An interesting thing about this election cycle compared to previous ones I've lived through is the vast amount of data available, through various channels. It is a far cry from 24 years ago, in the fall of 1996, when I had to go into my office in Austin, log int the NeXT desktop machine in the Stat Mech Center and peruse the text-interface of the USENET news groups like alt.politics for poll results for Clinton and Dole coming off the wire services. It was my morning obsession when I came into campus, when I was working on my dissertation. Even then I mostly only got national results (which are not very helpful), and only a few state by state results. As everyone knows, in reality the U.S. Presidential election is 51 separate elections (the way it should it be!).
This time the entire presidential election is fought in about fourteen states at most, which is typical. That doesn't mean votes in other states don't count. They do, in a big way. For example, Trump turnout in California will determine whether he gets a popular mandate, which this time will be very important. In that sense, California voters are very important.
Anyone jawboning about Hillary's supposed "popular vote" victory last time should be forced to reckon the huge number of California and other "blue state" Republicans who stayed home last time---arguably way more than the number of "red state" Democrats, who by contrast were encouraged to turn out and told that Hillary could sweep the Electoral College. In California, the only statewide race was for the U.S. Senate, and because of the primary system there, the election was between two Democrats running against each other! Why even bother!
This year we have lots of polls published in the news, some much better than others. Perhaps the biggest divide of information in our country is who is allowed to see "insider" poll data, or poll data of that caliber, and who is only allowed to see only "public" polls (which are mostly propaganda). The insider polls are the kind of data the campaigns actually use to allocate money.
More importantly, now that is October, for various crucial states, we have lots of beautiful broken-down-by-demographic data for both early voting (EV) and vote-by-mail (VBM), updated and shared daily by diligent amateur journalists, who are doing the real work of news reporting.
As of yet we don't know who people are voting for, but we can make good statistical guesses based on who these voters are. This is hard data, not poll numbers, and one good thing about it is that one can use it to weed out "bad" polls and confirm the "good" ones. The good ones seem to be converging well.
With early voting underway for a couple days now in critical states, the picture is becoming quite clear, and more clear every day, about what to expect when the smoke clears on election evening..
Based on the opinion and analysis of the smart people I follow, who are pouring over EV/VBM, as well as last minute party registrations and switches, the map looks like this:
 |
| This map indicates the states Trump has already locked up. It is the lower bound on his electoral votes. Based on the last election, he will likely win most if not all of the remaining unshaded battleground states. |
Trump has a solid 306 electoral votes that he is are certainly going to win at this point. This is exactly his winning map from last time. One could quibble if early trends actually make Pennsylvania so solid at this point, but each passing day makes it look better, so the trend is unmistakable. Florida, North Carolina, Georgia, and Arizona are almost certainly out of reach for Biden. Michigan is looking a solid Trump win from early data, more solid than people figured. Look for this trend to continue as other states' data comes in. Ohio and Iowa were never in doubt. Texas? Ha ha ha!
Trump is strong in Nevada and New Hampshire, and is likely to win those states, giving him at least 316. He is saying he is going to win Nevada, with great conviction, and strangely I believe that he knows more about the situation than I do.
Minnesota, Virginia, Maine, Colorado, Oregon, and New Jersey (!) are all legitimately too close to call. We don't have enough good data yet for those states (especially in Minnesota, where Trump appears to be leading). Biden is weak in all of those states. Nowhere is he close to topping 50%. Last time in 2016, undecideds broke hard for Trump at the last minute and he won every single battleground that was up for grabs. If he does that this time, and wins all the remaining true battlegrounds that are unshaded in the map above, he will go way north of 350, at least to 362, and to 377 if he wins New Jersey (the only battleground where he isn't currently tied or leading in good polls).
Another state to watch is Illinois, where Biden is leading but is weak (<50% in good polls). Trump could conceivable steal another 20 EV there if turnout breaks his way strongly. Alas even that would put him just shy of the magical 400 number!
New York, California, Washington, and Connecticut are all within about ten points in the good polls, but are longshots for him, since we are getting close to election day. Turnout there will drive up Trump's national popular vote victory, however, a result which is almost guaranteed at this point. That's why he was tweeting to voters in those states today, appealing for their voites. As anyone knows, who has been watching his rallies lately, Trump knows he has won this and is now going for the popular landslide. He will probably top 50% nationally, which will be very healing for the country, I think. The map I made in the previous vote is not out of the question, if turnout breaks the right way.
There will be no lingering chaos for days afterward. The election will be called on election night as a Trump victory. Trump will have enough states called for him that VBM counting will not matter for late-arriving votes.
Electoral College:
Baseline for Trump: Trump 306, Biden 232 (same as 2016)
Most probable: Trump 362, Biden 176
High end for Trump: 400+
Popular vote prediction:
Trump 50.4%, Biden 44.8%
No comments:
Post a Comment