Clinton Voters Like Obama More Than Sanders Supporters Do

A month ago, the talk about Hillary Clinton’s campaign was of deja vu, with Bernie Sanders adopting Barack Obama’s 2008 model to challenge her from the left. Moreover, the Sanders campaign is fueled partly by support from liberals and young Democrats. That’s similar to the January 2008 dynamic when New York Times columnist Gail Collins quipped, “Basically, everybody’s son is with Obama.”

But where some observers have stressed the continuities with the 2008 nomination fight, others see the opposite: a role reversal. In this take, Clinton has inherited the coalition that vaulted Obama to the presidency. While black voters were a crucial source of support for Obama in 2008, they’re rallying behind Clinton this year in similarly impressive numbers. Geographically, Clinton’s 2016 support looks like the inverse of her 2008 performance. In 2008, she won New Hampshire and Oklahoma while losing Iowa, South Carolina, Virginia and much of the South; in 2016, she has done the precise opposite.

So which is it: Does Clinton’s 2016 support look more like her own 2008 support or like Obama’s? Maybe neither.

Answering that question is more difficult than it may seem at first. The overwhelming majority of political surveys are cross-sectional, meaning that they survey people at a single point in time. In theory, we could ask people who they supported in 2008 at the same time that we ask them about their current preferences. But it’s notoriously hard to elicit accurate memories of a moving target such as support for Clinton at particular moments eight years ago.

Still, with my University of Pennsylvania colleague Diana Mutz and other researchers, I have been conducting an ongoing panel survey through the Institute for the Study of Citizens and Politics that offers a great opportunity to answer that question. Panel surveys return to the same respondents repeatedly, letting us see how their views evolve over time.

The panel’s first wave took place between October and December of 2007, the period before the caucuses and primaries began on Jan. 3, 2008. The comparable wave during this cycle covered the weeks before and after the Iowa caucuses, with most surveys having been completed beforehand. Looking back eight years, we see that Clinton won a plurality of support in the pre-Iowa poll: 47 percent of Democratic respondents backed Clinton, as opposed to 26 percent for Obama and 18 percent for John Edwards.

Let’s look at the 236 voters who backed Clinton in late 2007 and who still thought of themselves as Democrats as of the 2016 wave. That group stayed primarily in the Clinton camp, with 72 percent backing Clinton eight years later and just 26 percent favoring Sanders. So overall, Clinton managed to maintain most of her support from the pre-election period eight years ago. But she also outperformed Sanders among 2007 Obama supporters, winning 55 percent of their support to Sanders’ 44 percent.

Winning neither Obama’s 2007 supporters nor Clinton’s doesn’t seem like a recipe for a competitive campaign, so where does Sanders’ support come from? For one thing, it is important to remember this is a panel survey, so there are eight years worth of voters who were younger than 18 in late 2007 and didn’t participate. Surveys show that Sanders does exceptionally well with younger voters: in New Hampshire, his margin over Clinton in that demographic was a whopping 67 percentage points. The latest wave of the panel was also completed before Sanders’ overwhelming victory in New Hampshire.

It’s also important to keep in mind that in late 2007, the Democratic nomination wasn’t a two-candidate race. In the period when Rielle Hunter was a little-known campaign videographer, Edwards was a credible contender for the presidency, having been the party’s vice presidential nominee four years before. (In fact, here’s an example of the perils of political memory: How many of us remember that Edwards finished second behind Obama in the 2008 Iowa caucuses?) And it is among Edwards supporters that Sanders outperforms Clinton, winning 55 percent of their backing. People who leaned toward Edwards and his more populist-themed campaign are an important but overlooked pro-Sanders demographic.

Another way to make use of the panel data is to look at the attitudes in 2007, 2008 or 2012 that make people likely to support Sanders or Clinton. The advantage of this approach is that we measure the relevant attitudes years before voters knew who the candidates would be, and so can rest assured that people are not simply giving voice to the platform of the candidate they already back. If a poll comes out tomorrow finding Sanders supporters talking about a rigged economy, for example, it might be that his backers first came to like him and then adopted his position on that issue.

The table below shows the average score for Clinton and Sanders support on each survey item as well as the p-value for whether those averages are different. All the measures have been scaled from 0 to 1.

First, consider ideology, coded from “extremely liberal” to “extremely conservative.” Clinton backers are a bit more moderate, coming in at 0.41 as opposed to 0.35. There is little difference between the two groups on abortion as of late 2007, but Sanders backers were more supportive of gay marriage and — importantly — of raising taxes on the wealthy.

One source of division between Clinton and Sanders has been the Iraq War, and our nation’s international posture generally. In 2007, their supporters differed little on the Iraq War, as both found themselves between 0 (“U.S. should withdraw all troops from Iraq as soon as possible”) and 0.5 (“The U.S. should set a deadline for withdrawing its troops”) on the one-point scale. Notice, too, that very few supporters of either candidate described themselves as “hawks” in the 2008 post-election survey. While the question of military action abroad divides the candidates, it does not appear to have divided their supporters.

The bottom half of the table shows attitudes that were measured in the wake of the 2012 election, between November 2012 and January 2013. We see, for instance, that Clinton and Sanders supporters differ little on questions about overall government spending or targeted assistance for African-Americans. Like the Iraq War, the Affordable Care Act may divide the candidates (paywalled), but it does not seem to split their supporters. Nor do immigration attitudes or measures of prejudice against blacks or Latinos.⁸

But notice the Obama feeling thermometer. Even back in the aftermath of Obama’s 2012 victory, those who would become Clinton supporters in 2016 rated Obama at 0.82 on the 0-1 scale, as opposed to just 0.72 for those who would become Sanders supporters. That difference remains sizable even when looking only at non-black respondents. In recent debates with Sanders, Clinton’s embrace of Obama has been striking. But more than a tactical play for black voters, that embrace may reflect deeper differences in the two candidates’ bases of support.

And it’s not just about Obama. As the table’s final line shows, Sanders voters were markedly more likely to agree that “at present, I feel very critical of our political system,” scoring 0.66 on a scale from 0 (“strongly disagree”) to 1 (“strongly agree”). Sanders backers are notably more disaffected, a fact which might explain their reluctance to back a longtime insider like Clinton.

It is always tempting to see the current campaign in light of past ones, especially when one of the same candidates is on the ballot. But the divisions between Clinton and Sanders are not simply a reflection of those between Clinton and Obama eight years ago, with Sanders stepping into the role of Obama. Voters’ orientation toward the political establishment — in this case, Clinton and Obama — is an issue that wasn’t present in 2008. And while Edwards has left the political stage, the appeal of his economic populism lives on.