 |
|
Title: Event History Analysis : Regression for Longitudinal Event Data (Quantitative Applications in the Social Sciences)
ISBN: 0803920555
Author:
Paul D. Allison
Publicate Date: 1984-11-01
Publish: 1984-11-01
List Price: $16.95
Average Customer Rating: 4.0
Format: Paperback
 |
|
|
|
|
Amazon Lowest New Price: $14.39
Amazon Lowest Used Price: $8.62
Amazon Merchant Price: $15.25
|
|
|
|
|
|
|
|
|
|
|
| Customer Review: |
|
1: Good but dated
Allison, the author of this book is one of the few people in this country who is able to write in easy to understand language quite complicated technical stuff. I believe this is one of the greatest qualities of a book writer. So, because this book is written by him, it is a very nice and easy to understand introduction to event history analysis for anyone who would need such an introduction. However, in my opinion this book has two big problems: (1) it is quite useless from a programming point of view for most people since it uses examples of dated programs that nobody uses anymore in social sciences(maybe except SAS), (2) there is no mention of STATA, the easiest and one of the most powerful programs to use for event history analysis, that is at least 4-5 years more advanced than SAS. For instance, one of the reason Allison says discrete time models could be preferred over continuous time models is because they can incorporate time-varying covariates. However, STATA could now easily incorporate time-varying covariates into continuous time models. Thus, the reasoning of the book suffers from the fact that it fails to incorporate the newest and easiest methods available today.
|
2: Sometimes, it is more complicated than a pre/post test
Often, if we want to examine the effect of some particular event on something we are measuring/observing, we often resort to a simple pre/post test for all of the wrong reasons: it is easy and/or we did not know any better. This short text is a discussion on performing regression analysis when longitudinal events are involved. This text is not a 'how to' book.
By 'discussion', I mean that this book really is a discussion. This short book is primarily text where the author uses words, not equations [less than one per page] to discuss events and things that we should consider when examining them using regression analysis. In the text, the author makes a very strong effort to remind us that "censoring" and time-varying explanatory variables are prone to severe bias and the loss of information.
The author does an nice job of explaing his points though the use of well-considered examples. The chapters of the book are:
1) Introduction
2) A Discrete-Time Method
3) Parametric Methods for Continuous-Time Data
4) Proportional Hazards and Partial Likelihood
5) Multiple Kinds of Events
6) Repeated Events
7) Change of States
8) Conclusion
At the end of the text, there are several appendicies and references listed. One appendix is a listing of computer routines that can be used in this analysis. Though they are rather old now, I did a web search and found that they are still in use.
|
|
|
|
|
