Wednesday, April 30, 2008

A post I regret

About a month ago I wrote a post where I took the numbers from our tracking poll, distributed the undecideds in the proportion of current candidate support, and used them to help assess which races were headed for runoffs.

My intent was not to say that x candidate would beat y candidate by z margin if the election was that day, but to show how the relative standing at that point might or might not imply a runoff was in the future.

I should have known, however, that in contests for offices like Treasurer and Lieutenant Governor where there is little or no media coverage any bone I threw a campaign would be used for maximum advantage.

Later the same day an e-mail came out from Janet Cowell's campaign claiming based on the analysis that Janet was the candidate best positioned to cross the 40% mark.

Then Friday of that week Walter Dalton declared that we said if the election was held that day he would win by ten points. That made me cringe because the reality is I did not say that. I said that if the undecideds were distributed in the levels of current candidate support that Dalton would have 33 and his closest opponents would be at 23. Undecideds breaking in the same percentages as current candidate support is a big if and is a lot different than saying Dalton would win by ten if the election was that day. Hampton Dellinger fairly disputed Dalton's characterization in an email to the press later that day.

What really got me was an email from David Young today. He claimed that his 19-16 lead in this week's tracking poll was outside the margin of error! Their claim was based on their own extrapolations of the undecideds to 100.

I'm sorry but taking a poll with over 50% undecideds, distributing those voters in the same percentages as the 43% who are decided, and then claiming to have a lead outside the margin of error just is not correct. A three point lead with a +/- 2.8 MOE is a statistical tie and it is not accurate to claim it as anything else.

I'm not mad at the Cowell, Dalton, or Young campaigns for the claims they've made. It's my fault for writing that stupid post in the first place. But I hope campaigns will be more accurate in their portrayal of our numbers from here on out.

No comments:

Web Statistics