Review of "An Introduction to Data Management in the Behavioral Social Sciences, by Sheldon Blackman and Kenneth Goldstein", Wiley, 1971

Abstract

This is to register a strong disagreement with the review in the last SIGSOC Bulletin, which favorably described the Blackman and Goldstein book. The book was described as "a useful book to recommend to people beginning to work on dissertation research projects or in a programming course aimed at social scientists." While the book is quite readable, it glosses over the techniques involved in setting up analysis for computer data processing and omits problems of even minimal complexity. Its chief weakness is the perspective that it leaves the reader. For one thing it suggests that the BMD and P-Stat packages are the major packages for social data analysis, but even more disturbing it suggests that one can get along quite well using the ccmputer without much understanding of techniques and principles involved. It most certainly would not be appropriate for "a programming course aimed at social scientists," because not only does it suggest that programming skills are unnecessary, but it avoids interesting problems that programming is required to resolve. The negative consequences of teaching package usage alone have been previously stated at last year's Dartmouth Conference on ccmputers in undergraduate curriculum and at the ACM 1971 Panel on Social Science Computing Curricula. It is not clear to me why a book of the Blackman and Goldstein character should even appear in print given a more urgent need for other approaches in the social sciences. It certainly is justified as a mimeographed manual to be used as a supplementary resource for undergraduates utilizing computing for course projects. But the institutionalization of this kind of approach is most certainly questionable. I feel that much of the attitude of social science toward computing as low level, undesirable tedium results from an approach which over simplifies the techniques of computer utilization and implies that the most reasonable social scientists can go about their way quite well without understanding any computing. How can we stimulate students to tackle computing problems creatively, to optimize the potential of computing for the future of social sciences if we limit their exposure to control cards setups?