The Impact of Learner Variables on Language Test Performance

Abstract

Research has shown that learner variables such as age, language background, field of study, and educational background have an impact on non-native speakers' scores on language tests. However, does this impact reflect a natural advantage of one group over another, or is it indicative of test favoritism or test bias? This article looks at how the learner variables of age, sex, language background, high school graduation status, and length of stay in the United States affect performance on six language texts: two doze tests, a reading test, a structure test, and two listening tests. A multiple regression approach to the problem, posed by analysis of variance, was taken to determine 1) the amount of variance accounted for by each of the learner variables on all tests combined and on each test individually and 2) the interaction of variables with test type. Such an interaction would be evidence of either test favoritism or test bias. It was found that age, language background, and high school graduation status accounted for small but significant amounts of variance on the tests. Furthermore, there was an interaction between the test and learner variables. Therefore, both test advantage and test favoritism/bias were found. Researchers in ESOL have found that learner characteristics such as age, language background, field of study, sex, and high school graduation status are related to non-native speakers' performance on language tests (Hisama 1980, Ilyin, Spurling, and Seymour 1983). Other sources (Resource Consultants, Inc. 1981) have found that number of years of education correlates positively with performance on listening and reading tests, whereas number of years out of school correlates negatively. Some researchers (see, for example, Pike 1979) have found differences in test scores among different populations but have not accounted for such differences. Others (Krashen, Sferlazza, Feldman, and Fathman 1976) have found that