I went to library to look up some statistics books this week, and my eye was caught by several copies on the shelves of Statistics without Mathematics for Psychology, by Dancey and Reidy. I did a double take. I thought statistics was a branch of mathematics. Wikipedia agrees.

But if statistics is a branch of mathematics then statistics without mathematics looks like an oxymoron, a bit like a kosher pork sausage, or vegetarian chicken breast. I took a look and the book is aimed at psychology undergraduates who don’t like maths. Fair enough. But there is a real sense in which statistics cannot function without numbers, at all. If the result of the psychology experiment is that, say 65% of people are happy, then that’s a number, and so that’s unavoidably maths. And so the title is a little misleading, a slightly more accurate title would be Statistics in which you let a computer do the Mathematics for you for Psychology.

Idea being that you put the numbers into a program, and press the button to get an answer. In terms of making stats accessible this is good. But the risk is that you don’t understand what the computer is doing. Any statistical analysis (on a computer or not) makes assumptions, and if these assumptions are wrong, so are the results. If you don’t understand the assumptions, how do you know if they are right or wrong?

So it is good to make stats accessible, but I think if I see a book called Statistics with Mathematics for Engineers, I’d worry. Not sure I like the idea of crossing a bridge designed by someone who lets the computer do all the work.

Posted in Life in the Department | Comments Off