A couple of weeks ago I wrote a post where I assert that there are many reasons Trump wins, maybe even in a landslide, and few that Hillary can even win at all. I suppose you think I’m coming back now to say “see?” in light of the polls tightening—as well as the recent blows taken by the Hillary campaign. Nope. I’d have the same position even if the polls were more strongly in Hillary’s favor.
I have been saying since 1992 that Hillary will never be president and I’m more sure now than ever. Since the very first Republican debate in August of 2015, I suspected Trump would capture the nomination. Not certain, but either him or Cruz. By January, I knew Trump would get the nod, and I knew he’d beat Hillary or Sanders…didn’t matter, but I was sure he’d beat Sanders way worse. I saw that it was 1980 all over again. I first mentioned that in February on Facebook. Only recently have I seen more mainstream commentators like Rush draw that comparison.
The polls are baloney. In other words, they’re always right, except when they aren’t. Too many times have they been unreliable, including in 1980 Reagan-Carter where polls showed Carter up by double digits shortly before the election, only to lose by massive landslide. …And you know what case particularly makes them unreliable? Volatile, visceral, stark-contrasts, such as Reagan-Carter and now, Trump-Clinton.
Moreover, it’s an odd way of looking at things in 2016. People with landlines who actually answer them. I haven’t answered anything but my cell phone in 15 years, and haven’t even had a landline for about the last 6 years. Who has, and answers landlines? AARP members…of the sort that never saw a ripoff of young people and young families for their benefit they didn’t like and feel entitled to. But I digress…
They’re also flawed for three primary reasons:
- They rely on “likely” voters, not people who actually for sure vote (more on that at the end of this post).
- The shifting of numbers seems to rely on a seemingly endless supply of “undecideds;” but in my opinion, “undecided” is more about whether they’ll vote or not, and not who they’ll actually vote for.
- People lie; or, to put it congenially, like to tell others what they sense others want to hear.
The bottom line is that owing to the Internet, there is now lots and lots of data available to gauge actual enthusiastic engagement, not just what someone happens to tell a stranger calling on the landline.
So let’s get down to the data points I’ve been looking at. My analysis and chewing never has anything to do with polls, and only marginally to do with all the various “looks bad” scandals on either side. I don’t think people care about the quotidian scandals, beyond whether they’ll get off their ass and vote, or not. I view polls as a starting hypothesis. It tells me what to analyze. It’s why I don’t give a hoot about the Libertarian or Green.
- Google Trends predicting a Trump win in exactly the same way they predicted Obama over McCain in 2008 and Romney in 2012. Part 1. Part 2. What makes these even stronger than they appear is that if you drill into the search data, many of the top Hillary searches are for negative items like scandals and corruption; whereas, the negative searches for Trump show up way later in priority.
- Trump’s huge rally crowds vs. Hillary’s meager crowds (they’re even photoshopping them, now)
- Trump’s hugely greater social media following and engagement
- Trump’s internal polling, vastly skewed from MSM polling.
- CafePress sales for Hillary-bashing paraphernalia are 814% higher than sales for Trump-bashing stuff.
- Data gathering apps indicate yuge Trump win.
- Landslide Indicators. This covers items 2-6, above.
- Allan Lichtman, a professor of history at American University has a model he developed that has predicted the winner for 30 years, since 1984. Here’s a September article, and one from just a few days ago, Trump prediction unchanged.
- Halloween masks. Trump’s way outsold Hillary’s, which is a longtime flawless predictor. What I liked was that people who bought Hillary masks did so mostly to be scary, while those who bought Trump masks did so to be funny.
I myself have been lucky enough to call it since 1976 (I was 15) on my own visceral sense and gauge of general boredom with status-quo, enthusiasm for the candidates, general desire for change, etc. Meta-markers. I was really lucky in 1992 and 2000 when I just guessed correct (all other times I was sure).
In the former, because in spite of Clinton’s appeal, it’s an uphill battle to oust an incumbent (unless you’re Ronald Reagan, and the economy is complete crap, as in 1979-80). On the other hand, Bush senior was already the 3rd term of Republican incumbency following Reagan; and he was no Reagan in the minds of Republicans. In the latter case, Bush junior, while representing more of solid dude and and unlikely to be embroiled in bimbo eruptions for eight years, or an impeachment, nonetheless was just not very exciting. But Al Gore had “The Hillary Factor,” in that people like me can’t even stand to hear the guy talk. It was very close, and Gore actually won the popular. I call that one 50/50 for me.
What if there was a model that predicted the winner of the popular vote for more than 100 years (to 1912), with a single exception (Nixon-Kennedy, 1960)? Now, winner of the popular vote is not necessarily the winner of the presidency, but exceptions are rare, and it has to be a very tight race—and this is not likely to be.
Introducing professor Helmut Norpoth, PolySci, Stony Brook University. He developed this model, which he calls the Primary Model, that rather than using polling data—where people can say anything (like a dietary survey)—uses data from primary elections where people actually got off their asses and voted.
It has two basic elements:
- Which party is incumbent, and whether 1st or 2nd term (hard for the opposing party to oust 1st term incumbent—rarely happens—easy to switch parties after 2nd term—usually happens)
- How well the party nominee performed in their respective primaries against the next most competitive candidate
So, you can read about the generalities of the model here, and very much do read his entire discussion as to the specific prediction for the 2016 election. Here’s some elements you might find interesting.
- Trump beats Hillary by 87% likelihood. That was predicted March 7, 2016 and hasn’t changed.
- Had Sanders been the nominee, Trump would have beat him with 99% likelihood.
- Had Cruz been the nominee, he would have lost to Hillary, but beat Sanders.
Thank lucky stars that Cruz’s creepy weird-guy factor was just too much for most people.
Alright, I’m on the record.