![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
Previous studies have documented a gender gap in the study of economics in Canada, the UK, and the US. One important factor may be women's low expectations about their ability to succeed in economics courses. Women in our sample expect to do less well than men in an introductory microeconomics course, even after controlling for variables relating to family background, academic experience, and mathematics experience. These expectations are partly self-fulfilling, since expected grades have an important and positive effect on class performance. We also find that having taken an economics course in secondary school actually has a negative effect on performance. We observe this negative effect for women and men, but it is more pronounced for women. When we control for both expectations and secondary-school experience with economics, the independent effect of gender is small and insignificant.
ACKNOWLEDGMENTS
We thank Scott Adams, Byron Brown, Nancy Burnett, Kevin McGee, Rowena Pecchenino, Peter Schmidt, Diana Strassmann, Jeffrey Wooldridge, and four anonymous referees for helpful comments and suggestions. In addition, we thank Kelly Funk and Barbara Steidle for help in obtaining data for our research and Omar Ahmad for research assistance. Errors are our responsibility.
Notes
JEL Codes: A0, A14, A22
However, Robb and Robb (Citation1999) find that the gender of the instructor had no effect on the decision of women to take additional economics courses.
In the courses from which our data are taken, all exams consisted exclusively of multiple-choice questions. This format is not necessarily the best pedagogically. However, in view of the logistical difficulties of grading exams for 1,200 students, the decision was made to use multiple-choice exams. The multiple-choice format does have the advantage of lending itself to quantitative analysis.
Brown and Liedholm (Citation2002) compare student performance in introductory microeconomics courses taught in different ways. In a traditional class with live lectures, they find that women do substantially less well than men, and the difference is statistically significant. However, they find that the size of the gender gap is much smaller for online courses or for courses that are a hybrid of the live and virtual formats; here the gender gaps are no longer significant. In this paper, we only consider courses delivered in the traditional format with lectures given in person. Thus, we cannot assess the extent to which our results are an artifact of the format in which the course is delivered. Nevertheless, the results of Brown and Liedholm are interesting, and the issue deserves further investigation. However, we note that Brown and Liedholm find the overall level of achievement to be higher in the live course than in the online or hybrid versions. Thus, in a sense their results show that the online format reduces the gender gap by harming men more than it harms women. We would prefer to look for ways to reduce the gender gap by improving the performance of women, rather than by damaging the performance of men.
It is possible that women are simply aware of penalties relating to the social construction of gender roles, where the primary burden for childcare falls disproportionately on women, or that they have some expectation of discrimination in labor markets, both of which could lower the expected payoff to women from entering various professions. While this is a very interesting possibility, it is not the focus of this paper.
Thus, we cannot assess the role of the gender of the instructor, since there is no variation in the gender of the instructor in our data set. In any event, we are mostly interested in expectations formation and in the effects of expectations, rather than in the effects of the gender of the instructor. Moreover, since men form a substantial majority of college instructors of economics, it is essential to try to understand the determinants of student success, even for women who are taught by men.
The four sections were of approximately equal size. In each of the two semesters, the classes met consecutively on the same days of the week. Both sections in each semester were given the same exams, although different exams were given in the different semesters. We include a dummy variable for semester that is significant at the 10 percent level. In the regressions reported below, we do not include a dummy variable for sections within a semester; if it is included, we find that the section dummy is not statistically significant and the estimated coefficient is close to zero.
Institutional policy at our university allows students to retake up to five classes with no penalties (other than the costs of tuition for additional courses). Thus, students who do poorly in a course can choose to retake the course. In this case, the original course grade is replaced by the grade earned in the retake. In our sample, 57 students were retaking the introductory microeconomics course. Of these, 28 were women (4 percent of all women in the course) and 29 were men (3.8 percent of all men in the course). There was no significant difference between men and women in the likelihood of retaking the course.
