• Machine learning’s growing role hasn’t fixed how teams watch, compare, and judge talent, because human heuristics still slip into scouting and draft decisions even when data sets look clean.
• Building bias checks into everyday evaluation lets teams catch patterns and adjust judgment in real time, directly countering the shortcuts that make high-stakes calls unreliable.
• Designing more reliable metrics and decision architecture gives scouts cleaner inputs to watch, sort, and compare—helping to refine noisy or incomplete data that no amount of training can fix.

Machine learning isn’t a peripheral tool in professional sports anymore—it’s becoming part of how teams operate.¹ Across an array of sports including basketball, football, baseball, and hockey, predictive models now sit alongside traditional scouting reports, guiding decisions that once depended almost entirely on experience.² These systems can process years of player statistics, physiological measurements, and team outcomes. They can also uncover patterns humans might overlook, especially when evaluators are juggling pressure and time constraints. In one large-scale study using ten seasons of National Basketball Association (NBA) data, ensemble models predicted which draft candidates would become elite players with 87.3% accuracy, far above the 68.5% associated with scouting alone.³ The promise is clear: machine learning can expand what evaluators see rather than replace what they know.

Still, analytics didn’t appear from nowhere. The shift traces back to Moneyball,⁴ where the Oakland Athletics showed that better information could help teams compete with wealthier rivals. Since then, front offices have invested heavily in analysts, tracking systems, and performance databases. The stakes justify it: NBA teams spent more than $4.5 billion on player salaries in the 2023–24 season, and a single misjudgment on draft night can ripple through a franchise for years.⁵

Even with more data than ever, drafting remains fundamentally uncertain. Research in behavioral economics and decision science shows why. Under pressure, people can default to heuristics that simplify judgment but distort accuracy.^6,7 Decades of draft research confirm this pattern: NBA executives can be influenced by a player’s high school reputation,⁸ the prestige of their college program,⁹ draft-combine statistics that don’t always predict future performance,¹⁰ or nationality cues.¹¹ These tendencies may help explain why draft position only shows a modest relationship with later career success,¹² and why roughly 17% of the variance in career games played across major leagues can be linked to draft slot,¹³ rather than serving as a reliable predictor on its own. In practice, even experienced evaluators can’t fully avoid uncertainty, and analytics doesn’t remove the need for judgment—it just changes where it’s used. Debiasing talent decisions therefore depends on pairing these systems with structured bias checks, transparent data practices, and continuous awareness of how heuristics can shape decision-making under pressure.

Talent identification is a prediction problem. Scouts and coaches must estimate how a teenager or young adult’s current traits might translate into professional success years later. These decisions are rarely made with perfect information, which means people often rely on experience-based shortcuts to simplify judgment. In behavioral science, these shortcuts are called heuristics—mental rules that can speed up reasoning but may also introduce systematic errors known as cognitive biases.⁶

A prototype forms when repeated exposure teaches decision-makers what a “successful athlete” tends to look like. Over time, that mental model becomes a reference point: a preferred body type, personality, or developmental path that feels intuitively right. The problem isn’t that prototypes exist—they’re unavoidable—but that they can steer attention toward athletes who fit the mold and away from those who don’t. In other words, a coach may think they’re judging performance when they’re actually comparing a player to an internal template.

Since heuristics operate automatically, bias may slip in even when people believe they’re being objective. Experienced evaluators often assume their intuition reflects hard-won expertise, yet research shows that expertise itself can reinforce bias.¹⁴ Overconfidence, confirmation bias, and the blind-spot bias—the tendency to see bias in others but not oneself—can all shape selection choices.¹⁵ What feels like seasoned instinct may, in practice, be a pattern of mental shortcuts refined through repetition rather than accuracy.

The relative age effect is one of the clearest examples of how such shortcuts manifest in sports.¹⁶ In youth programs, athletes born earlier in the competitive year are consistently over-represented on elite teams.^17,18 Coaches may not consciously prefer older players, but early physical maturity can signal “readiness,” which then becomes a proxy for long-term potential. Studies across Brazil, France, Germany, Japan, North America, Poland, Portugal, and Spain all show that relatively older athletes are selected more often and receive more coaching attention.^17-20 Over time, these advantages compound—older players gain confidence, training time, and visibility—while later-born athletes exit the system. By adulthood, the developmental pipeline has filtered out a large portion of potential talent, not through deliberate exclusion, but through an unnoticed heuristic.

