If you’ve watched or read any presidential election coverage over the past month, you’ve probably heard that Hillary Clinton is beating Donald Trump in the polls. And that’s absolutely true – according to the RCP average, she leads Trump in by 6.8 percentage points in a one-on-one contest and by 5.6 points with third parties included.
But something weird is happening in the state polling.
State polls – especially in traditional swing states – aren’t consistent with a seven-point Clinton lead and our familiar Electoral College map. And they’re not exactly consistent with a larger or smaller lead for the Democratic nominee – they’re all over the map.
The State Polls Are Weird
This table shows two data points. In the right-hand column, it takes what we call a “uniform swing” from the 2012 results to the current national poll average, and applies them to the 2012 state results. The polls currently show Clinton’s lead to be about three points larger than President Obama’s win in 2012. So we just take all of the state-level results from 2012, and shift them toward the Democrats by three points. If Clinton really were up by seven points nationally, and the underlying map hasn’t shifted, this is what we should expect to see.
The second data point (in the left column) is more straightforward: It shows the current RCP averages in those states.
At first glance, some of these results don’t make much sense. Clinton is barely beating Trump in Arizona, where (based on Mitt Romney’s 2012 showing) he should be leading. But Clinton’s lead in Pennsylvania is much narrower than one might expect given her nearly seven-point national lead.
But not every state has a puzzling result – Clinton and Trump are about where we would expect them to be in Florida and Georgia given her national lead. So why would some states shift significantly and others behave as if it were still 2012?
Making Sense of the Situation
There are many valid ways to think about these discrepancies, but we’ll limit ourselves to three reasonable, straightforward approaches.
First, one could focus on the national polls and mostly ignore the state polling for the time being. Right now there aren’t that many swing state polls, and there are a number of swing states (e.g. Wisconsin and Nevada) where there aren’t even enough polls to calculate an RCP average. And many voters are still undecided. So it might be prudent to give state polls time to fall in line with national polls and not sweat the discrepancy too much yet.
Second, it could be that Clinton really is up by seven points nationally, but that the electoral map is being remade. For example, suppose that Trump’s coalition relies more on non-college-educated whites than Romney’s, but that Trump underperforms traditional Republican candidates with Hispanics. This could push his margins down in heavily Hispanic Arizona while helping him in Pennsylvania. Additionally, Trump has been performing relatively poorly in some conservative states – he leads Clinton by seven points in the most recent Utah poll, but Sen. John McCain won there by almost 30 points in 2008 and Romney won the heavily Mormon state by nearly 50 points four years later. So maybe Clinton’s underperformance in some of the swing states will be balanced out by an over-performance in states where social or ideological conservatives refuse to vote for Trump.
It’s important not to get too carried away with this approach. Every political junkie and election wonk would find a new map fascinating, so it might be tempting to look at the data and convince yourself the map is radically changing. But it’s still early, and there are relatively few polls. And if Clinton and Trump do shift the map somewhat, the changes will likely be at the margins – most deeply red states will probably still go for Trump and most deeply blue states will likely still go for Clinton.
Finally, you could recognize that Clinton leads but let the state polls add some uncertainty to your thinking. This is subtly different from the first approach. The idea here is to see that Clinton is leading, recognize that something is going on in the state polls, but decline to speculate about what that difference will mean and where the state polls will go. Maybe the polls are rightly signaling a slightly unusual map. Maybe by the conventions the electorate will polarize along familiar lines, and the final result will look very similar to the 2012 or 2008 map. Or maybe this gap between state and national polls will persist and we won’t know which is the right explanation until after Election Day. The point here is that “I don’t know, and I’m baking that lack of knowledge into my thinking” can be an honest, viable answer to a question.