Maxwell and Lopus (Citation1994) refer to the tendency of students to over-report their GPA as the “Lake Wobegon Effect,” after the fictional Minnesota town in which all the children are above average. However, in our sample, the extent of the overstatement is fairly small for GPA and ACT scores (see note 9 below). We interpret this as meaning that most students feel constrained to be approximately honest when reporting information on past performance since that information is objectively verifiable. On the other hand, when students report their expected grades in the course, they are very over-optimistic. They can be hopeful about their expected grade without “lying,” since the true grade is not yet known at the time of the survey. In spite of the fact that expected grades are a biased predictor of actual grades, the expected grades are still useful statistically.
On average, all students in our sample (men and women together) over-report their ACT score by 0.42 points, and they over-report their GPA by 0.10 points. The correlation coefficient between actual and reported ACT score is about 0.77, and the correlation coefficient between actual and reported GPA is about 0.91. The small size of the overstatements is partly caused by the fact that a large number of students report with precision. In our sample, 529 students reported their official ACT score exactly, and 193 students reported their university GPA accurately to two decimal places.
We do not know whether men and women are absent for different reasons. Investigation along these lines would be an interesting avenue for future research.
Fisher, Guilfoyle, and Liedholm have complete records on attendance at every class for every student. Thus, they have much more information on attendance than we do. We did not take attendance regularly because of the logistical difficulties of taking attendance in sections with 600 students. Thus, our only information on attendance is whether the student participated in the survey (which indicates attendance on the day the survey was administered) and the student's self-reported attendance habits. We find no statistically significant difference between men and women in the self-reported attendance patterns. (For information on the quantitative magnitudes of the differences in self-reported attendance patterns, see .)
We are missing ACT scores for 106 students. This is because transfer students with over 28 credits are not required to provide an ACT score for admission to the university.
The results in column 1 of are based on a sample of 1,457 students. The performance regressions reported in are based on a sample of 1,462 students. The difference is that we use age as an explanatory variable in our expectations regressions but not in our performance regressions. Five students did not report their age, and these observations are dropped from the regressions in . The observations for students who did not report their age are also dropped from the regressions reported in .
Introductory microeconomics is required for all students who want to major in the College of Business at our university, but it is also required for students in a variety of other disciplines.
Mother's education and father's education are highly correlated. If we enter both of these variables in a regression, the multicollinearity causes both of them to lose significance. However, if we enter either of these education variables by itself, the variable is statistically significant. In this paper, we report regressions in which mother's education is included among the explanatory variables. In regressions using father's education, the coefficients turn out to be fairly similar.
The coefficient is not statistically significant. However, we recognize that an important part of the reason for this difference is that the regressions in column 2 of are based on a smaller sample than the regressions in column 1.
This negative effect could arise in any of several ways. For example, it could result from adverse perceptions of the subject matter itself, from perceptions of the level of difficulty of economics, or from negative feedback received in the secondary-school course. Research to distinguish among these channels would be desirable.
If we replace “Mother's Education” with “Father's Education,” we find that father's education level positively influences the expectations of both men and women. However, in this specification, the effect is not statistically significant, either for men or for women.
Lewis Karstensson and Richard Vedder (Citation1974) provide support for including attitudes toward learning in education production functions.
The results for Specifications 2 and 3 of are both consistent with Esther Redmount (Citation1995), who argues that the gender differences between men and women cannot be captured effectively with a simple binary variable.
In another regression, we also control for whether students had taken introductory macroeconomics in college. (This regression is not reported explicitly in .) We find that students who had taken macroeconomics before microeconomics did nearly one percentage point worse in the class than students who had not taken macroeconomics, all else equal (p < 0.05). (This is in spite of the result from , indicating that expectations were improved by having taken macroeconomics.) One possible explanation is that, at our university, the first few weeks of introductory microeconomics are somewhat similar to the first few weeks of introductory macroeconomics. As a result, some students who have already had the introductory macroeconomics course may do well on the first examination, but they may then be lulled into complacency and do less well later in the semester.