list page were not seen, so the lesson attributes the label to bookseller pages. The Porchlight page supports the summary of the argument. The book’s own text was not read.
Open the published source (opens in a new tab)What it supports: Tuck School of Business’s report and MIT Sloan Management Review’s pages support that 79 experts assessed 77 of Christensen’s cases and that 9 percent fitted all four elements tested; and that a response by Juan Pablo Vázquez Sampere disputed the method, including that the theory has more than four elements. The authors say they do not advocate discarding the theory. The full article could not be opened by the checker either (permission request not answered), so volume and pages are omitted, and it is not a conventional peer-reviewed research paper.
Open the published source (opens in a new tab)What it supports: Supports that 36 technologies across 7 markets were studied and that lower-attack technologies were introduced by incumbents about as often as by entrants, were not cheaper and rarely disrupted firms. Abstract inspected.
Open the published source (opens in a new tab)What it supports: Supports that the theory began with the disk-drive study, its core components including overserving customers and incumbents constrained by existing customers; that the original theory was a statement of correlation lacking a causal mechanism; that follow-on research supplied causal pathways; and, hedged by the authors, that disruption research is adept at framing problems but mostly empty-handed on solutions. Written from inside the theory.
Open the published source (opens in a new tab)What it supports: Supports the 25,000 dollar contract fee, that Ibuka told Western Electric he planned a radio, and that it advised against this because of high-frequency difficulties and said a hearing aid was the best that could be done. It gives no licence date, names no individual author, and 2016 is its copyright year. It does not cover the TR-55, TR-63 or Regency.
Open the published source (opens in a new tab)What it supports: Supports publication in May 1997, sales rising sharply after a Forbes cover story of 25 January, 76,500 copies in print and number 2 on Amazon’s Top 100. It is the only sales figure seen and dates from 1999. The page text shows no headline, byline or date, so title and date are taken from its URL (1 March 1999).
Open the published source (opens in a new tab)What it supports: Supports the 1914 appointment by the Surgeon General; 30,000 South Carolina cases with a 40 percent fatality rate by 1912; the Thompson-McFadden Pellagra Commission’s conclusion, dated 1914; the Jackson orphanage diet change (more milk, buttermilk, eggs, beans, peas, restricted corn) and few recurrences, with a control group impracticable at the orphanages; Goldberger’s dates (1874 to 1929); and the 1937 identification of niacin. It calls the commission one set up by the US government, but three other pages describe it as private, so the lesson does not call it either. It does not cover the prison study in detail. An earlier version is on the James Lind Library.
Open the published source (opens in a new tab)What it supports: Supports the symptoms of pellagra; that Rankin Prison Farm had no pellagra; Governor Brewer’s pardons; the diet from 19 April 1915; 11 men pardoned on 1 November 1915 (12 were chosen, and the article does not explain the twelfth); Goldberger’s count of 5 men with pellagra, in a letter to his wife dated 30 October 1915; the spring 1916 self-exposure; that coercion concerns were largely absent then, though scientific criticism was sharp; and that the research was pattern-setting for prison studies into the 1970s. Its count of 5 differs from the NIH history’s 6, and the lesson gives both.
Open the published source (opens in a new tab)What it supports: Supports that inmates of mental hospitals and orphanages contracted pellagra while staff did not; 6 of 11 prison volunteers showing pellagra rashes after five months; the 26 April 1916 blood injections between Goldberger and Wheeler, and swallowed scabs; and that others including Mary Goldberger took part and none got pellagra. The exhibit is undated; 2009 is the capture read, and a 2008 capture is also listed. Only page 5 of 8 of the exhibit was read.
Open the published source (opens in a new tab)What it supports: Supports that of 363 articles testing standard of care in one high-impact journal from 2001 to 2010, 146 (40.2 percent) reversed the practice and 138 (38.0 percent) reaffirmed it. It does not give a rate for medicine as a whole. The caution that one journal is not all of medicine is the lesson’s reading.
Open the published source (opens in a new tab)What it supports: Supports that of 45 highly cited studies claiming effectiveness, 7 were contradicted, 7 had stronger effects than later studies, 20 were replicated and 11 remained largely unchallenged, and that 5 of 6 non-randomised studies were contradicted or had stronger effects, against 9 of 39 randomised trials.
Open the published source (opens in a new tab)What it supports: The book argues that readers can learn how health claims are tested, using homeopathy, placebo effects, health scares and drug trials. Bestseller support is the publisher’s: its author page says the book reached Number One in the nonfiction charts and has sold over 500,000 copies, and its product page says over 400,000 in the UK alone. These may be different dates or territories; neither is dated. No chart page was inspected. The 2008 date comes from Wikipedia (September 2008, with a Telegraph review dated 27 September 2008) and a Journal of Scientific Exploration review (first published in 2008 in the UK by 4th Estate), used for the date only; the publisher pages give no date. The publisher’s blurb raises alternative therapists and drug companies using the same tricks as a question, so the lesson does too. The MMR and journalism points draw on that blurb and Wikipedia’s chapter summary. The book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports that published literature showed 94 percent of trials as positive against 51 percent in FDA analysis, and that the authors could not determine whether authors, sponsors or editors caused the gap. Covers antidepressants only.
Open the published source (opens in a new tab)What it supports: Supports that after accounting for bias there was weak evidence for homeopathic effects and strong evidence for conventional interventions, compatible with placebo effects. Its choice of trial subset is disputed by M024I.
Open the published source (opens in a new tab)What it supports: Supports that Shang’s result varies with the trial subset and threshold chosen, and that it is less definite than presented. PubMed lists Rutten’s affiliation as the Association of Dutch homeopathic physicians. It does not show homeopathy works.
Open the published source (opens in a new tab)What it supports: Supports that at 18 he returned late after playing banjo at a Paris dance hall; that a candle set fire to celluloid flowers his wife made to sell at cemeteries (told by the biographer Michael Dregni); that he was burned saving his wife; more than a year in and out of hospitals; that no one expected him to play again; full use of only index and middle fingers; and hours of painful practice leading to a new technique using those fingers and a very quick right hand. Not a scholarly source. The date is taken from the page URL and fits its line that Saturday marks the centennial.
Open the published source (opens in a new tab)What it supports: Supports birth on 23 January 1910 in Belgium, Roma descent, learning violin, guitar and banjo while travelling, a 1928 caravan fire leaving two left-hand fingers unusable, and coleading the Quintette du Hot Club de France from 1934.
Open the published source (opens in a new tab)What it supports: Supports, citing Charles Delaunay’s 1961 biography, that by 1928 Jack Hylton had signed him for London and that the fire came before he could meet that obligation; about two years of self-conducted therapy; and, citing Art Wrightman, that he was strictly a two-fingered player who could rarely hook the little finger for certain chords; and meeting Grappelli in Paris in late 1931. It reports that sources conflict on his injuries, including which limb was considered for amputation, so the lesson leaves that out. Encyclopedia.com stamps the page as updated in 2018, which is the site’s date, not the entry’s publication date.
Open the published source (opens in a new tab)What it supports: Supports the operation on his 19th birthday being only partly successful, full use of only his first two fingers, laborious relearning, and that the 1934 quintet’s string line-up was unusual in jazz at the time.
Open the published source (opens in a new tab)What it supports: Supports deliberate practice explaining 21 percent of variance in music performance, and the conclusion that practice is important but less than has been argued. It does not study injury.
Open the published source (opens in a new tab)What it supports: Supports 39 violinists against the original 30; practice alone explaining 26 percent of variance against 48 percent in the original (which compared groups differently); and that the best group had not accumulated more practice than the good group. The 1993 figures (30 violinists, 48 percent) are taken from this paper; the 1993 paper was not opened. It is a small sample, and the authors say small samples should be interpreted with caution.
Open the published source (opens in a new tab)What it supports: Supports that within identical twin pairs differing in practice there was no difference in rhythm, melody or pitch discrimination ability. It tests those abilities, not instrumental performance.
Open the published source (opens in a new tab)What it supports: The listing dates the book 3 May 2016, names Scribner, and describes its argument, including that grit can be learned and what happens when you fall down. It does not mention music or injury. It calls the book an instant New York Times bestseller. A library record shows the same date, ISBN and wording. Wikipedia’s Duckworth entry says the book spent 21 weeks on the list, citing a New York Times page for 3 February 2019, but that page could not be opened, so the lesson attributes the label. The publisher’s page could not be opened, and the book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports grit accounting for an average of 4 percent of variance in success outcomes, predicting beyond IQ and conscientiousness, in samples including West Point cadets and National Spelling Bee contestants. It does not focus on musicians.
Open the published source (opens in a new tab)What it supports: Supports grit being only moderately correlated with performance and retention and very strongly with conscientiousness; perseverance of effort outperforming consistency of interest; and that interventions may have only weak effects. It does not focus on musicians.
Open the published source (opens in a new tab)What it supports: Supports the 2 March 1959 ceremony in the rain, with Cahill screwing in Utzon's bronze plaque and jackhammering beginning at once; that Cahill was 68, defended the expense for two years, narrowly won re-election in March 1959 and pushed ahead before critical design elements were settled because he feared obstruction, giving the project irreversible momentum; his death seven months later and his request to Norman Ryan; the inaccurate survey, the loose water-soaked ground, the roughly 700 shafts and the unbudgeted work; that the roof's weight was unknown and its anchor points only vaguely discernible; that a more prudent approach would have resolved design problems first; and that Stage One finished in February 1963, two years late, still needing substantial modification. It gives the estimate as 3.5 million pounds. The page's 'almost five years after construction began' does not fit its own 1959 date and implies a start around 1958, which fits M026M; the lesson does not use that phrase. The entry is undated, so the year field records the year of access.
Open the published source (opens in a new tab)What it supports: Supports that in 1958 engineers asked Utzon to define the roof curves, that he traced them by bending a plastic ruler against a table and sent the drawings to Arup in London, who needed the shapes expressed mathematically; that in late 1961 no workable way of building the shells had been found; and that the spherical geometry appeared in the Yellow Book of January 1962. It gives no year for the start of the podium: it says only that the podium began as the team explored the roof, and dates the roof's evolution 1958 to 1962, so it does not itself conflict with M026A. The entry is undated, so the year field records the year of access.
Open the published source (opens in a new tab)What it supports: Supports Utzon's win in January 1957; the 7 million dollars committed in July 1957 with completion targeted for December 1963; Utzon's resignation on 28 February 1966; the opening by Queen Elizabeth II on 20 October 1973, 10 years overdue, at 102 million dollars, 14 times the original budget. It also supports Bennelong Point and that Utzon was Danish. The sums are in dollars with no inflation adjustment, and the page gives no start date for construction. The 7 million dollars equals M026A's 3.5 million pounds at the 1966 conversion rate of 10 shillings to the dollar (Reserve Bank of Australia Museum, seen by the checker). M026M gives other overrun figures, so the lesson attributes these to this page's account.
Open the published source (opens in a new tab)What it supports: Supports that designing the roof shells took Utzon and Ove Arup and Partners more than four years, that earlier work then had to be revised by rebuilding and strengthening the foundations, and that the build took 16 years. It is independent of the building's own account. The entry is undated, so the year field records the year of access.
Open the published source (opens in a new tab)What it supports: Supports that 258 rail, bridge, tunnel and road projects in 20 nations, worth about 90 billion U.S. dollars, showed cost escalation in 9 of 10 cases, averaging 28 percent across all 258 (rail 45, bridges and tunnels 34, roads 20), with no sign of decline over time; that escalation rose about 4.64 percentage points per added year of implementation, defined as decision to build until operations begin, in a model of 101 projects, or about 3.28 in regional subsets; and that the authors advise resolving planning problems before committing to build. It does not analyse whether design was complete at the start of construction. The authors call the database imperfect and several findings tentative. Costs cover construction only.
