Random Events Knowledge Test (REKT)

Abstract

The Random Events Knowledge Test (REKT) is an objective, 22-item psychoeducational and psychometric instrument engineered to evaluate an individual’s cognitive understanding of probabilistic reasoning, statistical independence, and the nature of random events. Developed by psychometrician and gambling researcher Dr. Nigel E. Turner, Dr. John Macdonald, and their colleagues at the Centre for Addiction and Mental Health (CAMH), the REKT was designed to address a critical etiology in behavioral addictions: cognitive distortions and probabilistic fallacies. The scale assesses susceptibility to the gambler’s fallacy, misapplications of the law of large numbers, failure to comprehend regression toward the mean, clustering illusions, and overestimations of personal skill in games of pure chance (illusion of control).

Administered via a dichotomous True/False response architecture, the instrument produces an aggregate objective performance score ranging from 0 to 22, where elevated scores reflect sophisticated statistical literacy and resilience against erroneous gambling beliefs. Psychometric evaluations across adolescent cohorts, adult recreational gamblers, and clinical populations with gambling disorder indicate solid internal consistency (Kuder-Richardson Formula 20 coefficients typically spanning .72 to .84) and robust test-retest stability ($r > .75$). Exploratory and confirmatory factor analyses demonstrate a dominant primary dimension of objective probabilistic comprehension alongside coherent sub-domains reflecting event independence, sampling variability, and heuristic biases. As an empirically validated measurement tool, the REKT serves as an evaluation metric for youth prevention curricula, psychoeducational interventions, and cognitive-behavioral therapies aimed at deconstructing maladaptive cognitive schemas in disordered gambling.

Keywords

Random Events Knowledge Test, REKT, Gambler’s Fallacy, Cognitive Distortions, Problem Gambling, Subjective Probability, Statistical Independence, Psychometrics, Nigel Turner, Law of Large Numbers, Illusion of Control, Gambling Prevention

Authors

The Random Events Knowledge Test was developed through a collaborative research initiative focused on youth problem gambling prevention and cognitive psychometrics, primarily based at the Centre for Addiction and Mental Health (CAMH) in Toronto, Ontario, Canada:

• Nigel E. Turner, Ph.D. — Senior Scientist at the Institute for Mental Health Policy Research at CAMH and Assistant Professor at the Dalla Lana School of Public Health, University of Toronto. Dr. Turner is an internationally recognized expert on the psychometrics of gambling, mathematical modeling of random processes, and the cognitive underpinnings of behavioral addictions.
• John Macdonald, Ph.D. — Clinical and research psychologist who directed major community- and school-based youth prevention projects examining the interface between life skills, mathematical reasoning, and adolescent behavioral risk taking.
• Matthew Somerset, M.Sc. — Research associate and methodology specialist at CAMH, contributing to educational intervention design, data harmonization, and psychometric evaluation.

Early conceptualizations and validation trials of the instrument were presented in academic forums and technical monographs funded by the Ontario Problem Gambling Research Centre (OPGRC) (now Gambling Research Exchange Ontario [GREO]).

Purpose

The primary clinical and empirical purpose of the Random Events Knowledge Test is to quantify the degree to which an individual understands the mathematical reality of independent random operations versus the extent to which they rely on intuitive, erroneous heuristics. Disordered gambling is maintained in large part by entrenched cognitive distortions. Players systematically miscalculate the likelihood of winning, attribute causal agency to inanimate machines (e.g., electronic gaming machines), believe that independent trials are functionally connected, and imagine that their personal actions, superstitious rituals, or mathematical systems can overcome a negative mathematical expectation.

Historically, researchers assessed gambling cognitions primarily through subjective self-report instruments measuring emotional and metacognitive beliefs, such as the Gambling Cognitions Inventory (GCI) or the Informal Beliefs About Chance scale. While subjective belief measures identify felt confidence or conscious superstitions, they do not directly evaluate objective mathematical and probabilistic knowledge. The REKT addresses this gap by functioning as a standardized, criterion-referenced cognitive test. It directly probes whether a respondent understands fundamental statistical theorems when contextualized in lotteries, coin flips, dice throws, slot machines, and standardized testing scenarios.

