Customer Reviews

Theory Based Data Analysis for the Social Sciences (Undergraduate Research Methods & Statistics in the Social Sciences)

5 star:		(0)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

In 1968 Morris Rosenberg published a wonderfully non-technical methodological text titled The Logic of Survey Analysis. Over the years, this informative volume, which relies entirely on percentaged cross-tabs for statistical information, taught thousands of students to think about models with more than two variables. Rosenberg's concepts have became part of the social scientists' everyday language -- extraneous and component variables, intervening and antecedent variables, suppressor and distorter variables, spuriousness and non-spuriousness, and so on. These and other concepts introduced by Rosenberg are still invaluable aids in making sense of even the most complex statistical models.

Unfortunately, Rosenberg did not publish an updated version of the Logic of Survey Analysis, and, in time, his examples came to seem antiquated and lost their once compelling character. Happily, though, Carol Aneshensel has written a follow-up to Rosenberg's text. Using multiple regression analysis rather than cross-tabs, and providing an abundance of diagrams and tables to illustrate crucial concepts such such as statistical control. Aneshensel's book is a worthy sequel to Rosenberg's 1968 text.

Aneshensel's presentation is more than just a regression-based replication of Rosenberg's work. She has new concepts of her own, such as focal relationship, which fit neatly into Rosenberg's schema and provide a more complete language with which to discuss the elaboration of relationships involving more than two inter-related variables.

I think it would have been best if Aneshensel has stuck to OLS regression equations, which she presents and discusses admirably. The addition of logistic regression equations, while useful for those who understand logistic regression, makes the book seem harder than it is. Conceptually, moreover, the logistic regression equations add nothing to the logic of elaboration that could not be done with OLS.

In any case, this is a good book that deserves to be selling a lot better than Amazon's ranking indicates.