This won't be much use to you for January but I'm writing the final 2 chapters of what will probably be called "Psychological Assessment - a hands-on approach" which I'm delivering by the end of this year and which will be published by Routledge in mid 2018. It certainly considers Michell's points, and I find I keep on saying words to the effect of "if you've read this far you will have decided to ignore Michell's critique for some reason of your own - so this is how one goes about calculating x ignoring the assumptions that...".
The other unusual thing about the book is that it comes with a selection of Open Office spreadsheets many of which generate simulated data. These are fully referenced and integrated into the text - they're there because in my experience students only feel comfortable with psychometrics when they start to get their hands dirty and play with data. (I'm working on the external bias one as I write this.) It's taught me a few things too. Apart from the many well-known problems with coefficient alpha, I'd no idea how huge the 95% confidence intervals were unless sample sizes are truly enormous; I'd always relied on Nunnally's claim that a N of 200+ was OK...
The book contains al the usual stuff (including an introduction to IRT) though it's a bit heavy on exploratory factor analysis (principles, how to do it, hierarchical models and alternative designs - Cattell's P, Q S etc.). I've kept it as non-technical as possible with few formulae, because I know they scare both undergrads and practitioners!
Colin
|