Using data from 10 states, Hyde, Lindberg, Linn, Ellis, and Williams (2008) found gender similarities in performance on standardized math tests. The present study attempted to replicate this finding with national data and to extend it by examining whether gender similarities in math performance are moderated by race, socioeconomic status, or math ability in Grades 8, 10, and 12. Data from 9,813 students from the National Education Longitudinal Study (NELS) were used. The current results showed gender similarities in performance on standardized math tests across the three grades examined. Furthermore, this finding was not modified by race, socioeconomic status, or math level. These results provide empirical evidence for teachers and parents to encourage girls to persist and excel in math.
Stereotypes that girls lack math ability are widely held by teachers, parents, and even girls themselves (Cavanagh, 2008). These negative stereotypes, however, are contrary to reality. In a cross-sectional analysis of scores on standardized math tests for Grades 2 through 11 from 10 states (California, Connecticut, Indiana, Kentucky, Minnesota, Missouri, New Jersey, New Mexico, West Virginia, and Wyoming), Hyde, Lindberg, Linn, Ellis, and Williams (2008) found a weighted mean effect size of 0.0065, which indicates no gender difference in math performance.
The current study built on Hyde et al.'s work by examining whether gender similarities in math performance would be replicated in a national, longitudinal sample of students and whether gender similarities in math performance would be moderated (i.e., modified) by race, socioeconomic status (SES), or math level from middle school through high school. We chose these potential moderators because prior research has found that they are related to math performance (Halpern, Benbow, Geary, Gur, Hyde, & Gernsbachers, 2007; Yang, 2003). In addition, we focused on middle school through high school because it is at this stage that students' conceptions and attitudes toward career development influence the kinds of courses that they enroll in. If students do not persist and excel in math courses from middle school through high school, then they forego the opportunity to pursue careers that require strong math skills.
In sum, we expected to replicate Hyde et al.'s (2008) finding of gender similarities in math performance on standardized math tests because the 10 states in their study appear to be representative of the United States. We, however, made no specific predictions of whether gender similarities in math performance would be moderated by race, SES, or math-level because this part of our analysis was exploratory.
We used data from the National Education Longitudinal Study (NELS), which began in 1988 when it assessed a sample of students (n= 24,599) in Grade 8 from schools across the United States. Then NELS conducted follow-up assessments of these participants in 1990 (Grade 10), 1992 (Grade 12), 1994 (two years after high school), and 2000 (eight years after high school). NELS has data on various topics related to these participants' education (for details, see National Center for Education Statistics, 2002). For the purpose of the current analyses, we focused on the students' scores on standardized math tests in Grades 8, 10, and 12.
We selected only students with complete data for the variables under investigation. Of the 9,813 students selected, 51.0% were female. The racial distribution included 1.2% American Indian or Alaskan Native; 6.1% Asian or Pacific Islander; 11.2% Black, not Hispanic; 68.8% White, not Hispanic; and 12.8% Hispanic. The mean age of the selected students in Grade 8 was 14.38 years (SD = .59).
Standardized Math Tests
Test items clustered by level of difficulty (see Rock & Pollack, 1995). Math Level 1 tested proficiency in performing simple arithmetical operations on whole numbers. …