Locating, restoring and recompiling my digital records and documents, following crashes of two of my external hard disks, I have come across many documents that I hadn’t viewed in a long time. Among them are the papers I wrote for different classes as an undergraduate student at Grinnell College in the United States of America between 1990 and 1994. I have decided to reproduce some of them here in my personal blog.
This first paper I am publishing is one I wrote for the Statistics class. It’s about how the Scholastic Aptitude Test (SAT) is biased against females.
* * * * * * * *
Women and SAT
What seems to be an innocuous aptitude test, the Scholastic Aptitude Test (SAT), many studies have found to be biased against women (Rosser, 1989; Loewen, Rosser and Katzman, 1988). Every year 1.5 million students take the test to fulfill the requirements for enrollment at nearly 1500 four-year colleges in the US (Rosser, 1989). About 52 percent of the 1.5 million are women. From the very first time it was administered in 1926, men have always obtained higher grades. Women had a slight edge on men in the verbal section, scoring slightly over men by 5 points, until 1972 after which, following changes in the contents of the section, men have outscored women (Rosser 1989). One can imagine some of the damaging consequences of administering such a test. With half the test takers placed at a disadvantage and so many of the educational institutions basing their admission decisions on its score, many have begun asking if the test is serving its purpose at all.
The SAT is scored on a scale of 200-800 in each of the two sections: verbal and math. The verbal section is made up of 85 items each worth 7 score points and the math section 60 items each worth 9.5 score points (Loewen, Rosser and Katzman, 1988). As the name suggests, SAT is supposed to measure the aptitude of the test taker and predict how well the test taker will do at college. No wonder many colleges use the SAT score as a measure of academic competency of an applicant.
An extensive study report published by the American Association of University Women (AAUW) on “How Schools Shortchange Girls” cite SAT as being biased (AAUW Report, 1992). In the report, specific items in the test that bias the test takers on the basis of gender are identified. The report also touches on the effect of SAT results on the educational opportunities of women. What constitutes gender bias in SAT and how is it assessed? What are the implications of a gender biased test, specially SAT? The paper will attempt to answer these and other questions.
To begin with, what is bias in testing? Slightly different definitions of bias have been used in different studies. One that is very succinct is offered by Diamond and Tittle (1984): “Bias is considered to occur when two individuals of equal ability but from different groups respond differently to a test item and therefore do not have the same probability of success.” Studies of SAT have established SAT to be biased against women. A significant difference in performance of the two sexes have been found.
Assessments (Rosser, 1989; Selkow, 1984; Diamond and Tittle, 1984) of SAT focused mainly on the content of the test and the use of the result. To begin with, until 1972, women had scored higher in the verbal section by about 5 points (Rosser 1989) while males had higher total scores. Following systematic changes “to better balance the scores in between the sexes,” (Donlon and Angoff, 1971, pp. 25-26) after 1972, men have overtaken women even in the verbal section. Over the same period, no attempts had been made to balance the difference in math scores (Rosser, 1989).
One of the content factors many scholars point to as contributing to bias is the imbalance in the reference of the two sexes in the test: males are found to be referred to far more than females (Rosser, 1989). Where references were made, the items depicted the people in their stereotypical roles with men pursuing a career outside home and women working at home looking after the house and children. It is worth noting however, that studies showing no relationship between performance and extent of reference and stereotyping of people are plenty to be found (Diamond and Tittle 1984). Just the same, it has been pointed out that even though such depictions do not affect test scores, they serve to “[perpetuate] cultural bias against females” (Rosser, 1989). Yet in some studies women have been found to perform better when female or “neutral” people are depicted (Selkow, 1984). Still in others “both sexes found feminine-oriented…[question] equally difficult” (Graf and Riddell, 1972).
Assessment of specific SAT items can also be carried out by studying the difference in performance of the two sexes on the particular item (Diamond and Tittle, 1984). Loewen, Rosser and Katzman (1988) used 10 percent difference between the number of men and women that gave the correct answer to an item as an indication of significant difference arising from bias. In a study of 1112 students (Loewen, Rosser and Katzman, 1988) who were given SAT, the verbal scores of the two sexes were found to be about equal. In math however, men outscored women by 3.5 items (35 points). Nationally, men have been found to score higher by 10 points in the verbal section and 47 in math. An item by item examination of the test revealed significant differences between the two sexes. Between the two sexes, 7 items in the verbal section and 10 on math had greater than 10 percent difference in the number of correct answers given. In the same study, an additional 5 percent difference on 22 verbal and 16 math items were found. The relevance of the content of the item to either one of the sexes appeared to have affected how the two sexes performed on a given item. However, the degree of difficulty of an item did not cause performance differences (Loewen, Rosser and Katzman, 1988).
Studies (Rosser, 1989) have revealed bias in the use of the SAT scores. It does not predict first year college grades for women, as it purports to, and, in addition, under-predicts their high school GPA (Loewen, Rosser and Katzman, 1988). Despite that, SAT is still used extensively in admission decisions. At college, women have been found to perform as well or better than their male counterparts in all subjects. If the SAT scores were a good predictor of grades then women would be doing a lot poorly at college. Also, high school grades predicted using the SAT score are a lot smaller than they actually are. Evidently, SAT underscores the capability of the women test takers.
The women’s lower scores in SAT has meant their being left out of many merit based scholarships (Rosser, 1989). Many state and private merit based scholarships, like The National Merit Scholarship Corporation, are awarded on the basis of SAT scores. The National Merit Scholarship Corporation, select eligible candidates from PSAT scores (Rosser, 1989). The biased nature of the test resulting in lower score for women jeopardizes their eligibility.
Works cited:
P. Rosser, The SAT Gender Gap, Center for Women’s Policy Studies, Washington, DC 1989, pp 3-5, 21, 29-35.
Diamond, E.; Tittle, C. “Sex Equity in Testing,” in Handbook for Achieving Sex Equity Through Education, S. Klein, ed., (Baltimore, MD: John Hopkins, 1985), p. 167-173.
Selkow, P. “Assessing Sex Bias in Testing” (New York: Greenwood Press, 1984) pp. 3-6, 25-30.
Lowen, J.; Rosser, P.; Katzman, J. “Gender Bias in SAT Items,” paper presented at the annual meeting of the American Educational Reseacrh Association, New Orleans, LA, 1988. pp. 1-16.
The AAUW Report: How Schools Shortchange Girls, American Association if University Women Educational Foundation and National Education Foundation 1992, pp. 54-7.
Donlon, T. F., Angoff, W.H. 1971. “The Scholastic Aptitude Test.” The College Board Admissions testing Program, William H. Angoff, Ed., New York: College Entrance Examination Board, pp. 15-47.
