Review of "Wrapping up the Package: Critical Thoughts on Applications Software for Social Data Analysis, by Ronald E. Anderson, and Edwin R. Coover", Computers and the Humanities, 1972

Abstract

This article is, as its title indicates, a series of critical observations on the software available to the social scientist for statistical analysis of data. In it, the authors set out to discuss, specifically, the usefulness of currently available packages, and the probable directions the design of such systems will take in the near future, and to suggest improvements in both the composition and distribution of these programs. The first part of the article is a historical overview of the evolution of systems for statistical analysis, and a critical description of design features common to most of the current generation of statistical software packages, with general suggestions for future improvements. The article continues with a discussion of the factors that determine the popular success of packages, and some speculations on the potential usefulness of interactive mode in social data analysis. The final portion, a series of observations on the institutional environments in which such systems tend to be used, focuses on the university computing center and stresses the instructional aspect of software use in this context and the need for more extensive education and more effectively user-oriented programs. At the end of the article, a glossary of packages gives a general picture of the types of programs that have been available through the recent history of software packaging in this field. All in all, this seems to be, for its length, a comprehensive overview of the current state of software packaging for the social sciences, and one which offers quite a bit in the way of thoughtful and constructive criticisms and recommendations. The Data-Text Project recently completed the final version (Release 3) of Data-Text for the IBM 360 and IBM 370, and distribution has been under way for the past six months. For those not familiar with older versions, Data-Text is a social science computer language for data processing and statistical analysis. It represents one of the earliest attempts to implement the "integrated system" principle with development starting at Harvard in 1962. While the new 360/370 version preserves many of the principles and features of earlier IBM 7090/94 versions, it has also been substantially expanded and modernized in keeping with third-generation concepts.