Does The VP Pick Even Matter?

Is it Pence? Is it a Gingrich? Is it a real pirate or a pretend pirate? Are the stormy GOP waters calm now by the soothing presence of a not particularly well-known office-holder? How would've Trump's General matched up against Clinton's Admiral? Would doubling-down on Trump (by picking Ivanka) be a better move than doubling-down on women (by picking Warren)?

These are exciting political questions, eliciting no shortage of educated guesses and snarky tweets. But what's harder to find is real data. The public polling -- while not flawed -- ultimately just measures underlying partisanship or the hard ID of the different candidates.

So my colleagues and I at PSB research tried to use the same tools we use for our corporate work to answer this political question. We did an online survey (n=1,001 likely, conducted 7/1 to 7/5), with an initial 3-way horserace, Max Diff scaling, and Hierarchical Bayesian Estimation to figure out the added impact of eight different potential Dem picks and eight different potential GOP picks. We even put the tool online, so you could play around with it yourself, and look at hundreds of different combinations of Dem pick, GOP pick, and audience.

Here are some key takeaways:

Elizabeth Warren is the best pick of the Dems tested. She does better than other Dem picks in just about every matchup, and across subgroups--even White men. And she is the winner with former Bernie supporters. The theory that two women would somehow turn people off is not borne out by the data.

Meanwhile, doubling-down on Trump w/ VP Ivanka would be a massive mistake. She is the one VP pick we tested who can really drastically change the race--for the worse--right away. Even though she's popular overall and with Republicans, and is better-known than most of the GOP field.

Don't underestimate the power of political celebrity.
Maybe not Ivanka, but Gingrich & Christie do quite well with moving Republicans, including those who were non-Trump supporters, as well as the undecided. And Al Franken is one of the best Dem picks against Gingrich & Christie (aside from Warren). Meanwhile lesser-known candidates do little to help a ticket.

A VP pick doesn't automatically bring over a whole targeted population. Ernst (before she took herself out of the running) doesn't do as much to help Trump as do Rick Scott and John Thune, for example, even with White women, and even against a male Dem VP pick. On the Democratic side, Booker, who is more unknown than not among African-Americans, doesn't move this group much. Latino candidates actually do better with African-Americans, and it's one of the only groups that doesn't among whom Warren is not the top pick.

We couldn't test all possible candidates floated, and sadly we didn't include Pence (I know!) but merely Pence-adjacent candidates like Corker and Thune. But there doesn't seem to be a category of VP pick who can immediately solve Trump's various problems, even when he begins just four points down. Among white women, for example--a group Romney won +14--Trump is up only by single digits, and with Ivanka as the VP the race with this group is even.

By the same token, there isn't a Dem VP pick--not even Warren--who can immediately put away the race against Trump. Among independents, Trump leads slightly in most of the matchups. And among those initially undecided Clinton leads in most matchups. But she doesn't get to 50% or even 45% outside of her partisan base.

So the real danger here for both candidates may not necessarily be driving voters into the other's camp, but a concern that people move to Johnson, or decide to stay home altogether. There isn't a single matchup that doesn't help Johnson, and at this rate he could very well get to the 15% threshold he needs to participate in debates. Among independents and undecided he could exceed 20%. And nearly half of undecided, and lots of other voters say they'd might not even vote--even for a ticket they deemed the best from a series.

Remember as you pore over old tweets, Facebook Live, Trump-family backstory, and plane logs, the VP picks are unlikely to matter much, even less so in a year when both candidates have historically low favorability ratings. But don't take my word for it, or your gut's. Just look at the data.