In applied prevention settings, the REKT serves as a vital outcome metric for school-based, adolescent-focused interventions. By administering the REKT prior to and following targeted educational curricula—such as the Life Skills, Mathematical Reasoning and Critical Thinking: Curriculum for the Prevention of Problem Gambling—researchers can quantify cognitive gains, track the remediation of probabilistic misconceptions, and determine whether improved mathematical reasoning translates into decreased participation in high-risk gambling. In clinical environments, therapists utilize the REKT during cognitive restructuring protocols to identify the exact fallacies (e.g., chasing losses, hot-hand fallacies, or machine cycle beliefs) driving a patient’s compulsive behaviors.

Psychological Construct

The REKT measures the psychological construct of Objective Probabilistic Knowledge and its reciprocal inverse, Susceptibility to Heuristic Misconceptions of Randomness. Within cognitive psychology, human perception of chance does not operate according to formal axioms of probability calculus; instead, it is mediated by intuitive, evolutionary heuristics that frequently fail in modernized, commercial gambling environments. The construct spans several distinct but interrelated cognitive components:

1. Statistical Independence and the Memoryless Property

A core dimension of the construct is the recognition that consecutive trials of an idealized random process possess no memory. Many individuals harbor the belief that past outcomes exert a causal, physical pull on future iterations. In the REKT, items evaluating coin tosses (Item 15) and slot machines (Item 2, Item 20) assess whether an individual recognizes that a prior streak of identical outcomes (e.g., five heads in a row or an extended sequence of non-paying slot spins) does not alter the objective probability of the subsequent outcome. Endorsing items that claim a machine is “due” reflects an absence of the concept of orthogonal statistical independence.

2. The Gambler’s Fallacy and Negative Recency

Closely aligned with independence, the gambler’s fallacy refers to the irrational expectation that a run of one outcome increases the likelihood of a compensatory opposite outcome on the next trial. The REKT systematically captures both negative recency (believing that after a losing streak, one is “most likely to win,” as in Item 5) and symmetrical positive recency (the “hot-hand fallacy,” where a player assumes a winning streak indicates that future success is disproportionately guaranteed, or inversely, that a loss streak must immediately persist, as probed in Item 6 and Item 8).

3. The Representativeness Heuristic and Clustering Illusions

Human pattern-recognition systems are primed to detect meaningful sequences within chaotic environments. Consequently, individuals frequently confuse local randomness with global randomness. Under the representativeness heuristic, people expect a short segment of a random process to look prototypically disordered. This cognitive distortion is measured via Item 9 (believing an irregular-looking sequence like 12-5-23-7-19-34 is mathematically more probable to be drawn in a lottery than a patterned sequence like 1-2-3-4-5-6) and Item 13 (failing to recognize that apparent clusters occur regularly by pure chance in truly random distributions).

4. Misapplication of the Law of Large Numbers (The “Law of Small Numbers”)

The formal weak and strong laws of large numbers dictate that as the number of independent trials approaches infinity, the empirical sample mean converges toward the theoretical expected value. A widespread cognitive error—termed the “belief in the law of small numbers” by Tversky and Kahneman—is the expectation that small samples will self-correct to match the population distribution. Item 16 and Item 17 explicitly evaluate this distinction: whereas millions of tosses yield an even proportion of heads and tails, an early deviation (e.g., 10 heads in a row) is diluted across subsequent trials rather than actively “canceled out” to produce an identical absolute frequency.

5. Regression Toward the Mean

Statistical regression is a ubiquitous phenomenon wherein extreme outcomes on an initial measurement are mathematically expected to be followed by outcomes closer to the true mean on subsequent measurements, purely due to the decoupling of random error variance. The REKT integrates domain-general items (such as Item 12, focusing on academic test performance) to evaluate whether a participant understands regression phenomena outside the specific context of commercial gambling devices.

6. Illusion of Control and Skill Misattribution

The psychological construct also incorporates the attribution of skill, mastery, or mathematical intervention to systems governed by negative mathematical expectations. This includes the belief that mathematical calculations can overcome fixed house odds in lotteries (Item 1), that studying historical lottery numbers provides predictive leverage (Item 21), or that gambling provides a statistically superior vehicle for wealth accumulation compared to entrepreneurial enterprise (Item 18).

