The Gambler's Fallacy, Measured in Real Casinos

Study Overview

The gambler's fallacy — the mistaken belief that a losing streak makes a win "due" — had mostly been demonstrated in controlled lab experiments with students before this study. Croson and Sundali wanted to know whether the same bias shows up when real gamblers bet their own money in an actual casino, not a lab.

Methodology Summary

The researchers used videotape footage of real patrons playing roulette in a casino, recording the sequence of outcomes (e.g., red or black) alongside each patron's actual bets. This let them measure two related but opposite biases directly from real betting behavior:

• The gambler's fallacy: betting against a streak (e.g., betting black after red has come up several times), expecting the streak to "correct."
• The hot hand fallacy: betting with a streak, expecting it to continue.

Because this was field data rather than a lab experiment, the bets were real money, placed by people with no knowledge they were part of a study.

Key Findings

• Both the gambler's fallacy and the hot-hand fallacy were present in real casino betting behavior — patrons' bets shifted in ways consistent with both biases, depending on the specific outcome sequence.
• The biases were described as "small but significant" — real, measurable, and consistent with lab findings, but not so large that every bet was driven by them.
• A companion analysis (Sundali & Croson, 2006) found a positive correlation between the two biases at the individual level — people prone to one tended to be prone to the other, suggesting a shared underlying cause.

Limitations

• The study focused on roulette specifically — results may not generalize identically to other games with different pacing or feedback (e.g., slots, blackjack).
• Field data can't control for every variable a lab study can (e.g., how much a given patron had already won or lost that session), so the estimate of bias size is noisier than a controlled experiment would produce.
• This study documents that the bias exists and affects real betting — it doesn't measure how much money, on average, it costs bettors over time.

Practical Meaning

This is the research basis for the "well documented" claim in our Gambler's Fallacy guide. The practical takeaway: this isn't just a lab curiosity — it's a bias that measurably shows up in how real people bet real money at real roulette tables, which is exactly why it's worth checking your own instincts against the math rather than a "feeling" about a streak.

Related Reading

• Gambler's Fallacy Explained — our full guide built on this research
• Illusion of Control in Gambling
• Gambling Psychology Hub