Open the published source (opens in a new tab)What it supports: The abstract record supports that budgets for large hydropower dams are systematically biased below actual costs. The figures of 245 dams, 65 countries, 1934 to 2007, an average cost 96 percent above budget, 3 in 4 over budget and 44 percent longer than scheduled were seen only in press reports of the study, four of which agree on 245 dams, 65 countries and 96 percent. The arXiv link returned only the abstract page, and the publisher's page could not be opened; the abstract states no figures. It studies dams; the abstract says nothing about design completeness.
Open the published source (opens in a new tab)What it supports: The publisher's page supports the date of 7 February 2023, the principles described (know your odds, plan carefully before acting, start from the goal and work backward, build from smaller parts, handle unknown unknowns) and the statement that no less than 92 percent of megaprojects run over budget, schedule or both. Bestseller evidence seen: a 'Best Seller' badge on the Penguin Random House page, and an Oxford Saïd Business School news page saying the book has sold over 100,000 copies and calling it a business bestseller. No New York Times or similar list page was seen, so the lesson attributes the label. The badge is not explained on the page, the Oxford page is undated and comes from the author's former school, and a search found no list page naming the book. The Oxford page says the book is read across business and management, and gives Flyvbjerg's aim as advice for projects big or small; it does not otherwise report how readers use it. Both publisher pages list the Sydney Opera House among its examples. 'Know your odds from similar past projects' joins the publisher's 'understand your odds' to the method named in M026L. The book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports that reference class forecasting was compared with Monte Carlo simulation and earned value management on real project data and performed best for both cost and time. Only the abstract and record were seen, so the sample size and error sizes are not used. No DOI was seen on the pages opened, so none is recorded.
Open the published source (opens in a new tab)What it supports: Supports that the authors studied 258 projects worth 90 billion U.S. dollars, found estimates systematically misleading, and argued this is best explained by deliberate misrepresentation rather than honest error. Used here only as the claim its critics attack. Only the abstract record was seen.
Open the published source (opens in a new tab)What it supports: Supports that the authors attack the 2002 paper’s framing of delusion or deception as a false choice that does not reflect how large projects are procured. It is an argument about methods and interpretation. Only the abstract record was seen.
Open the published source (opens in a new tab)What it supports: Supports that cost overruns and benefit shortfalls can arise from selection alone, with no bias in the predictions, whenever predictions help decide which projects are built, and that the authors show with a project database that this effect may be substantial. The database’s size and country were not confirmed, so they are not stated.
Open the published source (opens in a new tab)What it supports: Dated 27 April 2018, by Andrew Forster. Supports that Love and Ahiaga-Dagbui call the 258-project sample statistically unrepresentative and cite Terrill and Danks's 2016 analysis of 836 Australian transport projects of AU$20 million or more, saying 66 percent were on or under budget and 34 percent overran. That split is the critics' reading, not confirmed by Flyvbjerg; the report itself was not read, and a separate 2017 paper led by Love gives 24 percent over and 9 percent under for the same study. Flyvbjerg replies that complexity and scope change are not root causes, that reference class forecasting does better on average, that the 2002 paper acknowledged its sample was probably not representative, and that a later set of 2,062 projects supports the results.
Open the published source (opens in a new tab)What it supports: Used only to disclose disagreement among accounts: it dates stage 1, the podium, 1958 to 1961, and says the 16-year build was 6 years longer than scheduled and 10 times its original estimated cost. It gives no reason for these figures differing from M026C.
Open the published source (opens in a new tab)What it supports: Supports that Grattan analysed 836 transport projects of 20 million Australian dollars or more, planned or built since 2001, measured from the first funding promise, and found 28 billion dollars of overruns, about a quarter of the projects' total value. It does not give the share of projects over or under budget. The page shows no byline, and its year and title are taken from the day and date given, the search listing and the page address. The full report could not be opened.
Open the published source (opens in a new tab)What it supports: Supports that in the first 45 minutes after the market opened on 1 August 2012 a faulty router sent more than 4 million orders while trying to fill 212 customer orders; that Knight traded more than 397 million shares and lost more than 460 million U.S. dollars; that 97 automated emails before the open flagged the error and Knight did not act on them; and the 12 million U.S. dollar settlement, in which Knight neither admitted nor denied the findings. The findings are settled charges, not a court ruling.
Open the published source (opens in a new tab)What it supports: Supports, as a summary quoting SEC Release 34-70694: that Knight stopped using Power Peg in 2003 and left the code on its servers behind a flag; that in 2005 the cumulative-quantity function, which stopped child orders once a parent order was filled, was moved and Power Peg was not retested; that new Retail Liquidity Program code reused the flag; that staged deployment began on 27 July 2012 and a technician did not copy the code to one of eight servers; that no second technician reviewed it and no written procedure required that. It says nothing about the response on the day. It is a secondary account, and the blogger’s own criticisms of the SEC are not used.
Open the published source (opens in a new tab)What it supports: The only inspected source for: Power Peg as a test program that bought high and sold low; the 8:01 emails; New York Stock Exchange analysts seeing doubled volumes by 9:34; no documented incident procedures; about 20 more minutes of diagnosis. It says Knight went back to the old code and reinstalled it on the others, which put the fault on all eight; the count of seven is arithmetic, and Kar (2013) separately supports the fault spreading. Its statement that engineers shut SMARS down at 9:58 does not sit easily with the SEC’s 45 minutes, so it is not used. The inspected text shows no post heading, date or byline; the author is inferred from the site name, the date from the archive path, and the title is descriptive, in square brackets.
Open the published source (opens in a new tab)What it supports: Supports that the 97 automated emails were not designed as system alerts and so were generally not reviewed on receipt, and that the SEC cited code no longer intended for use being left on the servers.
Open the published source (opens in a new tab)What it supports: Supports the 400 million U.S. dollar rescue by an investor group for a 73 percent stake, and the fall of about 70 percent in Knight’s stock from the previous Tuesday. It gives the loss as about 440 million U.S. dollars, a figure from before the SEC’s later figure of more than 460 million, which the lesson uses.
Open the published source (opens in a new tab)What it supports: Supports that 1,247 news articles and post-mortems covering 597 unplanned outages at 32 services from 2009 to 2015 were analysed; that among outages with a known cause, upgrades were the largest group at 16 percent (54 of 340 cause tags; an outage can carry several), with network (52) and bugs (51) at 15 percent each, so the three are close to level; and that 355 outages had no publicly known cause. It relies on public reports and leans towards popular services, so its figures are minimums and it does not test any release practice.
Open the published source (opens in a new tab)What it supports: Supports that a web survey of 7,522 technical DevOps professionals produced high, medium and low performing clusters, and that throughput and stability moved together, with no trade-off found. It is a self-selected, self-reported, correlational survey, and the effects of individual practices were left to a later phase. Two authors are closely tied to the DevOps research movement.
Open the published source (opens in a new tab)What it supports: The publisher’s description supports the plot (Bill, an IT manager at Parts Unlimited, has 90 days to rescue an over-budget project or see his department outsourced), the Three Ways, and the idea that IT work resembles manufacturing and must be streamlined across departments. A bookseller record gives first publication as 10 January 2013. Bestseller evidence: the publisher’s own resources page and distributor, library and bookseller copies of its description say over half a million sold, undated and unaudited; the library copy also calls it bestselling, in the publisher’s voice. No list entry was seen; the Wall Street Journal label on these pages belongs to the sequel. So the lesson attributes the figure and does not claim a list. The figure says nothing about why people read it.
Open the published source (opens in a new tab)What it supports: Supports that the survey identified 167 relevant papers and analysed 50 core ones; that it reports Erich et al. concluding that no quantitative studies support claimed DevOps benefits; that it calls for more quantitative research; and that it reports a severe lack of industry consensus on how to measure software delivery. It does not assess The Phoenix Project and covers work up to 2019.
Open the published source (opens in a new tab)What it supports: Supports that the four key DevOps metrics are often collected through surveys, which yields few data points, and that the authors built a prototype to measure them automatically and practitioners at a company judged the results valuable. Only the abstract page was inspected, and sample details were not seen.
Open the published source (opens in a new tab)What it supports: A second account, six years before the case study, of the response: staff removed the new RLP code from the seven servers where it had been deployed correctly, so that further incoming orders activated Power Peg. It does not say how long the diagnosis took. It gives its own loss figure of 465 million U.S. dollars in the title; the lesson uses the SEC’s figure.
Open the published source (opens in a new tab)What it supports: Supports that by 1958 Weikart was the new director of special education in Ypsilanti (a photo caption says director of special services, 1957-58); the poor outcomes he saw; the Perry School and all-white school comparison from 1948 to 1957 (no Perry class above the 10th percentile, no class at the other school below the 90th); a principal’s words about ability; principal Eugene Beatty’s support and district officials’ refusal; the unusual idea of preschool for 3- and 4-year-olds; professors’ warnings; and his framing it as an experiment. The year is inferred from the documentary’s audio files, dated 2009; the page shows only a 2018 copyright footer. It has no coin-flip detail.
Open the published source (opens in a new tab)What it supports: Supports a coin flip deciding who went to the preschool, with 58 of 123 children attending and 65 not, without describing the matching or later moves (see M028L); the IQ figures of 94.8 and 83.6 at the end of the second preschool year; and no significant IQ difference by about age 8. Its high school graduation figures (77 and 60 percent) differ from those in the HighScope release (65 and 45 percent), so the lesson gives no graduation figure. Same year inference as M028A.
Open the published source (opens in a new tab)What it supports: Supports the four-teacher team, the three-hour morning session five days a week, and the weekly home visits, and that the preschool ran from 1962 to 1967 under the Ypsilanti schools’ special education department. The September 2016 date comes from the Part II page; the Part I page shows only that Part II follows in November 2016. Volume and issue were not seen. Its three-hour session differs from the 2.5 hours in M028J and M028L, and the lesson gives both. The authors’ personal involvement is a strength for detail and a limit on independence.
Open the published source (opens in a new tab)What it supports: Independently confirms that the study began in 1962 with 123 children randomly assigned, with programme at ages 3 and 4 or none, followed through age 40 with low attrition (a median of 6 percent of cases missing). It does not mention a coin flip. It is written by a colleague and is favourable.
Open the published source (opens in a new tab)What it supports: Supports the age-40 figures: median annual earnings of 20,800 and 15,300 U.S. dollars, and 36 and 55 percent arrested five or more times. These are the study team’s own reported figures. Education Week (2004) repeats the arrest figures.
Open the published source (opens in a new tab)What it supports: Supports that, accounting for compromises in the randomisation, social rates of return mostly fall between 7 and 10 percent, usually well below earlier estimates, and are statistically significant for both sexes. Returns are also reported as above the historical return on equity. The abstract was read; the full paper was not.
Open the published source (opens in a new tab)What it supports: Supports that across 22 high-quality studies, early education was associated with 8.1 percentage points fewer special education placements, 8.3 fewer grade retentions and 11.4 more high school graduations. Studies span 1960 to 2016; only the abstract and summary were inspected.
Open the published source (opens in a new tab)What it supports: Supports that children randomly assigned to Tennessee’s state pre-K had lower state test scores in grades 3 to 6, strongest in grade 6, and more disciplinary infractions, lower attendance and more special education services, with no effect on grade retention. The authors say it is an open question whether other state programmes would show the same, and that no distinctive feature of the programme has been identified as an explanation. It also notes that Perry’s children had few intellectually stimulating alternatives, while its own control children mostly stayed in home care (63 percent of a 306-child subsample); it does not test whether that explains the difference.
