It's our mission here at The Washington Post to keep our readers as well-informed as possible. Often that means presenting new information that reporters have discovered. However, it can also mean helping readers avoid unimportant or misleading information.
With the latter objective in mind, we have put together a set of guidelines to keep in mind when providing information about elections. We are working here at a slightly higher level than beginner level. For example, I won't go into the fact that elections are decided by the Electoral College rather than the popular vote. (Well, that's a different story.) Instead, we'll explore common pitfalls people fall into when presented with seemingly interesting information.
“Red” and “Blue” states are often less red or blue than you think
A standard map of election results shows the winner of a presidential election in a state by color-coding the state red (if Republicans received more votes) or blue (if Democrats received more votes) . Although this convention is less than 25 years old, it has become a standard. That's fine.
The 2020 contest will produce the following maps:
But it's strange to present Florida and West Virginia, for example, as equivalent. Not only because the former had about six times as many presidential electors, but also because Florida's overwhelming support for Donald Trump was much lower.
With support gradients, your map will look like this: There are fewer brightly colored states.
You can also look at this another way. Here are the 2020 results, from the strongest supporters of Joe Biden to the strongest supporters of Trump. For example, did you know that Texas gave Trump the third-closest victory?
This scale is clearly difficult given that the results are off scale due to the inclusion of DC. But the important point remains the same. There are relatively few states where the election was a landslide in favor of one candidate or the other.
Biden and Trump won at least 40% of the vote in 38 and 40 states, respectively. Biden won at least 45% of the vote in 30 states. Trump is 33 years old.
Past results are not as predictive as we think
In the past nine presidential elections, most states have voted heavily for Democrats and Republicans at least once. Twenty states (plus D.C.) consistently vote for candidates from only one party. The remaining 30 of his have been flipped at least once. Over the past nine elections, each state has voted Republican an average of 4.7 times, or just under half of the time (on a scale of 0 to 9).
(Each state below contains a 9-square grid corresponding to these nine elections; they are identified by the key in the top left next to Alaska.)
Certain geographic locations were sometimes emphasized as “bellwethers” of presidential election results. Ohio is a prime example of this, with candidates supported by the state winning the presidency from 1988 until 2016. (And before that, you get the idea.) And in 2020, that wasn't the case.
Some Trump supporters sought to hold up the discrepancy as evidence that something strange had happened. But leaders are leaders not because they drive results, but because they happen to reflect results. And when they don't, it means they are no longer in the vanguard. Even though he allowed a hit in the ninth inning, that doesn't mean the hit was questionable and didn't count. That means the perfect game is gone.
Swing states work similarly. If the results in a particular state are expected to be close, we consider that state a “swing” state. Usually, this is because the past results were close and, importantly, the results in those states were conclusive.
But as you can see, there have been many times over the past nine cycles where a state has had close results (defined here as less than 5 points between the two parties). Over the past nine cycles, there have been 92 state-level results that are considered close to that definition. Whether it becomes a “swing” state depends largely on whether we collectively agree to talk like that.
For example, before 2020, there wasn't much discussion about Georgia being a battleground state. Then it went into a swing state. So it's a swing state right now. Until it doesn't.
Voting is almost over and will continue to close
Perhaps the most important thing to remember about elections is that we often try to get polls to do things they aren't intended to do.
For example, on Thursday, Fox News released new polls in four battleground states (see above) that showed Trump with a narrow lead in three states. However, if the margin of error is that he is talking about a 1-2 point difference in a poll where he is 3 points, there is no difference between a “narrow lead” and a “tie”, and “a narrow lead”.
Fox exaggerated the poll results by comparing the new numbers to each state's 2020 results and recent past polls. If you remove the labels and include the error bars, the comparison looks like this:
Take Wisconsin, for example. Biden won there by 1 percentage point (after rounding), and new polls in January showed the two candidates tied. So who will win? No one knows.
Well, what about Michigan? President Trump rose 2 points in February and is now up 3 points. The appropriate response to this is: “So?” Does this indicate traction for President Trump? No, because the difference stems from the vagaries of when and how polls are conducted.Needless to say, it's been a few months actual This means that Trump's lead could grow significantly or slip back at some point.
It's easy to absorb all this information and fall into observational inertia. You look at another close poll with misty eyes and think, “Well, this doesn't really matter.” And you're about right.
What, did you think I was trying to dissuade you from that idea?
Certainly, there is much we can learn from polls and observations about elections. But very little of what we're likely to hear in the coming months will answer the most popular question: Who will win? To do that, they'll need to rely on actual voting, rather than some kind of cheating based on past vote margins in purple states.
