Once more, dear friends: Johnson helps Clinton

On the one hand, I’m sure faithful BHL readers are sick of this. On the other hand, people are writing in widely-read political news sites that Clinton should go to extraordinary lengths to blow up the Johnson-Weld ticket. (Note: Actually offering a cabinet post in exchange for an endorsement would be a federal crime.) And the Washington Post editorial Board is repeating the claim that “Mr. Johnson takes more support from Hillary Clinton in three- and four-way polls than he does from Mr. Trump” as part of an ominous warning to Weld that he could be the Ralph Nader of 2016, swinging the election to the Republican.

So: unto the breach.

In the new Fox News poll Clinton leads by 5 in the 2-way race, by 3 in the 4-way race. Net of -2 from adding the third parties.

Stein polls at 4%. On any reasonable view of where her votes come from, she accounts for the whole loss to Clinton– probably and then some. Johnson’s 8% could be neutral, but most probably draws 1-2% net from Trump, making up some of the Stein effect.

And this is the pattern in almost every poll I’ve looked at that compares the 2-way and 4-way races. It’s the pattern in a large majority of states including most swing states. It’s the pattern you would expect from two former Republican governors running a ticket against a deeply unpopular (und un-conservative and unhinged) Republican nominee. It’s the pattern you would expect from the fact that (at this writing) six traditionally-Republican newspaper editorial boards, including the very conservative swing state New Hampshire Union Leader and the rock-ribbed Chicago Tribune and Detroit Free Press, have endorsed Johnson. I’m happy that Johnson and Weld have run a more leftward campaign than Libertarians usually do, and have emphasized Libertarians’ liberalism on drugs, crime, policing, imprisonment, and war. I’m happy that they’re bringing some young liberals into the movement, and I think that’s good for the future of libertarianism. But it’s still the case that the low-tax, free-market, free-trade candidates running against a protectionist Republican consistently bleed off a few more disaffected Republicans than they do Democrats. And the punditry, commentary, and strategy that pits Clinton’s interests directly against Johnson-Weld’s is mistaken.

The Post editorial mentioned above doesn’t itself offer any support for the claim that Johnson hurts Clinton. It just links to this John Fund piece as its footnote. But Fund provides no valid evidence for that claim, just a combination of ecological fallacies and fallacies of division. Johnson is strong among young people and out west, therefore… (without doing anything to see who those young people or westerners would otherwise support, or weighing their effect against those of other Johnson supporters). Clinton’s lead shrinks in a four-way race, therefore… (without distinguishing Stein’s effect from Johnson’s).

But what’s true of [Johnson + Stein] is not automatically true of Johnson, and in fact it is not true of Johnson. It’s the same mistake over and over again, in all these articles and commentaries, and now they’re feeding on each other, citing each other, and building a conventional wisdom out of something that isn’t true.

Update: We can quickly see the same thing in the Monday morning Morning Consult poll.
2-party: 46-39, Clinton +7
4-party: 42-36-9-2, Clinton +6

Yes, Clinton is hurt (net of -1) by including the third-party candidates. But on a reasonable assumption about where Stein’s 2% of voters go in the 2-party matchup, they fully account for that change of 1 point. That leaves Johnson somewhere between neutral and helping Clinton by 1 net point.

  • Why do you consider that the polling data is worth analyzing? In the Michigan primary, the polls had Hillary winning by 20% and Bernie won by 1.5%. People lie to the pollsters. A large portion of the population do not use land phone lines, and pollsters adjust their data based on historic voting patterns that may no longer be valid. Garbage in garbage out. The Republicans I know who say they will vote for Johnson tell me they would vote for Hillary or not vote, rather than vote for Trump. So I doubt Johnson is taking many votes from Trump.

    • jtlevy

      A) Polling is imperfect, but it’s a lot better than nothing. New problems arise and new work is done to try to keep up with them. The Michigan result sticks in your memory precisely because it’s so unusual for so many pollsters to be so far off.

      B) I’m a political scientist and we’re not in the habit of just shrugging our shoulders and saying “eh, whatever” about one of the most important sources of information about opinion and voting.

      C) I’m responding to an ongoing wave of articles and commentaries that misanalyze and misreport the same poll data. My primary concern throughout this series is to stop people from confidently saying something that is not supported by the data they’re relying on to say it. If everybody starts shrugging their shoulders and saying “we don’t really know the effects, polling is too imprecise, the numbers are too small,” I’ll stop the argument too; my concern is not to carry the ball for the value of high-precision claims about small poll effects. It’s just: if you’re going to make such claims, you need to actually pay attention to what the polls say.

      D) In these switches from 2-way to 4-way polls, Trump always loses voters– 4% in the Fox poll linked to above. (The argument is about whether Clinton loses more, and to whom, not about the fact of Trump losing some.) That is to say, people *who self-report that they would vote for Trump* in the 2-way give a different answer in the 4-way. I see no reason to think that they’re lying about that or that it’s a phantom effect; your anecdata don’t strike me as persuasive evidence against this consistent poll finding.

  • Joseph E. Knight

    Clinton and Trump are both so despicable that it really doesn’t matter.

  • Pingback: My vote in the four-way race | Hit Coffee()