SooperAnalysis: GOP Candidate Poll Results II

Now comes the gloating.

I wanted to test the accuracy of twitter flash polling – this is not scientific in almost any way except in some quasi-metaphysical twitter way that science does not understand yet.

The Method

I bombarded my followers and the conservative hashtag with a simple poll asking those who supported Sarah Palin who they would vote for after she announced her leaving the race. I closed the polling pretty quickly the next morning, as I saw that the Ron Paulestinians had reared up on their hind legs and started to skew the results.


After gathering the results, I averaged results from three polls right before the announcement, and added a percentage of support according as my twitter poll results predicted would be added to each nominee.


These were the results of my twitter poll predictions:

Forecast of GOP candidate results from SOOPER-ANALYSIS

This is the exact text of my prediction:

Here, we clearly see an enormous bump for Herman Cain, who gets the majority of Sarah Palin’s supporters. They even know that his name isn’t ‘Herb’! (maybe I shoulda ran “Herb Cain” also…)

Romney loses 2 points, Perry only gains one, but comes within striking distance of Mitt. If he can improve his debate performance, maybe he’ll take the number two slot. Gingrich moves down part of a percentage point, and Michelle Bachmann sees a small loss in support.

And now, the first results since the announcement from Democratic pollster, PPP polls!

click on the image to see the PPP poll results

PPP saw the exact bump I measured through my twitter poll – Cain getting enormous increases from Sarah Palin’s departure. They have him at 30% and I have him at 28%. The other accurate prediction are Rick Perry‘s numbers – we both have him at 14%.

Twitter polling was less accurate with Romney, as PPP has him at 22%, while the prediction was a very low 17% (rounded up). Gingrich also had a big difference – 5% from my prediction of 10%, with PPP at 15%. These could be representative of the undecideds choosing between the two candidates, as my “undecideds” was at 14%, and theirs was about less than 10%.

Another explanation of the poor prediction is simply that both Romney and Newt are surging for reasons other than Sarah Palin’s exit.

Of the secondary candidates, we both have Bachmann at 5%, twitter has cruhrazy ROn PaUl at 8% to PPP’s 5%, twitter Santorum at 3%, PPP Santorum at 1%, and similar results for Huntsman and that other guy.

Considering how unscientific the poll is, I’m satisfied with the results and the ability to measure public sentiment through twitter. In this case at least, the pool seemed to reflect well the overall conservative mood.

What do you think? Should we do some more twitter poll Mexperiments? Maybe I’ll poll you all about it!