Republican presidential candidate Donald Trump speaks at the California Republican Party 2016 Convention in Burlingame, Calif. (Jeff Chiu/Associated Press)

With presidential candidates Donald Trump and Hillary Clinton seemingly headed for a general-election showdown, a lot of people want to talk about the electoral map and, in particular, whether Trump could “re-draw” it or even win it.

But any conversation about the map has to start with this key fact: Over time, there are smaller year-to-year swings in the states. This places a big constraint on any presidential candidate’s ability to “re-draw” it.

In a given year, the presidential-election outcome in each state is now increasingly predicted by the outcome in that state in the last presidential election. States do shift, of course. But they tend to do so in a similar fashion.

Sometimes political scientists call this “uniform swing” — the states “swing” from election year to election year in mostly the same way and by similar amounts, depending on the national conditions in each year.

For example, consider the shifts between the 2008 and 2012 presidential elections:

Graph by Andrew Gelman
Graph by Andrew Gelman

In 2012, President Obama did worse in nearly every state — and by roughly a similar amount — than he did in 2008. This makes sense: National conditions were more favorable to the Democratic Party in 2008 than in 2012.

Of course, the “swing” between these two elections is not literally uniform. For example, Obama did significantly worse in Utah in 2012, compared to most other states — as you might expect because he was running against a Mormon. And he did better in Alaska in 2012 than in 2008, perhaps because an Alaskan governor was no longer the Republican vice-presidential nominee.

Over time, year-to-year swings within the states are actually getting smaller:

Graph by Andrew Gelman
Graph by Andrew Gelman

This trend is nothing new. One of us (Gelman) noticed it early in 2009. The political scientist Simon Jackman has written that uniform swing is now a better approximation of presidential elections at the state level than it has been since 1944.

Saying the electoral map is increasingly stable from election year to election year is not saying that there are no gradual, secular changes in individual states. Ben Highton has argued, for example, that the electoral map now affords Democrats an edge.

Rather, the stability from year to year implies that it is now harder for a candidate to “scramble” the map. And that’s the challenge facing Trump. If winning means he has to do better in some states than Mitt Romney, even as he might do worse in others — well, that’s increasingly difficult to pull off when the states all tend to move in the same direction from year to year.

To win, Trump really needs national conditions that are favorable to the Republican Party. He needs “uniform swing” in his party’s direction. Ironically, potentially favorable national conditions do exist. The problem for Trump seems to be, at least for the moment, Trump himself.