Retrieval practice with short-answer, multiple-choice, and hybrid tests

Abstract

Retrieval practice improves meaningful learning, and the most frequent way of implementing retrieval practice in classrooms is to have students answer questions. In four experiments (N=372) we investigated the effects of different question formats on learning. Students read educational texts and practised retrieval by answering short-answer, multiple-choice, or hybrid questions. In hybrid conditions students first attempted to recall answers in short-answer format, then identified answers in multiple-choice format. We measured learning 1 week later using a final assessment with two types of questions: those that could be answered by recalling information verbatim from the texts and those that required inferences. Practising retrieval in all format conditions enhanced retention, relative to a study-only control condition, on both verbatim and inference questions. However, there were little or no advantages of answering short-answer or hybrid format questions over multiple-choice questions in three experiments. In Experiment 4, when retrieval success was improved under initial short-answer conditions, there was an advantage of answering short-answer or hybrid questions over multiple-choice questions. The results challenge the simple conclusion that short-answer questions always produce the best learning, due to increased retrieval effort or difficulty, and demonstrate the importance of retrieval success for retrieval-based learning activities.

Keywords:

• Retrieval practice
• Testing effect
• Learning
• Question format
• Short-answer
• Multiple-choice

The writing of this paper was supported in part by grants from the National Science Foundation (DUE-0941170 and DRL-1149363) and the Institute of Education Sciences in the US Department of Education (R305A110903). The opinions expressed are those of the authors and do not represent views of the Institute or the US Department of Education. A portion of this project was an undergraduate honours thesis completed by Megan Smith under the instruction of Jeffrey Karpicke. We wish to thank Althea Bauernschmidt for numerous constructive discussions, Emily Boyne, Cathrine Brattain, Kait Cross, Kelli Olifirowicz, Samantha Ostler, Victor Panfil, and Nikita Saoji for help with data collection and scoring, and Philip Grimaldi for technical assistance.

The writing of this paper was supported in part by grants from the National Science Foundation (DUE-0941170 and DRL-1149363) and the Institute of Education Sciences in the US Department of Education (R305A110903). The opinions expressed are those of the authors and do not represent views of the Institute or the US Department of Education. A portion of this project was an undergraduate honours thesis completed by Megan Smith under the instruction of Jeffrey Karpicke. We wish to thank Althea Bauernschmidt for numerous constructive discussions, Emily Boyne, Cathrine Brattain, Kait Cross, Kelli Olifirowicz, Samantha Ostler, Victor Panfil, and Nikita Saoji for help with data collection and scoring, and Philip Grimaldi for technical assistance.