finding a good book is hard

May 7, 2009 1:45 PM

I need a good statistics book!

I an looking for a good probability and statistics book (or multiple, if necessary). I have a good math background (math undergrad, engineering grad school) but my statistics background is lacking. I want a book that is relatively accessible but does cover all the major topics, preferably with proofs. I want to gain a solid understanding of applying t tests, ANOVA and the like while understanding the fundamentals. The library is full of books but most don't seem right. Which book would you recommend?

posted by Brennus to Science & Nature (14 answers total)

"The Probabilistic Method" by Alon and Spencer is the greatest book on techniques in probability in the history of mankind. I consider it very accessible to someone with your proposed background.

It may be too "pure-math" for what you want to use it for, but check it out anyway. The first result on this page is the book I'm talking about, and there's a limited preview so you can see if it lines up with your criteria.
posted by King Bee at 1:52 PM on May 7, 2009

Not cheap, but good: Hays.
posted by fcummins at 1:56 PM on May 7, 2009

Even You Can Learn Statistics.
posted by curlyelk at 2:34 PM on May 7, 2009

Manga guide to statistics?
posted by selenized at 2:40 PM on May 7, 2009

My favorite 'undergrad'-level stats book is "Statistical Inference" by Casella & Berger; it's absurdly expensive on Amazon but maybe you can find it somewhere else. Lehman's book is a classic but probably too detailed and not broad enough for you.
posted by bsdfish at 3:18 PM on May 7, 2009

Not sure about your application, but if its time series data then I would look at a Guide to Econometrics by Kennedy. It is much more of a birds eye view, why are we doing this operation as opposed to a cookbook-style book full of formulas. Econometric Analysis by Greene is the standard text for people working in the field.
posted by shothotbot at 4:04 PM on May 7, 2009 [1 favorite]

Your math background sounds more advance than mine when I was first starting with statistics, but the book I always recommend for beginners (although, I'm in the social sciences) is Statistics For People Who (Think They) Hate Statistics.
posted by quodlibet at 8:42 PM on May 7, 2009

I was going to recommend Statistics for people who think they hate statistics too, but it doesn't have proofs, and I checked my copy - it really breezes past ANOVA.

But it goes through each relevance test in a step-by-step fashion with fairly basic math.

I have the "For Excel" edition, which is great for people who will be using excel (or similar) to work through these problems.
posted by MesoFilter at 10:01 PM on May 7, 2009

I know you're asking for books, but the Vassar stats website might be just what you need - it's laid out much like a statistics textbook.

http://faculty.vassar.edu/lowry/webtext.html
posted by primer_dimer at 2:47 AM on May 8, 2009

http://www.stat.cmu.edu/~larry/all-of-statistics/index.html and http://www.stat.cmu.edu/~larry/all-of-nonpar/index.html
posted by blueyellow at 4:46 AM on May 8, 2009

Statistics in a Nutshell is a quick applied overview. Whereas the Wasserman books above give you theory and more/different material.
posted by blueyellow at 4:58 AM on May 8, 2009

I used Probability and Statistics for Engineering and the Sciences by Jay Devore for my first two undergraduate statistics courses. It was, easily, the best text I used during college. It's certainly dull and old-school, but I found the approach to the material to be extremely logical and easy to follow. I use it regularly as a reference text. It will get you through probability, distributions, estimation, hypothesis testing, ANOVA and basic DOE, and regression. Although if you are very interested in regression at all, I recommend switching to an econometrics text, as mentioned above.
posted by prozach1576 at 12:09 PM on May 8, 2009

Casella and Berger is pretty good, but only covers anova and regression in a basic way since most of the book is on the fundamentals of statistics. Applied Statistics by Neter Wasserman and Whitmore is good for more depth on those. I also liked Applied Regression Analysis by John Fox.
posted by a robot made out of meat at 1:09 PM on May 8, 2009

Just a heads up. If you search for the title of a book and perhaps the authors name along with "pdf" or "djvu" you can usually find pirated copies online.
posted by delmoi at 1:24 AM on May 12, 2009