The order of questions on a test affects how well students believe they performed

We know that students are more optimistic about their performance after they take a test that progresses from the easiest to hardest questions than after taking one that progresses in the opposite order¹. In fact, these “Easy-Hard” students are more optimistic than “Hard-Easy” students even when the two groups perform equally. The literature explains this question order bias as a result of students’ failing to sufficiently adjust, in the face of new information, their extreme initial impressions about the test. In the first two of six studies, we investigated the possibility that a biased memory for individual questions on the test is an alternative mechanism driving the question order bias. The pattern of results was inconsistent with this mechanism, but fit with the established impression-based mechanism. In the next four studies, we addressed the role that the number of test questions plays in determining the size of the question order bias, discovered that warning students is only a partially effective method for reducing the bias, and established a more precise estimate of the bias’ size. Taken together, this work provides evidence that the question order bias is a robust phenomenon, likely driven by insufficient adjustment from extreme initial impressions.  ¹ Although the research in this thesis is my own, I conducted it in a lab and supervised a team comprised of research assistants and honours students. I also received advice and direction from my supervisors. Therefore, I often use the word “we” in this thesis to reflect these facts.