It's amazing to me how little polling there actually is of Iowa right now. We're about two weeks away from the caucuses, but according to Real Clear Politics as of this writing, there have only been six polls this year. So when we're 'ooh'ing and 'ahh'ing at the rough-and-tumble between Donald Trump and Ted Cruz, we don't actually have very many data points to look at.
But let's compare apples to apples, and see where each pollster (with his own house effects and biases) has the caucuses swinging.
[caption id="attachment_258992" align="aligncenter" width="600"] Photo by Gage Skidmore on Flickr[/caption]
It turns out not all pollsters have even polled Iowa twice over the last few months. So I've taken the polls since November, which is when Cruz started to take off. The results look pretty good for Cruz, and show a stable race.
We're comparing apples to apples. By looking at the way each pollster has the race showing, we can subtract out the pollster's house effects and differences in voter model, in order to get the essence of each candidate's recent changes.
Well as it turns out five pollsters haven't actually polled close enough together for it to matter. NBC/WSJ, Gravis, CBS/YouGov, Loras College, and Monmouth all show giant swings toward Cruz, which simply means they've only had one poll since Cruz jumped to the top tier. We can ignore those.
The other five pollsters show smaller changes. CNN/ORC, Fox News, Quinnipiac, and PPP* show no meaningful change in the race between Trump and Cruz. Changes of Cruz's lead of +-2 are so far within the margin of error of both candidates' totals, that we can't draw meaningful conclusions from them. Bloomberg however did show a 7 point drop in Cruz's lead. Cruz remains ahead, but it went from a 10 point domination, to a 3 point slugfest.
So the bulk of the polling says that Cruz is holding his own against Trump, even if we factor in the new OAN poll. We don't know if this poll used quite the same methodology as Gravis's previous polling, as this poll was conducted for OAN Network. However if it did, then we have four polls virtually even from a month ago, and two polls with Trump gaining. Note that even the Gravis poll doesn't have Cruz crashing so much (He drops from 31 to 28), but rather it just has Trump jumping a bit (29 to 34).
Conclusion: Cruz supporters shouldn't panic, but Trump's attacks have moved the needle a bit. With a bit over two weeks to go, game on.
* PPP is a left-wing, troll poll outlet. I don't trust them, and am only including them here because if I excluded them, somebody would start complaining, I know it.