Carnegie Mellon University

Eberly Center

Teaching Excellence & Educational Innovation

Eberly Center for Teaching Excellence & Educational Innovation

Is Predicting and Explaining Better Than Answering Practice Questions?

Predicting and explaining improves transfer of learning relative to answering practice questions.

Introductory Social Psychology

Students in introductory psychology courses need to learn about various psychological mechanisms and the evidence supporting them. It is not clear whether answering practice questions after reading about a research study or predicting outcomes as they are reading is a more effective active learning strategy.

Carvalho and colleagues conducted a study in which 75 students in a social psychology course completed homework assignments under one of two conditions. In the Practice Questions condition, students first read a summary of social psychology studies and then answered questions about them. In the Predict-Explain-Observe-Explain (PEOE) condition, instructors provided the research question and experimental design for a study and asked students to predict and explain the outcome of the study, before seeing the data. Next, students in the PEOE condition observed the actual results and refined their explanations of the findings. All students experienced two homework assignments under the Practice Questions condition and two homework assignments under the PEOE condition. To control for topic difficulty, the order of the conditions was counterbalanced across students.

On exams, students performed better on topics practiced using the PEOE method than the Practice Questions method. This is especially encouraging given that students completed the PEOE tasks quicker than the Practice Questions tasks. This team is now investigating how different variations of PEOE affect student learning.

Carvalho Graph

Students scored higher on exam questions for topics learned with the PEOE method (M = 90.80, SD = 10.01) compared to those learned using the Practice Questions method (M = 87.78, SD = 9.71). Error bars are the 95% confidence intervals for the means. The mean difference of 3.02 was significant, 95% CI [5.29, 0.76], t (74) = 2.66, p < .05, d = .30.


Dependent Variable: learning
Pedagogical Strategy: active learning
Study Design: category 3