Thursday, October 31, 2024

Trying to combat herding by looking at the most extreme polling results of each state

 So, if most pollsters are herding and smoothing out the averages to follow trends rather than risking being wrong by posting an outlier, then maybe in order to get a good idea of what's going on, we should look at the spread between the polls. After all, not ALL pollsters will herd, and if we get an idea of what the most extreme pollsters are saying, as they stick out like a sore thumb, maybe the outcome will be more of the more extreme ones rather than the averages. 

With that said, I'm gonna look at the polls on each of the 7 swing states and say what direction I feel like the energy is going in.

Michigan

 So, the average is Harris +0.4, but the polling results range from Harris +5, to Trump +2. And I know that there actually are a couple michigan multi candidate polls that RCP hasnt included despite otherwise not keeping up their multi candidate data since RFK dropped out (hence why I stopped tracking it), and they also seemed to be in the Harris +3 to Harris +5 range.

So...it seems like the spread here...favors Harris. And if we're gonna see an overperformance either way, I'd say it would be...for harris. Most of the trump polls are riding the average or are Trump +1 or Trump +2. The Harris polls tend to be relatively extreme. The reason the average favors trump as much as it does is because there are simply more of these Tie-Trump +2 polls. 

Wisconsin

Wisconsin is nominally Harris +0.2%. However, the results range from Harris +6 to Trump+2. So...this is basically mirroring Michigan in a way. Most results seem to be herding from Harris +1 to Trump +2, and then you got an outlier in Harris +6. It's very possible that this is more blue than it looks, if we assume the most extreme outlier polls are onto something.

Nevada

Here we have Trump +0.5 average, with results ranging from Harris +1 to Trump +2. So...basically, uniform, slightly in favor of Trump. I'm gonna call the average probably dead on. It's kinda symmetrical, but slightly favors Trump. But is still kinda a tossup. It's a tossup, that tilts trump. So basically it is what it is.

Pennsylvania

Trump +0.8% average...with the results ranging from Harris +3...to Trump +3. This is a true tossup, much like Nevada. Slightly in favor Trump, much like Nevada, but still a tossup. Most polls also seem to trend trump. So...idk. NV and PA are both "tilt trump" in practice here for a reason. Very close. Could go either way, but with trump having a slight edge.

North Carolina

Average is Trump +1.0, but the results range from Harris +3 to Trump +5. All in all, the Trump +1 average seems well earned all things considered, as the spread is 4 points in either direction. So...I'm gonna say this one tilts/leans trump. 

Georgia

This has an average of Trump +2.4%, with a spread between a Tie and Trump +6. I think it's very much in the lean trump territory and the odds of it going Harris are relatively low. 

Arizona

Here we have an average of Trump +2.5%, and a spread between Harris +1 and Trump +8. The range very much leans in favor of Trump.

The takeaway

As such, what can we assess from these polling ranges and averages?

Well, it seems clear AZ and GA are in the trump column. Let's just say that. They are. The majority of polls, if not all polls go toward trump, and it ranges from a tie to at most harris +1, to Trump +6-8. The average is +2.5 so yeah...that range of polling is very much in Trump's direction, and I feel comfortable calling those states for Trump.

The other five are nominal tossups. However, Wisconsin and Michigan both have polling ranges that favor democrats. Republicans can take them. I mean there are results that are as high as Trump +2 for trump, but there are also results as high as Harris +5 to +6. The herding is going toward the center, but yeah. That's the realistic 6-8 polling range that we see.

North Carolina is doing that in the other direction for Trump. We see Harris +3 to Trump +5, but that range generally favors Trump. Harris can win, but Trump is the obvious favorite. 

Nevada and Pennsylvania are the real head scratchers. Nevada goes Harris +1 to Trump +2, and slightly favors trump, so im tempted to call it a tilt trump state. But it is a tossup. 

Pennsylvania is the crown jewel and the state that basically makes or breaks both campaigns. It's Trump +3 to Harris +3, so perfectly balanced. But the average has more polls leaning toward trump. Could this be herding? Maybe. Could it be what the results are? maybe. But yeah. Tilt trump.

All in all, this gives us...just about the same map we already have. With MI and WI being blue tilting tossups, NV, PA, and NC being red tilting tossups, and GA and AZ being red leaning not tossups. The averages....end up KINDA accurately reflecting the range of the polling. And my current prediction...stands as it is. 251-287 is the most likely outlook, although the error in WI/MI/PA/NV/NC can make ANYTHING happen. AZ and GA, again, i feel more confident to write them off to the red column, but the other 5 is where the real magic is gonna happen. If things tilt toward the dems, they'll win. If things tilt toward the reps, they'll win. It's just a matter of which side the error actually favors. And again, keep in mind, a point in either direction can make or break this entire thing. So yeah, if anything, this exercise just confirms to me that...this thing is a huge tossup right now and the odds my formal prediction has are...roughly accurate.

No comments:

Post a Comment