ACCOUNTING GRADUATES' PERCEPTION AND EVALUATION OF TYPES OF ACCOUNTING ORGANIZATIONS: AN APPLICATION OF NONMETRIC MULTIDIMENSIONAL SCALING TO THE RECRUITMENT OF ACCOUNTING MAJOR

This research used multidimensional scaling techniques to develop a model which was used to predict the job choice behavior of college graduates seeking first-time entry-level positions in accounting. Two different samples were used in the research design. SAMPLE (N = 34, Fall 1981 graduates) was used to develop the model while SAMPLE II (N = 37, Spring 1982 graduates) was used to test the model's predictive ability. Three types of data were collected from SAMPLE I: (1) demographic data; (2) organization type similarity/dissimilarity data; and (3) organization type attribute data. The similarity data were collected to determine the perceptions of similarity between seven types of accounting organizations: international, national, regional and local public accounting firms, federal and state/local government agencies and private industry. The attribute data were used to estimate the dimensions used by SAMPLE I to make their similarity judgments. The analysis of the similarity and attribute data indicated that accounting graduates evaluate types of accounting organizations on three dimensions: (1) The job itself; (2) Job stability; and (3) Social aspects of the job. Stage II of the research design consisted of collecting demographic, preference, and organization choice behavior data from SAMPLE II. Personal interviews and interviewer administered questionnaires were used to collect the demographic and preference data from SAMPLE II members before they graduated and left campus. The organization choice data were collected by mail questionnaires and telephone interviews after the subjects graduated and had obtained a position in accounting. The preference data were used to predict each subjects' choice of the particular type of organization they would elect to join. The model's predictions were then compared to actual organization choice decisions to establish the predictive ability of the model. The model correctly predicted 9 of 30 organization choice decisions made by the subjects in SAMPLE II. A binomial probability test indicated that the probability of achieving 9 correct predictions out of 30 decisions could be expected to occur about 184 times out of 10,000 replications of this research if the model had no predictive ability. Implications of the research are discussed.