Biases like these rarely arise from ill intent. They emerge because forecasting human performance is uncertain and cognitively demanding. When time is short and accountability is high, evaluators may rely on cues that feel reliable—height, speed, pedigree, composure—while underweighting traits that are harder to quantify, such as adaptability or late-stage growth. The same process can explain why a technically gifted but unconventional player might be dismissed as a “poor fit.”

A qualitative study with twelve national-level coaches and selectors illustrates this limited awareness.²¹ When asked about cognitive bias, participants generally associated the term with overt prejudice—race or gender discrimination—rather than with everyday decision shortcuts. Few recognized subtler forms such as anchoring bias (over-reliance on first impressions), availability bias (judging likelihood by how easily examples come to mind), or their own blind-spot bias. This suggests that the concept of bias may be well known but poorly understood within high-performance environments.

The 2010 NBA Draft shows how these patterns can play out. The Houston Rockets’ data model identified Jeremy Lin, an Asian American guard from Harvard, as a strong candidate. Despite favorable metrics, the team passed on him. Lin later excelled with the New York Knicks, and the Rockets publicly acknowledged that implicit assumptions about his background likely shaped their judgment.²² The data were sound but the prototype wasn’t.

Reducing bias in talent identification isn’t about policing intuition, it’s about improving predictive accuracy. Because talent forecasting happens with limited data, every system operates under what behavioral economists call bounded rationality—decisions made with partial information, time pressure, and noise.²³ The question isn’t whether bias exists, but how teams can measure it, learn from it, and adapt. Three decision-science strategies can help organizations make judgments smarter: bias audits, bias profiling, and blind evaluation defaults.

Bias Audits as Feedback Mechanisms

A bias audit is simply a structured way to test whether selection patterns might be favoring traits that don’t actually predict performance. It turns bias from a moral issue into a diagnostic one. Teams can start by collecting data on each selection decision—age, region, physical maturity, and background—and linking those details to later outcomes. Over time, that record can reveal where decisions drift from reality. When analysts compare selections against base rates, they might notice base-rate neglect: overlooking how often success really occurs across groups and relying on more vivid cues like height or early maturity.

The next step is all about feedback. Rather than labeling decisions as biased, analysts can show the calibration gap—how predicted potential compares with actual performance. Saying “your forecasts for early-maturing athletes weren’t more accurate than for others,” invites reflection, not blame. Additionally, repeated audits don’t just measure accuracy, they teach it. Each cycle can update what scouts believe about which cues truly predict success—a kind of data-driven feedback loop where intuition learns from evidence. 

Bias Profiling as System Calibration

While audits reveal where bias shows up, profiling helps teams understand how it moves through a group. Everyone’s judgment carries patterns, and some cognitive styles may be more vulnerable to certain biases. That’s not a flaw—it’s human diversity in reasoning.

Research by Ceschi and colleagues shows that gaps in analytical reasoning, also known as mindware gaps, can lead to statistical mistakes and are more prevalent in those with lower cognitive ability, while valuation biases like overconfidence may appear more in people with high self-esteem.²⁴ Anchoring effects might also be stronger among open or exploratory thinkers, leading them to fixate on first impressions. None of this means people are “problematic,” it just means bias risk isn’t evenly distributed.

In practice, teams could invite everyone—scouts, coaches, analysts—to complete short cognitive or reasoning profiles. It’s not a test, it’s a mirror. The results can help map the team’s collective bias tendencies so they’re managed system-wide. From there, evaluators can be paired with complementary colleagues or have their input weighted differently in aggregate assessments. It’s a simple optimization process that lets diverse thinking styles cancel out one another’s blind spots.

When the process feels inclusive and developmental, people don’t feel targeted. Everyone participates, everyone gets feedback, and everyone improves. Bias profiling, handled this way, turns diversity of thought into a measurable advantage rather than a source of friction.

Blind Evaluation Defaults

A third strategy tackles the design of the decision environment itself. According to information-framing research, the way information is presented can unconsciously shape how people interpret it.²⁵ When visible cues, like a player’s name, club, or accent, enter the frame, they can activate representativeness heuristics that distort evaluation.

