Friday, October 20, 2006
On the Statistics Primer Posts
I stayed up a few nights trying to think of a way to give the gist of the statistics used in surveys without lying or doing violence to the theories. I didn't end up with any very brilliant ways of doing that. Not very surprising, of course, but then I always try to reinvent the wheel.
Those of you, my dear readers, who are statistics geeks can see where I skate on a fairly thin eye and get off just before the ice cracks. But I think I avoided any outrageous lies. I hope I avoided them.
The problem is in approaching the whole field as someone who hates statistics might do. Statistics is actually not difficult, but it can be tedious, and that turns a lot of people off. So I tried not to be tedious but that came at a cost of being a bit eellike. If you want a slower pace and firmer instruction, try one of the online statistics courses. Like this one.
The final post in this series will talk more about the confidence intervals in surveys in situations where the study doesn't use simple random sampling (like drawing names from a hat). It will also return to some other things that you should worry about in reading survey results.