AN ATTEMPT TO DEVELOP A MODEL FOR PREDICTING NO-SHOWS AMONG THE APPLICANTS TO WAUKESHA COUNTY TECHNICAL INSTITUTE (WISCONSIN)
Abstract
This study was justified as a result of two factors: (1) general decline of the population of college-aged students predicted for the 1980's and 1990's and (2) administrative inefficiency experienced by the admissions and instructional staff at Waukesha County Technical Institute (WCTI) in Southeast Wisconsin because of inability to predict which applicants would actually show up to register. The purpose was to bring understanding and control to the no-show problem by creating a model for predicting which applicants to WCTI would likely be no-shows. For purposes of developing a prediction model and a body of information relating to the model, data already available at WCTI were used: information on the application form. Half of the total sample of 300 were application forms from no-show applicants which were randomly chosen from forms on file for the fall semesters in the academic years from 1976 to 1982. Half of the total sample of 300 were application forms randomly selected for applicants who sought admission for the same academic years and who did register. The total random sample of 300 was stratified by program. It was hoped that a set of statistically significant predictor variables could be attained. The data were analyzed using the Statistical Package for the Social Sciences. A series of hypotheses were established to determine the independence of the registered/no-show criterion against the independent variables. A matrix of the intercorrelations of the dependent and independent variables was created using point biserial coefficients, phi-coefficients, and contingency coefficients; linear discriminant analysis was used to test Hypothesis 12 (the prediction model). In Hypotheses 1 through 11, chi-square analysis of the correlations required that the null hypotheses be retained at the .05 level. The chi-square value for the linear discriminant analysis of the Predictor Hypotheses was significant at the .05 level. The null hypothesis for the Predictor Hypothesis was rejected. However, since all 11 of the predictor variables taken simultaneously accounted for only about .08 of the variability in the dependent variable, the result of the linear discriminant analysis is of no practical administrative value.
This paper has been withdrawn.