Thursday, September 18, 2008

 

Not all pollsters are created equal

My poor father comes rushing in "CNN did a poll and blah, blah, blah". Meanwhile, CNN/Time's poll done by Opinion Research Corp is in the bottom half of pollster accuracy (http://www.fivethirtyeight.com/search/label/pollster%20ratings). Of course, CNN is going to use its own poll in its reporting because not only do they have full use of the polls internals (party registration of poll participants, women/men ratio, black/white ratio, etc.), but also because it is their poll.

And of course, many other news outlets will publish the reports (oh CNN did this poll, oh it is from a swing state). About the only thing, the polls have going for them is that they have a relatively large sample size, 800+ voters (reducing the margin of error).

Take for example, a recent CNN poll done in the state of Indiana (http://www.cnn.com/2008/POLITICS/09/17/battleground.poll/). McCain led by 6 (or 5 depending on using third party candidates) points (a reasonable result). But later that evening, a Selzer poll come out in the same state (http://www.indystar.com/apps/pbcs.dll/article?AID=/20080918/NEWS0502/809180460). This poll.... has Obama up by 3. Any of those that follow politics know that Ann Selzer (based in Iowa) has a record that most pollsters could only dream of (http://www.fivethirtyeight.com/search/label/pollster%20ratings).

Question.... which poll do you think will be repeated by the mainstream media?

I thought so.

Comments: Post a Comment

Subscribe to Post Comments [Atom]





<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]