Open the published source (opens in a new tab)What it supports: Publishers Weekly’s record, Kirkus and a bookseller’s description support that the book argues success depends more on traits such as perseverance, curiosity, optimism and self-control than on intelligence or test scores; that it draws on James Heckman’s research; and that it looks at schools in Chicago and New York. Publishers Weekly is the source for the brain chemistry and society points, in the reviewer’s words. Bestseller evidence: Family Action Network’s speaker profile, a Knowledge@Wharton excerpt on a Penguin listing and a 2016 Bridgespan interview page say more than a year on New York Times lists, and a bookseller labels it a New York Times bestseller. These are descriptions of the list record; no list page was opened, so the lesson attributes the claim. Those pages give 28 and 27 languages, so no figure is used. The book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports that changes in personality traits induced by the Perry programme explain a sizable portion of adult effects; that externalising behaviour improved substantially; that IQ gains had negligible explanatory effects, with the IQ effect essentially disappearing for males; and that academic motivation rose mainly for girls. A small, borderline positive IQ effect remained for females. It also gives sessions of 2.5 hours, and group counts of 58 and 65. The traits are childhood behaviour measures, not the perseverance or curiosity the book names, and Heckman’s work is a basis for the book, so it is not independent of it. No numeric decomposition is used.
Open the published source (opens in a new tab)What it supports: Supports that the link between waiting at age 4 and age-15 achievement was half the size in the original studies and fell by two thirds with controls for family background, early cognitive ability and home environment; and that links to age-15 behaviour were much smaller and rarely significant. The 552 is from the article (total 918). It measures a single test of waiting, is not an experiment, and does not test the book directly.
Open the published source (opens in a new tab)What it supports: Second source for the coin flip, and its limit: younger siblings took their older sibling’s status; the rest were ranked by IQ and split into two groups, adjusted for gender and family status; a single coin toss labelled the groups; then some children of working mothers were swapped, with 5 ending up moved from preschool to control. Also gives 2.5-hour sessions. The lesson says the groups were matched and later economists noted compromises. The protocol is described from Weikart, Bond and McNeil (1978), which was not read.
Open the published source (opens in a new tab)What it supports: Supports Lovelock finding CCl3F in all air samples on the R.V. Shackleton voyage from England to Antarctica (figure dated 1971) in amounts roughly comparable to all manufactured to date; Rowland’s 1973 question about predicting its fate; Molina joining later in 1973 and choosing the problem; rapid removal in the lower atmosphere (photolysis, rainout, oxidation) being ruled out; Farman’s May 1985 report, with Halley Bay October average below 200 Dobson Units by 1984 against 300 to 320 in the 1960s (reported in May 1985). Also gives about 100,000 ozone molecules removed per chlorine atom and the reference for the first paper (Nature, 249, 810–812, 1974). A participant’s own account. The lesson says “around 1971” because the date comes only from a figure caption. The 1974 Nature paper itself was not opened.
Open the published source (opens in a new tab)What it supports: Supports one chlorine atom destroying about 100,000 ozone molecules, publication in Nature in 1974, the push for a ban on consumer products, products earning billions of dollars a year, Aerosol Age’s suggestion about the KGB, some scientists feeling he overstated the hypothesis, the National Academy of Sciences verifying in 1976, the 1978 U.S. ban on CFC aerosols, and the 1995 Nobel Prize shared with Molina and Crutzen. The National Academy, 1978 ban and Nobel details come from photo captions on the page, not its body text; there is no byline. Comes from Rowland’s own university.
Open the published source (opens in a new tab)What it supports: Supports the 13-year gap before international action, the British discovery of the Antarctic ozone hole leading to the Montreal Protocol, and the decision to advocate for a ban. The year 1987 is confirmed in the Nobel press release (M029D) and Rowland’s lecture (M029A). Comes from Rowland’s own university.
Open the published source (opens in a new tab)What it supports: Supports CFCs being man-made and chemically inert, their gradual transport to the ozone layer, intense ultraviolet light splitting them and freeing chlorine that decomposes ozone catalytically, the Montreal Protocol being signed in 1987, and the 1995 prize to Crutzen, Molina and Rowland. Does not give lifetimes or depletion figures.
Open the published source (opens in a new tab)What it supports: Supports September healing of the Antarctic ozone layer since 2000, with dynamical and temperature changes also contributing and the Calbuco eruption interfering in 2015, when a record October hole occurred. Studies the atmosphere, not people. PubMed abstract and record read.
Open the published source (opens in a new tab)What it supports: Supports estimates that a no-restrictions scenario gives up to a quadrupling of skin cancer incidence by 2100 and that the Copenhagen Amendments scenario gives a peak relative increase of almost 10 percent. A projection, not observed patients. PubMed abstract and record read.
Open the published source (opens in a new tab)What it supports: The book argues that the misuse of new chemicals in farming and pest control harms the environment. Publication details (Houghton Mifflin, 27 September 1962), hundreds of thousands of copies sold, attacks from chemical and agricultural spokesmen, and “thirty-one months” on an unnamed best-seller list come from the Environment and Society Portal of the Rachel Carson Center. A Middlebury College archive record titles it a New York Times bestseller for 31 weeks in 1962 to 1963, crediting a college history, not the list. The conflict is unresolved, so the lesson gives both; 31 weeks looks the more plausible, but no New York Times list page was inspected. Best-seller status itself is shown by Time’s list (M029N). A public library catalogue confirms Boston: Houghton Mifflin, c1962 (2002 edition). The book itself, a publisher page and a national library record were not read.
Open the published source (opens in a new tab)What it supports: Supports Carson selling millions of books, her premise that synthetic pesticides enter air, water, soil and the food chain, poison animals and people and disrupt natural relationships, and critics’ charges of exaggeration, which it quotes from a 1964 Time obituary. Title, journal, volume, issue, pages and DOI come from the PubMed record; the archived page shows no title and only a web update date of 30 July 1998. PubMed has no abstract. The author’s name appears at the end of the archived page.
Open the published source (opens in a new tab)What it supports: Supports the American Chemical Society’s landmark designation, Carson challenging widespread, indiscriminate pesticide use and arguing for stricter regulation, and the book contributing to the 1972 ban on domestic use of DDT, stated in the magazine’s own voice. The page gives a 26 October ceremony without a year and no byline; the Pittsburgh section of the society records the dedication as 26 October 2012, which fixes the year. The title and issue come from the URL. The ACS’s own landmark page returned no readable content.
Open the published source (opens in a new tab)What it supports: Supports DDT concentrations rising with trophic level from 0.04 part per million in plankton to 75 parts per million in a ring-billed gull in a Long Island salt marsh. One location. PubMed abstract read.
Open the published source (opens in a new tab)What it supports: Supports a summary odds ratio of 0.97 (95 percent confidence interval 0.87 to 1.09) for p,p’-DDE and breast cancer. The authors leave open early-life exposure. Does not address other health effects. PubMed abstract read.
Open the published source (opens in a new tab)What it supports: Supports a statistically significant 5-fold increased risk with high serum p,p’-DDT among women born after 1931, and no association in women not exposed before age 14. One California cohort. PubMed abstract read.
Open the published source (opens in a new tab)What it supports: Supports that DDT and DDE may be associated with adverse health outcomes, and a lack of knowledge in communities where DDT is sprayed for malaria control. PubMed abstract read.
Open the published source (opens in a new tab)What it supports: Shows Silent Spring at number 2 on the nonfiction best sellers, number 1 the week before. The page does not say how the list was compiled. It does not support any length of run or sales figure.
Open the published source (opens in a new tab)What it supports: Supports the 28 August 1945 meeting at Montague Street, Robinson agreeing to sign with Montreal, the announcement on 23 October 1945, political pressure bringing it forward, and Rickey’s argument that Negro League players lacked valid contracts, with no intention of compensating their clubs. It also describes reputational, moral and financial concerns for Rickey. It does not mention the Ives-Quinn Law. The web date is 18 February 1989; it credits a June 1988 SPORT article.
Open the published source (opens in a new tab)What it supports: Supports the April 1945 Fenway Park tryout and Robinson’s later doubt about it, his Monarchs figures (34 games, .375), the Brooklyn Brown Dodgers cover in May 1945, and Monarchs co-owners Wilkinson and Baird being happy for Black players but wanting white baseball stopped from taking theirs. The page says the legislature passed the Ives-Quinn Law in July 1945; a 1945 report shows it was signed on 12 March and took effect on 1 July (M030M), so the lesson does not use that wording. The page reports a secret 15 to 1 owners’ vote; the lesson leaves that out because it was not confirmed elsewhere. Opening text signed by Dave Bohmer, other sections unsigned. No publication date is shown, so the year is that of the copyright notice.
Open the published source (opens in a new tab)What it supports: Supports more than 60 years of segregation, Rickey extracting a promise that Robinson would hold his tongue and fists, Robinson keeping it in his first season despite baiting, threats and pitches near his head, and his .349 average leading the International League in 1946. It says only “well over .300” for 1945, without the .375 in M030B. The page gives no date, so the year field records when it was read.
Open the published source (opens in a new tab)What it supports: Supports the 15 April 1947 debut before 26,623 at Ebbets Field, Rookie of the Year, and the National League pennant. Also reports slurs and threatening letters. Credits Thomas J. Brown Jr. in its introduction. No publication date is shown, so the year is that of the copyright notice.
Open the published source (opens in a new tab)What it supports: Independently supports Robinson first playing for Brooklyn on 15 April 1947, Rickey signing him in October 1945, and a year in the minor leagues. The year field records when the undated page was read.
Open the published source (opens in a new tab)What it supports: Supports treating integration as an innovation, data from Major League Baseball and Atlantic Coast Conference basketball, and the finding that winning teams led. Only the abstract and record were read, so no sample sizes or effect sizes are stated.
Open the published source (opens in a new tab)What it supports: Supports the test and the finding of statistically significant evidence of discrimination in this sense. Only the abstract was read, in two records, so no effect size is stated. Volume and issue come from a library discovery record; the publisher’s DOI page could not be opened.
Open the published source (opens in a new tab)What it supports: The bookseller’s description covers the Oakland Athletics written off in 2002, the “humble base-on-balls”, a cache of numbers collected by amateur enthusiasts that the game ignored, Billy Beane’s experiment with a team nobody else wanted, and a David and Goliath framing. It does not use the word “undervalued”; that is the lesson’s summary. A bookseller’s record of the Norton paperback (2004) was read, not the book, nor Norton’s own page. A public library catalogue confirms New York: W. W. Norton, 2003. Penguin Random House’s page for the audio edition shows a Best Seller flag. Bestseller evidence is in M030I.
Open the published source (opens in a new tab)What it supports: Supports the book marking its second week on the magazine’s nonfiction list, being atop the San Francisco Chronicle’s bestseller list, just under 200,000 copies in print after its May 12 release, and the Oakland record on a low payroll. The date and title come from the URL, as the page shows only “May 12 release” with no year; “with reporting by Dick Donahue” is the only credit. The list and Chronicle positions are the magazine’s report, not list pages. A New York Times list page was not inspected.
Open the published source (opens in a new tab)What it supports: Supports some skills being undervalued early in the period and the market correcting as knowledge spread. The abstract was read; the full text could not be opened (DOI page not reachable), so its figures are taken only through M030K.
Open the published source (opens in a new tab)What it supports: Reports Hakes and Sauer’s win coefficients (on-base 2.032, slugging 0.900), closely repeats the win results (the salary models only in part), and with data to 2017 calls these valuations very volatile from year to year, with no distinct trend. Used because the original’s full text could not be opened. A student work, and the lesson says so.
Open the published source (opens in a new tab)What it supports: Supports top draft picks being overvalued despite high stakes and many chances to learn. Abstract read, so no figures are stated.
Open the published source (opens in a new tab)What it supports: Supports the Ives-Quinn Bill being signed on 12 March 1945 and taking effect on 1 July, as a measure against discrimination in employment. It says nothing about baseball.
