A colleague let me glance over this book by Chang. I have to say it is one of the better introductory GIS texts I have seen. I've owned a few. The one for my undergrad courses was weak, but in fairness it was one of the least expensive textbooks I had to purchase. Longley et al. produced a nice one as well. I would go with Chang over Longely, though. Chang's book is ESRI-biased, but I think it is of value to everyone. There are a number of worked examples that show how the little black box works, e.g. affine transformations, and ordinary kriging. I'm not talking about point and click, I mean actual mathematics. My colleague also said the section on modeling is excellent, but I didn't look through it in great detail.
Unfortunately the book budget is a little low right now, as I've made a number of purchases recently so I don't own a copy. Speaking of which, I also recommend Bivand et al.'s spatial statistic book for R. This fills a void in Spatial Statistic books that has been growing. Most spatial stat books focus heavily on theoretical, which is fine, but for someone like me that is not in a course there is a lack of worked examples. Waller and Gotway's book is theoretical and has exercises, but now answer key so who knows if I produced the "correct" results....Andy Mitchell's book is a great start, but doesn't go into much depth (doesn't talk much about first and second order effects, doesn't talk about inhomgeneous k-functions, etc...)