Masking those cues can stop bias before it starts. Early-stage video clips could hide identifying information; for example, the order of clips might be randomized to avoid anchoring, with contextual data like competition level added only in later review rounds. These are small procedural tweaks that keep evaluations centered on observable performance.

Evidence from outside sports backs this up. When orchestras introduced blind auditions, women’s chances of selection rose by nearly 50%.²⁶ Academic and grant panels now rely on the same logic to keep assessments objective. Sports teams can adopt similar masking methods, giving every athlete an equal first impression without overhauling their entire system.

Behavioral interventions like defaults and feedback loops can each be used in talent identification to make selection decisions fairer and more consistent. They’re designed to improve human judgment, not to fix flawed data. A team could have every bias-audit protocol and decision framework tuned perfectly, but if its testing batteries or predictive models are biased, error enters the process long before a scout or coach ever weighs in. In other words, when measurement tools are misaligned with the realities of performance, they may distort evaluations before human judgment even begins.

The Hidden Bias in Testing Batteries

Testing is meant to make talent decisions objective. Yet in many programs, the data driving those judgments don’t fully reflect the game itself. For decades, identification systems have relied on physical or technical testing—speed, agility, endurance, reaction time, or ball-handling.²⁷ However, there’s still no consensus on which of these measures genuinely predicts long-term success.

Most tests remain unidimensional and context-poor. They capture traits like sprint speed in isolation, without showing how those traits interact under competition. That narrow focus can end up favoring certain body types, maturational timelines, or early-developing athletes. A fifteen-year-old who excels at an acceleration test might outperform peers in youth leagues, but those advantages may fade once others catch up physically.

In behavioral terms, this reflects base-rate neglect: evaluators treat test scores as absolute indicators rather than probabilistic signals that vary by age, role, and environment. If those data aren’t validated across multiple cohorts, small distortions compound. The results appear objective but can bias entire talent pipelines toward athletes who fit the measurement frame rather than the game demands.

Cognitive Load and Dual-Process Decision-Making

Scouting combines two modes of reasoning: System 1 (fast, intuitive pattern recognition) and System 2 (slow, analytical evaluation).²⁸ Under uncertainty, time pressure, or high stakes, decision-makers naturally lean on System 1. Data is supposed to steady that instinct, but if the data themselves are biased, they can anchor System 1 in the wrong direction.

When numerical rankings appear, coaches may defer to them, assuming the model has already weighed factors they’d struggle to re-evaluate under time pressure. This is automation bias—excessive trust in algorithmic or quantified outputs—where as the draft clock counts down, decision-makers may follow a model’s ranking over their own scouting notes, even when  those notes flag stronger tactical awareness of adaptability in a lower-ranked player. Under time pressure or other stressful conditions, the model’s judgement can feel safer than contradicting it.

The Limits of Machine-Based Models

Machine learning and predictive analytics were developed to widen the lens of evaluation and reduce bias . In practice, they often favor athletes who’ve already been seen, measured, and tracked. Predictive models learn from past selection records and tracking data rather than from future potential. When those records come from the same leagues, regions, and development systems year after year, the model learns to favor athletes who move through those pipelines. Players outside those pathways generate less data, appear less often in comparisons, and may receive lower rankings from the model even when coaches observe strong in-game skills. 

This reflects a classic garbage-in, garbage-out problem. Models can evaluate enormous volumes of data, but they can’t correct for missing coverage or skewed inputs. When certain pathways dominate the data, the system reinforces those pathways, even when performance indicators suggest talent elsewhere.

Making talent identification fairer isn’t just about teaching people to spot bias—it’s about improving the information that feeds every judgment. Even the best-trained scouts can’t make reliable calls if the data behind them are flawed. Behavioral science offers two complementary ways to strengthen those foundations: triangulating multiple evidence streams and aggregating independent judgments. Both may help teams make more consistent, data-informed, and context-sensitive decisions.

Triangulating Evidence and Auditing Metrics

Research on standardized testing batteries shows that even the most established assessments may not capture the full picture of athletic potential.²⁹ Studies across several sports have found little agreement on which tests reliably predict success.^30,31 Many programs still depend on narrow, one-dimensional measurements—speed drills, endurance runs, or technical skill tests—taken out of game context. These isolated metrics might highlight short-term readiness but miss how physical, tactical, and psychological traits interact over time.

