Monday, August 28, 2006

Polls and a grain of salt

The New York Times has produced a nice primer on the strengths and weaknesses of polls, and how to read them with a skeptical eye in order to draw insights while rejecting spurious data.

The main lessons: Look at the size and makeup of the sample and how it was selected; beware of attributing much significance to subsamples, most of which are too small to be valid; know what "margin of error" means; and look at how the questions were phrased.

As well roll into the election season, you will hear myriad polls quoted supporting one side or the other. A good rule of thumb is to treat polls the way you should treat horoscopes: "for entertainment purposes only." But if you want to take them seriously, do your homework first.

, ,

2 Comments:

Blogger KnightErrant said...

The order of the questions, seldom revealed, can also influence the results.

8/29/2006 1:43 PM  
Blogger Sean Aqui said...

Agreed. But most reputable polls release the questions exactly as asked. It annoys me when poll stories don't link to the poll itself; but you can find the details pretty easily on the Web.

8/29/2006 1:52 PM  

Post a Comment

Links to this post:

Create a Link

<< Home