Hillary Clinton and Donald Trump. (Getty Images, Reuters)

Lots of commentary has accompanied the “Trump Bump,” as the alleged bump in Donald Trump’s poll numbers has been called. While most polls had Hillary Clinton, the presumptive Democratic nominee, up by a solid 10-point margin as late as March, this lead seemed to have shrunk. Huffington Post’s Pollster counts about 100 head-to-head polls for Hillary Clinton versus Donald Trump since the start of 2016. Their model estimates that Clinton’s lead has shrunk to 4 points.

Is this shift real? We do not think so. Based on data we have collected using Pollfish, we see the opposite: a slow and steady rise for the eventual Democratic nominee.

Here’s why the “Trump Bump” is misleading. 

With Trump having secured his party’s nomination, it is not surprising that he would narrow the gap in so called match-up polls that explicitly name the candidates. Clinton’s poll numbers are likely temporarily deflated until she secures her party’s nomination or Sanders drops out of the race, because wayward Republicans have returned to the fold more quickly than Democrats.

Moreover, with the Democratic primary still underway, some Democrats may be answering polls strategically. In other words, Sanders supporters have an incentive to try to boost Sanders’ standing by saying they would vote for Sanders over Trump, but not Clinton over Trump. This could be why polls appear to show that Sanders is a more viable general election candidate. But the vast majority of Republican voters, 93 percent, supported in Romney in 2012, despite a contentious primary. And the vast majority of Democratic voters supported Obama in 2008 — 89 percent — despite their contentious primary.

So we’ve done a different kind of polling. 

For these reasons, it is good not to put too much stock in these head-to-head polls. This is why we’ve done something different in our polling.

We have been fielding just one question since the start of the year: Who are you most likely to vote for in the upcoming presidential election? Respondents choose among these answers: definitely Republican candidate, likely Republican candidate, likely Democratic candidate, definitely Democratic candidate, or not voting. This question focuses on the party, rather than the individual candidate, because we believe the question will more accurately reflect voting in November than does polling in the spring. (For more on our methodology, see here.)

The graph below takes the Democratic poll share and divides it by the sum of the Democratic and Republican poll shares. We do the same for the Pollster trend.


Those who say they will vote for the Democratic candidate over the Republican candidate in the fall. Data: Pollfish; Figure: David Rothschild

Rather than a shrinking lead for Clinton, we show a slow and steady rise for the Democratic nominee.

Why should we believe that this trend is real? Besides the problem of strategic responses, the head-to-head polls suffer from noise in a way our data do not. Noise can stem from two sources. One is the random noise that’s inherent in all polling. This is why it’s valuable to average polls. The other is more serious: mismeasuring what individuals actually believe.

Most Americans do not focus on politics all day, as hard as it for everyone reading this column to believe. Gallup’s May 13-15 poll showed just 40 percent of adults paying very close attention and this number could even be inflated if people don’t like to admit they’re not really following politics. This can create a disconnect between the answers that people give to a pollster and their true underlying beliefs.

With the noise reduced, the trend is clearly toward the Democratic nominee

Our polling not only avoids the problem of strategic responses, but more accurately reflects people’s underlying beliefs and likely voting choice. The strategic polling part is explicitly eliminated by focusing on the party, not the individual candidates. The noise is reduced, because people are more likely to gravitate toward the parties who they routinely vote for election after election rather than specific candidates. Thus, the question we ask in our poll is one that voters are familiar with and can answer more reliably. This is in contrast to the hypothetical head-to-head matchups, which require dealing with hypothetical outcomes, a notoriously difficult task.

Our data is clear: the trend is in favor of the Democratic nominee. Donald Trump is starting his presidential campaign with a serious deficit, not a surge.

Tobias Konitzer is a Ph.D. candidate in communication at Stanford University.

David Rothschild is an economist at Microsoft Research. Find him on Twitter @DavMicRot.