Discrepant LSAT Subscores (RR-93-01)
Lawrence J. Stricker


This study investigated the overall prevalence of discrepant LSAT subscores, their differential incidence for subgroups of examinees, and the psychometric properties of alternative measures of discrepant performance. Subscore differences, often very substantial ones, were frequent; statistically significant differences affected about a third of examinees; and significant and rare differences involved a tenth of test takers. The incidence of these discrepancies did not vary with the examinees’ sex, ethnicity, familiarity with the LSAT, or the number or selectivity of the law schools to which the examinees were applying. But the prevalence was greater for examinees who had high total scores on the LSAT or were older, primarily reflecting these test takers’ deviantly poor performance on the Analytical Reasoning subtest. Reliability was appreciable for two of the three measures of observed differences but minimal for the more important measures of significant or significant and rare differences. Subscore discrepancies appear to have no viable role to play in interpreting examinees’ LSAT performance.

Unlike other tests employed in graduate admissions, the Law School Admission Test (LSAT) reports only a total score. In contrast, the Graduate Record Examination General Test provides Verbal, Quantitative, and Analytical scores (and no total score), and the Graduate Management Admission Test gives Verbal and Quantitative scores as well as a total score.

A recent LSAS survey suggested that there was growing interest in the law school community about the potential usefulness of the LSAT subscores (Analytical Reasoning, Reading Comprehension, and Logical Reasoning) in admissions, especially for minority applicants, supplementing what is yielded by the test's total score. One approach to this issue is to identify examinees with markedly different performance on the test's subscores. Previous research has studied examinees’ discrepancies between their LSAT performance and their college record, as reflected in large differences between LSAT scores and undergraduate GPAs, in an effort to determine the implications of these differences for the validity of the test and college record in predicting law school grades. Similarly, examinees may exhibit discrepant performance on the various sections of the LSAT, with large differences in their subscores. These discrepancies may shed light on the examinees’ overall test performance, useful in interpreting the LSAT total score and in making admissions decisions about these applicants. For instance, discrepant performance could reflect language difficulties or atypical education.

“Profile” or “scatter” analysis of discrepancies in the subtest scores on individually administered intelligence tests, notably the Weschler Adult Intelligence Scale-Revised (WAIS-R), Wechsler Intelligence Scale for Children-Revised (WISC-R), and Stanford-Binet Intelligence Scale: Fourth Edition have long been employed in interpreting performance on these tests. For instance, differences between the Verbal and Performance IQs on the WAIS-R or WISC-R are examined. Subtest scores on these tests are also compared, such as the difference between the highest and lowest scores for the entire set of subtests, as well as the difference between scores for each pair of subtests.

Accordingly, the purpose of this study was to provide basic information on discrepant test performance on the LSAT: the overall prevalence of discrepant performance as well as its differential incidence for relevant subgroups of examinees, and the psychometric properties of alternative measures of this characteristic.

Discrepant LSAT Subscores (RR-93-01)

Research Report Index