From a decision-science perspective, that’s a cue validity problem: the signals being measured don’t always map onto the outcomes teams care about. When tests lack ecological validity, people can start overinterpreting weak signals as proof. Over time, the data may reinforce stereotypes about what “elite potential” looks like rather than revealing it.

Triangulation can reduce that risk. Instead of letting one test or model dominate, teams can draw on at least three independent evidence streams, each analyzed separately. For example, physical metrics, perceptual-cognitive tasks, and longitudinal behavioral data can all be used to inform decisions. Diverging results don’t mean the process failed, they reveal where uncertainty lies. Converging results can, over time, identify which measures truly forecast pro-level adaptation and which ones don’t.

Regular audits can keep those measures honest. In healthcare, diagnostic systems are continually revalidated against real outcomes.³² Talent pipelines might benefit from a similar loop, recalibrating metrics annually and flagging when predictive accuracy starts to drift. Publishing internal reliability coefficients or false-positive and false-negative rates can seem tedious, but they’re how organizations learn which indicators stay useful and which quietly decay.

Technology may accelerate this process but can’t replace scrutiny. Machine-learning tools might spot complex interactions, yet they’re still bound by the data they’re fed. If that data is biased or incomplete, the algorithm may simply replicate the same blind spots. Triangulation doesn’t replace analytics—it makes them accountable.

Aggregating Independent Judgments

Bias won’t disappear just because the data improves. It lessens when diverse perspectives are aggregated in structured ways. Research on the wisdom of crowds shows that combining multiple independent forecasts can outperform individual experts.^33,34 The math is simple: independent errors cancel out, while shared insights remain.

In talent identification, this means expanding evaluation beyond a single decision-maker. One coach’s intuition might favor physical maturity, another’s might highlight tactical intelligence, or a psychologist might notice adaptability under stress. When those insights are collected independently and then averaged, the group’s final judgment often lands closer to reality.

However, independence here matters. Each evaluator can record observations before group discussion so that social influence doesn’t distort results. Aggregating afterward turns individual noise into collective accuracy rather than conformity. Adding even two or three additional raters might significantly reduce error—especially when their expertise spans different domains.

Some teams worry that broader input could slow decisions, but experience suggests it often has the opposite effect.^33,34 When accountability is shared, evaluators feel safer admitting uncertainty, and disagreements can be resolved through data, not hierarchy. Over time, that shift might create a culture where learning replaces defensiveness, and where feedback feels like calibration, not criticism.

Together, triangulation lets teams verify what their data actually measure, while aggregation helps them interpret it more accurately. When policies embed both, organizations can detect bias earlier, recalibrate faster, and make stronger evidence-based selection decisions.

Implementing bias mitigation in high-performance sports might sound straightforward, but it’s rarely that simple. Machine learning models now guide national pipelines, yet many are still trained on historical rosters that underrepresent late maturers, smaller markets, and women’s programs. Unless organizations regularly examine who the data represent and who they leave out, models could reinforce historical inequities, narrow recruitment choices, and undermine return on investment.

Likewise, mandating bias audits or blind reviews doesn’t mean people will use them as intended. In high-stakes environments, coaches and scouts can feel that new systems question their expertise rather than refine it. Without clear communication, even strong policies might stall before they start.

Still, these obstacles aren’t reasons to stop, they’re signals to plan smarter. When policymakers frame equity as precision and performance, rather than bureaucracy, buy-in can rise. Bias reduction isn’t just a compliance exercise, it’s how teams stay competitive in a world where data and judgment must learn from each other.

Throughout this article, we’ve explored how decision science can help teams see, test, and improve the way they judge talent. The first challenge showed how intuition,while valuable,can steer judgment toward familiar types of athletes rather than genuine performance. The second revealed that even precise-looking data can mislead when it’s built on incomplete or outdated measures. Together, they show that fairer decisions don’t come from more information alone, but from learning how to question and recalibrate it.

The opportunities we outlined aren’t just theory. Bias audits, data triangulation, and independent evaluations give teams tools they can actually use. They let scouts compare evidence, spot where instincts overreach, and update their models before errors grow. These steps won’t remove uncertainty fully, but they can make choices steadier and more transparent. When accuracy improves, confidence follows—and so does trust across every level of performance.

