Let me walk you through this poll as an explanation of why bad polling is sometimes better than no polling. It’s an Idaho press release that confidently says: “Donald Trump will likely win Tuesday’s Idaho Republican presidential vote, a new Idaho Politics Weekly poll finds,” despite the fact that the poll is… actually saying the exact opposite. Why?
- First, dated Feb 17-26. Which is to say, before Super Tuesday and the ongoing series of dumpster fires for Trump. That means that it’s stale, and should have been released last week. No idea why it wasn’t. I suspect that there are other stale polls out there, too.
- More to the point, here’s the topline: Trump 30, Cruz 19, Rubio 16, Carson 11, Don’t Know 11, Other 9, Kasich 5. Which means that even before the ongoing attacks – which we know are working, thanks to this weekend – Trump was only at 30%. Between the undecideds and the Carson voters, either Cruz or Rubio would find it trivial to make up that gap.
And if you think that I’m being a Pollyanna with that last point, consider: it’s essentially what happened in Kansas and Oklahoma, which are two states that have the same general ‘feel’ as Idaho. Or the state races generally: the polling averages tend to put a ceiling on Trump’s support, and one that he typically doesn’t reach. If he’s getting 30% a week out, the question becomes: who will he be fighting for second place?
And that’s why bad polling is better than no polling. You can look at bad polling – although to be fair, the really only bad polling practice here was the decision to sit on this thing for a week – and try to figure out what the heck is going on. With no polling you’re just a rat on the RCP polling page, feverishly hitting refresh. Well… maybe not ‘you.’ You probably don’t have my condition, after all.