Job preferences of employees and applicants have become more and more widely recognized as important factors in the success or failure of personnel and industrial relations programs in industry. According to Jurgensen, accurate informatibn on job preferences is a valuable aid in designing and revising personnel policies and practices, including recruitment programs, in supervisory training, in diagnosis of employee morale, in collective bargaining, and in interviewing job applicants ( 7).The study reported in this paper was designed to measure job preferences of employees in two department stores by means of the Jurgensen Job Preference Blank (7) and, in particular, to measure the test-retest stability-of these preferences. Most of the above applications of the Blank require high test-retest stability of mean rankings of job preference factors by groups. Use of the Blank for selection, placement, and transfer of individuals would require high test-retest stability of job preference rankings by individuals.

THE following article presents the details of a study which attempted to investigate differences in responses to a pencil and paper questionnaire administered to two groups of individuals. In one group the respondents remained anonymous while in the second, respondents were identified. Three conclusions were reached. (1) Essentially the same responses were obtained for individuals in the group whether identified or anonymous. (2) Mean scores for the groups did not differ significantly.(3) These results may have been influenced by the fact that the questionnaires were administered by the staff of a university organization and respondents were repeatedly assured that the questionnaires would only be used for confidential research purposes. PURPOSEMany attempts have been made, and are presently being made, to measure employee attitudes in industry. It is often desirable to identify the respondents in such a survey so that the validity and reliability of their responses may be checked. It is possible, however, that identified respondents may give biased answers to attitude questions. Several writers in the field have expressed the belief that attitude questionnaires