Open the published source (opens in a new tab)What it supports: Supports that The Seagull premiered in St Petersburg on 17 October 1896 (Old Style), was almost hissed off the stage, that Chekhov left the auditorium during the second act and vowed never to write for the stage again, and that the Moscow Art Theatre revived it two years later with considerable success. The name of the Alexandrinsky Theatre comes from M031E and M031D, not this page.
Open the published source (opens in a new tab)What it supports: Supports that the theatre was founded in 1898 by Stanislavsky and Nemirovich-Danchenko, and that its revival of The Seagull, which had failed in its original staging, became the company’s first major success.
Open the published source (opens in a new tab)What it supports: Supports the June 1897 meeting of Stanislavsky and Nemirovich-Danchenko and their 18-hour talk; the plan for actors with no stars; that Nemirovich-Danchenko persuaded Chekhov to release the play although Stanislavsky strongly disagreed; and that the theatre was in financial crisis with its existence depending on the play. It gives 21 June 1897 while the theatre’s own history gives 19 June, so the lesson says only June 1897. It does not date the crisis, placing it only before the 29 December 1898 opening, and it says the other first-season failures came from a season that opened on 14 October, so the lesson places the crisis by December and after the September events. It calls the Tsar Fyodor opening a success, which sits awkwardly with Britannica’s “first major success”, and describes the audience as celebrating before the curtain closed, which differs from the long silence in M031D.
Open the published source (opens in a new tab)What it supports: Supports that Chekhov refused a Moscow staging and his friend Nemirovich-Danchenko overcame the refusal; 24 rehearsal sessions and 80 hours in total; that Stanislavski threatened to remove his name from the posters when the opening was not postponed by a week; that Chekhov judged Trigorin weak in September 1898, Stanislavski took the role and Nemirovich-Danchenko apologised; valerian drops; applause after a long silence; press praise; opening on 29 December (Old Style 17 December) 1898. The related The Seagull article was read for Chekhov’s later criticism of Stanislavski’s Trigorin. Tertiary and the weakest source in the lesson; the poster threat, apology, valerian drops, silence and 80 hours have no second source that could be read, and the lesson says so.
Open the published source (opens in a new tab)What it supports: Supports that the Aleksandrinsky production was billed for a benefit, which generally anticipated lightness of tone, and was performed as an outright comedy; and that Chekhov disliked Stanislavsky’s Trigorin. It says that in December 1898 Nemirovich-Danchenko ran 15 rehearsals and Stanislavsky 9, the same 24 that M031D gives as the total but dated to December alone, which is why the lesson says “one count”. It does not say Chekhov attended rehearsals or discuss finances. Page numbers are not in the record. ISBN 9781474284431.
Open the published source (opens in a new tab)What it supports: Supports that the 1896 production had only a few rehearsals with not all actors at each, that it was a benefit night, and argues from Chekhov’s letters that the premiere was “not a failure” and less of a tragedy than later portrayed. It quotes no published reviews. Not peer reviewed. The rehearsal detail has this single source.
Open the published source (opens in a new tab)What it supports: Defines team psychological safety; 427 members of 51 teams at one office furniture manufacturer completed surveys (86 percent response); learning behaviour mediated the link between psychological safety and observer-rated team performance. One company, cross-sectional, so causality is not established, and not theatre.
Open the published source (opens in a new tab)What it supports: Over 22,000 individuals and nearly 5,000 groups. Task performance: group level corrected correlation .29 (k = 33); individual level .43 (k = 18), with .40 on subjective and .07 on objective criteria. At group level the subjective versus objective moderator was not significant (Table 6: estimate −.02, p = .83), and no separate group figures are given. The authors say the literature lacks longitudinal designs and objective criteria and that it is less clear whether psychological safety is necessary and/or sufficient for behavioural and attitudinal work outcomes. The “link, not a proven cause” wording is the lesson’s inference, not the authors’ sentence.
Open the published source (opens in a new tab)What it supports: Publisher’s page (on sale 30 January 2018, Bantam, 304 pages, ISBN 9780804176989) and a bookseller copy of its description support the groups named and the claim that culture is built through action. A discussion guide by The Main Idea, hosted on the publisher’s image server and not written by the publisher, names and describes Build Safety, Share Vulnerability and Establish Purpose. Bestseller evidence: the publisher’s page labels it NEW YORK TIMES BESTSELLER with a Best Seller badge, and its author page calls Coyle the New York Times bestselling author of The Culture Code. No New York Times list page, date or rank was inspected, so the lesson attributes the label to the publisher. The book itself was not read.
Open the published source (opens in a new tab)What it supports: Names the three skills and presents them as a foundation for leaders who guide teams. A short review, not a test of the book’s claims. The journal has since been renamed Journal of Higher Education & Student Affairs on its repository page; DOAJ records the 2020 title used here.
Open the published source (opens in a new tab)What it supports: Task conflict and team performance were positively associated under high psychological safety. Abstract read through the PubMed record (PMID 21728397); correlational. The full paper is paywalled; that the teams were five students each, working over a semester, comes from the British Psychological Society’s Research Digest summary. The link to the founders’ fights is the lesson’s reading, not the paper’s.
Open the published source (opens in a new tab)What it supports: Stable negative relationships between relationship and process conflict and group outcomes; no strong negative association between task conflict and group performance, with a complex moderated picture. Abstract only, read through the PubMed record (PMID 21842974). It does not test psychological safety itself.
Open the published source (opens in a new tab)What it supports: Added by the checker as a second, independent source. Supports that Stanislavsky took over Trigorin at Chekhov’s suggestion, the 29 December 1898 opening, and acclaim from critics and public. It does not mention September, the apology, the poster threat, valerian drops or the silence.
Open the published source (opens in a new tab)What it supports: Supports the patient’s age of 50, about 21 years at Bicêtre, admission to the infirmary on 11 April 1861 with a gangrenous right leg, the single syllable tan, answers given with his left hand, anger when misunderstood, Broca’s judgement that he understood nearly everything, death on 17 April 1861, showing the brain to the Anthropological Society, storage in alcohol, deposit in the Dupuytren Museum, and leaving the inner parts unstudied so as not to destroy the specimen. It is one doctor’s account, and Broca also noted that some answers were inaccurate and judged the man’s intelligence reduced by the lesion and fever. Broca says the man answered tan, tan to almost everything and added one swearword when angry, so the lesson says almost all he could say. This translation gives his trade as hat-form maker. Broca places the original lesion in the left frontal lobe, probably the third convolution, but does not use the phrase speech centre, does not claim the left side is special, and treats the finer localisation as unsettled.
Open the published source (opens in a new tab)What it supports: The abstract supports that the patient’s identity was traced through French archive registers, that he had an almost complete loss of speech, and that the area was later named Broca’s area; it describes Broca as reporting a cortical speech centre in the middle part of the left frontal lobe, a retrospective framing Broca’s own text does not use. The full name and the trade of making wooden forms for shoemakers were read only in the British Psychological Society’s Research Digest summary of the article, not in the article, and the trade conflicts with Broca’s translated hat-form maker, so the lesson reports both. The same summary is the only source inspected for the name. Domanski’s 2014 follow-up (PMID 24433196, abstract only) adds details of the hospital history, so the lesson attributes the 21 years to Broca.
Open the published source (opens in a new tab)What it supports: Supports that MRI of the preserved brains of Broca’s 2 patients found lesions extending significantly into medial regions beyond the surface lesions Broca observed. Only the abstract could be read, so the lesson names no specific deep structures and gives no scan date, only the 2007 publication. A history article in Hektoen International separately reports that the brains were kept intact in the Musée Dupuytren and were rescanned, with an earlier 1984 CT of Leborgne’s brain. A second Hektoen article says Pierre Marie dissected the brain about 40 years after Broca, which conflicts with kept intact; this was not resolved, and the lesson says only what Broca chose and that the brain could still be scanned.
Open the published source (opens in a new tab)What it supports: The abstract supports that the London physician Thomas Barlow published in 1877 the case of a 10-year-old boy who lost his speech, regained it and lost it again, with softening in Broca’s area and, weeks later, in the matching right-side region, and that the article critically evaluates whether the case supports right-side takeover. Only the abstract was read, so the lesson does not state the authors’ final verdict. The abstract does not say when the boy lost speech relative to each lesion, so the lesson does not state that sequence as cause.
Open the published source (opens in a new tab)What it supports: Supports 57 randomised trials (74 comparisons) with 3,002 participants. Twenty-seven comparisons with 1,620 participants tested therapy against no therapy, with benefits for functional communication (standardised mean difference 0.28, called clinically and statistically significant), reading, writing and expressive language, but benefits not evident at follow-up on smaller numbers. Nine comparisons with 447 participants found no evidence of a difference from social support and stimulation in functional communication. High intensity, high dose or longer therapy did better, but high intensity and high dose had more dropouts. Trials were generally small. A PubMed title search on 2026-11-06 listed no later version; Cochrane’s own site was not checked.
Open the published source (opens in a new tab)What it supports: Supports that among patients aged 70 or younger with aphasia for at least 6 months, 3 weeks or more of therapy at 10 or more hours a week improved verbal communication compared with deferral (Cohen’s d 0.58), on a primary endpoint measured straight after treatment. An NIHR summary shows the trial also had a six-month follow-up period, whose outcomes were not read, so the lesson says this main result does not show how long gains last, not that the trial has no later data. The author list is shortened here.
Open the published source (opens in a new tab)What it supports: Penguin’s UK page for the paperback of 7 August 2008 supports the subjects of the book, including a stroke survivor who relearned movement and speech and a woman with half a brain that rewired itself, the sentence that brains can repair themselves through positive thinking, and a profile of scientists challenging older ideas; it speaks of the brain’s adaptability, not of adults specifically. Bestseller evidence: the same page states over one million copies sold, a publisher’s undated figure, and the lesson attributes it. No New York Times list page was inspected. The author’s own site says New York Times Bestseller 2008, unverified; the New York Times label on the Penguin and Harvard pages describes Doidge, and the Harvard page ties it to his later book. A Harvard Book Store record gives a Penguin paperback dated 18 December 2007, and Wikipedia gives Viking and 2007 as first publication, so 2007 rests on those, not a publisher record of the first edition. The book’s full text was not read.
Open the published source (opens in a new tab)What it supports: Supports that 2 weeks of constraint-induced movement therapy, 3 to 9 months after a first stroke, gave greater arm function gains than usual care at 12 months. It tested arm function, not speech, and the lesson does not claim it measured brain change. The author list is shortened here.
Open the published source (opens in a new tab)What it supports: Supports extensive evidence of improvement on trained tasks, less on closely related tasks, little on distantly related tasks or everyday cognitive performance, and major design shortcomings in many studies. It reviews cognitive tasks and games, not stroke rehabilitation, and it is not a review of this book.
Open the published source (opens in a new tab)What it supports: Supports the delayed vaccine shipment months away, the outbreak in a remote area, the daily 7 p.m. ham radio check-in, runners, and five-day markets used to trace routes. The page opens this episode in December 1966 and then says one day, so it does not date the outbreak exactly. The feature text ends with Evelyn Strauss’s name; the page metadata shows only a last-modified time of 27 February 2021, so the year is that date, not a confirmed first publication. It also says about 7 to 8 percent of the area was vaccinated, a figure the lesson does not use because a University of Washington magazine profile says less than 50 percent, and it says the outbreak was contained in two weeks.
Open the published source (opens in a new tab)What it supports: Supports the early 1967 epidemic in Ogoja Province, the limited doses used on close contacts, rapid elimination, the Abakaliki mass campaign (88 percent effectively vaccinated, then a 33-case outbreak), and the Eastern State’s move in late 1967 to surveillance and containment, which interrupted transmission in about 5 months with 750,000 of 12 million vaccinated. Lane co-wrote the 1971 paper with Foege (PMID 5110547), so he is not independent. Lane also says a 1964 WHO committee called for 100 percent coverage rather than the earlier 80 percent level, so the lesson calls 80 percent the thinking then, not a clear rule. The chapter counts 33 Abakaliki cases and does not say who they were; the lesson uses the primary report for that (M033L).
