Customer Reviews

Hierarchical Linear Models: Applications and Data Analysis Methods (Advanced Quantitative Techniques in the Social Sciences)

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

 

 

This book gives a detailed description of the use of an advanced method to deal with nested data sets.
At a general level the constructs and ideas are well written and can be followed reasonably easily.
However the mathematics is often written very dense, which makes reading and understanding complex.
My main problem with the book, is that in many of the examples they provide, the given formula's, and data skip rapidly to the solution. Thus it is often not insightfull at all, how the data led to the numerical outcome (and I and several of my colleagues could not reproduce all of the example outcomes). A more extensive discussion and a more step-by-step construction of the examples would have been helpful there.

So in short: Conceptually this book is fine, but for practical use mathematics are too dense, and examples are too hard to follow

I had taken a class in HLM before, and I bought this book to refresh myself on the details. It takes a good deal of attention to detail and concentration to really get the full measure from this book, although it's all in there. Despite the authors' best efforts, there is a good bit of stats jargon in the book, so a reader who is not familiar might have some difficulty. If you're at a point where learning HLM is a logical next step, you'll be fine and I recommend this book. However, if your over-eager faculty advisor told you to learn HLM, despite your minimal experience in stats, you're better off enrolling in a class or workshop.

The book is not bad. But need so much improvement. I send a letter to the authors with my comments.
For example:

A basic topic such as "assumptions" is not clear presented. You have to "discover" them on your reading.

You will find things like "as we can see this will create a problem" ok. but what kind of problem, why are these a problem?

I got the book, and for each chapter I read, I had to go online to look for additional information, and clarifications.

It is clear that the authors are experts and the topic, and things are "so clear and obvious" for them, but the people that is reading the book might have problem following it.

Conclusion. After 2 weeks I decided to return the NEW book and get a USED one.

I also got the "manual" for the HLM6 software, dont bother. It is not a good manual. Actually, it is not a manual because it does not teach you how to use the software, it does not explain its different options, it just show you some examples. You can find similar things online. I returned the manual as well.

The second edition of this texbook by Raudenbush and Bryk has achieved near-biblical status in the world of multi-level modeling. It is quite comprehensive, and the chapter on centering, an unexpectedly important and complex topic, is the best I"ve seen.

Nevertheless, Raudenbush and Bryk make what I take to be a serious error when they fail to acknowledge the strengths and weaknesses and breadth and limitations of their likely audience. For all but the best trained mathematical statisticians, this book is inaccessible and, for the reader, money poorly spent. Raudenbush and Bryk must know that most sociologists, political scientists, program evaluators, policy analysts, and numerous others will find their book too difficult to use as a self-teaching tool. Thus, in fairness to those trying to keep up with important methodological developments, the authors should, at the very least, conspicuously acknowledge the demands their book places on the reader.

For most readers, there are much better ways to a make a start on multilevel modeling. If one wants to, he or she can then work toward meeting the demands imposed by Raudenbush and Bryk.

This book describes important advances in statistical analysis of social science data, circa 1992. Much of this data has a natural hierarchical grouping. But traditional statistical methods proved inadequate at coping. The biggest drawback was the failure of the assumption of independence. If at the lowest level, Items I1,...,In are grouped into sets J1,...,Jm, where m<n, then all the items in Jk have that value k. They are not independent in this dimension.

To handle this, Hierarchical Linear Models were developed. The book gives a detailed treatment. A very comprehensive discussion. Including the handling of meta-analysis, where we wish to combine results across different studies. Which then involves using empirical Bayesian estimates. This method has also seen important usage in evaluating medical studies, conducted by different researchers on the same topic.

The book also illustrates the essential development of non-trivial computer programs to perform the gruntwork.

You will need a solid background in statistics to find this book useful. At a minimum, a year of statistics at the undergraduate level.

Raudenbush & Bryk's text is a must-have reference for those who use hierarchical models in professional research, but not the best introduction for beginners. As others have said, it is extremely dense at times, but I don't necessarily see that as a drawback as long as readers are aware of what they need to know in advance. For those with a solid foundation in general linear modeling (i.e. all the various forms of regression, MANCOVA, etc.), as well as some basic knowledge of what hierarchical models can do, this is the right book for you. If not, choose something more basic and work your way up. That said, HLM is not as daunting as it may seem at first, and those who do research in multiple settings simultaneously now have little excuse for "cutting corners" by simply throwing in covariates or assuming homogeneity, no matter how many variables look the same.

By far, the greatest improvement since the first edition is the extended discussion of HLM in longitudinal designs. Raudenbush & Bryk are at the cutting-edge here, and anyone who does growth-curve analysis will find this book to be a great resource. On the downside, their discussion of the unique factors to take into account in 3-level models is a bit sparse (and the only reason for 4 stars instead of 5).

I return to this book constantly, and strongly recommend it to those who do (or, more often, should) take a wider, more accurate accounting of the many possible sources of variance in their research.

I've only read the first chapter and half of the second. but it seems like a decent text. The major bonus is that it has data that comes with it, so you can check and see if you've done the activities correctly or not. A definite plus.

I have to admit that this book helped me a great deal to understand hierarchical and nested models. However I find the examples and applications used in this book rather poor. I would like to see a similar book with the same great insight but with better examples and clearer discussion especially the assumptions and some theoretical aspects behind the logic of hierarchical modelling.

I like this book - one of the most popular text book for HLM. Pretty clear math equations are well described with proper examples. Especially for those in social science, this is a good place to begin. I enjoyed and learned a lot.
For the purchasement, I got some discount - which was nice - as well as fast shipping.

Basically if you buy this book, you don't need anything else on HLM. It's comprehensive, as the technique stands. But you can't learn HLM from this book - you'll need a teacher.