267. How to Make a Bad Decision

Episode Summary

Executive Summary: The episode examines the gambler’s fallacy and related sequence effects in high-stakes real-world decisions. Using baseball umpires, Indian bank loan officers, and U.S. asylum judges, Toby Moskowitz and colleagues show that prior outcomes can bias subsequent judgments, but stronger incentives and more deliberate processing reduce the error. The discussion broadens to contrast effects in financial markets and everyday decision-making.

Main Topics: The gambler’s fallacy as a real-world bias (Priority: 5/5): The episode defines the gambler’s fallacy as the mistaken belief that random events should alternate, leading people to expect a reversal after a streak. Baseball umpire decision-making (Priority: 5/5): Researchers use PitchFX data to test whether umpires’ strike/ball calls are influenced by previous calls, finding sequence-based bias especially on borderline pitches. Loan officers and incentives (Priority: 5/5): A field experiment with Indian bank loan officers shows that sequential ordering of applications affects approvals, but strong incentives sharply reduce the bias. Asylum judges and high-stakes sequencing (Priority: 5/5): Analysis of federal asylum cases finds that prior approvals or denials influence later rulings, with meaningful consequences for applicants’ lives and futures. Contrast effects in financial markets (Priority: 4/5): Kelly Hsu’s related work shows earnings announcements are judged relative to what came immediately before, affecting stock-price reactions. How to reduce sequence bias (Priority: 4/5): The episode argues that time gaps, breaks, stronger accountability, and slower deliberation can lessen reliance on cognitive shortcuts.

Key Arguments: Humans see patterns in randomness and often expect streaks to “balance out,” even when events are independent. Sequence matters in decision-making: the order of cases or pitches can bias even experienced professionals. Baseball umpires show gambler’s-fallacy-like behavior most strongly on borderline calls, not obvious ones. Loan officers’ sequence bias drops dramatically under stronger incentives, suggesting effort and accountability improve judgment. Asylum decisions are especially consequential because a small sequencing effect can change whether someone remains in the country. Contrast effects show that judgments are often relative, not absolute, in both courts and financial markets. Breaking up sequences with time or task changes can weaken the effect of prior decisions on later ones.

Data Points: Coin toss probability: 50-50 - Used to illustrate that independent random events do not become “due” after a streak. Baseball games analyzed: Over 12,000 - PitchFX data set used to study umpire decisions. Called pitches analyzed: Roughly 1.5 million - Included only pitches where the batter did not swing and the umpire made the call. Umpires in sample: 127 - Home-plate umpires covered in the baseball analysis. Umpire accuracy on borderline outside pitches: About 64% - Umpires were correct on close pitches just outside the strike zone. Effect after a previous strike: About 0.5% less likely to call the next pitch a strike - Overall effect across the universe of next pitches. Effect on corner pitches after a strike: About 3.5% less likely to call strike - Borderline pitches showed a larger sequence effect. Effect after two prior strikes: About 5.5% less likely to call strike - Further strengthening of the sequencing effect on borderline calls. Loan officer error from sequencing: Roughly 8% - Misjudgments attributed to application order in the Indian bank experiment. Loan officer error under strong incentives: Down to 1% - Sequence-based error largely reduced when compensation strongly rewarded accuracy. Asylum data span: 1985 to 2013 - Long-run dataset of immigration court decisions. Asylum cases analyzed: About 150,000 decisions - Large sample across U.S. immigration courts. Asylum judges analyzed: 357 judges - Judges in courts with sufficient case volume. Effect after one prior approval: Almost 1% less likely to approve next case - Sequence bias in asylum rulings. Effect after two prior approvals: About 1.5% less likely to approve next case - Bias increases after consecutive approvals. Effect when prior cases were on same day: Closer to 3% or about 5% in larger sequences - Temporal proximity strengthens the sequencing effect in asylum decisions. Contrast effect in earnings: Unrelated large firm’s prior good earnings make today’s earnings seem less good - Financial-market application of sequential contrast effects.

Pivotal Quotes: "We like to tell stories and find patterns that aren't really there." — Toby Moskowitz: Explaining why people misread random streaks as meaningful patterns. "If you make a quick decision, you rely on these simple heuristics... If I'm forced to actually just use information and think about it slowly because I really want to get it right... then I don't rely on those simple heuristics as much." — Toby Moskowitz: Describing why strong incentives reduce sequence bias among loan officers. "If you miss something, the worst thing to do, you can never make up a call. People, oh, that's a make-up call. Well, no, it's not, because if you try and make up a call, now you've missed two." — Hunter Wendelstatt: Explaining why umpires should not compensate for prior mistakes with later calls.

Implications: Sequence bias can affect legal, financial, and sporting decisions with real consequences. Better incentives, more deliberation, and separation between cases can reduce—but not eliminate—these human errors.