Rumors of Gary Johnson's Poll-Slippage Have Been Greatly Exaggerated
If voters are "moving away from third-party options," as FiveThirtyEight claims, the data isn't there yet
As Scott Shackford noted earlier, the Washington Post's Callum Borchers made the claim yesterday that Gary Johnson's polls are "flat-lining," a term that can be evocative of death. Over at the number-crunching site FiveThirtyEight, Harry Enten sounds a similarly pessimistic note about the Libertarian Party nominee:
Johnson's support in national polls — in both our polls-only and polls-plus forecasts — has been trending downwards slightly.3 That suggests voters may be moving away from third-party options[.]
While one tangles with Harry Enten & co. at one's peril, I don't think the numbers at all suggest that voters are moving away from third parties. Post-convention polling for Gary Johnson and Green Party nominee Jill Stein looks almost identical to pre-convention polling.
If you go to RealClearPolitics' poll-aggregation pages for three-way and four-way presidential surveys, you will see find 11 polls conducted completely since July 29 that include Johnson, and nine that include Stein. Here is what they average for Hillary Clinton, Donald Trump, GJ and JS:
HC 43.5% DT 36.5% GJ 8.8% JS 4.0%
When you take those exact same 11 polling outfits (in order to create apples-to-apples comparisons across polling methodologies), and locate their most recent survey that took place wholly before the conventions, you get this:
HC 40.4% DT 36.7% GJ 8.9% JS 3.8%
Third-party candidates were pulling a combined 12.7 percent before the conventions, 12.8 percent afterward. There's no movement there.
Now, without question FiveThirtyEight does all kinds of complicated weighting and averaging and methodology-parsing that are beyond my immediate competence. The site's rolling "polls-plus" number for Johnson has, for instance, has gone like this:
June 8-11: Between 7.7 and 8.0 percent.
June 12-18: 7.8-8.4%
June 19-25: 8.2-8.5%
June 26-July 2: 8.3-8.7%
July 3-9: 7.9-8.3%
July 10-16: 7.9-9.4%
July 17-23: 8.5-9.9%
July 24-30: 7.6-8.8%
July 31-Aug. 6: 7.5-7.9%
Aug. 7: 7.7%
Basically, Johnson had a peak from July 15-24, a period during which results from a number of his most favorable polling outfits were bunched (while a number of his least favorable didn't produce), and has otherwise stayed around the same level before and since.
But go back above to Harry Enten's suggestion that "voters may be moving away from third-party options." See that footnote in the text? It goes to a fact of some relevance to that very conclusion: "FiveThirtyEight does not include Stein's percentage of the vote." Hard to make judgments about third parties when you don't measure 'em.
Jill Stein is going through a level of scrutiny right now unprecedented for a Green Party candidate in modern history, partly out of Democrats' irritable desire to bring the last of the disgruntled Bernie-or-bust crowd into line. How that heightened exposure and skepticism translates into polling numbers is, I think, still an open question.
Speaking of which, here I am on MSNBC's AM Joy yesterday talking about these issues: