Gender gap on concept inventories in physics: What is consistent, what is inconsistent, and what factors influence the gap?

Abstract

Listen

Translate article

We review the literature on the gender gap on concept inventories in physics. Across studies of the most commonly used mechanics concept inventories, the Force Concept Inventory and Force and Motion Conceptual Evaluation, men's average pretest scores are always higher than women's, and in most cases men's posttest scores are higher as well. The weighted average gender difference on these tests is 13% for pretest scores, 12% for posttest scores, and 6% for normalized gain. This difference is much smaller than the average difference in normalized gain between traditional lecture and interactive engagement (25%), but it is large enough that it could impact the results of studies comparing the effectiveness of different teaching methods. There is sometimes a gender gap on commonly used electricity and magnetism concept inventories, the Brief Electricity and Magnetism Assessment and Conceptual Survey of Electricity and Magnetism, but it is usually much smaller and sometimes is zero or favors women. The weighted average gender difference on these tests is 3.7% for pretest scores, 8.5% for posttest scores, and 6% for normalized gain. There are far fewer studies of the gender gap on electricity and magnetism concept inventories and much more variation in the existing studies. Based on our analysis of 26 published articles comparing the impact of 30 factors that could potentially influence the gender gap, no single factor is sufficient to explain the gap. Several high-profile studies that have claimed to account for or reduce the gender gap have failed to be replicated in subsequent studies, suggesting that isolated claims of explanations of the gender gap should be interpreted with caution. For example, claims that the gender gap could be eliminated through interactive engagement teaching methods or through a ``values affirmation writing exercise'' were not supported by subsequent studies. Suggestions that the gender gap might be reduced by changing the wording of ``male-oriented'' questions or refraining from asking demographic questions before administering the test are not supported by the evidence. Other factors, such as gender differences in background preparation, scores on different kinds of assessment, and splits between how students respond to test questions when answering for themselves or for a ``scientist'' do contribute to a difference between male and female responses, but the size of these differences is smaller than the size of the overall gender gap, suggesting that the gender gap is most likely due to the combination of many small factors rather than any one factor that can easily be modified.