Different Question Types Can Impact Students' Exam Predictions

Instructors can choose to use several different types of questions on exams, some more open-ended than others (e.g., true-false, multiple choice, fill in the blank, short and long answers, etc.). Students’ ability to predict or analyze their performance on exams can vary depending upon which format of questions are used. In this study, the researcher investigated how true-false, multiple-choice, and fill-in-the-blank questions affect students’ ability to predict their exam scores. The researcher applied Koriat’s (1997) idea that students use different clues from each question type to make these predictions. The study also considered the ease of predicting and postdicting correctly for each type of question. The goal was to find out which question types helped students make the most accurate predictions about their exam performance. 

The researcher asked students in an undergraduate psychology course to predict and postdict their performance on three types of questions—true-false, multiple-choice, and fill-in-the-blank—across three exams. Students made predictions about their performance before taking the exams and postdicted their performance afterward. The study measured how accurate these predictions and reviews were by using two different methods: bias scores (measuring the difference between what students predicted their scores will be and their actual scores) and gamma correlations (measuring how well students' predictions and reviews matched their actual performance, showing the strength and direction of this relationship). 

The accuracy of students’ predictions of their exam scores varied across different question types. When looking at bias scores, students’ predictions and reviews were most accurate for multiple-choice questions. Students predicted the lowest scores for fill-in-the-blank questions. However, when using gamma correlations, students’ predictions remained most accurate for multiple-choice questions, but their reviews were most accurate for fill-in-the-blank questions. The results indicate that different question types can affect how accurately students are able to predict and postdict their exam performance, at least in relation to course content.