How much are third parties impacting the election?

Okay, as we know, we hear it all the time. Third party candidates are spoilers. They're going to cost Clinton the election, blah blah blah. But is it really true? Let's examine the data.

Trump vs. Clinton

As of now (July 29), Trump and Clinton are effectively tied at 44.3%.

This means about 11.4% aren't being counted, and it seems strange there are so many undecideds. These guys may be looking at third parties, or they could be staying home. The best way to find out is to look at the polls with third parties included.

Trump vs. Clinton vs. Johnson

Here, Trump has 39.6%, Clinton has 39.2%, and Johnson has 8.6%. Johnson has effectively drained about 5% from each candidate, with Trump only ahead by 0.4%, which is very negligible in polling. Still, even with Johnson included, we're seeing about 12.6% refusing to respond. Unless Stein has a lot of support, that 11-13% or so who refuse to support any candidate. Support for Johnson seemed to directly come out of both candidates, and it did so about equally.

Trump vs. Clinton vs. Johnson vs. Stein

Here, Clinton has 40.4%, Trump has 40.2%, Johnson has 7.2%, and Stein has 3%. In this scenario, Clinton is ahead by 0.2% (negligible), and once again, the two parties seemed to take about 4% from each candidate respectively. Johnson is down 1.4%, but both Clinton and Trump are up almost a point from the three way scenario. Here, we see the least undecideds, about 9.2%, so we see a drop of about 2-3 points in this category when Stein is included.

Still, all in all, the net results are about the same for Clinton and Trump, and while third party candidates are taking from them in some ways, they're doing so pretty much equally between the two candidates. You might see Trump or Clinton up a few tenths of one percent, but the overall impact on the election is arguably negligible. This is, admittedly, at the national level. Things might play out differently in various swing states. So just to be safe let's check how things are working in some of the larger, closer swing states.

Florida

2 way - 0.3% Trump

4 way - 0.2% Trump

All in all, the presence of Johnson and Stein make a pretty negligible difference.

New Hampshire

 2 way - 3.7% Clinton

4 way - Tie

I wasn't going to bother with New Hampshire since it's such a tiny state, but based on my election scenarios I've been releasing, New Hampshire could be the state that decides the election if the election becomes close enough where a single state flipping is essential (say, my Trump +3 and +4 scenarios).

Here we see a much larger difference. However, looking at the data carefully, we see that the 2 way has way more polls (data points) to consider, whereas there is only one poll for the 4 way. Comparing the data directly, since the same polling agency had a 2 way and a 4 way scenario for that day, we see that the 2 way was +2 Clinton, whereas the 4 way was a tie. So we can argue here, Clinton loses 2 points due to Johnson and Stein in net. Still, there's only one data point to compare here and that's not good for larger trends. Polls have margins of error. They are not accurate in and of themselves. This is why some states have polls ranging from like +6 Trump to +9 Clinton. I mean, individual polls can be wrong. So while I will point out there is a trend here based on limited data that suggests that Clinton lost a point or two here, it's kind of hard to say how this will actually impact the election.

North Carolina

2 way - 2% Clinton

4 way - 2.7% Clinton

Ohio

2 way - 0.8% Clinton

4 way - 1.4% Clinton

Similar story as Florida and North Carolina. The effects are negligible and well within the margin of error.

Pennsylvania

2 way - 4.4% Clinton

4 way - 3% Clinton

Here we see a larger difference, with Clinton losing 1.4% over the introduction of Johnson and Stein, but I would still argue this is well within margins of error. If you recall my analysis from the other day, Pennsylvania was at 3.2%, so Clinton's lead actually increased. A single poll either way can have a difference of a point or so, and the only reason the 4 way has less of a lead is because one Quinnipiac poll leaned more strongly toward Trump, and because they averaged four polls for the four way vs. five polls for the two way. When you have four pro Clinton and one pro Trump poll vs three pro Clinton vs one more strongly pro Trump poll, that will make a difference in the overall averages. As such, this may still be attributable to just the fact that polls have margins of errors. We've seen states and election scenarios flip dramatically in my election analyses over new polls coming out and shifting entire scenarios, with candidates losing like 30 electoral votes because a handful of new polls came out that shifted the averages.

Virginia

2 way - 5.3% Clinton

4 way - 5.0% Clinton

Nothing particularly interesting going on here either.

Conclusion

All in all, despite the hyperventilating from rabid Clinton supporters, there's little evidence that third parties are really causing a single candidate to have a massive deficit that could realistically swing the outcome of the election. Most of the differences third parties are causing are fairly negligible and come down to less than one percentage point all things considered. There are scenarios in individual states in which this may not be the case, but even this can be at least partially attributed to mere polling errors and/or lack of reliable and consistent polling in those states. After all, polls have errors. They can be off by as much as 3-5 points, or even more if you take into consideration the possible error of both candidates. You only have a good idea what's going on looking at trends and averages from many polls, and if there isn't a lot of data, or there's less data in one situation than there is in another, that can make a difference in and of itself. As such, I have yet to see strong evidence that will suggest that third parties are influencing the election either way.

At best, we might see them have an effect if we have, say, a repeat of the 2000 election where the entire thing comes down to a handful of votes in one state. This could theoretically happen in my Trump +3 or Trump + 4 scenarios where the entire election could come down to New Hampshire or Pennsylvania or something, but that is pretty much like a worst case scenario for Clinton. All in all, assuming the current data is realistic and unreliable, it is impossible for such minor polling differences to have a significant effect on the race that will dramatically change the course of the election. You might lose a really close state or two MAYBE, but considering how far Clinton is ahead in the electoral college it won't make a difference.