8 of 8 people found the following review helpful:
5.0 out of 5 stars
A fine text on Test Theory, September 6, 2007
This review is from: Test Theory: A Unified Treatment (Hardcover)
I had the fortune of having one Roderick P. McDonald as my instructor in this topic, in the very last semester he taught at the University of Illinois before he retired and returned to his first love, Sydney. This textbook is, essentially, an expanded version of his lecture notes. (As an aside, Rod writes pretty much like he speaks---dry British wit and all---so if you read his book you will get a decent idea of what he is like in person.) Test Theory takes McDonald's basic position that test theory is an integrated field, ranging from classical test theory as a computationally feasible special case of the Spearman factor model all the way to item response theory, which is a nonlinear model that adjusts for the nature of binary data. The factor model is the glue that holds everything together. McDonald's perspective that the confirmatory model makes more sense and that once you have learned it, the exploratory model will be simple to understand, almost as an afterthought, is a very nice bonus. Other books on this topic, by contrast, do not give the reader a sense of the unity of the topic.
There are a few downside to the book. First is that it has a lot of typos. You should definitely make sure to get the errata, which used to be available on the publisher's web page. Unfortunately, the book was written in a fairly antiquated typesetting system and not done in LaTeX; it could really stand an update. Second, the book has no exercises. I have some worksheets from my days at Illinois passed from TA to TA until they got to me. I should get them scanned and put on the web. Third, the publisher also indicates that a suite of free programs (Confa, Cofa, Cosan, NOHARM) written by Colin Fraser (one of McDonald's students) is available, but I cannot find them at present. If you want to do EFA, you can get Michael Browne's CEFA, but having free and simple CFA, SEM and MIRT software would be awfully nice....
Update (2/20/09): After having taught psychometrics all on my lonesome with no one else's lesson plan this Spring 2008, I believe this book is an excellent capstone work. It's too much for my students, which is why they got Allen & Yen combined with Streiner & Norman, but my lectures were based heavily on TT:aUT. It is on the required reading list for comprehensive exams.
Also, Colin Fraser's web page http://people.niagaracollege.ca/cfraser/download/ has COFA, CONFA, NOHARM, COSAN free to download.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
