But much of what you’ve been told about the importance of those usual factors is wrong.
It’s true that voters may like a running mate more because s/he comes from the same state or belongs to the same demographic group as those voters. But rarely does this affect their votes. Except in extraordinary circumstances, citizens vote based on the presidential candidate, not the running mate. Vice presidential candidates are significantly more popular among home state voters (measured via 0-100 “feeling thermometers”). Yet this does not translate into votes.
VP candidates don’t deliver more votes from their home states
Now, occasionally, running mates can deliver a statistically significant home state advantage. But that effect is conditional: It only happens when s/he comes from a relatively less-populous state and has served that state for many years as an elected official. Think Joe Biden.
The problem, of course, is that less-populous states have very few electoral votes, thus making them unlikely to flip the outcome in the Electoral College.
It’s a different story at the top of the ticket. Presidential candidates are also significantly more popular among home state voters. But unlike the running mates, in most of our analyses they get significantly more votes in their home states. Depending on the method and time frame of analysis, as well as the candidate’s party, generally the advantage is between 3.6 percent and 7.4 percent.
So what about choosing a VP nominee from an oft-slighted demographic group?
Already this year, we hear speculation about Hillary Clinton selecting Housing and Urban Development Secretary Julian Castro or Labor Secretary Tom Perez, in hopes of attracting more votes from Latinos.
We can’t test that directly, since no one has yet nominated a Latino — or, for that matter, an African American or Asian American — vice presidential candidate. But we can analyze the performance of some other would-be breakthrough candidates: women (Geraldine Ferraro, 1984; Sarah Palin, 2008); Catholic (Sargent Shriver, 1972; Ferraro, 1984; Joe Biden, 2008, 2012; Paul Ryan, 2012); and Jewish (Joe Lieberman, 2000) running mates. Did those candidates attract greater support from voters within their underrepresented demographic group, in comparison with previous elections?
No. In fact, we find the same pattern of results as for home state nominees. In each case, female and religious minority voters rate running mates from their own group significantly higher on feeling thermometers. That’s true for women evaluating Ferraro and Palin; for Catholics rating Shriver, Ferraro, Biden and Ryan; and for Jews rating Lieberman.
But again, more positive feelings toward the running mate do not necessarily translate into more votes. Controlling for a range of relevant covariates (such as age, income, party identification), gender is not a statistically significant predictor of vote choice in 1984 or 2008. Nor is Catholic identification in 1972, 1984, 2008 or 2012. The one exception is in 2000, when Jewish voters were significantly more likely to vote for the Gore/Lieberman ticket. But in a subsequent pooled analysis of presidential vote choice, Jewish respondents to the ANES proved to be significantly more likely to vote for the Democratic candidate in each presidential election since 1960. Further analysis shows that Jews didn’t feel more warmly toward Lieberman than toward the average Democratic running mate. So it’s not clear that Lieberman’s candidacy actually brought more Jewish votes to the 2000 Democratic ticket.
By contrast, presidential candidates typically do win a higher percentage of the vote from fellow members of a demographic minority group. Catholics were significantly more likely to vote for the Democratic ticket in 1960 when John F. Kennedy was the presidential nominee, as were African Americans in 2008, when Barack Obama ran as the first presidential candidate of color nominated by a major party. John Kerry, who ran in 2004, is the exception; Catholics were no more likely than non-Catholics to positively evaluate or vote for Kerry in 2004. One possible explanation is that Catholic identification with Kerry was undercut by some church leaders’ high-profile denunciation of his views on abortion during the campaign.
But the overall pattern seems clear: Shared demographic identity does increase voting for presidential candidates, but not for their running mates — the same pattern that we find when examining geographic identity.
So there you have it: Voters like vice presidential candidates who come from the same home state or demographic group, but it doesn’t change their votes for president. By contrast, they are indeed more likely to vote for a presidential candidate who comes from the same region or demographic group.
This evidence points to an obvious conclusion that’s consistently lost in veepstakes speculation: Presidential elections are, first and foremost, about electing a president — not a vice president.
Kyle C. Kopko is assistant professor of political science at Elizabethtown College. You can find him on Twitter @KyleKopko.