Ambivalent Support, Part 2: Supporting A Non-Preferred Candidate
By Mark White
Much has been written about candidates’ “electability,” and pundits often wonder how these concerns affect the Democratic presidential primary this year. Are primary voters planning on voting for the person they want to be president, the person they think is most likely to beat Donald Trump, or are these things one and the same for voters?
To understand how concerns of “electability” are influencing voting in the primary, Data for Progress asked likely Democratic primary voters two different questions about their vote: first, a traditional “horse race” item that asked who the respondent plans to vote for, and second, a “magic wand” item that asks who the respondent would want to be president in an ideal situation. These survey items were first asked in a survey last June, which collected 2,953 likely voter responses. These respondents were recontacted in a second wave in January, and 1,619 of them completed another survey about the primary. Both surveys were conducted online by YouGov Blue.
Analyzing these two questions together, it is possible to tell who plans to vote for a candidate who is different from the one they would ideally prefer. In the June wave, 27% of respondents were planning to vote for a non-preferred candidate; in the January wave, this number is down to 22%.
We consider those whose ideal candidate was not the one they said they would vote for as being a “mismatch.” This post follows up on the analysis of the June 2019 data, examining how mismatching has changed since the first wave, who mismatchers support, and why people might mismatch. This provides some empirical evidence into the strategic decision-making process voters make when deciding for whom they will vote.
“Mismatching”
We distinguished between vote intention and preference for candidates in two separate questions. First, we asked respondents to indicate whom they would vote for in the 2020 Democratic primary if they “had to choose today.” A few questions later, respondents were then asked the “magic wand” question:
Now, imagine that you have a magic wand that could make any of the candidates President. That candidate would not have to compete in the general election, and would automatically become the President. If you could use that magic wand to make any one candidate President, who would you choose?
These questions were also asked in the June 2019 wave of the survey, and the response options were all of the candidates running at the time of fielding, plus a “none of these” option. The response options differ in the January wave: Three candidates in the current survey had not yet announced in June and several candidates have dropped out since the first wave. The figure below shows the magic wand and vote intention numbers for each candidate, broken out by time. Only candidates appearing in the January 2020 wave are shown.
Not everyone is planning to vote for the candidate that they would make president with a magic wand. In June 2019, 27% of respondents had a mismatch between their choices; this is 22% in January. This was not due to differential dropout—mismatchers in the first wave were equally as likely to participate in the recontact as those whose choices matched.
We can think of a candidate as having a “mismatching advantage” if more people say they are voting for a candidate than indicating them on the magic wand question. The figure below plots this advantage. The y-axis shows the percentage of people saying they would vote for the candidate minus the percentage of those selecting them with the magic wand. For example, 7% of respondents indicated an intention to vote for Pete Buttigieg in June 2019, while 9% said they would make him president if they had a magic wand. This means he was at a mismatching disadvantage of -2 points.
Former Vice President Joe Biden is the only candidate with an advantage reliably over zero, showing this in both waves. Senator Bernie Sanders crossed the zero threshold in the second wave only. We also see the “none of these” response dropping in this metric: Even though those ideally wanting someone else has remained at 5% in both waves, fewer people in January 2020 are indicating that they will vote for someone else.
Which candidates do mismatchers prefer? In the initial wave, 770 respondents had mismatching preferences; there were 363 in the second wave. The table below shows candidate choice under both the horserace and the magic wand questions among these mismatchers for both the June and July survey. Responses are only shown for candidates earning at least 10% in one column, so columns will not add to 100%.