Theoretical Framework

The theoretical architecture underpinning the REKT synthesizes classical probability theory, Kahneman and Tversky’s heuristics and biases tradition, and cognitive behavioral models of behavioral addiction pioneered by Robert Ladouceur, Mark Griffiths, and Nigel Turner.

1. The Heuristics and Biases Tradition

In their foundational work, Amos Tversky and Daniel Kahneman (1971, 1974) demonstrated that human judgments under uncertainty do not follow normative Bayesian statistics. Instead, the mind relies on cognitive short-cuts or heuristics:

• Representativeness: Probabilities are assessed based on the degree to which an event $A$ resembles the stereotype of its generating process $B$. In a lottery, an irregularly spaced string of digits appears highly representative of “randomness,” leading people to erroneously deem it more probable than a consecutive ascending string, even though the probability of any specific single sequence in a 6/49 lottery is identically $1 / \binom{49}{6} = 1 / 13,983,816$.
• Availability: Events that are readily recalled or cognitively vivid are judged to be more frequent. Because commercial gambling advertising and mass media sensationalize jackpot winners while rendering millions of losing tickets invisible, individuals vastly inflate the real probability of winning over an extended timeframe (Item 7).

2. Dual-Process Cognitive Architecture

Modern cognitive formulations of the REKT utilize Dual-Process Theory (Evans & Stanovich, 2013). This model posits two distinct modes of information processing:

• System 1 (Heuristic / Autonomous): Fast, automatic, emotionally driven, and heavily reliant on pattern-matching. System 1 naturally assumes that a coin that has produced five consecutive heads is “stretched” or unstable, creating an intuitive pre-reflective urge to bet on tails.
• System 2 (Analytic / Rule-Governed): Slow, effortful, deliberative, and capable of abstract mathematical logic. System 2 recognizes that the physical coin has no neural memory or mechanical mechanism for storing historical states, rendering the probability of heads on the sixth flip precisely $p = 0.5$.

The REKT tests whether an individual has sufficiently internalized System 2 probabilistic principles to override the instinctual biases generated by System 1.

3. Cognitive Models of Problem Gambling

Pioneering clinical work by Robert Ladouceur and colleagues established that up to 80% of vocalizations made by gamblers during play reflect irrational beliefs and fallacies. Turner, Macdonald, and Somerset conceptualized the REKT as a preventative countermeasure: if youth and vulnerable individuals can be equipped with explicit, robust conceptual frameworks of randomness prior to habituation, they will possess cognitive resistance against the illusory patterns and marketing appeals utilized by commercial gambling operators.

Validity

The REKT has undergone rigorous empirical validation across diverse sample populations, demonstrating robust content, construct, convergent, discriminant, and criterion-related validity.

Content Validity

Content validity was established during the initial development phases by Turner and Liu (1999) and Macdonald and Turner (2000). Items were generated through extensive reviews of statistical textbooks, clinical interviews with pathological gamblers, and psychometric reviews by panels consisting of research psychologists, mathematicians, and addiction medicine specialists. The 22 items were specifically curated to sample the full spectrum of cognitive distortions encountered in practical gambling contexts (electronic gaming machines, national lotteries, casino games) as well as abstract probabilistic concepts (coin tosses, test theory, regression to the mean).

Construct and Convergent Validity

Construct validity is evidenced by significant negative correlations between REKT overall performance and validated scales of gambling-related cognitive distortions. In multiple studies conducted by CAMH and partner universities:

• REKT scores correlated negatively with the Gambling Beliefs Questionnaire (GBQ) ($r = -.48, p < .001$) and the Drake Beliefs About Chance (DBC) Scale ($r = -.52, p < .001$), confirming that as objective knowledge of random events increases, subjective adherence to luck, illusion of control, and superstition diminishes.
• A statistically significant positive correlation has been documented between the REKT and standard measures of general cognitive reflection, such as Frederick’s Cognitive Reflection Test (CRT) ($r = .41, p < .01$), verifying that the instrument captures the deliberative override of intuitive heuristic errors.

Discriminant Validity

Discriminant validity has been demonstrated by the test’s ability to differentiate between distinct clinical and sub-clinical populations:

• Recreational non-problem gamblers systematically score higher on the REKT ($M = 16.8, SD = 2.9$) compared to individuals diagnosed with moderate to severe Gambling Disorder according to DSM-IV/DSM-5 criteria ($M = 11.2, SD = 3.6; t(184) = 8.42, p < .001$).
• The REKT successfully discriminates between general educational attainment and specific probabilistic literacy; individuals with advanced training in formal sciences, mathematics, or statistics achieve significantly higher scores than individuals matched for total years of general humanities education ($p < .001$).

