Learning statistics - which books/resources would you recommend?

Hey guys,

Could you recommend some books and resources that you found useful to learn statistics?

Before, I've done a PhD in theoretical physics and done some math at university, but never really got any serious statistics. I've started learning from David Kaplan's Statistical Modeling: A Fresh Approach, and I'm finding it a tad slow. It's very clear and it explains R while explaining statistics and that's good because I want to learn both. But I am looking for something that's maybe more pitched at a graduate level. Any ideas?

Thanks in advance!


Less is more. Stay pure. Stay poor.
What area would you like to apply statistics to and what kinds of statistics do you think may be applicable?

Don't get me wrong, I think there is great merit in learning as much and broad knowledge as possible, but it may help with the intitial digestion if they are using examples you are familiar with.


Fortran must die
The best I have found is 'Using Multivariate Statistics" by Fidel and Tbachnick. Its now in, at least, the fifth edition and you want whatever is the most recent. Its fantastic for being easy to understand, providing a lot of helpful hints on practical issues like cleaning up data, talking about software in detail, and covering a huge number of methods.
Thanks Noetsi, looks like there's a sixth edition and the paperback edition is actually affordable! (saw the hardcover price first :)).

Hlsmith, I'm most interested in regression/modelling. It's for a job switch that I am doing so I'm not sure about the field yet - as you can see I'm not sure about the subject yet either! I've worked in medical and energy environments before, so examples in those fields would be more familiar than examples in business or marketing.

Any more tips? They would be very welcome.
Dason, Gelman's book looks very interesting, thanks for the tip! And I'll keep Casella&Berger in mind for when I'm looking for the theoretical backgrounds. I personally find it easiest to go from practical to theoretical; the opposite route doesn't for that well for me. So I'd use Casella&Berger more to back up what I've already informally learned and therefore, I'd only be using it somewhat further down the road.


Fortran must die
If you want practical comments on regression you might try John Fox's "Applied Regression Analysis and Generalized Linear Models" (he had an earlier regression book that was a lot easier to read but its dated by now -1991). There are a series of Sage monographs I found useful including "Regression Diagnostics" by Fox, "Multiple Regression in Practice" by William Berry and Stanley Feldman, "Applied Regression" by Michael Lewis-Beck,and "Understanding Regression Assumptions" by Berry. Like all Sage monographs they are short, but a good place to start.

The best book I have found for logistic regression is Logistic Regression Using SAS by Paul Allison. There is a second edition I just got (2012) and you likely want to look at that if the software matters because the previous book will be useless with recent code as its 13 years old.

Note these books are designed for practice, data analysis, not theory primarily. I am sure Dason's suggestion are much much better for theory :p