Wednesday, November 4, 2009

An Exit Poll Calculation Update

I added the following on Nov 9...


OK, I have to thank Patrick Murray for pointing out very nicely that I am sometimes a doofus. In my zeal to talk about how turnout matters, I mis-used the exit poll data. What I didn't think about (and should have) was how the data are weighted. They are weighted either to a vote "guess" - in the case of the early polls this time around to an estimated tie vote - or to the actual vote results (for the final exit poll numbers).

So no surprise that the calculations came out the way they did in terms of vote estimates - they were WEIGHTED that way! Duh.

BUT, the concept that relative turnout makes all the difference still is the point I wanted to make. Christie won at least partly because the Republicans turned out in greater proportions. And the independents who did turn out were more conservative and voted for Christie.

Thanks Patrick for pointing out my mis-use of the exit poll. I will try to do better next time. But I'm leaving the original posts because I don't want to pretend I didn't do it!


So this morning's NY Times published the exit polls. And they look different from the exit polls reported in The Atlantic right after the polls closed last night. The immediate exit polls calculated to a <1 point margin for Corzine. Of course Christie actually won by 4+ points.

Well the (more complete) exit poll shows this:

Democrats: 86% Corzine, 8% Christie, 5% Daggett
Republicans: 91% Christie, 6% Corzine, 3% Daggett
Independents: 60% Christie, 30% Corzine, 9% Daggett

And this exit poll shows turnout shares as 41% Democrat, 31% Republican, and 28% Independent.

Republicans turned out +11 over their share of registered voters. Democrats only managed +7 over theirs. And independents were -18. And Republicans were somewhat more cohesive: 91% voted for Christie; 86% of Democrats voted for Corzine.

But Corzine lost it with the independents, 60% of whom voted for Christie (Corzine got only 30%). If Corzine had gotten 35% of Independents and Christie 55%, it would have been a tie. The result may be that simple.

We can do the (new) math on the exit poll:

Corzine: 45.5% (.86*.41 + .06*.31 + .30*.28)
Christie: 48.3% (.08*.41 + .91*.31 + .60*.28)
Daggett: 5.5% (.05*.41 + .03*.31 + .09*.28)

So the final exit poll calls for a 2.8 point Christie win, 48.3% to 45.5%.

The final results with 99% counted (from today's Star Ledger): 48.8% to 44.6%

Pretty darn good in the end if you ask me!

No comments:

Post a Comment