When it was first reported that the final CNN-Des Moines Register poll before the caucuses wouldn’t be published, the Des Moines Register issued this explanation on why it wouldn’t be published:

The Des Moines Register, CNN and Selzer & Co. have made the decision to not release the final installment of the CNN/Des Moines Register/Mediacom poll as planned Saturday evening. Nothing is more important to the Register and its polling partners than the integrity of the Iowa Poll. Today, a respondent raised an issue with the way the survey was administered, which could have compromised the results of the poll. It appears a candidate’s name was omitted in at least one interview in which the respondent was asked to name their preferred candidate.

Jazz Shaw’s article asks some important questions that stop well short of conspiracy theories:

How big would the impact have been on the final results? The sample size for the last Des Moines Register Iowa poll was 701 likely caucus-goers. Let’s say there were ten people making the calls. If one person’s font size was off, there might have been roughly 70 calls where a name was left off the list of choices. But they randomize the order of the names, so all the candidates would have missed being listed, likely less than a dozen times each. Wouldn’t that randomization balance out?

I’m not a statistician but couldn’t this be fixed by weighting the results differently? To the statisticians reading this, feel free to offer insights into this question in the comments section.

UPDATE: Powerline has an update on this story that’s worth checking out:


I won’t say that I’ll trust this information. I’ll just offer this opinion: if it’s true, then the Klobuchar campaign ends in Iowa.

The other thing worth noting is that the DMR poll is the gold standard in Iowa polling, much like the Marquette Law School poll is the gold standard in Wisconsin. We’ll still have to wait on the results but the DMR poll is usually accurate.

3 Responses to “What’s behind the unpublished CNN/Des Moines Register Poll?”

  • eric z says:

    Klobuchar will stay in no matter what. She’s running for VP. She’s noting to lose. She keeps Biden and Buttigeig percentages down by being female and a clone of the other two; with cosmetic differences between the three: Biden’s slipping; Buttigeig being clearly smart but South Bend is no proof of mettle to justify the man’s galling ego; and Amy IS the best of those three. Voters may disagree, but she’s honest and not arrogant.

    Finally, if those ghost [unofficial] numbers you give turn out spot-on, it would be great. Possibly the start of reckoning within the Republican wing of the Democratic Party. Sooner is better for that to happen. Honor the Bern.

  • eric z says:

    Just rotored over to the Powerline item. Thanks for posting the link. I like the “Biden sinking like a stone,” thought. However, it’s not a fully kosher impression to take, this early, and, after all, Iowa. If not the first from the gate, nobody would care a thing about Iowa. Iowa is about as unrepresentative as you can get, or it seems so. Does anybody want to suggest it makes any real sense to have this Iowa stuff first, as with South Carolina being early.

    Super Tuesday will tell, and Bloomberg’s spending there has sense to it. If he does well it may redefine campaign strategy, both parties, from now on. Billionaires buying the office for themselves instead of buying politicians and then seeing some that don’t stay bought.

  • Gary Gross says:

    Eric, I think it makes perfect sense to start with Iowa & NH because they’re states that measure up the prospective candidates, their strengths & weaknesses, then evaluate their organizing strengths. Finally, they test whether a candidate’ supporters would run through a brick wall for their candidate or whether their support is half-hearted.

Leave a Reply