Open the published source (opens in a new tab)What it supports: Supports Foege going to Yahe in 1966 to run a Lutheran mission medical centre, the CDC asking him to consult on a smallpox programme, and the thinking that at least 80 percent, perhaps more, had to be vaccinated. It is a tribute and written in praise. The year is the award year; the text says this year’s recipient, and the document’s own date was not seen.
Open the published source (opens in a new tab)What it supports: Supports ham radio, runners, 4 affected villages identified within a day, 3 more villages within a 15-mile radius with the virus already present in 2, and the last recorded natural cases in Somalia in 1977. The page shows no publication date and says Foege died on 24 January with no year; other pages confirm 2026 (WHO notice of 25 January 2026, Emory Wheel, January 2026), so the year is well supported though not shown on this page. Small (M033E) says one or two nearby villages, not 3, so the lesson marks this as one account.
Open the published source (opens in a new tab)What it supports: Supports the shortwave radio reports from missionaries, vaccinating affected villages and one or two nearby villages, the mixed reception of the method because mass vaccination was the accepted approach, and that eastern Nigeria was still free of smallpox when Foege returned after more than a year away during the civil war. It is a review of Foege’s and Henderson’s books, not a primary record.
Open the published source (opens in a new tab)What it supports: Supports listing rings of contacts and contacts of contacts after a confirmed case, random assignment of 51 and 47 clusters (4,539 and 4,557 people) to immediate or delayed vaccination, and 0 cases from day 10 among those vaccinated in immediate clusters against 16 among eligible people in delayed clusters. It tested a vaccine delivered by rings, not rings against mass vaccination. The author list is shortened here.
Open the published source (opens in a new tab)What it supports: Supports that adding ring vaccination at the outset of the West African epidemic might not have led to containment, while in later stages or outbreaks with less intense transmission or more effective control it could help eliminate the disease. It is a model, not an observed result.
Open the published source (opens in a new tab)What it supports: Bookseller pages carrying the publisher’s description of the Sceptre editions (hardback 3 April 2018, paperback 27 June 2019) support the authors, the 2018 first publication, Hans Rosling’s death in 2017, the ten instincts, and the argument about predictable errors. Bestseller evidence: those descriptions state over a million copies sold, number 1 Sunday Times bestseller and New York Times bestseller, and the Flatiron edition is billed as an instant New York Times bestseller. No list page itself was inspected, and the publisher’s own page could not be opened, so the lesson attributes the label to the publisher’s description. The book’s full text was not read. The opinions-and-facts idea, the us-and-them and getting-worse examples and the warning about worry and fearful media were seen on the Flatiron description; the Sceptre description gives the sales and list claims.
Open the published source (opens in a new tab)What it supports: Supports that for a headline of average length each additional negative word raised click-through by 2.3 percent, in 22,743 randomised tests and about 105,000 story variations on one site, Upworthy. It studied clicks on one website and did not test the book.
Open the published source (opens in a new tab)What it supports: Supports that many countries improved child survival since 2000 and an estimate that 58 percent of child deaths from 2000 to 2017 in these countries could have been averted in the absence of geographical inequality. These are modelled estimates. It did not test the book. The author list is shortened here.
Open the published source (opens in a new tab)What it supports: Supports that the authors doubt that about 90 percent of people lived in extreme poverty before the 19th century, using real wages, height and mortality across five regions. Its abstract and repository record name neither this book nor Rosling. Whether it bears on the book’s own story of progress is this lesson’s inference, which the lesson signals, and the book’s own long-run claims were not read.
Open the published source (opens in a new tab)What it supports: Supports the February 1967 Abakaliki pilot, at least 88.5 percent of the population successfully vaccinated by assessment studies, and 32 smallpox cases, 30 of them among members of the Faith Tabernacle, a church that refused vaccination. Foege co-wrote it, so it is not independent. It reports 46,517 people vaccinated against a 1963 census population of 31,177 for Abakaliki urban, a gap it does not explain, so the coverage figure should be treated with caution. Lane (M033B) gives 33 cases.
Open the published source (opens in a new tab)What it supports: Supports that in 2006 Doudna, in her office on the seventh floor of Stanley Hall at UC Berkeley, took a call from Jillian Banfield, whose lab’s data had turned up CRISPR and who wanted tools from genetics and biochemistry; that she had never heard of CRISPR and skimmed a few papers; that they met the following week at the Free Speech Movement Café; that Banfield sketched repeats and unique sequences, and that essentially every cell had a different array; and that Banfield did not know its function. It does not mention viruses (that detail rests on M034B). It is Doudna’s own recollection; the page’s byline names Doudna alone, and it credits the book to Doudna and Sternberg. The excerpt’s weather detail is not used. Nothing is quoted.
Open the published source (opens in a new tab)What it supports: Supports that Banfield, a geomicrobiologist, suspected CRISPR could be an antiviral defence; that Doudna recalled there was no experimental evidence for that at the time; that she followed her curiosity and began studying CRISPR; that in 2012 the labs found changing the guide controls where Cas9 cuts; and that Doudna and Charpentier shared the Nobel Prize in Chemistry on 7 October 2020. The page shows no author or publication date, so the year is inferred from the prize it reports (ca. 2020). Doudna’s own university, not an independent historian; the virus suspicion and the lack of evidence rest on this page alone.
Open the published source (opens in a new tab)What it supports: Supports that tracrRNA was uncovered in the human pathogen Streptococcus pyogenes and that it, RNase III and Csn1 (Cas9) are essential to protect the bacterium against prophage-derived DNA. The first affiliation listed is in Umeå, Sweden, but M034D also places Charpentier’s CRISPR work in Vienna, so the lesson says only Europe.
Open the published source (opens in a new tab)What it supports: Supports that she and Doudna met at a conference in Puerto Rico in 2011, a few months after her CRISPR findings appeared, and vowed to collaborate. It gives Charpentier’s move to Umeå as 2009 but also places her lab in Vienna at the meeting, an inconsistency the lesson avoids. The year comes from the 2016 copyright footer, and no publication date is shown. The meeting is confirmed by Charpentier’s own answer in the GEN interview and by Wikipedia, citing Abbott (2016, Nature 532, 432–434; PMID 27121823 matched, article not opened). “Vowed to collaborate” rests on this page alone.
Open the published source (opens in a new tab)What it supports: Supports that a two-RNA structure directs Cas9 to make double-stranded breaks in target DNA at sites complementary to the guide, that a single engineered RNA also directs this cleavage, and that the paper highlights the potential for RNA-programmable genome editing. It did not edit living human cells.
Open the published source (opens in a new tab)What it supports: Supports that 80 percent of cited research articles link forward to a future patent, 61 percent of patents link back to a research article, and linked papers and patents typically stand 2 to 4 degrees apart. It covers all fields, not biochemistry alone, and it measures citation links, not curiosity or benefits.
Open the published source (opens in a new tab)What it supports: Supports that about 10 percent of grants generate a patent directly and 30 percent generate articles later cited by patents, and that there was no systematic relationship between a grant’s basic or applied focus and its propensity to be cited by a patent. The number of grants is not stated in the abstract, so the lesson gives none.
Open the published source (opens in a new tab)What it supports: The publisher’s description on a bookseller page (paperback, 3 May 2022) supports that the book is a biography of Doudna, whose father left The Double Helix on her bed; that CRISPR and the race for coronavirus vaccines will hasten the transition from a digital age to the next great innovation revolution, in the life sciences; that it raises ethical questions; and that Bloomberg BusinessWeek, Time and The Washington Post named it a best book of 2021. Its argument is described from this copy, not from the book. Bestseller evidence: the unsigned Five Books page states it went to the top of the New York Times nonfiction bestseller list, and calls it highly readable. Wikipedia reports a number one debut, citing the Times combined print and e-book list dated 28 March 2021, and a mirror of Wikipedia’s number-one list shows the title on 28 March and from 11 April 2021. The New York Times list page itself could not be opened (permission request unanswered), so the lesson attributes the claim to Five Books.
Open the published source (opens in a new tab)What it supports: The abstract supports that the path to CRISPR genome editing was a 20-year journey and that the article sets out the stories of pioneers and lessons about the ecosystem underlying discovery. Lander is affiliated with the Broad Institute, a party to the patent dispute (see M034L). Only the abstract and record were read; the full text could not be opened. Not a study of the book.
Open the published source (opens in a new tab)What it supports: Supports that teams increasingly dominate solo authors in producing knowledge, across nearly all fields, and that they typically produce more frequently cited research. It does not test the book, and citation counts are not the same as importance.
Open the published source (opens in a new tab)What it supports: Supports that 44 patients received the CRISPR-Cas9-edited cell therapy, with a median follow-up of 19.3 months; that of the 30 with sufficient follow-up, 29 (97 percent) were free from severe vaso-occlusive crises for at least 12 consecutive months. It has no comparison group and covers one disease. Several authors are at the companies developing the therapy. PubMed lists 28 named authors and one further, unnamed entry.
Open the published source (opens in a new tab)What it supports: Supports that a heated patent dispute set Doudna and Charpentier, through the University of California, against Feng Zhang of the Broad Institute of MIT and Harvard, that Zhang won the first round, and that the rest was still being appealed when written. It is dated to 2017 and does not report later outcomes. The lesson states no current status.
Open the published source (opens in a new tab)What it supports: Supports, in Chang’s own words, that he felt sidelined at Texas Instruments and resigned without a job offer; left General Instrument after little more than a year; accepted the ITRI presidency in Taiwan in 1985 at much lower pay; that K. T. Li, a minister without portfolio, asked him a few weeks after arrival to build a semiconductor company and present a plan (a week or so, he recalled presenting in four days); that the RCA technology (he says 1975) had fallen behind; that Taiwan had little or no strength in research, design, sales or intellectual property, and that wafer manufacturing was the only potential strength; his reading of Carver Mead; the least evil and possibly fatal where’s-the-market remarks; that Philips was the only willing significant investor; that before 1991 TSMC relied mainly on leftovers from large companies; that it lost money in 1987 and 1990 only and not from 1991, as of 2007; that fabless firms took off around 1991 to 1992. It is his own memory, read through a fetch tool’s extracts, and the loss-year passage is garbled in the transcript. Philips: he names no other investors; Wikipedia and Caixin (not cited) give Taiwan’s government 48.3 percent and Philips 27.6 percent, and say Intel and Texas Instruments declined. Later profits checked only in part: M035C reports profit falling, not a loss, in 2001.
Open the published source (opens in a new tab)What it supports: Supports that Chang worked at Texas Instruments from 1958 to 1983, rising to group vice president of its worldwide semiconductor business, and that TSMC, founded in 1987, is a dedicated silicon foundry. The page has no date of its own; 2007 is the year he was named a Fellow. It puts his recruitment by Taiwan in 1984. Chang’s transcript, the SEMI biography and Caixin say 1985, and M035C implies about 1987, so the lesson gives no year.
Open the published source (opens in a new tab)What it supports: Supports that TSMC was founded in 1987 and was the first foundry, with backing from Taiwan’s government and further support from Philips. Its funding figure differs from another account and is not used, and its timing of his move to Taiwan is not used. The 2006 date is the Gale copyright line. It rests on trade publications from 1998 to 2001.
Open the published source (opens in a new tab)What it supports: Slide 5 charts wafer capacity by region for 2019, showing 92 percent of logic capacity below 10 nanometres in Taiwan, and its headline rounds this to about 90 percent. The slides are dated 6 April 2021 and name no company, so the figure says nothing direct about TSMC. The report treats this concentration as a supply-chain risk. An industry body hosts it, so it may reflect that body’s interests.
