How to Lie with Statistics

Huff, D. & Geis, I. (1954). How to lie with statistics. New York: Norton.

Time for another book review! I’m almost done Applied Statistics for Education and Psychology I that I’ve been taking this fall. One more week, one more unit, one final paper, then I’m done!

My dad gave me the musty old bright orange book called How to lie with statistics to supplement my reading for the course.  Sounds kind of unethical, doesn’t it? But this is a really funny book that puts the principles and concepts in Stats into real-life situations where you can see the consequences of getting it wrong! It’s an easy read, written in layman’s language. Certainly very old, and you’ll notice the numbers on salaries etc. in the examples are amusingly low. Still, it’s a great way to “take a break” from stats while still working on drilling the concepts into your brain.

The book addresses sampling issues, the different types of averages and how they can be used incorrectly (remember mean, mode, and median?), the acceptable way to create charts (makes you examine carefully charts you see in the news), probable errors and standard errors, and more.

And of course, what has been drilled into my head this past fall: causation is NOT correlation! Ah, bummer. I thought I’d be able to do some research on factors that influence how much my districts use videoconferencing, but even if I find a significant correlation, it still doesn’t necessarily mean those factors caused their high use of VC. Stats is a good reality check for what you can and can’t say with authority.

So, if you’re taking a stats class and it’s driving you crazy, I recommend this little gem. Dig it out from the recesses of your nearest library and enjoy a musty smell and an amusing read. All the while you’ll be learning stats a little better!

Oh, look! Amazon.com is selling a 1993 edition. I’m sure it won’t have quite the same aura of a 1954 edition. But probably still a great read.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.