Skip to main content
ABC News
What Early 2020 Ratings Can And Can’t Tell Us

You could be forgiven for thinking that the 2020 election has gotten started way too soon. Presidential aspirants are announcing their candidacies abnormally early in the cycle, and election handicappers like our friends at Sabato’s Crystal Ball, Inside Elections and the Cook Political Report have already offered their first race ratings1 for the Electoral College. Naturally, you might wonder how much to trust these ratings more than a year and a half out from the election. In general, I find them to be useful starting points; while you might quibble with a specific rating, historical and demographic trends make it safe to put states into rough buckets even this far out, without the benefit of polls or even knowing who the candidates will be. And these experts have great track records. But they would also be the first to tell you that the ratings aren’t yet precise enough to tell you who is going to win. So here’s a FiveThirtyEight guide to digesting race ratings with an appropriately sized grain of salt.

There are two main components to handicapping a presidential election: (i) predicting what the national popular vote will be (in other words, the political environment) and (ii) predicting how states will deviate from that baseline.

It’s hard for anyone to make confident forecasts about the former this early. General-election polls at this point in the cycle (the first half of the year before the election) frequently fail to predict the actual final outcome of the race.2 At this point in the election cycle, there is still just too much time in which major, race-changing events, such as the 2008 financial crash, can take place. And, of course, we still don’t know who at least one of the nominees will be. Whether Democrats pick a strong or weak candidate3 will affect the party’s overall standing.

But the second type of forecast — predicting the relative partisanship of the states — can be made with a reasonable amount of confidence, even this far out from the election. Many states are consistent in their relative partisanship from presidential election to presidential election. For example, in the five elections from 2000 to 2016, Florida has been between 1 and 4 points redder than the nation as a whole. That’s a state that will very probably vote a bit more Republican than the national average once again this year. And the states that are less consistent — think Hawaii, which was 11 points more Democratic-leaning than the nation as a whole in 2004 but 39 points bluer in 2012 — tend to be firmly entrenched in one party’s camp, so they are safe to handicap for a different reason.

The problem is that this second type of prediction may be easier, but it’s also less valuable. If one party wins the national popular vote by more than a few percentage points, the outcome of the election will be a foregone conclusion — there will be no need to finely parse electoral-vote totals. But the closer the election, the more precise you need to be to guess the outcome correctly. This level of precision just isn’t what race ratings are intended for — there’s a reason they group states into categories “Toss-Up” and “Likely Republican” instead of projecting exact vote margins. Ratings aren’t set up to answer questions like, “Will Michigan be bluer than Nevada (as it was in 2012) or redder (as it was in 2016)?” Yet these are the questions that decide close elections.

If the election is close, which party benefits from the Electoral College could be even more important than the national popular vote — as we saw in 2016. And the Electoral College advantage tends to be determined by small shifts — a couple of percentage points or less — at the state level. Over the last five presidential elections, 12 states have vacillated between being redder than the national average and bluer than the national average.

States that have flipped relative to the nation

For the 12 states that have been both redder and bluer than the nation overall at least once in presidential elections from 2000 through 2016, their bluest, reddest and average leans

lean relative to national average
State bluest Reddest Average
Oregon D+9 EVEN D+7
Michigan D+9 R+2 D+5
New Mexico D+8 EVEN D+4
Minnesota D+6 R+1 D+3
Pennsylvania D+5 R+3 D+2
Wisconsin D+7 R+3 D+2
New Hampshire D+4 R+2 D+1
Nevada D+5 R+4 D+1
Colorado D+3 R+9 R+1
Iowa D+2 R+12 R+1
Virginia D+3 R+9 R+2
Ohio EVEN R+10 R+4

Source: Secretaries of state, U.S. House of Representatives

Even if these states don’t shift around that much relative to the country overall, they shift around enough to change which party has an Electoral College advantage. For example, in 2016, President Trump’s appeal in states like Michigan, Pennsylvania and Wisconsin caused them to shift just a few points more Republican. It just so happens that that small shift was enough to push them from being more Democratic than the nation to more Republican, which meant the Electoral College gave Republicans a disproportionate advantage in 2016, even though it had given Democrats a disproportionate advantage in 2012. Indeed, the party that has benefited from the Electoral College has bounced around from election to election, seemingly at random.

How the states in the table above will lean relative to the nation in 2020 will likely determine which party has the Electoral College advantage in the next presidential election. Future variables, like the strength of the ultimate Democratic nominee, will help determine those state leans. But they are exactly the kind of thing that we lack the precision to predict this far in advance.

FiveThirtyEight’s 2020 draft: Episode 2


  1. In which they declare certain states “Likely Democratic,” “Solid Republican,” etc.

  2. Hillary Clinton was leading Donald Trump by double digits in multiple polls in early 2015, for example. Those polls obviously narrowed as we got closer to the election.

  3. Not that we, or anyone else, knows what constitutes an electable or unelectable candidate at this stage.

Nathaniel Rakich is a senior editor and senior elections analyst at FiveThirtyEight.