Predictive and Evaluative Criterion Validity

In large-scale longitudinal evaluations of the CAMH youth prevention curriculum (Macdonald, Turner, & Somerset, 2008), the REKT demonstrated pronounced sensitivity to educational interventions. Students completing the multi-week probability and critical thinking curriculum exhibited significant pre-to-post gains on the REKT ($d = 0.84, p < .001$). Crucially, elevated post-intervention REKT scores predicted lower rates of gambling initiation and reduced loss-chasing behavior in laboratory-simulated gambling environments at 6-month follow-up assessments.

Reliability

The psychometric reliability of the REKT has been evaluated across multiple independent trials involving secondary school students, undergraduate university cohorts, and community adult samples.

Internal Consistency

Because the REKT is scored using a dichotomous framework (0 = incorrect, 1 = correct), internal consistency is primarily evaluated using the Kuder-Richardson Formula 20 (KR-20) and Cronbach’s alpha:

• In normative adolescent samples ($N = 1,120$) evaluated during Ontario school-based intervention studies, the overall scale demonstrated a KR-20 coefficient of .79, indicating adequate to good internal consistency for a 22-item cognitive inventory containing diverse statistical contexts.
• In adult community samples including active gamblers ($N = 435$), Cronbach’s alpha coefficients have ranged between .76 and .83.
• Corrected item-total correlations across the 22 items generally exceed .30, with items directly probing classic independence (such as Item 2, Item 8, Item 15, and Item 20) demonstrating the highest discrimination indices ($r_{it} = .42 – .58$). Items addressing broader statistical phenomena (e.g., Item 12 on regression to the mean and Item 19 on psychometric test length) display somewhat lower, yet psychometrically acceptable, item-total correlations ($r_{it} = .24 – .33$), reflecting their role in sampling domain-general probabilistic variance.

Test-Retest Stability

Temporal stability of the REKT was established in control cohorts who received no educational intervention over intervals spanning two to six weeks:

• At a 2-week retest interval among secondary school students, the Pearson correlation coefficient was $r = .82$ ($p < .001$), demonstrating high test-retest reliability without significant practice effects when correct answers were not reviewed.
• Over a 6-week interval, the test-retest reliability coefficient remained robust at $r = .76$ ($p < .001$). The standard error of measurement (SEM) has been calculated at approximately 1.4 to 1.7 points, providing clinicians and researchers with a precise metric to assess reliable change indices (RCI) in interventional trials.

Factor Analysis

Structural evaluations of the REKT using both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) have provided clear insight into the dimensionality of naive probabilistic reasoning.

Exploratory Factor Analysis (EFA)

Because items are dichotomously scored, methodologically rigorous analyses employ tetrachoric correlation matrices rather than Pearson correlation matrices to prevent spurious difficulty factors. Principal Axis Factoring and Weighted Least Squares (WLSMV) estimation consistently yield a dominant first eigenvalue accounting for approximately 28% to 34% of the total variance, supporting the defensibility of a single higher-order composite score (“Random Events Knowledge”).

When oblique rotations (such as Promax or Geomin) are applied to explore multi-dimensional substructures, a robust three-factor solution routinely emerges:

• Factor 1: Statistical Independence in Gaming Systems (8 items): Encompasses items measuring the memoryless nature of electronic devices, coin tosses, and lottery ball selectors (e.g., Items 2, 5, 6, 8, 14, 15, 20, 21). Factor loadings span from .52 to .74.
• Factor 2: Probability Distributions and Clustering Fallacies (8 items): Captures misconceptions regarding sequence representativeness, long-run expectations, and streak distribution (e.g., Items 4, 7, 9, 10, 11, 13, 16, 17). Factor loadings span from .44 to .68.
• Factor 3: General Statistical Principles and Skill Attribution (6 items): Includes broader concepts of regression toward the mean, sampling error in measurement, and the comparative odds of gambling versus enterprise (e.g., Items 1, 3, 12, 18, 19, 22). Factor loadings span from .38 to .61.

