Comments on: Two More Polls: Where Things Stand in Iowa

By: Matthew Dickinson

Matthew Dickinson — Sat, 31 Dec 2011 14:47:19 +0000

Jeff,

Good points all. Alas, I’ve stopped reading Nate since the Times went to the online pay wall, but you are right that all pollsters are not alike when it comes to reliability. Still, you need to keep in mind that this doesn’t mean those at the bottom of Silver’s rankings are completely unreliable – people tend to overstate the significance of the distinctions on his list, particularly since there’s legitimate differences re: the criteria by which to rank pollsters. As for looking at internals – that’s about all we do here, as you know if you read regularly. And that includes data on second choice options, which I’ve discussed repeatedly during this polling cycle. As you note, the real world is messy – but this is particularly true when trying to poll caucuses.

By: Jeff Winchell

Jeff Winchell — Sat, 31 Dec 2011 11:50:04 +0000

Have you looked at Nate Silver’s poll rankings that he uses to determine the quality of predictability of various pollsters? If not, ARG and Insider Advantage are ranked at the very bottom (only Zogby Interactive was worse) out of roughly 40 pollsters. I prefer to look at the pollsters ranked above average in his list, and then look at internals (like CNN’s leaving out probably 15+% of the voters because they didn’t register as Republicans until after GOP gave CNN their list).

I also like to look at things like who is the 2nd choice candidate given that your first choice is X (PPP shows this, though the layout of that data has mislead more than one commentator) or what percentage of self-identified Evangelicals prefers whom. These details contradict the simple narratives many people have, but the real world is messy.