Franklin R. Evans

A study of test speededness as a potential source of bias in the quantitative score of the Admission Test for Graduate Study in Business

Minority and majority groups were administered a special quantitative section of the Admission Test for Graduate Study in Business (ATGSB) under varying time conditions to determine if increasing the time allotted for the test would eliminate any bias that may exist due to an irrelevant speed factor. By a commonly employed definition, the special section was found to be moderately speeded for all candidates under normal time conditions. Neither the main effects due to time condition nor the interaction between the ethnic and time factors was statistically significant, suggesting that increasing the time per item does not reduce any bias that may exist in the test. Although a substantial proportion of minority group scores fell at or below the chance level, these appeared to retain fairly high levels of reliability.

A Comparison of Predictor Selection Techniques Using Monte Carlo Methods.

able to predict some future criterion performance from immediately available predictor variables. Typically, several predictors, such as test scores are available which must be combined in some way to predict the criterion measure. The most common procedure in this situation is to use a linear regression model to make the predictions. More often than not the problem is further complicated by the fact that initially one has measures on a larger number of predictors than is convenient to include in the final test battery. Having a reasonable means of reducing the number of predictors is important from several viewpoints such as cost and examinee time.

The Validity of Written Simulation Exercises for Assessing Clinical Skills in Legal Education1

Four different groups of respondents completed a set of simulation exercises intended to reflect lawyering skills involved in client interviews. The written simulation exercises began with a brief description of a practical situation and required a series of interdependent steps be taken toward resolution of a specific problem in client interviewing. The four response groups were: (a) undergraduate prelaw students; (b) law students without any direct client experience; (c) law students completing programs of clinical legal education; and (d) law professors and attorneys. Scores on the simulation exercises showed successive, significant increases across response groups and seemed independent of academic preparation and ability. These results suggest both the appropriateness of simulation exercises as measures of clinical skills in legal education and the effectiveness of clinical programs in promoting the development of such skills within law schools.