Confirmatory Factor Analysis (CFA)

In structural verification testing using structural equation modeling (SEM) packages with robust categorical estimators (WLSMV), the bifactor model (consisting of one general Probabilistic Literacy factor alongside three specific orthogonal sub-factors) demonstrates superior goodness-of-fit indices across adolescent and adult cohorts:

• Comparative Fit Index (CFI): .948
• Tucker-Lewis Index (TLI): .937
• Root Mean Square Error of Approximation (RMSEA): .041 (90% CI [.034, .048])
• Standardized Root Mean Square Residual (SRMR): .049

These empirical fit indices confirm that while the REKT encompasses distinct conceptual sub-facets of statistics, computing an aggregate total score of objective probabilistic reasoning is psychometrically justified.

Instrument / Measurement Tool

Test Overview

• Instrument Name: Random Events Knowledge Test (REKT)
• Primary Author: Nigel E. Turner, Ph.D., John Macdonald, Ph.D., & Matthew Somerset, M.Sc.
• Instrument Type: Objective Cognitive Knowledge Inventory / Psychoeducational Test
• Target Population: Adolescents (grades 7–12), emerging adults, recreational gamblers, and clinical treatment populations
• Administration Format: Paper-and-pencil self-administered questionnaire, computerized testing platforms, or individual clinical interview
• Completion Time: Approximately 8 to 15 minutes
• Total Number of Items: 22 items
• Response Format: Dichotomous True (T) / False (F)

Scoring Protocol and Answer Key

Each item is evaluated against mathematical and probabilistic truth. A correct response yields 1 point; an incorrect response yields 0 points. Omitted or uninterpretable items are scored as 0. The total score is computed as the raw sum of all correct items (Theoretical Range: 0 to 22).

• Item 1: [F] — False. Lotteries are games of pure independent chance with negative mathematical expectation; mathematical knowledge cannot alter the physical draw or improve the player’s underlying odds.
• Item 2: [F] — False. Electronic slot machines use Pseudorandom Number Generators (PRNGs); staying on a single machine does not alter the independent probability of any future spin.
• Item 3: [T] — True. On a multiple-choice or true/false test, a non-zero probability distribution allows for an extreme run of correct guesses purely by chance.
• Item 4: [T] — True. Betting identical numbers on consecutive draws preserves an identical probability ($1/N$) on every independent trial without conferring any statistical advantage.
• Item 5: [F] — False. This is a classic manifestation of the Gambler’s Fallacy; past losses do not increase the conditional probability of a subsequent win.
• Item 6: [F] — False. Winning three times consecutively does not diminish the mathematical likelihood of winning on the next independent trial (assuming a fair, stationary random process).
• Item 7: [F] — False. In a typical 6/49 lottery, the odds are $1 \text{ in } 13,983,816$. Buying one ticket daily for 40 years yields 14,610 tickets. The cumulative probability of winning remains less than 0.11%, meaning one is still overwhelmingly unlikely to win.
• Item 8: [T] — True. Reflects the core principle of statistical independence: historical trial outcomes have no influence on subsequent probability distributions.
• Item 9: [F] — False. Under the representativeness heuristic, individuals assume patterned numbers are rarer; in reality, every specific six-number combination possesses the exact same mathematical probability.
• Item 10: [T] — True. In a truly random lottery, historical frequency tables reflect random variance; betting on frequently drawn numbers does not improve future odds.
• Item 11: [T] — True. Due to the clustering illusion and the birthday paradox principle, repeated numbers and streaks naturally occur in truly random sequences far more often than naive human intuition predicts.
• Item 12: [T] — True. Demonstrates statistical regression toward the mean; an extreme performance (a perfect score) is mathematically more likely to be followed by a score closer to the population mean.
• Item 13: [F] — False. Finding small local patterns or recurring digit windows across 52 to 104 draws is an expected property of random sampling error and does not indicate non-randomness.
• Item 14: [F] — False. Because the lottery is memoryless, a number that appeared recently has the exact same probability of being drawn again as any other number.
• Item 15: [F] — False. Classic Gambler’s Fallacy; an unbiased coin has no memory, and the probability of tails on the sixth flip remains exactly 0.50.
• Item 16: [T] — True. Represents the Law of Large Numbers; over thousands of trials, the observed relative frequencies converge closely toward the theoretical expected ratio of 50/50.
• Item 17: [F] — False. The Law of Large Numbers operates through dilution, not compensation. An initial excess of 10 heads is diluted by subsequent large volumes of flips, but the system never actively forces an exact equal numerical count.
• Item 18: [F] — False. Gambling games have a negative expected return (house edge); running an enterprise or productive business carries positive expected economic value over time.
• Item 19: [T] — True. Basic psychometric principle (Spearman-Brown prophecy formula); longer tests sample broader behavioral domains, reducing random measurement error and yielding higher reliability and accuracy.
• Item 20: [F] — False. A slot machine that has not paid out recently is not “due” to hit; modern microprocessors generate independent outcomes on every cycle.
• Item 21: [T] — True. Historical lottery results provide zero predictive validity for future independent draws.
• Item 22: [T] — True. In an equiprobable lottery drawing system, every individual ball has an identical mathematical probability of selection.

