Contrary to popular belief, whether you’re taking a test
or buying a stock, it may pay to rethink your choice.
By Derrick Wirtz, Justin Kruger, Dale T. Miller, and Pragya Mathur
ILLUSTRATION BY CHRISTOPHE VORLET
hether you’re an individual trying to build a college fund for your first-born, a hedge fund manager responsible for billions of dollars, or just a visitor to Las Vegas, at some point you’re going to wonder whether you should do a tad more research or just plunk your chip down on the red because, well, it feels right.
When it comes to decision-making dicta, going with one’s first instinct fits somewhere on a continuum from useful rule of thumb to guiding principle of life. A leader who “trusts his gut” is often admired for his self-confidence and decisiveness. That goes for CEOs, athletes, even presidents.
The common wisdom of going with one’s gut is parroted in a diverse range of situations, from those with potentially disastrous consequences – say, going to war – to the life-changing – for instance, college entrance exams – to the perfectly mundane – having to choose a checkout line at the local supermarket.
Second-guessing is considered an also-ran in many spheres. Psychologists have persistently linked self-doubt to a variety of negative outcomes, including diminished attention and self-esteem, excessive reassurance seeking, greater conformity under pressure, greater materialism, and poorer performance in competitive situations. Does confidence in oneself, then, mark the path to optimal decision-making? Are first instincts truly more accurate, insightful, or more likely to lead to optimal decision-making? Or is it possible that first instincts are often false instincts? We set out to answer these questions and to understand how the common wisdom got that way.
Multiple-choice test-taking offers a paradigm in which to examine the effect of second-guessing on objective outcomes. Among the most commonly endorsed strategies for such situations is, of course, to “go with your first instinct,” which appears as a staple in the test-preparation industry. As Barron’s GRE guide admonishes: “Exercise great caution if you decide to change an answer. Experience indicates that many students who change answers change to the wrong answer.”
Indeed, research shows that roughly three out of four students believe that changing their answers will result in lower test scores; instructors appear to hold this belief as well. In a survey of faculty at Texas A&M University, 55 percent responded that changing one’s first instinct to another answer was likely to lower test scores; whereas, a mere 16 percent thought that such a change would improve test scores.
The facts are to the contrary. The vast majority of the evidence on answer changing, we found, pointed to just the opposite: Answer changing actually improves test scores. In research spanning seven decades, most answer changes are from the wrong answer to the right answer, taking into consideration a wide variety of test formats (multiple-choice, true-false, achievement, aptitude, timed and untimed, computer-based, and paper-and-pencil). In a comprehensive review of the answer-changing literature, not one of 33 studies found that test takers fared worse because of changing their answers.
"Answer changes were twice as likely to result in correct answers as incorrect ones."
If sticking with one’s first instinct is such a demonstrably suboptimal strategy – we call it the “first-instinct fallacy” – why is it so widely accepted? Our hypothesis was that it stems from the regret produced when a test taker chooses an erroneous answer. As an analogy, consider the seemingly universal experience of changing checkout lines in the supermarket only to see one’s original line speed ahead. People seem to remember this frustrating experience far more often than they remember having failed to change lines when they should have. In much the same way, we argue that changing a correct answer to an incorrect answer is likely to be more frustrating and memorable than failing to change an incorrect answer to the correct one. When a test taker thinks about the consequences of past answer changing, he is most likely to remember those instances that produced the greatest frustration and self-recrimination, not the instances where he changed an answer appropriately or failed to change what turned out to be a wrong answer.
Setting the Stage
To evaluate this theory, we constructed four steps. First, we asked test takers about their belief in the first-instinct fallacy, and then compared their intuitions with objective outcomes by scrutinizing their actual introductory psychology midterm exams. Second, we examined whether the regret and frustration resulting from changing a correct first instinct to an incorrect second guess was greater than that produced by failing to change an incorrect first instinct to a correct alternative. Third, we investigated whether this bias in regret produced a similar memory bias. Finally, we tested the full model to see whether the differing levels of regret and how the situation was remembered affected the relationship between answer changing and belief in the first-instinct fallacy.
In our first step, independent coders scrutinized the multiple choice midterm exams of 1,561 introductory psychology students at the University of Illinois for evidence of answer changing. Of the 3,291 answer changes on the exams, 51 percent were from wrong to right, 25 percent from right to wrong, and 23 percent from wrong to wrong. In other words, answer changes were twice as likely to result in correct answers as incorrect ones.
id the very students who benefited from changing their answers realize this fact? Clearly not. Test takers’ beliefs about the outcome of answer changing were consistent with the first-instinct fallacy. They believed that the most likely outcome of answer changing was an incorrect response, that answer changing would overall hurt test takers, and that in situations of uncertainty between two answers, the first answer was likely to be correct. In reality, the outcome of answer changing was exactly the opposite: the most likely consequence was a correct answer; more people were helped than hurt by answer changing; and the willingness to doubt one’s first instinct, when another answer seems better, leads to objectively better results.
Je Ne Regrette Rien
Why does belief in the veracity of first instincts persist in the face of considerable empirical evidence to the contrary? We hypothesized that the regret and frustration that follows from switching a correct first instinct to an incorrect alternative (an action) is more intense than the regret and frustration that follows from not switching an incorrect first instinct to a correct alternative (an inaction), as prior research has demonstrated that actions are more easily imagined than inactions.
