According to conventional wisdom, a test taker should not change his/her first response to a multiple-choice, although empirical evidence has consistently supported such changes. Quizzes for masters level students in educational measurement and evaluation showed increments due to answer changing. Low anxious students tended to make more changes and to gain from the changes than did high anxious students. Relationships between test-taking behavior and other cognitive style, attitudinal and academic performance variables are also discussed. (Author/MV)