Customer Reviews

Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence

5 star:		(16)
4 star:		(4)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

This book is, bar none, the best book on longitudinal analysis in social sciences.

The book has three outstanding features that make it the must-have for researchers who conduct longitudinal studies. First, the book has numerous examples that use data from real studies, collected by prominent scholars in this area. With the help of the accompanying website at UCLA, you will learn how to set up data files, which is crucial in longitudinal analysis. The sample codes and data files in SAS, SPSS, Stata, MLwiN, Mplus, HLM, and Splus will allow you to replicate the analyses. The authors use every effort to explain the results in plain, understandable language. They use a lot of graphs and tables to compare different nested models and help you to choose the one that best describes your data. It feels like you have an excellent tutor by your side when you are reading this book.

Second, the coverage of this book is comprehensive. Part I covers the regular growth curve modeling and multilevel modeling, with a few chapters dealing with time-varying covariates, discontinuous and nonlinear change. Part II covers discrete-time and continuous-time survival analysis. If you are conducting a longitudinal study, chances are you will find a technique in this book that suits you just right.

Third, the book is quite deep. Although it gears toward applications of different longitudinal analyses, it is no cakewalk. You need at least some background in multiple regression and multivariate statistics. I think the treatment of mathematics (both concepts and formulas) is just right. In some sections you may need to revisit them often in order to fully understand the subject.

This is simply the best book for those analyzing longitudinal data (data measured at more than one time point). Singer's coverage of Hierarchical Linear MOdeling (HLM) is clear, well-written (sprinkled with humor, it's like a lecture by the most popular prof. at your school), and geared towards researchers who need their programs to run, not just learn the mathematical underpinnings. Singer and Willett (the coauthor, not listed above!) set the standard for presenting math/statistics book examples.

THe authors accomplish the latter by keying her examples to data located at a UCLA website; you can run the same programs on the same datasets used in the book (wow!), and compare your output, troubleshooting any problems you may have. Singer and Willett (her coauthor, not listed here!) provide outputs and programs correspoing to several of the most popular statistical programs, including SAS and SPSS.

SInger and Willet also explain the rationale for using HLM over more traditional techniques such as regression. Simply stated, regression aggregates at a level that cause one to lose information (and hence the power to detect differences.) HLM allows one to look at overall differences due to time, but also the trajectories of individual differences who are "nested" within those time points. It's the (relatively) new thing, and is increasing used by investigators, and desired by peer reviewers.

As supplements, I suggest using the UCLA website mentioned above, subscribing to an e-mail LISTSERV for interesting (though sometimes compicated discussions of "multilevel modeling" (MULTILEVEL@JISCMAIL.AC.UK), and searching for Judith Singer's website through Google or A9 (if you use A9--"Alexa"--enough you'll get a small discount at Amazon.com). Also, compare Amazon's and Judith Singer's (through her website) current prices on this book.

Singer and Willett is an absolutely wonderful book on longitudinal data analysis. It is divided into two main sections -- one on longitudinal analysis per se, and another on time-to-event, or survival analysis, models. The former is especially good on the basic setup and interpretation of multi-level statistical models.
This is a book for beginners in the sense that it emphasizes data analysis, rather than theory. But every statistician, and every user of statistics, can find something of value.
When I was only halfway through reading this book, I recommended it to my friends. Several of them have purchased a copy and are glad they did.
This is probably the most well-written statistics text I have ever read.

This book does a particularly good job of explaining the substantive meaning of the equations involved in multilevel modeling analyses. It spends a lot of extra time explaining what the equations mean in real world terms using examples from actual data sets. I teach a graduate level course on HLM and I much prefer this book to the Raudenbush & Byrk book because it not only does a better job of explaining the math (for graduate students less comfortable with statistics) but the chapters are also sprinkled with incredibly useful advice on actually running the analyses (getting them to converge, interpreting them, etc.) The Raudenbush & Bryk book probably does a slightly better job of presenting the equations, but it falls short on explanation and practical advice. If you were only going to buy one HLM book, I would start with this one.

One of the great features of this book is that it is addressed to the empirical researcher and it really tells you how to conduct good data-analysis with longitudinal data. It doesn't push one particular piece of software, either, but uses a variety of different software packages. The book is really easy to read, and clearly explained -- and, there's so much in it!

This is a great book, it tells you in a straightforward way how to analyze your longitudinal data to answer questions of critical importance in the social sciences. It's not wedded to one particular piece of software, as many books on statistical topics are, but uses examples of real data and different software (HLM, MLwiN, SAS, Stata) to conduct the analyses. An absolute must for the researcher who collects longitudinal data.

Others have have already said that this book is superb and I completely agree. If you have had a class that covers applied statistics (basic correlation and regression) you should be able to pick this up and read it with no trouble. There is math here but is is *well* explained and the algebra is always presented with a worked example.

The code at UCLA (sorry they will not let me post the link here) makes the incredibly good writing even more valuable because, not only will you understand the concepts behind Mixed Effects/Hierarchical Linear Models, you will be able to implement the ideas. If you already have some experience with Mixed Effects/HLM browse the code and you will quickly see this book covers a wide scope. I have worked with the SAS code a lot and even though the book is a bit old (by a programmer's standards) the code still works just fine.

While the book is written to be clear for non-mathemeticians, there are many "intermediate to advanced" statistical topics covered here. These are importantly areas which are typically unintelligible to non-statisticians or are glossed over or ignored by other authors. Here are some noteworthy examples. This book could/should be used as a text on data exploration and visualization. There are many case-studies throughout the book and they all begin with great visualizations (with the all important code supplements showing the novice how to make the plots in the book). Topics like fitting lines, splines, curves are covered clearly and are shown beautifully. The discussion on choosing between sets of models using deviance (-2log likelihood) and AIC has the best coverage of any book. The general discussion of likelihood estimation (maximum likelihood and restricted maximum likelihood) is superb. The coverage of data transformation for model fitting is explained well and is presented with wonderful plots. These "bonus" topics are interwoven into the great explanations of longitudinal data analyzes.

There is so much to like in this book and nothing to criticize (except perhaps the price). It makes the rest of the books in the field look bad.

Wonderful book. It does a great job in helping develop intuition using practical examples and easy-to-follow reading style. The codes for all the examples (for all the common software like SAS, STATA, R and SPSS) are available online that is very helpful. It does not contain any math, but if you want to see the math, there are plenty of books and papers out there.

Excellent book in general. Makes a complicated subject easy to understand. Great examples and the website helps with programming for SAS.

Clearly written text... and usefull for researchers.
I would recommend it to anyone starting to learn about the subject!