Sunday, December 28, 2003

Fun with numbers 

Here's another comment on that CBS poll. The poll shows that among Democratic primary voters, Dean leads with 16%, followed by Wesley Clark at 10% and Dick Gephardt at 9%. As in any poll, that numbers doesn't actually mean that Howard Dean has 16% of the vote among Democrats; what it does mean is that there is a 95% chance that Dean's actual level of support is between 10% and 22% (In this case, the margin of error is 6%, since there were 244 respondents*). Because of the margin of error, weird things can happen in individual polls, as it did in this one in which Carol Moseley Braun is fourth with 6%. Ignoring the slight possibility that Braun had a sudden surge in support in the last two weeks that I missed, this is definitely a statistical aberration. And 6% isn't particularly impressive when the margin of error is also 6%.

Of course, even less impressive is John Kerry at 4%. This is the second consecutive poll in which he is behind one of the fringe candidates. An amusing interpretation of his poll number is that Kerry's actual level of support might be 0, since the margin of error is bigger than his poll number.

* Margin of error = 1 / square root (number of respondents)
Comments: Post a Comment

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