Who we are Search Resources Submit a resource Links to sites Discussion Board Contact Us Return to Home
Multiculturalism and Social Work | San Francisco State University

Exploring gender differences on the NELS: 88 History achievement tests.

Author: Le, Vi Nhuan
Author Background: Stanford U., US
Date 7/2000
Type Dissertation
Journal Title: Dissertation-Abstracts-International:-Section-B:-The-Sciences-and-Engineering
Volume/Pages Vol 61(1-A): 83
Subject Matter Motivation; Standardized Tests, Hispanics, Gender, Academic Achievement
Abstract Gender differences on standardized measures of achievement often prompt questions concerning whether those disparities are due to actual achievement differences, bias in the test, or some combination of both. This dissertation investigated some possible causes underlying the male-female gap found on the National Education Longitudinal Study of 1988 (NELS:88) multiple-choice history items. It examined the attributes of items that showed differential performance between boys and girls, and explored the extent to which differences in motivation and educational experiences accounted for the discrepancy. Attention was also given to ethnic variations in the magnitude of the gender gap. The present study used a range of methodological approaches. Descriptive statistics and differential item functioning procedures identified items that manifested unusually large discrepancies. Relationships among test scores, student characteristics, and educational background were investigated through regression models. Finally, a small-scale survey provided additional information pertaining to male-female differences in motivation, sources of knowledge, and interest in particular historical content areas. Analysis of the NELS:88 history exam revealed that the content could be divided into two subtests reflecting stereotypically masculine and feminine interests. Both the descriptive statistics and the DIF analyses suggested that the male advantage was limited to masculine items, and did not represent a difference in overall proficiency. Hispanic boys, in particular, performed significantly better than Hispanic girls, but Black boys and girls demonstrated nearly comparable outcomes. The small-scale survey study indicated that performance differences might be linked to variations in interest and sources of knowledge. On masculine items pertaining to wars and politics, boys were more likely than girls to express greater interest, and to report having acquired knowledge from television, newspapers, magazines, and the Internet. The findings have several implications. Test developers should be aware that the kinds of items included on an assessment can have substantial influence on the extent to which gender differences are manifested. Researchers may also want to consider other criteria for achievement besides total test score, as more refined outcome measures, such as subtest scores, may reveal the relationships among student characteristics, educational variables, and performance in greater detail. (PsycINFO Database Record (c) 2000 APA, all rights reserved)