A job analysis survey was done on 39 entry-level professional classes (e.g. personnel officer, social worker, rehabilitation counselor, research analyst, accountant, computer programmer) to discover statistical factors on which the classes could be scored and to cluster the classes on the basis of these factor scores. A job analysis questionnaire was designed based on a literature review and a set of in-house job analyses; 738 incumbents completed these questionnaires. General recommendations for use of the study's results in selection and validation are made. In a reanalysis of the data, the task data were reclustered based on the percentage of time a class spent on a set of factor tasks. The author contends that percentages are a better metric for reflecting the existing relationships among the data than are standard scores and that clustering based on percentage of time is superior to that based on factor scores. A section entitled "Detailed Description of Individual Classes" gives an example of the six-page computerized report that was printed for each of the 39 classes included in the study. (Author)