This is the website of Abulsme Noibatno Itramne (also known as Sam Minter). Posts here are rare these days. For current stuff, follow me on Mastodon

Categories

Calendar

September 2016
S M T W T F S
 123
45678910
11121314151617
18192021222324
252627282930  

Electoral College: Poll Churn (mostly pre-debate)

States with new poll data added since the last update: All 50 states, and DC. Notable changes in the tipping point plus New Mexico, Maine (at large), Kansas, and Missouri.

National Summary

With the latest batch of updates, there was a lot of churn that caused some big movements if you look at the charts. For a while as I entered polls it looked like Trump’s peak was definite and a clear downward trend was in place, even before the debate. But then as I continued to enter polls, most of those Clinton gains were reversed.

The primary reason for this? Florida [29 EV], Ohio [18 EV], and North Carolina [15 EV] are all currently very close in my averages. Close enough that single polls can move them back and forth across the center line. As I worked through this round of poll updates, all three of these states moved from Weak Trump to Weak Clinton and back again. Now, the actual changes in the averages in these three states were not much. They just happened to pass the zero line, and have large numbers of electoral votes.

So if you look at the “Expected Case” chart for the electoral college, you see a big move toward Clinton, then a big reversal a few days later. Because of the nature of the electoral college, you should expect (and be careful interpreting) large electoral college moves like this so long as there are big states that are close. In these cases looking at the tipping point provides a different view with a bit less volatility.

In any case, looking only at the final net change from this round of polls, we see these changes in the national picture:

  • The expected case moves from Clinton by 8 EV to Clinton by 6 EV
  • Clinton’s best case improves from Clinton by 188 EV to Clinton by 208 EV
  • Trump’s best case declines from Trump by 106 EV to Trump by 92 EV
  • The tipping point moves from Clinton by 0.1% in KS to Clinton by 1.3% in NH

Now, this round includes a handful of post-debate polls, but almost all of the movement discussed here comes from before the debate. (The exception is Florida returning to Weak Trump after the debate.)

Looking at the charts…

chart-304

Notice the clear dip from about September 19th to September 28th when Florida, Ohio, and North Carolina went slightly blue, then moved slightly red again. In the end, the “expected case” remained flat. The movement Clinton saw from winning by 6 electoral votes to 8 electoral votes in the last update was erased by New Mexico and Kansas both returning to their “normal” colors.

Both “best cases” moved toward Clinton though, with New Mexico and Maine at large both moving out of Trump’s reach, while Missouri moved into Clinton’s reach.

chart-305

The tipping point chart doesn’t end up with huge moves when big states cross the center line, so it might be a better chart to look at right now to see how the race is moving. What we see is that after big gains starting with Trump’s campaign shakeup on August 17th, Trump seems to have peaked around September 7th… way up from where he had been, but never actually taking the lead… and he has been on a slow decline ever since… but we haven’t yet seen a BIG move back toward Clinton.

By most reports, Trump did badly in the debate, and has been doing things counter-productive to his campaign since the debate. But this does not mean large numbers of people have changed sides. Maybe this will happen, maybe it won’t, but the small number of post-debate polls already included here don’t yet show it. (In fact, as I mentioned, the single state category change post-debate so far is a move toward Trump.)

Additional post-debate polls are coming in quickly now, but I had to cut off the round poll updates somewhere to get this blog post out. There will be another update soon enough, and we’ll start to see a bit better what (if any) lasting change comes out of the first debate. Of course just about as soon as there has been enough time to see that in the polls, it will be time for the next debates…

39.4 days left until polls start to close.

State Details

For those digging into the details, here are the charts for all the states influencing the analysis above:

The big close states:

chart-306

chart-307

chart-308

Weak Clinton to Weak Trump

chart-309

Strong Trump numbers from Ipsos and CVoter balance out the strong Clinton numbers from Google and leave us with an essentially tied Kansas and erase the short blue period for the state I noted last time. (The Ipsos and CVoter results came out later, but covered an earlier time period, so they moved the lines in the past, not just the current number.)

As I stated last time I am incredulous about the Google numbers, and wish there was more additional polling in Kansas to clear up what is really going on there.

Weak Trump to Strong Clinton

chart-310

New polls in New Mexico erased the short foray into red territory for New Mexico. As with Kansas, since some of the “new” polls covered older date ranges, the shape of the curve in the past changed as well as the current number.

Strong Trump to Weak Trump

chart-311

Weak Clinton to Strong Clinton

chart-312

Other states that influenced the Tipping Point

chart-314

chart-315

chart-313

Note on placing polls on the timeline

I somehow forgot to mention this in my recent interview about this site, but it seems relevant to the discussion above… Many other election tracking sites place polls based on the END of the poll’s time in the field. I place polls on the timeline based on the mid-date of their time in the field. I think this does a better job at trying to identify when directional changes happen, because if an event and a mid-date are aligned, you know that the poll was half before and half after the event. When you place polls by the end of their field time, you can’t make any sort of statements like that about how the poll results match up with the event without referring to the specific start and end dates.

For more information…

This post is an update based on the data on ElectionGraphs.com. Election Graphs tracks both a poll based estimate of the Electoral College and a numbers based look at the Delegate Races. All of the charts and graphs seen in this post are from that site. Additional graphs, charts and raw data can be found there. Follow @ElectionGraphs on Twitter or like Election Graphs on Facebook to see announcements of updates or to join the conversation. For those interested in individual general election poll updates, follow @ElecCollPolls on Twitter for all the polls as they are added. If you find the information in these posts interesting or useful, please consider visiting the tip jar.

Edit 2016-10-01 00:34 UTC to correct the one state that moved after the debates. I had said it was North Carolina moving back from Weak Clinton to Weak Trump, it was actually Florida.

@ElecCollPolls tweets from 2016-09-29 (UTC)

@abulsme tweets from 2016-09-29 (UTC)