ERIC Number: ED210325
Record Type: RIE
Publication Date: 1978-Apr
Reference Count: 0
Predicting the Long-Term Performance in College of Minority and Nonminority Students: A Comparative Analysis in Two Collegiate Settings. College Entrance Examination Board Research and Development Reports.
Wilson, Kenneth M.
A longitudinal analysis of the records of performance of cohorts of minority and nonminority students in two different undergraduate settings was used to determine the predictive validity of admissions tests and school rank in determining long range and short range outcomes for college students. It was also hypothesized that the performance of minority students would improve as the college record accumulates. The results showed that substantial across-cohort increases in average grades relative to average ability levels were greater for minority students in the college setting. Consequently, in later cohorts, minority and nonminority students were less sharply differentiated by their average grades than were their counterparts in earlier cohorts, and grade point average trends suggested the possibility of "late blooming." The report slates that the results did not explain the increases in average grades across cohorts and suggest that increases in the average level of grades awarded across cohorts did not necessarily reflect increases or invariances in the average quality or quantity of academic achievement. The report also states that admissions variables were valid success predictors for both minority and nonminority students. The ambiguity of these findings are said to further point to the need for special consideration of problems involved in setting and maintaining standards for evaluation of student achievement in future, comparative across-cohort studies. (JCD)
Publication Type: Information Analyses; Reports - Research; Numerical/Quantitative Data
Education Level: N/A
Sponsor: College Entrance Examination Board, Princeton, NJ.
Authoring Institution: Educational Testing Service, Princeton, NJ.
Note: Some tables may be marginally legible due to reproduction quality of original document.