Is a vote for Stein/Johnson a vote for Trump/Clinton?
This meme continues to spread around the Interwebs, and unless it's understood to be mere hyperbole, it's not correct. Let's look at some reasons:
The Math
Always my favorite reason. Hopefully, I can show this meme to be false, without burdening the reader with dreaded equations. I'll address The Claim that "a vote for Stein is a vote for Trump", but the same logic applies to other claims about 3rd-party votes. I consider "a vote for Stein is tantamount/equivalent to a vote for Trump" to be essentially the same claim.Consider a hypothetical voter base with only 7 voters. Makers of The Claim start from the assumption that voters have a preference to vote for one of two leading candidates. I'm accepting this assumption for the example. Here are the preferences of our voters in a race with only two candidates:
What is a Vote for Stein Worth?
(a) What if one of those 4 Clinton-leaning voters voted instead for Stein? The election would be tied 3-3, with Stein collecting 1 vote. This is a net difference of -1 for Clinton's total, 0 for Trump's, and +1 for Stein. In terms of the margin between Clinton and Trump, this is a net value of -1, going from Clinton +1 (winning 4-3) to even.
What is a Vote for Trump Worth?
(b) What if one of those 4 Clinton-leaning voters switched to vote for Trump? The election would be won by Trump, 4-3. A net difference of -1 for Clinton as in (a), +1 for Trump, and 0 for Stein. In Clinton-Trump margin, we've gone from Clinton +1 to Trump +1, for a net value of -2.
(c) We don't like ties, though, so let's make example (a) clearer. What if all four Clinton leaners voted Stein? The election goes to Stein 4-3 over Trump.
The only way to conjure the result that a vote for Stein is even worth a vote for Trump is to:
1) Tally votes by taking into account what we assume voter preferences should be. Elections don't do this. You do not mark on your ballot who you might prefer in an alternate universe with only 2 choices. The vote counters simply do not know.
2) Use the net value calculation above, which yielded -1 in scenario (a).
But, in using net value math, we would also have to ignore that in net terms, a vote for Trump (scenario b) is worth -2. So, since 2 ≠ 1, The Claim is false.
The Politics
The math argument above is in fact a conservative debunking of The Claim. It cedes an important assumption to those making the claim. That assumption is that 3rd-party voters would even vote for President at all, if constrained to only 2 choices. Some obviously would not. About half of our eligible voters do not vote at all. Some of the votes won by 3rd-party candidates would have gone to neither Clinton nor Trump, on a ballot without any 3rd-party choices. Therefore, they do not even fully represent the -1 net value to Clinton in scenario (a).
In addition, we do not have a single national election. We have 48-50 separate state elections (Nebraska and Maine are special, as they potentially split electoral votes by district). In 48 states, the winner takes all of the electoral votes in our electoral college. This matters, a lot. Unfortunately, it means that even if a race is close overall, individual states will often not be close. What is "close"? In this context, close is close enough that 3rd-party votes could change the outcome. In fact, as of the day I wrote this post (9/24/16), in about 29 of 52 elections (counting NE and ME as 4 separate elections), the Clinton-Trump difference was more than 10%.
FiveThirtyEight.com Election Forecast
This, in a race that overall was polling with only a 2% difference in national popular vote (now 6%). Close elections don't mean most states' elections are close. In fact, most state elections aren't competitive. In those states, this argument is even more moot.
It also has to be noted that neither Stein nor Johnson have yet averaged polling even at that 10% threshold. By identifying 29 out of 52 races, I was extremely conservative in selecting competitive races. Furthermore, Stein and Johnson's effects do not add. To a large degree, they cancel one another. For example, today, in an average of 2-candidate polls, Clinton leads by 6.5%. In polls including both Stein and Johnson, that difference is 6.2%.
The net 3rd-party effect is likely taking less than one percentage point from Clinton's margin. Again, that's completely ignoring the electoral college effect.
Caveats
My conclusion, again, is drawn with multiple concessions granted to the point of view I'm arguing against. First, that these assumed Clinton voters would vote at all. The RealClearPolitics data above indicates many 3rd-party voters would not, if limited to two choices. Second, I only considered cases (a, b, and c above) where it was the presumed Clinton voter switching votes. Finally, the cases ignore our electoral college, which makes the argument moot for the 3rd-party candidates in a majority of states. If mathematically, The Claim is only 1/2 correct with the most generous assumptions granted, it's closer to false than true in the full political reality we live in.Alternate Expressions of This Idea
To say that The Claim is false is not to say that it has zero basis in truth. Here are some alternate phrasings that I would have no problem with:
- In a close election, a 3rd-party candidate might win enough votes to change the outcome.
- Liberals voting for Stein is better for Trump than if they vote for Clinton.
- Trump wants progressives to vote for Stein.
- Trump bad.
None of these expressions is equivalent to The Claim. None is even particularly close, in my opinion. Use them if you like, but if you resort to The Claim, prepare to be told why you're wrong (and you are).