To find out whether switching a correct first instinct to an incorrect second guess is more memorable than failing to switch an incorrect first instinct to a correct second guess, we asked a group of undergraduate participants to imagine a scenario involving a very important multiple-choice exam. The results confirmed our prediction. By a wide margin, participants clearly viewed changing a correct first instinct to an incorrect alternative as a more regrettable, foolish, and avoidable error than the (objectively) equally poor decision to stick with a first instinct when one’s second guess would have been right.
Our model posits that belief in the first-instinct fallacy stems not from anticipated (or experienced) regret directly, but rather from salient memories of regrettable and frustrating actions. To examine the actual and remembered outcomes of going with one’s first instinct versus switching to a second guess, we recreated a typical test-taking situation: a group of participants was given a 30-minute portion of the Scholastic Aptitude Test (SAT) or Graduate Record Exam (GRE), in which they attempted to answer a series of difficult multiple-choice questions. They were told that if they were unsure of the correct answer to a question, they could select two answers. In such instances, participants were then asked to indicate which of the two selected answers was based on their first instinct.
Our initial analysis focused on the actual outcomes of sticking with one’s first instinct in these instances. As expected, we found that participants were more likely to answer a question incorrectly if they stuck with their first instinct than if they switched answers. Next, we compared the actual outcomes of sticking versus switching with the outcomes participants remembered.
Overall, test takers remembered the outcome of sticking with their first instincts as relatively positive and the outcome of switching answers as relatively negative. Whereas the actual outcome of the exam was such that second guesses led to fewer wrong answers than first instincts, participants remembered exactly the opposite.
When second-guessing oneself turns out poorly, the outcome is both regrettable and memorable. Is it the case, as our model proposes, that the resulting disproportionate regret and memory account for faith in the first-instinct fallacy? To find out, we created a mock game show patterned after television’s “Who Wants to Be a Millionaire?”
“Overall, test takers remembered the outcome of sticking with their first instincts as relatively positive and the outcome of switching answers as relatively negative. Whereas the actual outcome of the exam was such that second guesses led to fewer wrong answers than first instincts, participants remembered exactly the opposite.”
We found that participants were more frustrated when the contestant repeatedly switched from a correct first instinct to an incorrect second guess than when the contestant repeatedly stuck with an incorrect first instinct.
Confirming our hypothesis, participants remembered the contestant as having been helped more by pursuing a strategy of sticking with his or her first instinct than by always second-guessing that instinct, even though we designed the experiment so that both strategies led to exactly the same number of right and wrong answers. Also, participants in the “stick” condition rated the contestant’s strategy more favorably than participants in the “switch” condition – again, despite equivalent outcomes.
Our data enabled us also to determine that going against one’s first instinct (versus sticking with it) led to greater frustration; that frustration, in turn, led to a memory bias for the negative consequences of switching; and that the memory bias reinforced participants’ belief in “going with your first instinct” as an effective test-taking strategy. When the contestant pursued a strategy of switching answers, greater frustration resulted. In addition, the more frustrated participants felt, the more they remembered the contestant’s strategy as hurting, versus helping. Of critical interest, the more participants remembered the contestant’s strategy as hurting them, for example, the more poorly they evaluated the strategy overall and confirmed their belief in the first-instinct fallacy.
A Pinch of Self-Doubt
As might be expected, the reticence to go against one’s first instinct is not unique to test-taking. For instance, in follow-up research, we have observed a similar tendency among investors. The effect of self-doubt, in general, on a variety of behaviors has been widely studied. Social psychological literature explores the fact that personal uncertainty is not only psychologically uncomfortable and must be managed or resolved, but also leads to such maladaptive behavior as an inability to act.
Our research on the first-instinct fallacy seems to suggest that contrary to conventional wisdom, self-doubt may, in some circumstances, be worth listening to. If more test takers or stock choosers were willing to second-guess their first instincts, our research makes clear that the outcomes of their choices would likely be objectively better. The hypothesis that individuals prone to doubt themselves would be more willing to second-guess a first instinct is intuitively appealing, but whether chronic high self-doubters would be more likely to attain objectively better outcomes than low self-doubters remains an open question.
It is not our contention that one should always second-guess oneself or that first instincts are always wrong. In many cases, a first instinct is correct – such as when one simply knows the answer to a test question. Rather, we advocate the careful weighing of decisional alternatives, accompanied by a strategy of revising one’s first instinct when one comes to doubt it in favor of a better alternative. This strategy is not one of generalized self-doubt, but of the willingness to consider that an option that one suspects may be correct on second, third, or fourth look is no less likely to be accurate just because it came after an initial decision.
DERRICK WIRTZ is assistant professor of psychology at East Carolina University. JUSTIN KRUGER is associate professor of marketing at NYU Stern. DALE T. MILLER is the Morgridge Professor of Organizational Behavior at the Stanford Graduate School of Business, professor of psychology at Stanford University, co-director of the Strategic Uses of Information Technology Executive Program, and director of the Center for Social Innovation. PRAGYA MATHUR graduated in May 2008 with a PhD in marketing from NYU Stern.
This article is forthcoming in the book, The Uncertain Self: A Handbook of Perspectives from Social and Personality Psychology.