The Washington PostDemocracy Dies in Darkness

A key election metric for Democrats is on a sign in your neighborhood

High gas prices are displayed at a Chevron gas station in Los Angeles on Oct. 5. (Frederic J. Brown/AFP/Getty Images)

A month ago, I noticed an unsubtle correlation: In recent months, the extent to which Americans said the country was on the right track moved in concert with gas prices. In other words, as gas prices fell, American optimism rose. As prices rose, optimism fell.

This seems counterintuitive. We tend to assume that people are complicated, albeit rational, actors who process information and come to intricate conclusions about cause and effect. But that may be overcomplicating things. Looking at the same data for President Biden’s approval rating and for the Democratic position in generic-ballot polling (that is, polling that asks which party people plan to vote for in November) shows similar correlations.

Sign up for How To Read This Chart, a weekly data newsletter from Philip Bump

One way we can visualize the relationship between two sets of data is to compare them directly on a chart. After all, if two numbers move in concert, they can be depicted literally moving in concert. One goes up, the other moves either up or down consistently. That sort of thing. So if we plot one set of numbers on the vertical axis of a chart (say, weekly gas price averages) and another on the horizontal axis (like FiveThirtyEight’s average of Biden’s approval polling), a correlation will be revealed as the two numbers are moving in concert.

As is the case with those particular numbers. Imagine the black line as being drawn from the dot at bottom center to the arrow near where it says “three-week shift.” That’s the weekly cost of gas and Biden’s approval three weeks later. The line goes up and to the left — as gas prices rise (moves up), Biden’s approval falls (moves left) — and then right and down. (Because as prices fall (move down), Biden’s viewed more positively (moves right).)

I included the same data without the shift; the correlation isn’t as strong. That makes some sense, given that it takes a while for lower prices to trickle out through the country, among other things.

If we look at monthly averages, though, we see a similar pattern. Biden’s approval falls a bit as gas prices rise from January to June (the dot at the top of the graph), then his approval recovers as prices fall.

I’ll hasten to note that correlation isn’t causation. Just because Biden’s approval rose as prices dropped doesn’t mean prices caused the drop. But it seems safe to assume some relationship, particularly given how often pollsters found that Americans were worried about energy prices as they expressed disapproval of Biden.

The pattern for Democratic support in the generic ballot (again from FiveThirtyEight) is different but, since prices peaked in mid-June, even more sharply correlated to the cost of gas. From January to that peak, Democratic support in the generic ballot didn’t really budge. Then prices started to fall and Democratic support started to increase.

There’s a measure called the correlation coefficient that is a mathematical determination of how strongly correlated two sets of data are. It ranges from 1 (perfect correlation) to 0 (no correlation) to minus-1 (perfect inverse correlation). The correlation between gas prices and Biden’s approval shifted three weeks is minus-0.80. The correlation since prices peaked with Democratic generic-ballot polling is minus-0.91.

Again this is confounded with other things. The Supreme Court decision in Dobbs v. Jackson Women’s Health Organization was released about a week after prices peaked. That probably helped boost Democrats’ position in the polls. But, again, the correlation to gas prices is really strong.

This offers a warning for Democrats. Gas prices are trending back up, just as we hit the three-week mark from the election. There’s a reason the Biden administration wants OPEC to kick out its planned production cut for a month. His team is certainly aware of how these numbers have been moving, too.

It used to be that savvy observers tracked FiveThirtyEight’s election projection pages and read detailed analyses at The Washington Post to figure out where the election was headed. Now they may need simply drive down the road and keep an eye out for the closest Chevron or BP.