On Saturday night, the world of political polling and elections analysis experienced a complete meltdown.
The Iowa caucus poll conducted by Selzer & Co. on behalf of the Des Moines Register and CNN, a survey that has generally set a very high standard for reliability, was set to be released live on CNN on Saturday night. Had everything gone according to plan, the results would have provided a key data point in what has been a largely under-polled contest. (Though it goes by many names, we'll refer to it as the DMR poll here.) But before the poll was released, a Pete Buttigieg supporter told the former mayor's campaign that Buttigieg's name had been left out of at least one interview. It's not clear how widespread the problems with the poll were, but the outlets collectively decided not to release the results, leaving the world of political data with a murkier view of the Feb. 3 caucus race.
This might initially seem like no big deal — other organizations have already polled this race, and more surveys were set to be released before Monday night. But, for better or worse, the DMR poll doesn't just measure reality: It helps create it. Candidates who want to claim late momentum and bring people onto the bandwagon have one less opportunity to do so, and the media and campaigns now lack a benchmark they could have used to evaluate who is overperforming expectations and who is falling short in the aftermath of the race.
In primaries, there's a well-documented bandwagon effect. Voters aren't divided into rigid partisan camps, since they're all in the same party, and a candidate who claims to be surging can attract voters who want to support a winner. Polls can be the grease on the wheels on the bandwagon — if candidates can point to a good, late poll showing that they're gaining momentum, they can get more media coverage and draw more supporters. But now the DMR poll, which could have helped someone gain speed, is no longer in play.
Primaries are also partially governed by expectations. If a candidate outperforms the expectations of the media, voters and other campaigns, they're likely to earn positive coverage and attract new supporters from candidates who didn't hit their marks. Given pollster Ann Selzer's reputation, the DMR poll would have helped everyone set expectations for who might win or lose Monday. Voters and journalists have enough data from other pollsters to effectively set expectations, but some will feel emboldened to substitute unsubstantiated rumors for the real DMR poll and think about the race in a much less rigorous way.
Other pollsters have released surveys, so no one would be going into Iowa completely blind. But this last-minute crisis hasn't just changed what we know about the contest: It's changed the dynamics of the caucuses, too.
David Byler is a data analyst and political columnist focusing on elections, polling, demographics and statistics.