28

I have become interested in statistics, but I must admit that it has been a long time since I have used math seriously. Sometimes I understand what equations mean, but sometimes I can't follow them.

I like the answer given here that uses the image with an arrow: Conceptual understanding of root mean squared error and mean bias deviation.

Would you have any advice regarding serious statistical books on modeling, PCA, simulation, etc., that use figures more than equations? I'm hoping for something like a 'stats for dummies' book, a book that would make stats accessible for everyone?.

mathieu_r
  • 4,211
  • 3
  • 14
  • 5
  • 1
    You could supplement your useful books with various statistics demonstrations from the [Wolfram Demonstrations Project](http://demonstrations.wolfram.com/topic.html?topic=Statistics&limit=20). For instance, [here](http://demonstrations.wolfram.com/StudentsTDistributionAndItsNormalApproximation/) is one showing the difference between the normal and Student's t distributions – Jason O. Jensen May 29 '12 at 19:32
  • I would recommend Andy Field's "Discovering Statistics Using SPSS, 3rd Ed.", or his latest R book, "Discovering Statistics Using R" (with co-authors Jeremy Miles and Zoe Field). – neuron May 30 '12 at 08:05

5 Answers5

25

The Cartoon Guide to Statistics covers the basics, including random variables, hypothesis testing and confidence intervals.

MånsT
  • 10,213
  • 1
  • 46
  • 65
6

I really like A Guide to Econometrics by Peter Kennedy. Some material in it will probably be irrelevant, but the conceptual info is excellent and useful for non-economists. For example, here's Kennedy on graphical intuition for omitted variable bias and multicolinearity in multiple regression using Ballentine/Venn diagrams. Each topic starts with a simple explanation, usually with diagrams, followed by technical notes with some math and references.

dimitriy
  • 31,081
  • 5
  • 63
  • 138
4

While reading the reviews for The Cartoon Guide to Statistics, I noticed one saying The Manga Guide To Statistics was better: http://www.amazon.com/gp/product/1593271891

The Manga Guide has fewer reviews, but gets better ones on average. (I.e. the mean number of stars is better; hopefully after reading either book you'd be able to calculate if that is a significant "better" or not ;-)

Darren Cook
  • 1,772
  • 1
  • 12
  • 26
3

Ram Gnandesikan's book "Methods for Statistical Data Analysis of Multivariate Observations" has some equations but also a lot of graphics. Duda Hart and Stork "Pattern Classification Second Edition" has a lot of nice graphics including some color. Hastie, Tibshirani and Friedman "The Elements of Statistical Learning" although filled with equations is loaded with beautiful graphics and heavy use of color (true for both editions).

Michael R. Chernick
  • 39,640
  • 28
  • 74
  • 143
3

One book that I really like is "The Statistical Sleuth" by Ramsey and Schafer. It does still have the formulas, but the more complicated formulas have arrows pointing to the different parts with explanations of what that part of the formula means, there are lots of good graphics to help explain the concepts. It also covers a lot more than the cartoon guide (which I also like, but someone else already suggested it).

One of the best parts of it is that every chapter starts with one or more case studies that describe a dataset in general terms and a question that is of interest related to the data and give an answer to the question in general terms, then the chapter goes on to show the methods that lead to the answer and give more detail. It is nice to see how the technique will apply to the real world as you learn the details.

Greg Snow
  • 46,563
  • 2
  • 90
  • 159