Shortly before the 2016 presidential election, Donald Trump's communications adviser Jason Miller proclaimed that both Michigan and New Mexico were in play. In both states, Miller said, internal polls showed a "dead heat," and he predicted that campaigning in both places would spike.
In Michigan, of course, that prediction was borne out. Trump won the state by the skin of his teeth, 11,000 votes. In New Mexico, though, he got blown out, losing by eight percentage points.
This story seems pertinent now because of another prediction by another Trump campaign staff member focused on a different election. According to Axios, Trump's 2020 campaign manager, Brad Parscale, is targeting two other blue states for victory in the next election: Minnesota and Colorado. As with Michigan and New Mexico, one of those seems more likely than the other to go in Trump's favor.
In considering Parscale's targeting, let's look at how Trump got to the White House. In particular, let's compare two states central to Trump's victory — Wisconsin and the aforementioned Michigan — with Colorado and Minnesota.
All four states were trending back to the Republicans after voting heavily for Barack Obama in 2008. Michigan and Wisconsin got over that middle line separating a Republican win from a Democratic one. Minnesota got close; Colorado, less so.
But that's not a great way to look at things. After all, there are national trends at play that influence individual states. If we factor out the national shifts, we can get a different sense for how the states voted in 2016, relative to years past. In other words, we can see how heavily each state voted for the Democrat or the Republican in each presidential election relative to how the country on the whole did.
In doing so, we see two things. First, that Minnesota voted more Republican in 2016 than the nation on the whole, a big shift from 2012. Second, that Colorado voted less Republican than the rest of the country and trended to the Democrats after 2012. In fact, the trend in Colorado since 1996 has been to vote increasingly less Republican, relative to the rest of the country.
That suggests that speaking of Colorado and Minnesota in the same breath might be somewhat ill-advised.
It's certainly true that Colorado has regularly elected Republicans to statewide office. In fact, looking at how states have voted in Senate races relative to the national average in an election cycle, we see that in the past three Senate races in Colorado, the state has voted more Republican than the national average for Senate races in that cycle. It currently has a Republican senator, Cory Gardner, which Minnesota does not. (Minnesota has voted more Democratic than the national average in each of its past four Senate races.)
But this brings up another important factor. Looking forward to the 2020 race from the vantage point of 2018, we have more information than we did looking forward to 2016 in 2014. Namely, we know one of the candidates: Trump.
Gardner is a very different Republican than Trump. And although we don't know whom the Democrats will nominate to face Trump, we do know that it is unlikely to be someone who has the history in electoral politics that Hillary Clinton did — and, therefore, fewer people will have entrenched feelings about the candidate. That's an X factor that will almost certainly work against Trump if his approval ratings remain mired in the low 40s.
It's hard to match up Trump against an invisible candidate, but we can match him up against himself using approval data. In January, Gallup released state-by-state approval ratings covering the span of his first year in office. Those figures reflect last year, and not the recent uptick in approval that Trump has experienced, but it still allows us to see how individual states view him - particularly relative to how much support he earned in the states in 2016.
So in Colorado and Minnesota, for example, Gallup had Trump at 37 percent approval. Comparing apples to oranges here, that's a drop of six points from how much support he earned in Colorado in 2016 and a drop of nearly eight points in Minnesota.
What that fraught comparison shows, though, is that Colorado and Minnesota are about in the middle of the pack when it comes to views of Trump and how those views have changed since the election. It's not the case, in other words, that either state seems likely to suddenly embrace Trump with newfound vigor, or even that either has suddenly grown less skeptical of him relative to the rest of the country.
Remember, these were both places in which Trump lost. He needs to improve relative to 2016, not just hold steady. The recent data aren't confidence-inspiring, and the long-term trend makes Colorado look even less likely still.
But, then, Trump winning Michigan seemed unlikely in late October 2016. Polls still showed Clinton with a decent-but-narrowing lead. A lot changed in the last few weeks of that race.
A lot can change in the last 30 months of a presidential race, too.