When polls ask voters to choose between Republican nominee Donald Trump and Democratic nominee Hillary Clinton, Clinton currently leads Trump by over 1 percentage point overall. When Libertarian nominee Gary Johnson is included in the question, that lead evaporates.
Johnson himself averages nearly 9 percent support when he’s named in poll questions.
Clinton loses over 4 percentage points to Johnson, going from 43.9 percent in questions without any explicit mention of him to 39.6 percent with the Libertarian in the picture. Trump drops more than 3 points, from 42.6 percent to 39.2 percent, when Johnson is a named option in the poll.
If it seems counterintuitive that a Libertarian candidate would pull more from Clinton than from Trump, keep in mind that we’re in the middle of the conventions. Johnson was taking more from Trump than Clinton a few days ago.
HuffPost Pollster has now made it possible for readers to see this comparison. We’ve created two charts on the national presidential race: One shows Trump vs. Clinton vs. “other” candidates, and a second shows Trump vs. Clinton vs. Johnson vs. “other” candidates.
Pollsters know that naming a candidate makes a difference in their results. But it’s not as simple as saying surveys that go beyond the major-party candidates show a truer picture of the race. Surveys that include third-party candidates often overestimate their support among voters, while polls that don’t include them tend to underestimate their support.
How Does This All Differ From the Old HuffPost Pollster Chart?
The Trump vs. Clinton chart we had been displaying aggregated results from the very first vote choice question asked on surveys. For some polls, that was a Clinton vs. Trump matchup; for others, it included Johnson and even Green Party nominee Jill Stein. The logic for focusing on the first answer was that responses to subsequent questions could be tainted by the order in which the questions were asked.
We’ve now decided the risk of question order effects is probably minimal compared to the benefit of seeing how the different questions compare. So that chart is gone.
We’ve replaced it with two charts based on two different sets of poll questions. The first chart looks at all questions that ask voters whether they plan to vote for Trump, Clinton or an unnamed “other.” The second includes all questions that name at least one of those other candidates, typically Johnson. That means polls that ask both versions of the vote choice query are represented in both charts. Polls that ask only one version are represented in only one chart.
Why Isn’t Jill Stein Named on the New HuffPost Pollster Chart?
Right now, we have Stein classified as “other” on the Trump vs. Clinton vs. Johnson chart. There’s a reason for this: Not all pollsters that ask about Johnson by name also include Stein by name, so there’s not as much data on her. That’s probably because Stein isn’t guaranteed to be on the ballot in all 50 states, as Johnson is expected to be.
That said, all of the poll questions we include have an “other” category, so if Stein is not named, chances are her supporters are in that “other” category. We will continue to evaluate the decision not to name her and will make changes if they’re appropriate.
Why Did It Take So Long to Create These Charts?
We’ve been working on these changes for a couple of months. Our database wasn’t originally set up to let us select and send multiple questions from a single poll to more than one chart. The database has to automatically and accurately update our models with new poll results, but previously it only had to pull one question per poll per chart.
So the entire data entry system had to be overhauled. Adam Hooper from our HuffPost data team worked closely with the polling team to make these changes and squash bugs that came up along the way. The result is a much more streamlined process for us and the ability to show you cool things like how including Johnson in the polls affects Trump and Clinton.