Open the published source (opens in a new tab)What it supports: Supports that learning rates averaged about 20 percent, that firms learned three times more from their own cumulative production than from others’, that learning spilled over as much between firms in different countries as within a country, and that spillovers across chip generations were weak. It covers memory chips, not foundries or logic.
Open the published source (opens in a new tab)What it supports: Supports that integrated firms perform better on complex, poorly defined technical problems, and specialised firms on simple, clearly defined ones. The sample size was not visible in the abstract, so the lesson gives none.
Open the published source (opens in a new tab)What it supports: Bestseller evidence: Harvard’s Davis Center page calls it a New York Times bestseller and a winner of a Financial Times business book prize; the praise section of the publisher’s copy and Wikipedia carry the same label. No list page was inspected, so the lesson attributes the label. Publishing Perspectives confirms the 2022 award and describes long, fragile supply chains and dependence on a few very large manufacturers. The publisher’s description on a bookseller page supports that military, economic and geopolitical power depend on chips, and Miller’s post as Professor of International History. Wikipedia supports the October 2022 Scribner publication, the Cold War origins, the strategies of Europe, Japan, China and Taiwan, and the focus on U.S.-China tensions. A UK publisher page failed to load and the book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports the argument that asymmetric global economic networks concentrate power in key nodes, which states can use to gather information or deny rivals access. The summary seen gives counterterrorism and nonproliferation as examples and does not show a study of chips. The full text was not read: the DOI page could not be opened (permission request unanswered). Wikipedia links the book’s subject to this idea; no inspected source says the book draws on this article.
Open the published source (opens in a new tab)What it supports: Supports that research effort behind Moore’s Law rose by a factor of 18 from 1971 while the growth in chip density stayed roughly constant (working paper; the published abstract gives more than 18 times the researchers). It does not study the book or firm concentration; the link to few firms at the frontier is this lesson’s own reading, and the lesson says so.
Open the published source (opens in a new tab)What it supports: Supports graffiti on every car by 1984, about 6,000 cars and over 3,700,000 riders per workday (stated in the present tense and undated in a 1990 article), derailments on average every 18 days and a third of the fleet out during the morning rush (both describing the late 1970s and early 1980s), the earlier failed measures (released offenders, cut fences, expensive paints, plans for attack dogs abandoned after media pressure), Gunn’s announcement in April 1984, the start in May 1984 with 2 trains, described as a manageable number, the rule of cleaning within 2 hours or pulling from service, cars removed during rush hour at least 10 times, rising yearly goals, and the last car on 12 May 1989. It is a programme account co-written by Kelling, without an outside evaluation, and does not test any effect on crime.
Open the published source (opens in a new tab)What it supports: Supports that added patrols had not cut graffiti, Gunn’s reasoning that taggers wanted to see their work on later trips, the last car on 12 May 1989, Bratton’s hiring in 1990, plainclothes sweep teams against fare evasion, Bratton’s own statement that 1 in 6 evaders stopped carried a weapon or had a warrant, and a 30 percent fall in subway felonies in his first two years. McMillian (2025; M036J) gives different ratios, about 1 in 7 with a warrant and 1 in 20 with a weapon, which count warrants and weapons separately: if the groups overlap, they can fit a combined 1 in 6, so they do not strictly conflict, but all are police-side figures, not audited counts. Not an evaluation; it does not show the sweeps caused the fall.
Open the published source (opens in a new tab)What it supports: Supports the counts of felonies against transit riders, 17,497 in 1990 and 12,199 in 1992, and Bratton’s focus on illegal behaviour underground such as fare evasion. It does not show what caused the fall.
Open the published source (opens in a new tab)What it supports: The PubMed record (PMID 19023045, read through PubMed’s data service, because the web page returned no content to the fetching tool and the DOI page could not be opened) supports the citation and the finding that people who see a norm or rule violated are more likely to violate others. The percentages, 33 versus 69 percent littering with a graffiti wall and 13 versus 27 percent taking an envelope from a graffiti-covered mailbox, were read in National Geographic’s report of the paper and agree with Reason’s and the Wilson Quarterly’s reports, not in the paper itself, and none of these gives sample sizes. It measures petty acts, not violent crime or policing.
Open the published source (opens in a new tab)What it supports: Supports, from the abstract, a modest but statistically significant overall fall in crime, the largest effects from community and problem-solving work that changes disorderly conditions at places, and no significant reduction from aggressive order maintenance aimed at individual behaviour. Checked through the Office of Justice Programs abstract record; the DOI page could not be opened and the full text was not read.
Open the published source (opens in a new tab)What it supports: Supports, from the publisher’s pages, the argument that behaviour and ideas can cross a threshold and spread like a flu epidemic, and that a small, targeted push can tip a trend or a fall in crime; also the on-sale date of 1 March 2000. Bestseller status: the publisher calls it a “widely acclaimed bestseller”, its author biography names it among Gladwell’s eight New York Times bestsellers, and its description says the book is changing how people think about selling products and spreading ideas. No New York Times page or sales figure was inspected. A trade site reprints a paperback business list, as ranked by the New York Times on 18 February 2008, with the book at number 1; this is a second-hand reprint, so the lesson attributes the label to the publisher. The book itself was not read and is not quoted.
Open the published source (opens in a new tab)What it supports: Supports the review’s account of the book: a subway of fires, derailments and heavy fare evasion, a sharp fall in crime from 1990, and Gladwell crediting the broken windows theory. It is used to describe the book’s New York chapter, and it does not test the claim. The title’s spelling is as on the page, and the byline is Miriam Lambert (her author note gives Miriam Udel Lambert). The review is itself critical of parts of the book.
Open the published source (opens in a new tab)What it supports: Supports the attribution of the theory to Wilson and Kelling in 1982; that precincts with the largest crack-era rises in crime had the largest later falls, which fits mean reversion better; that misdemeanour arrest changes did not predict falls in violent crime; a verdict of not proven on New York; and that Moving to Opportunity moved volunteer families to less disorderly areas in 5 cities without, on balance, reducing their criminal behaviour. The publisher’s record and the journal’s citation give only the start page; the end page comes from one library discovery record. The PDF text was read to about page 309, where the fetched text stops in a table note, so the final pages were not seen. It is a law review article, not a conventionally peer-reviewed paper.
Open the published source (opens in a new tab)What it supports: The TED page, opened only at the address listed here because the plain talk address could not be opened, supports the title, date and event; its summary says he was wrong and that his analysis contributed to the rise of stop and frisk. A third-party transcript (The Singju Post) has him say he was wrong and bears some of the blame for the misunderstanding that stop and frisk worked; it does not have him say his book caused the policy. It also gives his figures, about 700,000 stops in 2011 and under 50,000 after a 2013 ruling, and his statement that crime fell further. The stop counts match NYCLU’s table of police data (685,724 in 2011, 45,787 in 2014); his statement about crime was not checked.
Open the published source (opens in a new tab)What it supports: Supports the counter-argument that critics confuse broken windows with stop and frisk and zero tolerance, and that New York’s decline is evidence for it, with the concession that cities cannot be studied like laboratories. Also gives other ratios for the fare stops, 1 in 7 with a warrant and 1 in 20 with a weapon. It is an argument, not a study.
Open the published source (opens in a new tab)What it supports: Supports that Garfield joined the Johns Hopkins Welch Medical Library Indexing Project in March 1951; that the project was formed in 1948, reflecting the conclusion that bibliography was out of control; the citation of his 1955 Science paper (Citation indexes for science: A new dimension in documentation through association of ideas, Science 122(3159), 108–111, July 1955), given in the talk's references; that the decision to launch the Science Citation Index in 1964 was, in his words, a calculated risk that almost failed; that reception in the 1960s was conservative and cautious and some libraries were forbidden to buy it lest faculty be judged by citation counts; and that the printed edition began to show a small profit only around 1972, after the indexes went online with Dialog. It does not name the funders of the Genetics Citation Index. It is Garfield's own recollection.
Open the published source (opens in a new tab)What it supports: Supports that Garfield proposed a citation index in Science in 1955; that Joshua Lederberg contacted him in 1958 and their correspondence led to a meeting with the NIH genetics study section and funding for a Genetics Citation Index, which included a multidisciplinary index to the science literature of 1961; that the NIH and NSF declined the opportunity to publish that multidisciplinary index; that Garfield began regular publication of the Science Citation Index in 1964; that his firm took the ISI name in 1960; and that the index covered virtually all fields of science and uniquely indexed the references cited in the articles it covered. The page is undated and has no named author, so the year recorded is the year of access. The archive's Genetics Citation Index page confirms that ISI published that index in 1963, with a preface by J. Lederberg. A page by Clarivate, the index's current owner, says government sponsors chose not to fund a national citation database and Garfield published his multidisciplinary index privately, but dates the first availability to 1963; the lesson notes this one-year difference.
Open the published source (opens in a new tab)What it supports: Supports that Garfield founded the Institute for Scientific Information in 1960; that the Science Citation Index was developed there and officially launched in 1964; that Thomson Reuters acquired ISI and its citation index in 1992 for 210 million U.S. dollars and that it is now available through the Web of Science, maintained by Clarivate Analytics; that Garfield devised the journal impact factor and was wary of its misuse, opposing its use as a shorthand for ranking publications, researchers or institutions; and that he died on 26 February 2017 aged 91. The buyer in 1992 was the Thomson company (the article uses the later name Thomson Reuters), so the lesson does not name the buyer.
Open the published source (opens in a new tab)What it supports: Supports that the results of four studies suggest that, faced with difficult questions, people are primed to think about computers, and that when people expect future access to information they have lower recall of the information itself and enhanced recall of where to access it; and that the authors describe the internet as a primary form of external or transactive memory. Read from the PubMed abstract; sample sizes were not confirmed and are not stated in the lesson. Only the first experiment (computers primed after hard questions) was retested by M037E and M037H, and it did not replicate; the memory experiments were not retested in those sources.
Open the published source (opens in a new tab)What it supports: Supports that the project replicated 21 experimental social science studies published in Nature and Science between 2010 and 2015, with replication samples on average about five times larger than the originals; that 13 (62 percent) found a significant effect in the same direction; and that replication effect sizes averaged about 50 percent of the originals. It used two stages: if the first data collection failed, a second was run and pooled with it. Sparrow et al. (2011) is reference 35 and is not among the studies that replicated in either stage; the FLoRA database records failure both in the first round (p = 0.2647) and pooled (p = 0.4493). The article says the original authors did not provide materials or feedback, and after the replication noted some design differences, which the team cannot rule out influenced the result. The article page does not say which Sparrow experiment was retested; M037H says it was the first experiment (the 'Google Stroop effect'), not the memory experiments. The full author list (24 authors) was read from a pre-publication copy of the article hosted by a third party (gwern.net); the publisher page shows only the first three. The issue number was not confirmed and is not recorded.
Open the published source (opens in a new tab)What it supports: The publisher's page (Fourth Estate, hardback ISBN 978-0-00-722573-6) supports that the book traces how people create, transmit and preserve knowledge, from African talking drums to Wikipedia and from Morse code to the bit; that it argues information is the defining idea of the modern age; that it treats writing, dictionaries, navigational charts, electronic signals and the genetic code as information revolutions; and that it won the Royal Society Winton Prize for Science Books 2012. The publisher's page makes no bestseller claim for this title. Bestseller evidence: NPR's Hardcover Nonfiction Bestsellers lists, compiled with the American Booksellers Association from weekly surveys of close to 500 independent bookstores, show the book at number 6 for 24 March 2011 (3 weeks on the list) and number 7 for 7 April 2011 (5 weeks on the list). No New York Times list page was inspected. The publication year 2011 is consistent with these lists; the publisher page itself shows no date.
