Wednesday, September 14, 2016

2016 Presidential Election Prediction Further Explained

The Official Prediction...or is it?

So a few weeks or so ago I released my official™ 2016 Presidential Election Prediction. I've had comments on it, here and there, from email to phone to message boards on other sites, and there are all kinds of "interesting" criticisms. 

I don't necessarily feel the need to explain myself, as I think I did so pretty well in the original post, but I will anyway just because I want to make sure what I'm getting at is clear. I also don't consider myself any kind of infallible genius or expert, so forgive me if I ignore those particular points. All I did was crunch some numbers. Anyone could do it.

 I used a mathematical model based on the "primary model", which basically compared primary turnout from the 2012 Republican Primary and the 2008 Democratic primary (adjusted for respective increase in voter registration based on increase in national population) to the 2016 primary turnout for both parties. 2012 for Republicans because it was their last significant primary, and 2008 for Democrats because it was their last significant primary.

The thinking is that turnout in primaries gauges voter turnout for the respective parties in the general election. It is basically a measure of voter engagement. Voters traditionally don't show up in numbers for primaries as they do for the general election. Why this is? Voters don't see it as important, when they aren't all that concerned. Republicans will for the most part accept the Republican nominee when they are generally satisfied with things, and Democrats do likewise. When one or the other is very dissatisfied, they are more likely to show up for a specific candidate or candidates in the primary season, seeking to enforce change.

This happened overwhelmingly for the Republicans, while there was a drop in turnout for Democrats...again overwhelmingly. A (roughly) net zero change, adjusted for population-caused voter registration would indicate satisfaction within a party. A net positive would indicate dissatisfaction within the party and a desire to change it, and a net negative would indicate either a severe dissatisfaction with the party, apathy and a feeling of no control, or a switch to another party. Either way, bad for a party when assessing voter turnout in the general election.

The Republican turnout for the 2012 primary was low, but not horribly. The Republican turnout for the general election was anemic. This is a result, apparently, of the outcome of the Republican Primary. Voters didn't display much enthusiasm for the candidates they were presented with, and the eventual nominee was even less exciting. The turnout for the Democratic Primary in 2008 was intense. They wanted change, and Obama was that change. The general election turnout was also intense. It was lower, but still decent for 2012 in the general election for the Democrats. 

This is the basis for my prediction. It is based on math, and on absolute numbers of actual votes, not on polls. Could it be wrong? Sure. It could be totally wrong. But it follows logically. This is a weird election, so there is a chance it is totally off. I mean, this is a really weird election.

But again, it follows logic. When people want change, they show up. This year, they showed up for the Republican side of things. And they didn't show up for the Democrat side. In some states this was a pretty big swing. Florida, whose population is always in flux, had a huge increase in Republicans. They also had a small increase in Democrats. But a look at their population numbers tells why both went up. However that doesn't explain the huge increase in Republicans, other than that there were a lot of excited Republicans.

Ohio, on the other hand, is not exactly a retirement destination or beach home. Democrat turnout went waaay down (-50%), while Republican turnout went waaay up (+60%). Why is Ohio important? Because the same trend follows in Pennsylvania and Michigan. And pretty much all "Rust Belt" states. That is significant.

Florida and Ohio are states that have primaries during the "thick of things". Not all states are. Which is why I have New York pretty much static in this prediction. Yes, they had an 372% increase in Republican turnout compared to 2012. But in 2012, Romney was pretty much solid by the time New York had their primary. No point for Republicans to show up, and they didn't.

So when people ask me about New York and Trump, I shrug. I have no idea. The Primary Model doesn't work for that state.

When I said I used the "most significant factor", I mean I weighed factors based on the situation with that particular state. Early states got weighted the most. As well as "Super Tuesday" states with big populations. Ohio and Florida fall into that category. I assigned only the increase in projected voter registration to New York, data. Romney was "it" by the time New York had it's primary in 2012, so I can't take anything from that 372%. 

If New York goes for Trump, great. But I have no way to predict that using the Primary Method. This is all numbers, and all it shows is swings. There is also no way to predict actual numerical outcomes. I can only say that this state or that will "most likely" go to this or that candidate based on a trend in numbers. I called the numbers a trend because, as I just said, actual numbers are unpredictable. And the numbers I got, in some cases, are so wild I can't believe them to be actual.

So it stands to reason that any states I have "just barely" going one way or another might go the opposite. And it stands to reason that states in the later stages of the Primaries could go any way at all. I will say this, however: New York and California specifically have an enormous amount of votes to overcome for any Republican nominee, even Trump who seems to draw all kinds of people. So while I might accept the idea that he could take them, I wouldn't put any money on it.

Overall, I feel confident in my prediction of an overall Trump electoral victory. On the specifics? Not so much. There is an obvious swing towards a Republican, regardless of who it is. And Trump seems to attract certain voters of his own outside of that realm. I have a hard time believing the polls about his "likeability", based on how many people literally showed up to vote specifically for him, even in late states like California when he had it sown up.

Something strange is happening here...

No comments:

Post a Comment