Academic journal article College and University

Mathematics Placement Tests and Gender Bias

Academic journal article College and University

Mathematics Placement Tests and Gender Bias

Article excerpt


This study examined whether a mathematics placement system accurately predicted success in mathematics classes for both genders at a medium-sized, independent Northwest university using data on 1,388 entering students. The placement system used four variables to

predict grades a student would receive if placed in various freshman mathematics classes.

It has been widely reported that the Scholastic Assessment Test in Mathematics (sAT-M) is a poor predictor of women's success in college (Linn and Kessel 1996; Stricker, Rock and Burton 1993; Wainer and Steinberg 1992). Leonard and Jiang (1995) found that a "small" gender difference in predicted grades for Berkeley freshmen admitted between 1986 and 1988 resulted in excluding 200 to 300 women per year who would have been admitted to its freshman classes if unbiased predictors had been used. The Admissions Department at Massachusetts Institute of Technology (MIT) found that it was admitting more men to the school but that women were earning higher grades. To remedy the situation, MIT began to admit women with scores below the previous cut-off of 750 on the SAT-M. As a result more women were admitted and the gap between grades for men and women decreased (Johnson 1993). At both MIT and Berkeley, the predictor variable that caused biased admissions was the SAT-M.

Several studies have found that the sAT-M combined with entrance examination scores, placement tests, and/or information from the high school record is a better predictor of success in college mathematics courses than the SAT-M alone (Bridgeman and Lewis 1996; Bridgeman and Wendler 1989, iggi; Odell and Schumacher 1995). The mathematics department at the Northwest university where the current study was performed includes the SAT-M as one of four variables in its predictor equations used for placement of students in their first freshman-level mathematics course.

During 1997-98, we researched whether the gender bias found in SAT-M scores was effectively neutralized by the three other variables in the university's predictor equations. Bias in this four variable system would indicate potential bias at other colleges that use similar methods for mathematics placement.

There were three objectives for this study: First, determine whether the mathematics placement system was gender biased in the sense that the predictor equations underpredicted or overpredicted actual grade averages for either gender; second, determine if the SAT-M variable was the cause of the bias if it existed; and third, determine the effect of the other three predictor variables in the over/under prediction of actual grade average.

Mathematics Placement System

The purpose of the mathematics placement system is to determine the freshman mathematics courses in which a student is likely to be successful with an appropriate amount of hard work. Prior to registering for the first freshman mathematics course, the student must complete a survey and take the mathematics placement test. The first part of the instrument is a survey where students report the most advanced mathematics class they have taken and the grade they received in that course. The second part of the instrument is a multiple choice, computational mathematics placement test designed by the university's mathematics department. The placement test is scored differently for each of four freshman mathematics courses. The scoring is based on statistical findings concerning the test questions and student performance. That is, those questions which are useful in prediction of successful performance in a course receive a high number of points for that course and those questions which are not useful in prediction of successful performance receive low or zero point weight for that course. The three variables-past highest mathematics course, grade, and placement test score for a particular course-are used together with the SAT-M to predict success in each of four freshman-level courses. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.