slutdragon
8mo ago
Re: the Selzer poll.
I'd like to examine the wording of this poll and the effect had on the outcome. The wording was as follows:
Likely voters in Iowa were asked: "If the general election were held today and the candidates for president were Kamala Harris for the Democrats, Donald Trump for the Republicans, Robert F. Kennedy Jr. for We the People and Chase Oliver for the Libertarians, for whom would you vote? If you already voted, for whom did you vote?"
The September poll, by contrast, did not include language targeted at early voters:
Likely voters in Iowa were asked: "If the general election were held today and the candidates for president were Kamala Harris for the Democrats, Donald Trump for the Republicans, Robert F. Kennedy Jr. for We the People and Chase Oliver for the Libertarians, for whom would you vote?"
The 2020 wording of the Selzer poll was as follows:
Likely voters in Iowa were asked: "If the general election were held today, for whom would you vote?"
The 2016 wording was as follows:
If the general election were held today, and the candidates were Hillary Clinton for the Democrats, Donald Trump for the Republicans, Gary Johnson for the Libertarian Party and Jill Stein for the Green Party, for whom would you vote?
I believe the "If you already voted, for whom did you vote?" wording of the October 2024 poll skewed the results for a few reasons. First, it seems slightly confusing. It's possible that respondents could have interpreted the additional wording as being about their perception of how others would have voted. Given that Democrats tend to vote early and people tend to know this because of the 2020 election coverage, this may have skewed the result. Secondly, poll questions should be as short and simple as possible. The 2020 wording of the question was much shorter, and did not include multiple parts for poll respondents to think through. The 2016 wording, while structured similarly to 2024, also did not include language about early voting. One rule of survey design is to avoid asking multiple questions at once, and this violates that.
Lastly, the inclusion of this language may bias participation towards respondents who did vote early. And while that does tend to favor Democrats, Nate Silver has written about how early voting doesn't predict results. Therefore, the accuracy of this poll result may be skewed.
Overall: The markets moving from 60/40 to 50/50 might have happened anyway and this just happened to be the outlier poll that triggered it. It's probably a good thing that the betting markets now accurately reflect where the race has been according to aggregators like Nate. But the result seems to be due to the wording of the poll, rather than an underlying change in the Iowa (read: Midwestern) electorate.
EDIT: For contrast, the Emerson poll showed a 53%/43% lead for Trump and was worded as follows:
If the Presidential Election were held today, would you vote for Kamala Harris or Donald Trump?
This is one that bothers me. I'm allergic to dogs (not fatally, thankfully), but if a dog licks me or I pet one, I have a mild reaction. It's incredibly frustrating how prevalent dogs are in public spaces. This wouldn't be a problem if the dogs were well trained, but they are not.
Where is my tyranny of the minority when I need it?
More options
Context Copy link