Bias 

Bias isn’t just a buzzword—it’s a systematic pattern of thinking that drifts away from rational judgment. It happens when our brains try to simplify how we process and interpret information. You’ll find it everywhere: in healthcare, law, sports, and education. Read this article to learn how bias shapes decisions and what we can do about it.

What Are Heuristics?

We often make quick judgments without realizing it. In this interview, Dr. Sekoul Krastev, decision scientist and TDL co-founder, explains what heuristics are and why we rely on them. You’ll also learn how they were discovered and why these mental shortcuts can be both powerful and risky, depending on when we use them.

1. Horvat, T., & Job, J. (2020). The use of machine learning in sport outcome prediction: A review. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 10(5), e1380.
2. Abeza, G., O’Reilly, N., Nadeau, J., & Abdourazakou, Y. (2023). Big data in professional sport: The perspective of practitioners in the NFL, MLB, NBA, and NHL. Journal of Strategic Marketing, 31(8), 1413–1433. https://doi.org/10.1080/0965254X.2022.2108881 
3. Ajiboye, A. A., Gaffari, M. A., & Obamuwagun, O. E. (2025). Predictive analytics in sport management: Applying machine learning models for talent identification and team performance forecasting. Communication in Physical Sciences, 12(7), 2032–2048.
4. Lewis, M. (2004). Moneyball: The art of winning an unfair game. W. W. Norton & Company.
5. Spotrac. (2023). NBA team salary cap tracker: 2023–24 season. https://www.spotrac.com/nba/cap/_/year/2023/sort/cap_maximum_space
6. 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 
7. Raue, M., & Scholl, S. G. (2018). The use of heuristics in decision making under risk and uncertainty. In Psychological perspectives on risk and risk analysis: Theory, models, and applications (pp. 153–179). Springer International Publishing. https://doi.org/10.1007/978-3-319-92478-6_7 
8. Berger, T., & Daumann, F. (2021a). Anchoring bias in the evaluation of basketball players: A closer look at NBA draft decision-making. Managerial and Decision Economics, 42(5), 1248–1262. https://doi.org/10.1002/mde.3305 
9. Ichniowski, C., & Preston, A. E. (2012). Does March Madness lead to irrational exuberance in the NBA draft? High-value employee selection decisions and decision-making bias (Working Paper No. 17928). National Bureau of Economic Research. https://doi.org/10.3386/w17928 
10. Teramoto, M., Cross, C. L., Rieger, R. H., Maak, T. G., & Willick, S. E. (2018). Predictive validity of National Basketball Association draft combine on future performance. Journal of Strength and Conditioning Research, 32(2), 396–408. https://doi.org/10.1519/JSC.0000000000001798 
11. Motomura, A. (2016). MoneyRoundball? The drafting of international players by National Basketball Association teams. Journal of Sports Economics, 17(2), 175–206. https://doi.org/10.1177/1527002514526411 
12. Groothuis, P. A., Hill, J. R., & Perri, T. (2009). The dilemma of choosing talent: Michael Jordans are hard to find. Applied Economics, 41(25), 3193–3198. https://doi.org/10.1080/00036840701564459 
13. Koz, D., Fraser-Thomas, J., & Baker, J. (2012). Accuracy of professional sports drafts in predicting career potential. Scandinavian Journal of Medicine & Science in Sports, 22(4), e64–e69. https://doi.org/10.1111/j.1600-0838.2011.01408.x 
14. Kahneman, D., & Klein, G. (2009). Conditions for intuitive expertise: A failure to disagree. American Psychologist, 64(6), 515–526. https://doi.org/10.1037/a0016755
15. Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic decision making. Annual Review of Psychology, 62, 451–482. https://doi.org/10.1146/annurev-psych-120709-145346
16. Helsen, W. F., Baker, J., Michiels, S., Schorer, J., Van Winckel, J., & Williams, A. M. (2012). The relative age effect in European professional soccer: Did ten years of research make any difference? Journal of Sports Sciences, 30(15), 1665–1671. https://doi.org/10.1080/02640414.2012.721929
