Review of "SCSS A User's Guide to the SCSS Conversational System, by Norman Nie, C. Hadlai Hull, Mark N. Franklin, Jean G. Jenkins, Keith J. Sours, Marija J. Norusis, and Viann Beadle", McGraw Hill Book Company, 1980

Abstract

From the Preface-"The SCSS Conversational System... is the product of teachers, researchers, and programmers , all of whom believed that conversational computing could provide the most natural, efficient, and instructive approach to a broad area of data analysis... They sought a system that would allow users to interact not so much with a computer as with the information lying within their data... The SCSS system developed as a comprehensive set of data management and analysis facilities. "As a complement to the SCSS system, this manual incorporates many features to make it both useful as a teaching aid and convenient as a continuing reference to the system. Summary tables at the beginning of the chapters give an overview of the system's facilities, and the thumb index printed on the back cover should make those tables easy to find. Brief sample sessions give examples of how each facility works. For quick reference, equations, figures, and tables are numbered according to the section in which they appear. Lists of keywords with their definitions appear in a consistent format. As a special feature, "advisories" containing information and suggestions for using the system efficiently to accomplish research goals are set off from the main operational discussions." SCSS has been a long time in the making, 1973 or even earlier. The SCSS manual has also gone through many versions. I have personally worked with four prior to this "official" McGraw-Hill publication. September 1977 added only two, a total of 134. Release 3.0 of the Preliminary User's Manual, September 1978, expanded to 184, and now we have 583 pages, large size format, one and one-quarter inches thick, all to tell us how to use an interactive system! The format is easy to follow, with numerous examples, and the above mentioned "advisories" contain useful information such as "In a multiple bivariate plot, no attempt is made to use special symbols for multiple points at one printing position. Consequently, for a large file, the entire plot can be filled with points, especially if many vertical-axis variables are being plotted. Thus, a multiple bivariate plot may not be a useful tool for simultaneously examining a number of bivariate relationships in large files." Yes, a useful bit of information, and I appreciate such tidbits. However, more is not always better, and I continue to dream of an interactive system that would "carry me along", making every effort to "understand" my responses, …