75 years of service to our nation

The Correlation Between Gender and Confidence in STEM: A Closer Look

By Dr. Daniel M. Leeds, Research Analyst

A recent IES newsflash addressed gender-based differences in "how … students perceive their math and science abilities as they move through high school" (National Center for Education Statistics, 2006). The report contains several findings that support conventional wisdom on gender differences in STEM experiences. In particular, the number of female students who were confident in their ability to do an excellent job on math assignments fell between grades 9 and 12 (though the corresponding number for science assignments rose) and the percentage of female students who thought that others considered them "math people" dropped by a larger amount and from a lower baseline than the percentage of male students.

Although these results are interesting and useful to researchers and practitioners, they also raise several questions. For example, how is students’ confidence in their ability to do an excellent job on assignments related to the course from which those assignments are given? The data used for the IES newsflash (the High School Longitudinal Study of 2009, along with its 2012 follow-up survey), allow us to address these questions.1

Students’ confidence in their ability to perform excellently on assignments depended heavily on the courses they took. For each course, the percentage of male students who agreed or strongly agreed that they could do an excellent job on assignments was subtracted from the percentage of female students. Positive numbers imply that female students were more confident on average than male students, while negative numbers suggest the reverse.

In most courses, male students were more likely than female students to be confident in their ability to do an excellent job on assignments, and male students were frequently much more likely to be confident. In only four courses were female students more confident than male students in their ability to perform well on assignments; these were among the least advanced courses listed. Female students were more likely than male students to be confident in their ability to do well on assignments in Pre-Algebra (a difference of 6.72 percentage points); "Other math course" (5.21 percentage points); Algebra 1, 1A, or 1B (1.11 percentage points); and Integrated Math 2 (0.22 percentage points). The courses in which many more male students were more confident tended to be among the most challenging. Four courses had differences greater in magnitude than -10 percentage points: IB Mathematics Standard or Higher Level (-16.46 percentage points), AP Statistics or Probability (-16.10 percentage points), "Other Calculus" (-12.50 percentage points), and AP Calculus AB or BC (-10.16 percentage points).

These results may reflect a gender-based confidence gap with ramifications for postsecondary study, careers, and earnings. Such a gap might be caused by teachers, parents, other students, society as a whole, some other factor, or more than one of these; the proper solution depends heavily on which of these root causes are at work.

Dr. Daniel M. Leeds is a quantitative research analyst with expertise in quasi-experimental research design. His research focuses on student responses to education policies such as financial aid, guaranteed college admission, and state accountability labeling. Daniel holds a Ph.D. in economics from the University of Michigan, with a focus on the economics of education. He also earned an M.A. in economics from the University of Michigan and B.A.s in mathematics and in Greek, Hebrew, and Roman classics from Temple University.


Hanebutt, R., & Christopher, E. (2016). Student self-assessment of math and science ability in high school. Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics. Retrieved from http://nces.ed.gov/pubs2016/2016164.pdf.

The views in this piece represent those of the author alone.


1 Analyses use unweighted data for simplicity; adding sample weights would change results somewhat, but implications would likely stay the same.

Did you like this article? Sign up and we’ll send you more articles like this in our monthly newsletter.