Why We Think the Rich Will Get Richer and the Poor Will Get Poorer: Understanding Regressive Bias
Seventeen-year-old Aditya loved cricket statistics. After every match, he’d predict how players would perform in the next game. His pattern was consistent: if a batsman scored 150 runs in one match, Aditya predicted they’d score 140-160 in the next. If a batsman scored only 15 runs, Aditya predicted they’d score 10-20 next time. “Performance is consistent,” he reasoned. “Good players stay good, struggling players stay struggling.”
His friend Rohan, who was studying statistics, noticed something curious. “Aditya, your predictions are always wrong in the same direction. When someone scores really high, they almost always score lower next time than you predicted. When someone scores really low, they almost always score higher next time than you predicted. It’s like you expect extremes to stay extreme, but actually they tend to move toward average.”
Aditya was defensive. “But that’s just regression to the mean—a statistical concept. I’m predicting based on skill!” Rohan explained: “Exactly. Regression to the mean happens because extreme performances involve both skill and luck. An unusually high score involves good skill plus good luck. Next time, skill remains but luck varies, so the score tends to be lower. You’re ignoring this regression, overestimating how high the high performers will score and underestimating how well the low performers will do.”
Rohan continued: “This is called regressive bias—you overestimate high values and high likelihoods while underestimating low values and low likelihoods. You think the extremes will persist when actually they tend to regress toward the middle. It’s not just cricket. You do this with exam scores too—you think students who topped once will always top, and students who failed once will always fail, but actually both groups tend to move toward average performance over time.”
This bias—expecting extremes to persist and underestimating regression to the mean—affects how we predict everything from sports performance to stock returns to student achievement. Understanding it reveals why we’re systematically surprised when extremes don’t persist.
What Is Regressive Bias?
Regressive bias is the tendency to overestimate the persistence of extreme values while underestimating how much they’ll regress (move back) toward the average. When we see someone at a very high position, we expect them to stay very high. When we see someone at a very low position, we expect them to stay very low. We underestimate the natural tendency for extreme values to move toward the mean on subsequent measurements.
The phenomenon is related to fundamental misunderstanding of regression to the mean, identified by statistician Francis Galton in the 1880s. Research at University of California, Berkeley studying prediction errors found that people consistently make “extreme forecasts”—predicting that high values will remain high and low values will remain low—when actually subsequent measurements almost always regress toward the average. This creates systematic prediction errors where we overestimate extremes and underestimate how moderate subsequent values will be.
According to studies from Carnegie Mellon University, regressive bias operates because we intuitively attribute extreme performances entirely to stable characteristics (skill, ability, quality) while ignoring the role of variable factors (luck, circumstances, randomness). An extremely high performance gets attributed to “being extremely good,” leading to predictions of continued extreme performance. We fail to recognize that extreme outcomes typically involve both stable ability and variable luck, and the luck component won’t repeat reliably.
Research from Duke University demonstrates that regressive bias affects professional forecasters as much as casual observers. Stock analysts predict that high-performing stocks will continue dramatically outperforming when actually they tend to regress toward market average. Sports commentators predict that athletes with exceptional seasons will continue at that level when actually they typically regress toward their career averages. The bias is stubborn and widespread.
The Farmer’s Foolish Fortune-Telling
A teaching tale tells of a farmer who, after an exceptionally abundant harvest yielding three times his normal crop, visited a fortune-teller. “My harvest this year was extraordinary,” he said. “Tell me—will next year be even better? Should I expand my land and buy more equipment to handle the huge crops I’ll have?”
The fortune-teller, knowing the farmer’s region and typical yields, said: “Your next harvest will likely be good but not as extraordinary as this year. Probably closer to your normal yields.” The farmer was insulted. “You’re no fortune-teller! This year proved my farming skill is exceptional. I’ve improved my methods. Next year will be even better, not worse!”
The farmer borrowed money, expanded his land, and bought new equipment based on his expectation of continued extreme yields. The next year’s harvest was good—better than his historical average—but nowhere near the extraordinary yield of the previous year. The farmer couldn’t manage his debt and lost his farm.
A wise neighbor explained what had happened: “Your extraordinary harvest involved both good farming and exceptional weather—unusual rain at the perfect times, no pest problems, ideal temperatures. Your good farming was repeatable. The exceptional weather was luck and wouldn’t necessarily repeat. By expecting the extreme result to persist, you failed to account for regression toward average conditions. Your fortune-teller was actually wise—she understood that extreme results rarely persist because the lucky factors that contributed to them don’t reliably repeat.”
