The Prediction of Success of Three Ethnic Samples on a State Board Certification Examination for Nurses from Performance on Academic Course Variables and on Standardized Achievement and Study Skills Measures

Abstract

For each of three samples of 207 Caucasians, 61 Mexican-Americans, and 53 Negroes in a nursing training program at a very large metropolitan hospital, validity coefficients of 15 predictor variables available prior to training and of 13 predictor variables obtained concurrently with training were calculated relative to each of five subtests of a state board certification examination for nurses. In addition, stepwise multiple regression analyses relative to each of the same five criterion measures as well as with respect to an average score on the five subtests of the certification examination were carried out for selected sets of predictor variables in each of the three samples. Standardized test measures involving reading skills were the most valid of the predictor variables obtained prior to nursing training, whereas total grade point average in program courses as well as scores on the National League for Nursing (NLN) Achievement Tests provided highest validity coefficients for each ethnic group. In terms of the predictor variables selected, the stepwise multiple regression analyses served essentially to reinforce the other correlational findings. THIS paper represents an extension of a validity study that was concerned with the prediction of the success of three ethnic samples in the academic components of a nursing training program conducted at a very large metropolitan hospital (Haney, Michael, and Martois, at UNIVERSITE LAVAL on July 7, 2015 epm.sagepub.com Downloaded from

THE VALIDITY and FAIRNESS OF EMPLOYMENT and EDUCATIONAL TESTS FOR HISPANIC AMERICANS: A REVIEW and ANALYSIS

“…Then the expected value of the "chi-square" statistic would still be the nominal degrees of freedom, 100. To obtain the variance, we note that this statistic is the sum of 91 independent entries, 90 of which have the usual chisquare variance of 2, while the last entry is repeated 10 times and hence has a variance of 2(10) (10). Thus, the true standard deviation would be ,/2(90) + 2(10)(10) = 19.49, and an observed value of 128 would no longer be statistically significant.…”