The Multivariate Social Scientist
Introductory Statistics Using Generalized Linear Models
- Graeme D Hutcheson - University of Manchester, UK
- Nick Sofroniou - Saint Patrick's College
Mathematics & Statistics | Quantitative/Statistical Research | Research Methods & Evaluation
Using a unified conceptual framework (based around the Generalized Linear Model) the authors explain the commonalities and relationships between methods that include both the analysis of categorical and continuous data.
`I learnt a lot from this book, which is very clear and carefully explains a wide range of multivariate techniques, basing them all on a consistent explanatory framework. It will be invaluable for social scientists wanting to progress beyond an introductory level of statistics' - Nigel Gilbert, University of Surrey
`A good introduction to ordinary regression, logistic regression, and loglinear models for the analysis of categorical frequency data....A nice feature of the book and one that social scientists will find very useful is the detailed listing of computer codes for the implementation of these models within the SPSS and GLIM software' - Zentralblatt MATH
"The book combines a sound and clear exposition of GLMs with a wealth of practical tips. Even for those already familiar with most of the techniques covered it would be an excellent refresher course."
I find the book quite good. It describes a series of analytical approaches in a clear way. In principle the book is very suited for self-study by students, as most of the book is easy to follow and explains the analyses step by step.
The main problem however with the book is that it refers to a rather old version of SPSS and therefore is not easy to use in relation to the newer versions of SPSS where generalized linear models is a separate set of models. The book was published more than 15 years ago, so indeed is in need of an update. And I would love to see an updated version of the book.
A second point is that the explanation of the different types of models is rather difficult to understand - I would guess too difficult for most social science students. But understanding the models and the role of the link functions is important as general background. In a next edition, the authors may try to improve these parts.
More generally, I would welcome more instrumental discussion of model selection: how does the student select a specific model and why?
Overall, an updated version would be very useful.
Interesting but the approach was a bit challenging.
Very fine book, we have chosen it as the textbook for the course.
The book is a bit too mathematical for social sciences students who are weak in statistics, but useful for those who have solid backgrounds in mathematics and want to learn more
A very good overview of multivariate modelling techniques and associated issues.
Some explanations could be more explicit e.g. in multiple logistic regression how to interpret a TWO unit increase or a TEN unit increase or a ONE HUNDRED unit increase in variable x1, since this often causes concern and misunderstanding.
Also, the SPSS menu prompts should be replaced with SPSS syntax, since SPSS syntax is more likely to be used at the level which this book is aimed at. And the SPSS menu prompts are more likely to change from one version of the software to the next.