Open the published source (opens in a new tab)What it supports: Supports that across 63 conditions from 50 published and unpublished experiments (N = 5,036), the mean effect size of choice overload was virtually zero with considerable variance between studies, and that the analysis found several potentially important preconditions but no sufficient conditions for choice overload. It studies choosing among many options, not information overload or search directly, and it does not test the book.
Open the published source (opens in a new tab)What it supports: Supports that Camerer et al. (2018) retested the 'Google Stroop effect' from Sparrow et al. (2011), slower responses to internet-related words in a Stroop task after hard trivia questions, and did not replicate it; that the first author of the original study later pointed out problems with the replication's design; and that this study, designed closer to the original, found no conclusive evidence that the concept of the internet is automatically activated by hard trivia questions (117 tested, data from 89 analysed, Bayesian analysis favouring the null). It does not retest Sparrow's memory experiments and does not report any retest of them. Added by the independent checker.
Open the published source (opens in a new tab)What it supports: Supports that in 1938 University of Chicago researchers Nathaniel Kleitman (aged 43) and his graduate student Bruce Richardson spent about a month in Mammoth Cave; that Kleitman wanted to test whether a person's circadian rhythm could be altered without environmental cues such as daylight; that they tried to adjust to a 28-hour day; that they lived in Rafinesque Hall near the Historic Entrance, where stable conditions and no sunlight removed time cues; that the cave averaged about 54°F (12°C); that Mammoth Cave Hotel staff delivered food, drinking water and mail; that body temperature fluctuated through the day; and that Richardson adapted to the 28-hour cycle while Kleitman stayed anchored to 24 hours. It gives Richardson's age as 25, which differs from M038B, so the lesson gives no age for him. The page shows no publication date, only "Last updated: March 13, 2025", and it also describes a 2024 return visit; it is cited as n.d. and the year recorded is the year of access, as for other undated pages on the site. It gives no depth; the 140-foot depth comes from M038B. It describes a month-long stay; the exact dates come from M038B.
Open the published source (opens in a new tab)What it supports: Supports that Kleitman joined the University of Chicago physiology department in 1925 and launched what the article calls the world's first sleep laboratory, and often used himself as a subject; that he and his student Bruce Richardson lived in Mammoth Cave, Kentucky, from 4 June to 6 July 1938 (per a photo caption), about 140 feet underground with no natural light at a constant 54°F; that each day they slept 9 hours, worked 10 and rested 9, a 28-hour cycle; that they measured the daily rhythm of body temperature; that Richardson felt he adjusted after about a week; and that Kleitman's clock did not shift, as he still became tired around 10 p.m. and felt alert eight hours later. It gives Richardson's age as 20, which differs from M038A. The article is listed in the March 2016 print issue; part of the page is behind a membership wall.
Open the published source (opens in a new tab)What it supports: The PubMed abstract supports that earlier reports put the human body temperature rhythm at about 25 hours but were based on light levels sufficient to confound the estimate, and that in carefully controlled lighting the intrinsic period of the human circadian pacemaker averaged 24.18 hours in both young and older people, with a tight distribution. A Harvard Gazette report (Cromie, 15 July 1999) supports that 24 healthy young and older men and women lived in isolation for a month, going to bed four hours later each day to create a 28-hour day, and that their clocks ran on a cycle of 24 hours and 11 minutes. PubMed lists twelve named authors plus an incomplete final entry; the twelve shown are recorded here.
Open the published source (opens in a new tab)What it supports: The PubMed abstract supports that social jetlag quantifies the discrepancy between circadian and social clocks, and that in a large-scale epidemiological study social jetlag was associated with increased body mass index beyond sleep duration. The LMU Munich news release (10 May 2012) supports that the data come from the Munich ChronoType Questionnaire, which more than 130,000 people had completed; it does not give the number of people in the analysis, and the lesson does not state one. It is observational and does not show cause.
Open the published source (opens in a new tab)What it supports: Penguin's UK page supports that the publisher bills the book as the number one Sunday Times bestseller; that Walker is Professor of Neuroscience and Psychology and Director of the Sleep and Neuroimaging Laboratory at the University of California, Berkeley; that the book presents sleep as vital to health and increasingly neglected; and that the publisher says deficient sleep has strong causal links to diseases including Alzheimer's, cancer, obesity and diabetes. Its editions listed start in 2017. Simon & Schuster's page for the Scribner paperback (19 June 2018) calls it a New York Times bestseller and says the book argues that sleep serves many functions for brain and body, that chronic sleep loss raises the risk of serious illness and impairs learning, mood and decision-making, and that society faces a widespread sleep deficit. Bestseller evidence actually seen as a list: the American Booksellers Association's Indie Bestseller List for the week of 1 August 2018 shows the Scribner paperback at number 7 in paperback nonfiction. No Sunday Times or New York Times list page was inspected, so those labels are attributed to the publishers in the lesson. The Penguin page lists a 2017 ebook and audio edition and a 2018 paperback. The ABA list entry reads Walker, Scribner, $17 (the Simon & Schuster page now lists $19).
Open the published source (opens in a new tab)What it supports: Supports that 16 prospective studies with 27 independent cohort samples, 1,382,999 participants followed for 4 to 25 years and 112,566 deaths were pooled; that sleep duration was assessed by questionnaire; that short sleep was associated with greater risk of death (relative risk 1.12, 95% CI 1.06 to 1.18) and long sleep also (relative risk 1.30, 95% CI 1.22 to 1.38), with heterogeneity between studies. It is observational and does not show cause, and it does not assess the book directly.
Open the published source (opens in a new tab)What it supports: The PubMed abstract supports that the review searched studies published from 1960 to 2013 that recorded total sleep time objectively (polysomnography or actigraphy) in healthy normal-sleeping adults aged 18 to 88; found 168 studies with 257 data points representing 6,052 individuals; found no significant association of sleep duration with study year; and says the results are consistent with reviews of self-reported data that have challenged the notion of a modern epidemic of insufficient sleep. It covers healthy normal sleepers in laboratory and actigraphy studies, not whole populations, and it does not assess the book directly. PubMed gives an online date of 28 August 2015; the University of Arizona record gives volume 28, pages 69–85, 1 August 2016. Added by the independent checker.
Open the published source (opens in a new tab)What it supports: Supports only that in November 2019 Alexey Guzey published, on his own website, a long critique arguing that several of the book's claims are wrong or overstated, among them that shorter sleep means a shorter life (he cites meta-analyses showing a U-shaped relationship), that the World Health Organization declared a sleep-loss epidemic, and claims about cancer risk, fatal familial insomnia and sleep deprivation as a treatment for depression, and alleging that a graph was altered. The essay was revised after first publication and says Walker did not respond to it directly. It is not peer-reviewed, and the lesson presents it as one critic's argument, not as established fact; the checker did not verify its individual points against the book. Added by the independent checker.
Open the published source (opens in a new tab)What it supports: The Carl Bosch history page supports that Bosch was born in Cologne in 1874, did practical training at an ironworks, studied metallurgy and mechanical engineering in Berlin-Charlottenburg before switching to chemistry, and joined BASF in 1899; that Haber demonstrated his ammonia apparatus to BASF representatives one year after the 1908 partnership agreement (the page does not print 1909; the year is inferred and matches M039C’s 1909 patent date) and Bosch was appointed to lead its industrialisation; that in early 1911, after high-pressure steel tubes kept bursting (the page says the double tube was devised in early 1911), Bosch devised a reactor design that let infiltrating hydrogen escape outward; that the world’s first ammonia synthesis plant went into operation at Oppau on 9 September 1913; and that Bosch shared the 1931 Nobel Prize in Chemistry with Friedrich Bergius for high-pressure chemical methods. It also supports that in September 1914 Bosch promised Germany’s military command nitric acid production and that the first nitric acid plant started at Oppau in May 1915 (used only in the limitations). BASF’s Science around us page on ammonia supports that hot, highly pressurised hydrogen removed carbon from the steel, making it soft and brittle, and that Bosch lined the apparatus with thin low-carbon soft iron and drilled holes in the pressure-bearing wall. The year is the year the pages were read; no page date was shown. It is the company’s own account.
Open the published source (opens in a new tab)What it supports: Supports that 24 test apparatus were in continuous service for years and that the separate catalyst examinations grew to the number of 20,000; that the two tubes drawn by Mannesmann each did service for 80 hours and then burst, because hydrogen penetrated the iron and formed a brittle alloy; and that the solution was to fit a pressure-bearing steel jacket internally with a quite thin lining of soft steel, with a large number of holes drilled in the jacket through which hydrogen is free to emerge. It also dates the first converter with a lining tube to March 1911 and a related patent to 12 February 1911, and describes the lining as soft steel (BASF’s page says low-carbon soft iron; the lesson follows Bosch). The Nobel page confirms the lecture date of 21 May 1932; the lecture text was read in a copy of the Nobel lecture PDF. It is Bosch’s own account, given two decades later.
Open the published source (opens in a new tab)What it supports: Supports that Haber and Robert le Rossignol patented a process yielding about 15 percent ammonia at around 175 atmospheres and 550 °C with an osmium and uranium catalyst; that osmium was unsuitable because world supplies amounted to only a few kilogrammes; that BASF assigned the chemical engineer Carl Bosch to scale up the process; that his converter tubes burst after 80 hours because hydrogen made the steel brittle; that the fix was a pressure-bearing steel jacket with a thin soft-steel lining, the hydrogen escaping through grooves and holes; that jackets still buckled in prolonged operation until the reactor was heated from the inside; that Alwin Mittasch ran some 20,000 experiments on the catalyst; and that the first industrial plant started at Oppau in 1913. It is a secondary popular history, not a peer-reviewed article.
Open the published source (opens in a new tab)What it supports: Supports that the study used data on 37 chemical products; that it found a strong and consistent learning effect; that learning was a function of cumulated industry output and cumulated investment rather than calendar time; that standard economies of scale were significant but small relative to the learning effect; that the slope of the learning curve varied with R&D expenditure and capital intensity; and that market concentration influenced price flexibility. It measures prices, not plant costs, and does not study individual engineering decisions. No DOI was listed on the pages inspected, so none is recorded.
Open the published source (opens in a new tab)What it supports: Supports that the inverse relationship between unit cost and cumulative output is one of the best-documented empirical regularities in economics; that the Boston Consulting Group built a consulting approach (the experience curve) on it; and that the step from that regularity to an organisational learning curve with major strategic consequences has not been adequately supported by direct empirical evidence, since several explanations, including static economies of scale, fit the pattern. It is a review, not a new study of chemical plants.
Open the published source (opens in a new tab)What it supports: The publisher’s page (WH Allen, Penguin; UK paperback dated 1 August 2024, subtitle A Substantial Story of Our Past and Future) supports that the book examines sand, salt, iron, copper, oil and lithium, argues that they built the modern world, underpin homes, computers and medicines and are taken for granted, follows their hidden supply systems, and was shortlisted for the FT Business Book of the Year Award and the British Academy Book Prize. Publishers Weekly’s review of the Knopf edition, dated 4 October 2023, supports the 2023 publication and that the book visits a copper mine in Chile and discusses the Saltpeter War between Bolivia and Chile, sparked in 1879, over saltpetre in the Atacama Desert (the review does not use the name War of the Pacific or the word nitrate). The penguin.co.uk pages date the paperback 1 August 2024 and the ebook and audio editions 2023; the review ends with (Nov.), suggesting the Knopf edition appeared in November 2023. Bestseller evidence: the publisher’s pages and the author’s own page carry the label THE SUNDAY TIMES BESTSELLER; no Sunday Times list page or sales figure was inspected, so the lesson attributes the label to the publisher. The exact UK first-publication date was not confirmed.
