Clinton vs. Trump: 4 1/2 months out, the needle points to…


First of all, before we go any further, let’s stipulate that polls at this stage of an election are not very good at predicting final results, though not as bad as they are a year out. You really need to start looking at polling a week or two after both conventions are over, which would be mid-August or so. But you just can’t help it, can you?

Let’s start with this basic fact: Generally, if a candidate had this kind of net unfavorablity rating, you wouldn’t give that candidate a very high chance of being elected:


Unless they were running against this candidate:


Score one for Clinton! What about head-to-head polls? Despite my above caveat, there is some sensibility to looking at the polls now. Trump has eliminated his opponents, Clinton’s has gone quiet, nobody really disputes that either is the prospective nominee now, and the last week’s worth of polls even captures reactions after a major news event involving two hot-button issues, terrorism and guns. If we start with the day that Trump became the Republican’s all-but-certain nominee and run through today, we get:


You can see that Trump gained some ground after wrapping up his nomination, and Clinton lost some as her’s dragged on. But she’s now gotten a drift up following her consolidation of nomination, while Trump has dropped. Note that RCP employees a straight averaging of recent major polls:


However, not all polls are created equal. Different polling firms have different track records of reliability, and also some built-in tendency to skew either Democratic or Republican. 538 does a good overview of this, if you’re interested. Ideally, you’d do some re-balancing of your poll weightings based on historical accuracy and partisan skew. Huffpost Pollster does some version of this, and they’re showing the following:


It’s worth mentioning that in either polling aggregation, Clinton’s lead is well outside the average margin of error. Of course, the popular vote isn’t everything. In fact, in a U.S. Presidential election, it isn’t even the thing that determines the winner. RCP doesn’t yet have a “no toss-ups” version of their electoral map, but based on the latest state polls, they’re showing the following solid, leaning and toss-up states:


You see a lot of the usual “toss-up state” suspects here, but you also see two that indicate the Republicans are stretched in more territory than usual: Arizona and Georgia. Meanwhile, based on recent state polls where available and extrapolations from 2012 where there isn’t good recent polling, has an electoral simulator that can be fun and terrifying to watch. This simulator does a run of 10,000 simulations a night, for which the latest results are:


They’ve been doing this for three weeks now, and there hasn’t been a lot of variability:


In addition to silly things like polls and electoral votes, we can also look at the betting markets. These can be much handier to pay attention to than the opinions of pundits because they do the same thing (indicate the opinions of election watchers) but with the advantage of aggregating many thousands of those opinions, and literally asking the opiners to put their money where their mouth is., based on the results of several different betting sites, is currently showing:


Finally, we can look at the popularity of the incumbent party. This isn’t as reliable as when an incumbent is running for reelection themselves (for example, Bill Clinton was sitting on around 60% popularity in early November 2000, but Al Gore ended up with 48.4% of the vote), but it certainly has some value as an indicator. Obama’s approval ratings over the last 6 months look like this:


It’s certainly early days still, but looking at every statistical leading indicator we have, you would have to say that Hillary Clinton’s chances look pretty good.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s