The Prediction of Doctorate Attainment in Psychology, Mathematics, and Chemistry

Author(s):
Rock, Donald A
Publication Year:
1974
Report Number:
GREB-69-06aR
Source:
Document Type:
Subject/Key Words:
Predictors doctorate attainment indicators of success validity attainment measures

Abstract

The GRE Board-sponsored project utilized data from the National Science Foundation Fellowship applicant records and the NRC Office of Scientific Personnel Doctorate Records File to evaluate the potential of GRE Aptitude and Advanced Tests as predictors of whether or not the candidate attained the doctorate within a period of from seven to ten years. In addition, the study sought to determine whether there were particular subgroups within each field as described by variables such as age or "quality" of the institution or graduate department for which the GRE tests have varying degrees of predictive accuracy. Sample sizes ranging from 643 to 779 were obtained for the three fields and divided into two samples so that cross-validation could be performed. The results of the study by field are summarized as follows: Psychology — Of the predictor tests, the GRE Advanced Psychology Test had the most consistent relationship with the criterion. Undergraduate GPA had a surprisingly low predictive validity with Ph.D. attainment. Sex had a strong relationship to the criterion; women were less likely to attain the doctorate in psychology than men. Age level provided a basis for defining more and less predictable groups. A "U" shaped relationship existed with younger and older groups being more predictable than the middle or 25- and 26-year old applicants. There is a slight tendency for students attending "lower quality" psychology departments, as defined by Cartter (1964), to be more predictable. Mathematics — The criterion was generally more predictable for mathematics than for psychology. The GRE Advanced Mathematics Test was the single best predictor with correlations of .38 and .44 for the two samples. GRE Verbal and Quantitative followed in order of magnitude. It may be that the successful completion of the Ph.D. program in mathematics depends upon the assimilation of a relatively structured body of knowledge which in turn leads to more accurate assessment of any one individual. There was little or no consistently different prediction for groups defined by age or by departmental quality indices. When age and departmental quality were combined, however, the young who attend "lower quality" departments appear to be more predictable than the remaining groups. *Chemistry? The validity coefficients for chemistry were similar in level and pattern to those of mathematics. Again, the GRE Advanced Test was the single best predictor in both samples. Correlations of GRE Verbal and Quantitative, while lower than for mathematics, were still reasonably strong. Age, when included as a predictor, added significantly to the prediction, contrary to the case in psychology and mathematics. There was little or no consistently different prediction for subgroups based on age, "quality" indices or both together.

Read More