Tuesday, March 15, 2016

Missouri exit polling misses the mark

Yes, exit polling is subject to margins of error just as opinion polling is. But this is a "teachable moment" for understanding why the polls showing Trump performing worse against Hillary than the other Republican candidates should be taken with a healthy dose of skepticism. From the same exit poll and same pool of 1,363 respondents:

We get, according to the way the results were calibrated, adjusted, and extrapolated based on presumed voter profiles, within the same exit poll two separate breakdowns contradicting one another vis a vis the overall outcome.

Trump is shown leading by five points among men, who make an assumed slight majority of the Republican electorate. Conversely, Cruz is shown leading by five points among women, who comprise slightly fewer than half of the total. This suggests a Trump victory.

Yet when we look at income, it appears as though Cruz unequivocally won the state as a whole since he beat Trump by four points among those earning under $50k/year and by two points among those earning more than $50k/year.



Dan said...

Turnout is an interesting story.

The Missouri republican primary seems to have gotten 913.5 thousand votes, with 99% reporting.

In 2012, the Missouri republican primary had 252 thousand votes.

Dan said...

More GOP primary turnout comparisons:

In the Florida Republican primary, in 2012 there were 1.67 million votes as compared to 2.37M in 2016. Florida was much earlier in the calendar in 2012 and was hotly contested then. In 2016, Florida was never close.

North Carolina 2012 973K, 2016 1107K. I was surprised NC had such big primary turnout in 2012 (May 8th) considering that Mitt Romney had already tied up the nom by then. But then I remembered NC had marriage on the ballot that same day. They affirmed traditional marriage 61% to 39%. It is interesting that NC had better turnout for the GOP yesterday than on May 8th 2012.

Ohio 2012 1.21 million, 2016 2.01 million.

Illinois 2012 933K, 2016 1.39 million

One more detail. The reason Missouri had such poor turnout in the 2012 primary is because it didn't count! There was also a caucus that year, and only the caucus would count. Therefore, a direct comparison in turnout between 2012 and 2016 probably shouldn't count.

Dan said...

Europe is slowly getting its bearings.


Greece apparently finds itself outside of Europe.

Meanwhile Merkel urged unified action, gets it, with everyone but Merkel taking unified action.

Audacious Epigone said...


Primary exit polling should include questions about intentionality for the general election and whether or not the respondent voted in the last presidential election. If Trump is bringing out people who don't normally vote and won't vote unless he is the nominee, they may be describing as unlikely general election voters at the moment.