17. Fortin-Guichard, D., Huberts, I., Sanders, J., van Elk, R., Mann, D. L., & Savelsbergh, G. J. P. (2022). Predictors of selection into an elite-level youth football academy: A longitudinal study. Journal of Sports Sciences, 40(9), 984–999. https://doi.org/10.1080/02640414.2022.2044128
18. Fortin-Guichard, D., Tétreault, É., Paquet, D., Mann, D. L., & Grondin, S. (2023). Identification of “sleeping” talent using psychological characteristics in junior elite ice-hockey players. Journal of Sports Sciences, 41(7), 605–615. https://doi.org/10.1080/02640414.2023.2230709
19. Nakata, H., & Sakamoto, K. (2011). Relative age effect in Japanese male athletes. Perceptual and Motor Skills, 113(2), 570–574. https://doi.org/10.2466/05.10.11.PMS.113.5.570-574 
20. Leite, N., Borges, J., Santos, S., & Sampaio, J. (2013). The relative age effect in school and federative sport in basketball. Revista de Psicología del Deporte, 22(1), 219–222.
21. Ramanayaka, N. D., Dickson, G., Libich, J., & Rayne, D. (2025). Susceptibility to cognitive biases in athlete–team selection. International Journal of Sports Science & Coaching. Advance online publication. https://doi.org/10.1177/17479541251338508 
22. Mann, D. L., Fortin-Guichard, D., & Müller, D. (2025). A framework of cognitive biases that might influence talent identification in sport. International Review of Sport and Exercise Psychology, 1–28. https://doi.org/10.1080/1750984X.2025.2556393 
23. Simon, H. A. (1990). Bounded rationality. In Utility and probability (pp. 15–18). Palgrave Macmillan. https://doi.org/10.1007/978-1-349-20568-4_5 
24. Ceschi, A., Costantini, A., Sartori, R., Weller, J., & Di Fabio, A. (2019). Dimensions of decision-making: An evidence-based classification of heuristics and biases. Personality and Individual Differences, 146, 188–200. https://doi.org/10.1016/j.paid.2018.07.033 
25. Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453–458. https://doi.org/10.1126/science.7455683 
26. Goldin, C., & Rouse, C. (2000). Orchestrating impartiality: The impact of “blind” auditions on female musicians. American Economic Review, 90(4), 715–741. https://doi.org/10.1257/aer.90.4.715 
27. Lidor, R., & Lavyan, N. Z. (2002). A retrospective picture of early sport experiences among elite and near-elite Israeli athletes: Developmental and psychological perspectives. International Journal of Sport Psychology, 33, 269–289.
28. Kahneman, D. (2011). Thinking, fast and slow. Macmillan.
29. Johnston, K., & Baker, J. (2020). Waste reduction strategies: Factors affecting talent wastage and the efficacy of talent selection in sport. Frontiers in Psychology, 10, 2925. https://doi.org/10.3389/fpsyg.2019.02925 
30. Lidor, R., Côté, J., & Hackfort, D. (2009). ISSP position stand: To test or not to test? The use of physical skill tests in talent detection and in early phases of sport development. International Journal of Sport and Exercise Psychology, 7, 131–146. https://doi.org/10.1080/1612197X.2009.9671896 
31. Breitbach, S., Tug, S., & Simon, P. (2014). Conventional and genetic talent identification in sports: Will recent developments trace talent? Sports Medicine, 44, 1489–1503. https://doi.org/10.1007/s40279-014-0221-7 
32. Branson, C. F., Williams, M., Chan, T. M., Graber, M. L., Lane, K. P., Grieser, S., Landis-Lewis, Z., Cooke, J., Upadhyay, D. K., Mondoux, S., Singh, H., Zwaan, L., Friedman, C., & Olson, A. P. (2021). Improving diagnostic performance through feedback: The diagnosis learning cycle. BMJ Quality & Safety, 30(12), 1002–1009. https://doi.org/10.1136/bmjqs-2020-012456 
33. Surowiecki, J. (2005). The wisdom of crowds: Why the many are smarter than the few and how collective wisdom shapes business, economics, society, and nations. Little, Brown.
34. Fleenor, J. W. (2006). The wisdom of crowds: Why the many are smarter than the few and how collective wisdom shapes business, economics, societies and nations [Book Review]. Personnel Psychology, 59(4), 982.