Buddhist philosophy addresses regressive bias in teachings about impermanence and the middle way. The Buddha taught that extreme states—extreme pleasure, extreme pain, extreme success, extreme failure—are inherently impermanent and naturally return toward more moderate states. Regressive bias represents clinging to extremes, expecting them to persist when the nature of conditioned phenomena is change and moderation. Wisdom recognizes that extremes are temporary; the middle way is stable.
The Bhagavad Gita discusses this through Krishna’s teaching about the cycles of nature and the temporary nature of extremes. Krishna teaches Arjuna that just as seasons cycle and extremes naturally moderate, worldly fortunes rise and fall. Expecting extremes to persist represents ignorance of natural patterns. The wise person recognizes that high positions tend to fall and low positions tend to rise toward balance, and plans accordingly rather than assuming extremes will persist.
How Regressive Bias Misleads Our Predictions
In academic performance and student assessment, regressive bias makes teachers and parents overestimate how consistently top students will perform and underestimate struggling students’ potential for improvement. A student who scores 98% on one exam is predicted to score similarly high on the next, when actually they’ll likely score somewhat lower (still good, but regressed toward their average). A student who scores 35% is predicted to score similarly low, when actually they’ll likely score somewhat higher next time.
Research from Stanford University tracking student performance across multiple exams found that extreme scores (both high and low) on one test almost always regress significantly toward the student’s average on the next test. Yet teachers consistently predict that top performers will stay at the top and bottom performers will stay at the bottom, underestimating the regression and being surprised by the movement toward middle performance.
In sports predictions and athletic performance, regressive bias makes commentators predict that players with exceptional seasons will continue performing exceptionally when actually they typically regress toward their career averages. A basketball player who scores thirty points per game one season (well above their career average of twenty) gets predicted to score thirty again, when actually they’ll likely regress to around twenty-two to twenty-five—still above average but not as extreme.
Studies show that “Rookie of the Year” award winners typically perform worse in their second season—not because they got worse, but because exceptional rookie seasons involve both talent and luck, and the luck component doesn’t repeat. Regressive bias makes observers expect continued exceptional performance and then conclude the player “declined” when actually they just regressed to their true talent level.
In investing and stock market predictions, regressive bias makes investors chase “hot stocks” that have recently performed extremely well, expecting continued extreme returns. Similarly, investors avoid or sell stocks that have performed extremely poorly, expecting continued extreme losses. But stock returns exhibit strong regression to the mean—stocks that vastly outperform the market one year typically perform closer to market average the next year, and stocks that vastly underperform typically perform better subsequently.
Research from University of Chicago analyzing decades of stock returns found that buying previous year’s top performers and selling previous year’s worst performers (a strategy motivated by regressive bias) consistently underperforms simply holding market index funds. The bias toward expecting extremes to persist costs investors significantly.
In business performance and company valuation, regressive bias makes investors overvalue companies with exceptional recent performance and undervalue companies with recent struggles. A company that grew revenue by fifty percent last year gets valued as if they’ll continue that extreme growth, when actually they’ll likely regress toward industry-average growth rates. A company with negative five percent growth gets valued as if decline will continue, when actually they’ll likely regress toward small positive growth.
Studies demonstrate that growth stock investing (buying companies with extreme recent growth) typically underperforms value investing (buying companies with recent struggles) largely because regressive bias causes growth stocks to be overvalued based on extreme performance that won’t persist, while value stocks are undervalued based on extreme struggles that will moderate.
In health outcomes and medical treatments, regressive bias creates the illusion of treatment effectiveness when actually regression to the mean is operating. People seek medical treatment when symptoms are at their worst (an extreme state). Natural variation means symptoms will likely improve somewhat regardless of treatment—regression toward their average severity. But regressive bias makes patients and doctors attribute the improvement entirely to treatment, overestimating treatment effectiveness.
Research shows that many alternative medicine “successes” and spontaneous recovery stories are actually regression to the mean. People seek treatment during extreme illness episodes, then improve toward their average health state naturally, but regressive bias makes them conclude the treatment caused improvement when timing and regression explain it.
Predicting Regression, Not Persistence of Extremes
The most important principle for countering regressive bias is expecting extremes to regress toward the mean in subsequent measurements. When you see an extreme value—very high or very low—your default prediction should be that the next measurement will be closer to average, not that the extreme will persist. This doesn’t mean the entity is changing; it means measurement includes variable factors that won’t repeat identically.
Learn the concept of regression to the mean and recognize it in action. Extreme performances almost always involve both stable factors (skill, true quality) and variable factors (luck, circumstances, measurement error). On subsequent measurements, stable factors persist but variable factors change, causing regression toward the mean. This isn’t mysterious—it’s statistical necessity.