Open the published source (opens in a new tab)What it supports: Supports that human-made mass was equal to about 3 percent of global biomass at the start of the twentieth century; that it roughly equals and is surpassing global living biomass of about 1.1 teratonnes in 2020, plus or minus 6 years; that it has recently doubled roughly every 20 years; that concrete and aggregates such as gravel make up most of it; and that the result depends on definitions (on a wet-weight basis the crossover shifts to the 2030s) and on modelling assumptions. It does not test claims about geopolitics or conflict.
Open the published source (opens in a new tab)What it supports: Supports that global material extraction grew by a factor of 12 from 1900 to 2015, to 89 billion tonnes a year; that stock-building materials made up 59 percent of extraction in 2015; and that the extraction estimate carries an uncertainty of about plus or minus 16 percent in 2015 (23 percent in 1900), with many outflows modelled rather than measured. It does not test claims about specific materials’ geopolitics.
Open the published source (opens in a new tab)What it supports: Supports that on 17 June 1972 five men were arrested after breaking into the Democratic National Committee headquarters, where police found bugging devices, thousands of dollars in cash and rolls of film; that on 1 August 1972 the Washington Post reported a 25,000 dollar cheque meant for Nixon’s re-election campaign had been deposited in the bank account of one of the arrested men; that on 10 October 1972 Woodward and Bernstein reported FBI-found links between Nixon aides and the break-in; that Nixon won a second term on 7 November 1972; that on 23 March 1973 Judge John Sirica read James McCord’s letter describing a wider conspiracy in open court; that the Senate committee’s televised hearings began in May 1973; that on 13 July 1973 Alexander Butterfield revealed the taping system to Senate investigators; that on 24 July 1974 the Supreme Court ruled Nixon must surrender the tapes; and that Nixon announced his resignation on 8 August 1974. It is a popular reference timeline, not a scholarly history.
Open the published source (opens in a new tab)What it supports: Supports that the Washington Post won the 1973 Pulitzer Prize for Public Service for its investigation of the Watergate case, and that the entry centred on nine articles by Carl Bernstein and Bob Woodward dated 1 August to 15 October 1972, with 12 supplemental news articles from June to December 1972, covering a 25,000 dollar cheque, a secret campaign fund, the destruction of records after the break-in and a federal intelligence and sabotage effort against Democrats. The year is the year the page was read.
Open the published source (opens in a new tab)What it supports: Supports that Woodward and Bernstein reported the Watergate story for the Washington Post, beginning with the burglary at the Democratic headquarters; that Woodward met his secret source, later identified in 2005 as FBI Associate Director Mark Felt, in an early-morning parking garage on 9 October 1972; that a Post story the next day, based on that meeting, exposed the burglary as part of a larger plan; that the reporters won the Pulitzer Prize (the page links this to May 1973); that they then signed with Simon and Schuster and worked on the book at night and at weekends while still covering the story, trying other approaches such as the burglars’ perspective before telling the story of their investigation; that the book was published in 1974; and that the 1976 film gave the reporters as much publicity as their actors. The page misprints the story year as 1974 in one place; the lesson uses 1972, consistent with M040A. The archive’s claim that the reporting led to 40 indictments is not used.
Open the published source (opens in a new tab)What it supports: Supports that Bradlee was the Washington Post’s executive editor from 1968 to 1991 and authorised Woodward and Bernstein to keep investigating the Watergate scandal. The year is the year the page was read. It gives no detail of his rules on sourcing.
Open the published source (opens in a new tab)What it supports: Supports that the study uses the mismatch between newspaper markets and congressional districts, shifted by redistricting, to isolate the effect of lower coverage; that where coverage of a House member is exogenously lower, voters are less likely to recall and to describe and rate their representative, representatives appear to do less constituency work (testifying less at hearings and voting against their party less often), and federal spending is lower. The NBER working-paper version (w13878) reports 161 newspapers covering the 103rd to 107th Congresses (1991 to 2002) and American National Election Studies data; the published sample was not checked. The working paper notes weaker identification in some analyses and only marginal spending effects in the redistricting specification. U.S. only.
Open the published source (opens in a new tab)What it supports: Supports that after local newspaper closures, municipal borrowing costs rose by about 5 to 11 basis points, adding roughly 650,000 dollars per issue; that the authors argue the effect is causal and not explained by local economic conditions; and that the loss of oversight was linked to higher government wages, larger deficits and more costly financing choices. The article copy inspected reports closures identified over 1996 to 2014 and a baseline of 204 closure counties compared with 1,062 counties without a closure. It notes data limits in measuring substitute news sources. U.S. only.
Open the published source (opens in a new tab)What it supports: The author’s book page supports that the book, published in 1974, recounts the Watergate investigation by the two Washington Post reporters, from the burglary at Democratic headquarters to the revelations that preceded Nixon’s resignation, and labels it a bestseller. Bestseller evidence: Publishers Weekly’s century-of-bestsellers feature (print issue of 19 April 2022) records it as the number one nonfiction bestseller for the week of 12 August 1974 (the page itself shows no date; the print-issue date is taken from its URL). The publisher’s own book page could not be opened. The lesson’s statement of the book’s central idea is the author of this lesson’s summary; no source states why people read it.
Open the published source (opens in a new tab)What it supports: The bookseller record of the publisher’s description supports that this expanded second edition (University of California Press, 18 October 2016, ISBN 9780520291294) examines media-driven myths that overstate the news media’s power, including the myth that the Washington Post’s Watergate reporting brought down Nixon’s presidency. Campbell’s reasoning is taken from his 2022 article (M040I).
Open the published source (opens in a new tab)What it supports: Supports that Campbell, professor of communication studies at American University, argues that the reporters exposed financial links to the burglars and tied figures such as John Mitchell to the scandal but missed the hush-money payments and the existence of the tapes; that special prosecutors, federal judges, the FBI, congressional panels and the Supreme Court did the decisive work, with the Court’s unanimous July 1974 tapes order decisive; that Woodward called the idea that he single-handedly brought down Nixon absurd and Katharine Graham also rejected that framing; and that Mark Felt admitted in 2005 that he was Deep Throat. The date is the republication date; the original Conversation date was not shown. It is an opinion article by a scholar, not peer-reviewed.
Open the published source (opens in a new tab)What it supports: Supports that Kolff and Belding Scribner received the 2002 Albert Lasker Clinical Medical Research Award for hemodialysis; that in 1938 Kolff, a recent medical graduate at Groningen, watched a young man die of kidney disease, and that over the next several years he built four artificial kidneys (the page does not say outright that work began in 1938); that he took a post as an internist in Kampen and kept developing it; that he used sausage casing made of cellophane wound around a large horizontal rotating drum; that between 1943 and 1944 he treated 16 patients with acute kidney failure with limited success and that the first 15 died, the page says not from the therapy; that in September 1945 he achieved his first undeniable success, in a woman who had collaborated with the Nazis; and that after the war he donated all five of his artificial kidneys to hospitals in London, Poland, The Hague, Montreal and Mount Sinai Hospital in New York (this count and list come from Goldstein’s presentation; the unsigned essay says only that he sent artificial kidneys to England, Canada and the U.S.). The kidney-failure type (acute) comes from the presentation; M041B says some patients had acute and others chronic failure, so the lesson does not say which. The page gives the survivor as 67 and says she lived seven more years, which differs from M041B, so the lesson gives neither figure exactly. The page’s main text and the Goldstein presentation differ on the cellophane length, which the lesson does not state. The main text is unsigned.
Open the published source (opens in a new tab)What it supports: Supports that after Germany attacked the Netherlands in May 1940 Kolff moved to a small hospital in Kampen; that his 1943 rotating drum kidney used a Ford Model T water pump and a sewing-machine motor; that in October 1938 he watched a 22-year-old man die of kidney failure and became determined to find a treatment; that during the war his team treated 16 patients, some with acute and others with chronic kidney failure, 15 of whom died, with the effect of dialysis on the sole survivor inconclusive; that treatments were suspended for 14 months until the war ended while he continued to modify the device; that on 11 September 1945 he dialysed a 68-year-old woman in a coma for 11.5 hours, after which she woke and lived eight more years; that he was criticised for treating a Nazi collaborator and said a medical provider was not the one to decide whom to treat by background; and that after the war he donated his five artificial kidneys to hospitals around the world. Its figures for the survivor’s age and survival differ slightly from M041A, so the lesson says late sixties and for years. It is a short secondary review, not an archival study.
Open the published source (opens in a new tab)What it supports: Supports that Kolff worked at a small hospital in Kampen under German occupation with repurposed materials, and that the first artificial kidney was made from juice cans, sausage casings and parts from an old washing machine; also that in 1947 he shipped a device to Mount Sinai Hospital in New York. It names the 1945 survivor and counts her as the 16th patient, which differs from M041A and M041B; the lesson relies on it only for the materials.
Open the published source (opens in a new tab)What it supports: Supports that, in data covering 10 years, 71 cardiothoracic surgeons and more than 6,500 procedures using a new cardiac surgery technology, individuals learned more from their own successes than from their own failures, and more from others’ failures than from others’ successes. It is observational, in one surgical setting, and does not study device design.
Open the published source (opens in a new tab)What it supports: Supports, from the abstract (University of Colorado Denver press release agrees), that organisations in the orbital launch vehicle industry learned more effectively from failures than from successes, that knowledge from failure depreciated more slowly than knowledge from success, and that prior experience and the size of a failure affected learning. The sample (more than 4,600 orbital launches from nine countries, 1957 to 2004, including 426 failures) comes from a strategy+business summary of the paper, not from the article itself, which could not be opened. It is observational and covers one industry.
Open the published source (opens in a new tab)What it supports: The Pulitzer Prize record supports that the book (Scribner) won the 2011 Pulitzer Prize for General Nonfiction. Bestseller evidence: NPR’s Hardcover Nonfiction Bestsellers list for 28 April 2011, compiled from weekly surveys of close to 500 independent bookstores with the American Booksellers Association (sales ending 24 April), shows it at number 6 in its 7th week on the list; the publisher’s description of the updated edition (Scribner, 25 November 2025, ISBN 9781668047033, first published 2010) calls it a New York Times bestseller, and no New York Times list page was inspected. The publisher’s description supports that it is a biography of cancer from its first documented appearances thousands of years ago through the 20th-century effort to cure, control and conquer it. NPR’s 2011 interview supports that Mukherjee treats cancer. The book’s argument as summarised in the lesson (step-by-step progress through bold, often failed attempts, with radical breast surgery and more intensive chemotherapy for solid tumours among well-meant mistakes) rests on M041G and is the lesson’s own paraphrase; the book itself was not read.
Open the published source (opens in a new tab)What it supports: Supports that the book traces cancer from early nihilism and mysticism through surgery, radiation and chemotherapy to targeted therapy based on oncogenes and signalling-pathway mutations; that it weighs the achievements and failures of figures such as Halsted and Farber; that it presents radical breast cancer surgery and the push for ever more intensive chemotherapy in solid tumours as well-intentioned but mistaken; and that it covers smoking research and prevention. The reviewer’s only notable criticism is a Canadian omission. It does not give sales figures.
Open the published source (opens in a new tab)What it supports: Supports that a randomised trial begun in 1971 tested whether less extensive surgery, with or without radiation, was as effective as the Halsted radical mastectomy, in 1,079 women with clinically negative and 586 with clinically positive axillary nodes (1,665 in total); that after 25 years there were no significant differences in disease-free, relapse-free, distant-disease-free or overall survival; and that the authors concluded the findings validate earlier results showing no advantage from radical mastectomy, while differences of a few percentage points cannot be excluded. It does not test the book itself.
Open the published source (opens in a new tab)What it supports: Supports that U.S. cancer mortality continued to decline through 2021, averting over 4 million deaths since 1991, because of reductions in smoking, earlier detection for some cancers and improved treatment; that incidence is increasing for 6 of the top 10 cancers; and that wide, persistent disparities remain. It covers the United States only and does not assess the book.
Open the published source (opens in a new tab)