I am in a management PhD program and we have to read this book for our required class in psychometric theory. I totally agree with one of the other reviewers that almost no term is clearly defined by Nunnally and Bernstein in this 3rd edition. The book goes on and on and on talking about validity, reliability, scaling, ... without defining any single term in a concise manner. It is very frustrating!
So, my suggestion for everyone is to get the 2nd edition. I read it and was happily surprised. Nunnally is great, Ira Bernstein messed the 3rd edition up (Nunnally died a while ago and Bernstein was responsible for the writing of this edition). The previous edition is much, much shorter and has better organized chapters that go right to the point (well, relatively speaking ). In addition, I recommend several short Sage books (e.g., factor analysis from Kim and Mueller), which are much clearer.
In any case, this book or better the 2nd edition, is a must have for any social science researcher (or wanna be researcher ).