American polls for British politicos or ‘When do we need to panic?’
You’ve seen it by now, haven’t you? Trump is leading! Or he’s tied! Or he’ll soon be leading! But hang on, some people are saying Clinton is winning. Or she’s ahead nationally — but it’s all about the electoral college over there, right, and isn’t she behind in Florida?
Having seen lots of these types of comments from people citing random polls in recent days, I thought it was time to say something. From someone from the UK who has spent far too much of his life obsessing about US politics, I thought it might be helpful to people who have invested their time more wisely to have a quick, easy guide to how to know who’s winning.
So who’s winning the presidential race right now? (July 23rd)
Hillary Clinton is currently winning by every metric we have available; in the swing state polls and the national polls.
By a lot?
Eh, by a bit. She’s probably up about 3–4 points right now. For historical context; Barack Obama was up by 1–2 at this time 4 years ago (He went on to win by 4) and up by a similar 4–5 at this point 8 years ago. (He went on to win by 7) But after a bad week or two for Clinton, that lead is going down (Her 538 chances have fallen by about 12%) and there’s quite a while to go.
What’s the best way of knowing how the race is going?
The best two resources for working out who’s likely to win in November are the proper models. That means first of all, Nate Silver’s FiveThirtyEight forecast, which uses a pretty sophisticated forecast and currently gives Clinton about a 58% chance of winning and puts her about 3–4% ahead.
The other big model is the New York Times Upshot, which paints a bit of a different picture; it thinks there’s a 74% chance of Clinton winning. The difference here shows that even detailed forecasting can’t resolve everything — but both are consistent in who they think is ahead.
But you don’t need to get that sophisticated. You can also just check the averages. A bunch of sites do this in different ways but the most reputable and useful is the Pollster average, which at time of writing has her 4.4 points ahead.
RealClearPolitics’ model is also popular, although note that it is a bit more selective than other averages and has a worse record. Right now it has Clinton 3 ahead — or 4 including third party candidates.
Quick note on pollsters- Some pollsters are pretty good generally (Marist), some are good but produce results especially favourable to Clinton (Fox — yes, really) or Trump (Quinnipiac), some are garbage (Rasmussen). Sometimes a result that looks good for one candidate is actually bad if the trendline is bad, and vice versa. Unless you want to spend endless time learning all these individually, stick with the averages and try to ignore individual polls. It all gets tossed together and something better should come out.
So why am I hearing that Trump’s winning?
Probably because people love to cite individual polls rather than averages, especially when those give shocking results. And right now, Trump leads are pretty surprising! Only a handful of polls have had Trump leading nationally at any point and almost all of them are by a company with a terrible record for accuracy. Still, that doesn’t mean these are bad polls; with an average Clinton lead in the mid or possibly low single digits, it’s perfectly normal to have some outliers with Trump leads, just as it is to have some with double digit Clinton leads — which we also have. That’s why averages are best.
That said, there is certainly strong evidence on the date this is written that the race has narrowed recently. If it has though, expect averages to narrow too. If Trump really does pull ahead, it’ll be clear.
Ah, but everyone said he was going to lose the primary too
Not the polls! The polls had Trump winning the primary from almost the very beginning, July 2015 onwards. It was all the pundits who dismissed the polls who were wrong — and probably are again. It is true that even 538 believed Trump was going to lose — but they did so precisely because they didn’t trust the polls and didn’t have a model. Lots of people try to sound smart with this point but they’re actually missing the point;
If you thought Trump would lose the primary, you thought the polls were wrong. If you think Trump will lose to Clinton, you think the polls are *right*.
Isn’t it all about swing states anyway? Why bother having national polls?
At this point, national polls are actually more predictive than swing state polls. We don’t get state polls as often and the quality tends to vary wildly until around September. Use them, sure, but treat them with a grain of salt. States tend to move in unison anyway, like national swing in the UK. If, say, Trump is leading by 5 overall then he won’t lose North Carolina by 6.
When will we know who’s going to win?
November 8th.
Fine, the actual answer is that while we obviously can’t know for sure, we should have a great idea by September unless it’s extremely close — which it very well might be. The reason is that some big things happen in the next few weeks — which in turn means you really want to ignore most polls for the next few weeks. Once the polls settle down afterwards though, they rarely change all that much barring external circumstances.
Why shouldn’t I trust the rest of the polls in July?
In the next three weeks, both nominees will pick their running mates and hold their conventions. Historically speaking, both of these events give a poll boost to their respective party which then fades away, leaving us with a much clearer view. The conventions are also often the trigger for a lot of voters who aren’t following politics that closely to tune in.
That means you should expect to see Trump pull ahead in the polls in the next two weeks. I can’t stress this enough, because it is going to happen and people are going to freak out about it. But they shouldn’t. Even John McCain — who, as previously noted, lost by over 7 points in 2008 — pulled ahead after his vice presidential pick and convention. Ordinarily, the bump will be erased by the Democratic convention as well as fading itself. So there’s no need to panic at all just yet.
Ok but there are reasons to be worried, right?
Yes. God, yes. Hillary Clinton’s favourability ratings are extremely bad by historic standards, which may stop her taking advantage of how low Trump’s are. Despite running a shambolic, half-funded campaign and espousing open racism, Trump is still not losing in a landslide in the polls. And there are a historic number of undecided voters as well as a large number saying they’ll vote third party — all that plus Clinton has just had a very bad week with an FBI admonishment over her use of e-mails and her numbers are undoubtedly falling right now. Trump is behind, but a very long way from out. But he *is* behind. That part is worth bearing in mind… and thus, not panicking.
So when do I panic?
If Trump is running ahead on a consistent basis in averages by, say, mid to late August? Then you should start buying tinned goods.
Is Trump going to win?
Maybe! 538 give him a 42% chance after all, which is definitely not nothing and much higher than you probably want it to be. But it’s unlikely based on the polls at this point. Again, that’s July 23rd. We’ll see where we are soon.
