Unlikely Voter

Tweet

Quinnipiac University released a new poll on the Ohio Governor’s race that has caused a stir. It shows Governor Ted Strickland, Democrat, ahead of former Congressman John Kasich, Republican, while previous polling showed the reverse.

The swing is large at first glance. Should Ohio Democrats be excited? Should Ohio Republicans panic? Let’s unpack this.

First off, the poll itself. Strickland 43, Kasich 38, MoE 2.5. The pool was of registered voters. I come up with a 15% likelihood that Kasich is ahead. Small but not at all negligible.

How does this compare with previous polls? PPP on the 23rd showed much the opposite: Strickland 37, Kasich 42, MoE 3.9%. Pool is listed as “Ohio voters,” which I’m taking to mean registered voters. Model result: 73% chance Kasich is ahead. Before that, on the 8th, Rasmussen polled likely voters and got Strickland 38, Kasich 49, MoE 4.5. Kasich gets an 88% likelihood of being ahead.

This to me looks like it could be a legitimate shift, as it’s putting Strickland on the other side of the tipping point, but not giving him too large of a lead suddenly. Additionally, the PPP poll did show movement away from Kasich, which the Quinnipiac poll merely continues.

But at the same time it is reading the opposite of the previous two polls in this race, so I’d like to see another poll before making any conclusions. If Kasich leads in the next poll, then we dismiss the PPP poll entirely as a statistical anomaly. If Strickland leads again, then we conclude that the last month has seen a steady slide away from Kasich, because the chance of two consecutive outliers is only 1/400, which is probably too small to take seriously.

I would need to see a large body of data discrediting Quinnipiac’s methodology before preemptively dismissing the poll as wrong, though.