Dem
51
GOP
49
Yesterday, we had "Who's Gonna Win This Thing?, Part I: The Siena Poll," in which we wrote that there's a lot of wonky stuff going on in the New York Times/Siena swing-state polls that are getting so much attention. We wrote, at that time, that we planned a couple of companion items today. That has now grown to "three companion items," courtesy of reader M.M. in San Diego, CA. This very item is the unplanned one, and while we were a little critical of the Times yesterday, today we give them credit for running a useful iteration of a fairly well-known polling experiment.
Here is the setup, which was executed not by The Times/Siena, but instead by The Times/Ipsos. They asked poll respondents two key questions (among a bunch of others). Here they are:
Long Question
Thinking about the presidential election in November, if the election were held today, whom would you vote for if the candidates were...Short Question
- Joe Biden, the Democrat
- Donald Trump, the Republican
- Robert F. Kennedy Jr., the third-party candidate
- Jill Stein, the Green Party candidate
- Lars Mapstead, the Libertarian Party candidate
- William Davis, the third-party candidate
Thinking about the presidential election in November, if the election were held today, whom would you vote for if the candidates were...
- Joe Biden, the Democrat
- Donald Trump, the Republican
- Other (specify)
And now, here are the third-party results when the long question is asked first:
And here they are when the short question is asked first:
So, Kennedy picks up 6 points in the second condition, Stein picks up 2 and Mapstead loses 1.
The purpose of the experiment was to illustrate two well-known phenomena, and to gauge how much they might be influencing this year's polling numbers. First, when respondents are given only two choices at the outset (with many presumably disliking both), then 18% of those folks become "third-party voters" when given the longer list. But when respondents are given the longer list first, only 11% of them are "third-party voters." We put third-party voters in quotes, because history shows that many (in fact, most) people who say they are going to vote third party in May or June or July don't actually do so in November. The devotion of those 7% who are only third-party when the questions appear in a particular order has to be particularly suspect.
Second, you might be wondering why you haven't heard about this William Davis character, who seems to have the support of 1% of the electorate. You can be quite sure that 1% won't actually be voting for him, since he's not actually a candidate for president. No, he's an editor at the Times. Obviously, some respondents are currently in the "anyone but these guys" phase of the cycle. Now, is that 1% people who don't want ANY candidate or party they've heard of? Or is it more randomly distributed among all the third-party candidates? If it's the latter, it's possible that a meaningful chunk of Stein/Mapstead/Kennedy voters aren't really for those candidates at all.
Ultimately, the point here is that the numbers for Kennedy are wildly unreliable (and note this effect has shown up in many other polls this cycle that were not experiments; he always does better if he's in the second list than in the first). The numbers will get... somewhat better, but there's always some percentage of people who stick to their third-party guns until the very second they have to cast their ballots. The question is whether Kennedy will bleed support at the usual third-party rate, or at a lesser rate because he's a prominent name. (Z)