Evaluating the Impact of Vice Presidential Selection on Voter Choice
Grofman, Bernard, Kline, Reuben, Presidential Studies Quarterly
One holder of the office of vice president, John Nance Garner (1868-1967), reputedly said that "[t]he vice-presidency ain't worth a pitcher of warm piss." (1) However, as one heartbeat away from the presidency, in the modern "red phone" era, the importance of the vice president has generally been thought to have grown. And, certainly in 2008, there was almost as much hullabaloo about the Republican vice presidential pick as there was about John McCain himself, with a long period of time in which media and blogosphere coverage of Sarah Palin was at a fever pitch.
There have been only a few attempts to determine the effect of vice presidential selection on presidential vote totals (see Adkison 1982; Romero 2001; Wattenberg 1995; Wattenberg and Grofman 1993). As voters now must take the president and the vice president as a "package deal," it would seem that sometimes voters would be unhappy with the package, preferring the presidential candidate of the Republicans (Democrats) to that of the Democrats (Republicans), but having the reverse preferences for vice president. Or, voters might have strong preferences for one party's president over the other's, but no preferences vis-a-vis the vice presidency, or vice versa.
Adkison (1982) examines both aggregate and individual-level data on candidate preferences, but not systematically and only for a limited time period. Romero (2001) attempts to reconcile aggregate-level analysis indicating that vice presidential candidates offer little to the ticket in terms of home state or regional advantages with some individual-level analyses of voting propensities that suggest vice presidential candidates can have a nontrivial impact on vote choice. He concludes that the individual-level analysis overstates the impact of vice presidential preferences on vote choice, after controlling for "rationalization" of the vote.
Most similar to this current study are Wattenberg and Grofman (1993) and Wattenberg (1995). Our study improves on these earlier attempts by extending the data into four additional elections (all in a time period in which the office of the vice presidency has been widely considered to be increasing in importance and prominence) and by examining all possible preference pairings with respect to presidential and vice presidential thermometer ratings. We, too, find limited vice presidential effects, but we would emphasize that the effects we find are nontrivial in magnitude.
Using data from the National Election Study, (2) we look at the likelihood of voting for the Republican presidential ticket among nine categories of voters, whom we label DD, DN, DR, ND, NN, NR, RD, RN, RR, with the first letter indicating which party's presidential candidate is preferred, and the second letters indicating which party's vice presidential candidate is preferred (with N indicating no preference reported, or a tie). We hypothesize that presidential preferences should, on balance, be more important than vice presidential preferences, but we also expect that vice presidential candidate preference should matter. In particular, we expect to observe a generally lexicographic ordering among these nine preferences--that is, the likelihood of indicating a vote preference for the Republican presidential ticket should increase monotonically as we move from DD to RR. As shown in Table 1, this expectation is confirmed when we look at data averaged over the period 1968-2008, and, as shown in Figure 1, it is generally true for each of the individual presidential election years as well. Indeed, on average, those with fully consistent preferences vote consistently with such preferences between 96% (for the DDs) and 98% (for the RRs) of the time, and voters with no preference for either president or vice president (NN voters) behave, on average, rather like coin flips, with a 50% chance of voting Republican.
[FIGURE 1 OMITTED]
There are several different ways to parse the information contained in Table 1 and Figure 1 in terms of developing measures of the impact of vice presidential preferences on voter choices. …