Reliability of Students' Self-Reports of Scholastic Aptitude Scores: Data from Juniors and Seniors

Abstract

Goldman, Flake, and Matheson (1990) reported data on the reliability of student self-reports of Scholastic Aptitude Test (SAT) scores and obtained correlations between .74 and .85 between actual and self-reported scores among university freshmen and sophomores. These authors suggested that additional data should be reported on upperclassmen. These analyses are based on Career Center intake forms collected on a women's liberal arts college campus on which students were asked to report their quantitative and verbal SAT scores. One hundred twenty-nine juniors and seniors completed intake forms (52% of full-time upperclass students). These self-reports were compared to official score reports in the students' files. Thirty-the students left the item blank or indicated that they could not remember their scores. These students' total SAT scores (using highest scores for those who took the SATs more than once) were significantly lower (M = 802, SD = 164) than those of students who reported scores (M = 881, SD = 198; t = 2.04, p< .05). Of the 96 juniors and seniors who reported scores there were 49 (51%) marches of SAT-Verbal, 40 (42%) matches of SAT-Quantitative, and 27 (28%) matches of SAT-Total. Matches were counted if the students' reports matched any of the scores reported. Fifty-three of the self-reporting students took the SAT two or more times. In this group, there were 51 matches on individual tests, of which 42 (82%) were with the highest individual scores they had obtained. Of the 103 nonmatches on the individual tests, 71 reports (69%) were higher than their highest obtained scores. The correlations of self-reported scores with highest obtained scores were .70, .63, and .64 for SAT-Verbal, SAT-Quantitative, and SAT-Total, respectively. These correlations are lower than those reported for freshmen and sophomores by Goldman, et a/. (1990). The difficulty of relying on selF-report data can be demonstrated by comparing SATVerbal scores with cumulative GPA. Self-reported SAT-Verbal scores and GPA were correlated .37, while highest actual SAT-Verbal scores of the self-reporting groups were correlated .44 with GPA. When SAT-Verbal scores for the entire group were compared to their GPAs, a correlation of .58 was obtained, indicating a range of common variance (;) from 13.7% to 33.6%. These analyses suggest that research efforts which rely on self-report of SAT scores may be tainted by a low rate of reporting among the lowest scoring students, by reports favoring highest scores, and by a large number of overestimates.