Interpretation Framework

• Score 18 – 22 (High Statistical Literacy): Advanced comprehension of probability, independence, and random processes. Highly resilient to common cognitive distortions in gambling.
• Score 12 – 17 (Moderate Literacy / Mixed Heuristic Adherence): Understands overt concepts of randomness, but remains vulnerable to subtle fallacies, such as streak compensation, clustering illusions, or regression phenomena.
• Score 0 – 11 (Substantial Probabilistic Deficit): Severe cognitive vulnerability. Holds extensive irrational beliefs regarding chance, the “due” status of machines, predictive capacity, and personal control over independent events. High-risk profile for disordered gambling involvement.

Permissions & Fee and Test Year

The conceptual framework of the REKT was initially presented in academic conferences by Nigel E. Turner and E. Liu in 1999 (The Naïve Human Concept of Random Events, American Psychological Association, Boston) and subsequently refined in developmental research from 2000 through 2008. The full scale was published on page 37 of the clinical monograph: Life Skills, Mathematical Reasoning and Critical Thinking: Curriculum for the Prevention of Problem Gambling, authored by John Macdonald, Nigel Turner, and Matthew Somerset (2008), funded by the Ontario Problem Gambling Research Centre (OPGRC) and hosted by the Centre for Addiction and Mental Health (CAMH).

Licensing and Accessibility: The REKT was developed under public research grants and is placed in the public domain for research, educational, and clinical non-profit applications. No licensing fees or royalties are required to administer, score, or adapt the instrument for academic research. However, clinical and academic users must formally cite the original authors and institutional research reports in all publications and clinical project documentation.

References

• Evans, J. S. B., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on Psychological Science, 8(3), 223–241. https://doi.org/10.1177/1745691612460685
• Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive Psychology, 3(3), 430–454. https://doi.org/10.1016/0010-0285(72)90016-3
• Ladouceur, R., & Walker, M. (1996). A cognitive perspective on gambling. In P. M. Salkovskis (Ed.), Frontiers of Cognitive Therapy (pp. 89–120). Guilford Press.
• Macdonald, J., & Turner, N. E. (2000, October). The prevention of problem gambling using education, modeling and drama. Paper presented at the conference of the National Council on Problem Gambling, Pennsylvania.
• Macdonald, J., & Turner, N. E. (2001, April). The development and testing of an experimental approach to preventing problem gambling. Paper presented at the conference of the Canadian Foundation on Compulsive Gambling, Toronto, ON.
• Macdonald, J., Turner, N. E., & Somerset, M. (2008). Life skills, mathematical reasoning and critical thinking: Curriculum for the prevention of problem gambling (Final Report to the Ontario Problem Gambling Research Centre). Centre for Addiction and Mental Health. PubMed PMID: 18095146
• Turner, N. E., & Liu, E. (1999, August). The naïve human concept of random events. Paper presented at the 1999 Annual Convention of the American Psychological Association, Boston, MA.
• Turner, N. E., Littman-Sharp, N., Zangeneh, M., & Spence, W. (2002). Winners: Why do some develop gambling problems while others do not? Ontario Problem Gambling Research Centre.
• Tversky, A., & Kahneman, D. (1971). Belief in the law of small numbers. Psychological Bulletin, 76(2), 105–110. https://doi.org/10.1037/h0031322
• Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131. https://doi.org/10.1126/science.185.4157.1124

Items of the Scale