In both waves, mismatchers favor Biden by a wide margin:
| June 2019 | June 2019 | January 2020 | January 2020 | |
| Candidate | Magic Wand | Vote For | Magic Wand | Vote For |
| Biden | 9% | 35% | 3% | 43% |
| Warren | 20% | 16% | 23% | 15% |
| Sanders | 13% | 12% | 8% | 15% |
| Harris | 13% | 13% | -- | -- |
| Buttigieg | 12% | 5% | 14% | 7% |
| Bloomberg | -- | -- | 10% | 5% |
| "None of These" | 10% | 3% | 18% | 1% |
Primary Strategy
The decision to vote for a candidate one wouldn’t ideally prefer involves strategic thinking about the primary process. In both waves, we asked about this process. Respondents were asked to indicate which statement was closer to their opinion:
<1> Long primary election campaigns among Democrats weaken the eventual nominee
<2> Long primary election campaigns ensure the strongest Democrats emerge to advance the party’s agenda
<3> Don’t know
Those believing long primary elections weaken the eventual nominee were more likely to mismatch than those believing a long primary would ensure the strongest nominee. People are more likely to vote for a candidate who is not their ideal candidate when they believe a drawn-out primary process will be bad for the general election. This difference, however, is not huge: 9 points last June and 6 points this January.
We also asked respondents to indicate who they think will win the primary, regardless of who they support. The same pattern of results was found across waves. In June 2019, those indicating their magic wand candidate would win the primary mismatched 9% of the time, while 41% of those believing their magic wand candidate would not win indicated they were planning on voting for someone other than that candidate. In January 2020, these were 4% and 37%, respectively.
General Election
Voters appear to be making considerations about the general election when deciding whether to actually vote for their favorite candidate. Later in the survey, Data for Progress asked a question to understand how primary voters prioritized seventeen different characteristics when determining the candidate to support:
Thinking about your priorities, if you had to allocate 100 theoretical ‘points’ among the following candidate characteristics, which would you prioritize?
One of the characteristics was that the candidate “can beat Trump.” Examples of others were: having held elected office before, supports action on climate change, is younger than 50, and wants to increase taxes on the rich.
The more one prioritized beating Trump, the more likely they were to mismatch in both the first and second waves.
Democratic primary voters understand the scope and gravity of Donald Trump’s failure as a president. The average points allocation to beating Trump in this exercise was 21, which was double the amount for the issue with the second-most points allocated. And the more importance voters place on beating Trump, the more likely they are to decide to vote for a candidate that they wouldn’t ideally prefer. This trend has remained the same over time. Democrats want to beat Trump first and foremost, and voters are attempting to think strategically about electability in the general when casting their vote in the primary.
Conclusion
Voters consider strategies and the political process of both the primary and general elections when deciding who to vote for in a primary. Voters will make tradeoffs from their idealistic wants if they believe their candidate is not likely to win the nomination or if they are a poor fit for unseating the incumbent in the general election phase. Looking at the discrepancies between the horse race and magic wand numbers allow us to infer who the perceived “practical” candidate is from behavioral intentions, define the voters making these tradeoffs, and examine when and why people might mismatch.
Mark White is a Senior Political Analyst at YouGov Blue.
Methodology
This survey was conducted in two waves. The first included 2,953 interviews conducted from June 24th to July 2nd, 2019 by YouGov on the internet of registered voters likely to vote in the Democratic presidential primary in 2020. A sample of 6,116 interviews of self-identified registered voters was selected to be representative of registered voters and weighted according to gender, age, race, education, region, and past presidential vote based on registered voters in the November 2016 Current Population Survey, conducted by the U.S. Bureau of the Census. The sample was then subsetted to only look at respondents who reported they were likely to vote in their state’s Democratic primary or caucus. The weights range from 0.2 to 6.4 with a mean of 1 and a standard deviation of 0.5.
The second wave included 1,619 interviews based on recontacting respondents participating in the first wave (a 55% recontact rate). Respondents participated from January 18th to January 27th, 2020. This sample was weighted on gender, age, race, education, region, and past presidential to the weighted sample proportions from the first wave. The weights range from 0.5 to 4.5 with a mean of 1 and a standard deviation of 0.4. Significance tests were performed using the general linear model function from the survey R package, computing standard errors using bootstrapping with 5,000 replicates.