When predicting future performance from current extreme performance, adjust toward the average. If a student scored ninety-eight percent (extreme high), don’t predict ninety-eight percent next time—predict something between ninety-eight percent and their historical average, probably eighty-five to ninety percent. If they scored thirty percent (extreme low), predict something between thirty percent and average, probably forty to fifty percent. This regression-adjusted prediction will be more accurate than assuming extreme persistence.
Distinguish between stable traits and variable performance. A student who consistently scores ninety percent has demonstrated stable high ability—predicting continued ninety percent is reasonable. A student who usually scores seventy percent but scored ninety-eight percent once had an extreme performance involving luck or ideal circumstances unlikely to repeat—predicting regression to seventy-five percent is more realistic than predicting ninety-eight percent again.
Be especially cautious about decisions based on extreme recent performance. Don’t expand your business based on one exceptionally profitable year—wait to see if it persists or regresses. Don’t bench an athlete based on one terrible game—see if it’s an extreme fluctuation or true decline. Don’t invest heavily in stocks based on one year’s extreme returns—expect regression. Extreme recent results are poor guides to future results because of regression to the mean.
Remember Aditya always wrong in his cricket predictions because he expected extreme batting performances to persist when they consistently regressed toward average. Remember the farmer who lost everything by expecting his exceptional harvest to persist when it regressed to normal yields. Both fell victim to regressive bias—overestimating how long extremes last while underestimating how strongly values pull back toward the mean.
Regressive bias makes the world seem more extreme and persistent than it is. We think the rich will keep getting richer at extreme rates (they usually don’t—fortunes regress). We think the poor will stay extremely poor (they usually improve somewhat—poverty regresses). We think top students will always top (they usually regress toward good-but-not-extreme). We think failures will always fail (they usually regress toward moderate performance). The reality is that extremes are temporary. The high fall somewhat. The low rise somewhat. The middle is stable. Understanding regression to the mean and overcoming regressive bias means expecting this natural movement toward average rather than being perpetually surprised when extremes moderate, as they almost always do.
Frequently Asked Questions
Does regression to the mean mean everyone becomes average?
No—it means extreme measurements tend toward average on subsequent measurements, but this doesn’t eliminate true differences in ability or quality. A genuinely exceptional student who usually scores ninety-five percent might score ninety-eight percent one test (extreme high that will regress) but regress to ninety-five percent (still well above average), not to seventy percent (the class average). Regression to the mean affects measurement fluctuations, not true underlying differences.
If extremes regress to the mean, how do some people stay at the top consistently?
True exceptional ability creates consistently above-average performance with fluctuations around a high average. A truly great athlete has a high performance average (say eighty points) and fluctuates between seventy and ninety. When they score ninety (extreme), they regress to around eighty (their average, which is still exceptional). Regressive bias makes observers expect ninety to persist; regression makes it return to eighty; but eighty is still far above the overall population average.
Why don’t professional forecasters account for regression to the mean?
Many don’t understand it, even professionals. Those who do understand often face incentives to make exciting extreme predictions rather than boring moderate ones. Predicting “this extreme performance will continue!” generates more attention than “this will probably regress toward average.” Additionally, regressive bias is intuitive (extremes feel stable) while regression to the mean is counterintuitive (requiring statistical thinking that doesn’t come naturally).
Can regression to the mean be prevented or overcome?
No—it’s a mathematical necessity, not something that can be prevented. If measurement includes any variable component (luck, circumstances, randomness), extreme values will regress toward the mean on subsequent measurements. You can’t prevent it, only understand and expect it. The key is recognizing that regression doesn’t mean decline for high performers or improvement for low performers—it means their next measurement will likely be closer to their true average than their extreme measurement was.
How can I tell if improvement or decline is real versus just regression to the mean?
Look at trends over multiple measurements, not changes from one to the next. Regression to the mean affects single measurement pairs (one extreme measurement followed by regression). True change shows sustained movement in one direction across many measurements. If a student scored thirty percent, then sixty percent, then seventy percent, then seventy-five percent over four tests, that’s probably real improvement. If they scored thirty percent then sixty percent then fifty percent, the sixty percent was likely an extreme fluctuation that regressed, not sustained improvement.
Observer Voice is the one stop site for National, International news, Sports, Editor’s Choice, Art/culture contents, Quotes and much more. We also cover historical contents. Historical contents includes World History, Indian History, and what happened today. The website also covers Entertainment across the India and World.
