What the research says

Should You Change Your First Answer?

“Go with your gut” is old exam advice. On real exams, changed answers go from wrong to right about twice as often as the reverse. A 2005 study explains why the myth survives.

By the Giga Quizzes team4 min read
Two pencils, one blue and one patterned green, on a pale blue surface
Photo: Edz Norton, Unsplash, Unsplash License, cropped.
In this post
  1. 1What real exams show
  2. 2Why the myth feels true
  3. 3Testing the memory explanation
  4. 4What the authors do not claim
  5. 5The limits
  6. 6What to tell students

Ask a room of students whether they should change an answer on a multiple-choice test, and most will say no: your first instinct is usually right, and second-guessing gets you into trouble. Some test-preparation guides say the same, and so do many teachers. The research has said the opposite for decades. In 2005 Justin Kruger, Derrick Wirtz and Dale Miller set out to explain that gap: if changing answers usually helps, why is almost everyone convinced it hurts?

What real exams show

The researchers began with the evidence on actual tests. Independent coders examined the multiple-choice midterm exams of 1,561 introductory psychology students at the University of Illinois, looking for erasures: an answer marked, rubbed out and replaced.

Kind of changeShare of the 3,291 changes
Wrong to right51%
Right to wrong25%
Wrong to a different wrong23%
Kruger, Wirtz & Miller (2005), as reported by NYU Stern

Changes were about twice as likely to gain a mark as to lose one. This was not a surprise to people who study testing. A review of 33 studies across seven decades, cited in the same work, found none in which test-takers did worse overall by changing answers. Yet around three in four students believe changing will lower their score, and in a survey of faculty at Texas A&M University, 55% thought it would, against 16% who thought it would help.

Why the myth feels true

Kruger and his colleagues argue that the belief comes from how regret works. Imagine two students, both of whom got a question wrong.

  • The first had the right answer, changed it, and lost the mark.
  • The second had a wrong answer, thought about changing to the right one, and did not.

Both lost exactly one mark. But the first feels much worse. Changing was an action, and the loss feels caused by it: if only I had left it alone. Sticking was a non-action, and that loss feels like bad luck. In one experiment, students imagining these two situations judged the first as more regrettable, more foolish and more avoidable.

Painful moments are memorable moments. So the times that changing an answer went wrong stick in the mind, while the more numerous times it went right, and the times that stubbornly sticking went wrong, fade. Over years of tests, each student builds up a vivid, personal and misleading collection of evidence.

Testing the memory explanation

Two further experiments checked this directly. In one, participants worked through 30 minutes of difficult SAT or GRE questions, and whenever they were torn between two answers they marked both and noted which was their first instinct. Those who stuck with their first instinct got more of these questions wrong than those who switched. Yet afterwards, they remembered it the other way round: sticking had gone relatively well, switching relatively badly.

In the other, participants watched a mock game show in which a contestant either always stuck or always switched, with the outcomes arranged so that both strategies won and lost the same number of times. Viewers still remembered sticking as more successful and rated it more favourably. Switching produced more frustration when it failed, and the frustration distorted the memory.

What the authors do not claim

None of this means every doubt should be acted on. The authors are careful: they do not argue that people should always second-guess themselves, or that first instincts are always wrong. The finding is narrower. When a student, after more thought, has a specific reason to prefer a different answer, the evidence says that change is more likely to help than hurt. Changing for no reason, or changing back and forth, is not what the studies measured.

The limits

The exam data come from paper tests where changes could be seen as erasures. Changes made in the head before writing anything never show up, and nor do students who considered changing and did not. The participants were mostly university students. And the advantage depends on what prompts the change: a second reading of the question, or remembering something from the lesson, is a good reason; anxiety alone may not be.

What to tell students

The old advice can be replaced with something closer to the evidence.

A better rule for answer changes

  • If you have a real reason to change, such as rereading the question or remembering a fact, change it.
  • If you are only feeling nervous, leave it.
  • Use any spare time to reread questions you answered fast, not ones you agonised over.
  • Remember that the changes that went wrong are the ones you will remember. The ones that went right count just as much.

This matters most in formal exams, where students have time to go back. In a timed live quiz, answers are usually locked in, and the question does not arise. But going through the results afterwards is a good moment to talk about it: a student who switched to the right answer at the last second has just experienced the change that research says is the common one. For a related effect about confident mistakes, see the hypercorrection effect.

Sources

  1. Kruger, J., Wirtz, D. & Miller, D. T. (2005). Counterfactual thinking and the first instinct fallacy. Journal of Personality and Social Psychology, 88(5), 725–735.
  2. NYU Stern School of Business (2008). Think again (on the research of Kruger, Wirtz and Miller). Sternbusiness, Fall 2008.

More from the blog

All posts →