Earlier this week, the Census Bureau released its semiannual assessment of turnout during the most recent federal election. It includes data on who was registered to vote and who voted, by age, race and state. It overlaps turnout data with questions about why someone did or didn’t vote and background information such as types of residence.

It is, in other words, the most thorough official assessment of who voted that the country generates. And the new data show that a 30-year-long trend has continued: The electorate is getting less White.

There is a bit of wonkiness with the bureau’s data; specifically, it doesn’t adjust its numbers to account for people who don’t respond to the survey. Happily, Michael McDonald of the United States Elections Project does, so we can use his adjusted data to understand how the national electorate has evolved.

The graph below shows McDonald’s data. The larger circles indicate presidential elections.

You can see how the density of the White vote has seesawed in recent years, with the density increasing in lower-turnout midterm elections and dropping in higher-turnout presidential ones. Since 1990, the difference in the density of the White vote between midterm and presidential years has grown a bit wider.

But that seesawing can obscure the broader trend. For example, the 2014 midterms had a density of White voters that was nearly equivalent to the White density in 2008 — the year that saw Barack Obama elected as the country’s first Black president.

That’s the unexpected thing about this trend. Given that non-White voters also tend to be more heavily Democratic, we’d expect, at least on the surface, for the national results to skew more Democratic as the density of non-White votes increases. But that hasn’t really happened. In 2008, for example, Democrats won more votes in House seats nationally by a 10.6-point margin. Six years later, with the same non-White density, Republicans won by 5.7 points.

There simply hasn’t been a correlation between the density of White voters in an election and the results in the House contests. (The graph below shows the relationship of House results to White density in the electorate over time. It begins in 1986, at top left. From there, the density heads lower, but the House results flip back-and-forth.)

We highlighted one section on that graph, the period from 2010 to 2016. It behaves as we’d expect: In the midterms, the White density rose and Republicans did better; in presidential races, White density dropped and Democrats did better. But beyond that stretch, the correlation doesn’t hold up well.

From 1986 to 1990, the electorate was on average 85 percent non-Hispanic White, and Democrats won the national House vote by eight points. From 1998 to 2002, the electorate was 82 percent White, and Republicans won by an average of two points. From 2010 to 2014, the electorate was 76 percent White and the Republicans won by four points. In the past three House elections, Democrats have won by that same average even as the electorate has been an average of 73 percent White.

House races are wonky things. That many House districts are safe for incumbents (either because of gerrymandering or because of the density of partisans in the districts) might decrease votes from those districts. If we compare White density to the national presidential vote, though, the correlation is only a bit stronger.

From 1988 to 1996, the electorate was an average of 84 percent White, and Democrats won the national vote by two points on average. From 2000 to 2008, the electorate was 79 percent White, and Democrats won by about the same amount. In the past three presidential contests, the electorate has been 73 percent White, and Democrats have won the national vote by just over three points on average.

We’re only talking about a sample of nine elections here, so we should be careful about drawing too many sweeping conclusions. But it certainly isn’t the case that the increased density of non-White voters has appreciably shifted the results for Democrats. Bill Clinton won by six points and nine points with White density over 80 percent. Barack Obama won by seven points and four points with White voter density well under that mark.

A lot of caveats apply to those examples, such as the presence of a strong third-party candidate in Clinton’s races and the shift of some White Democrats to the right as the density of non-Whites has increased.

But the reminder still stands. American politics is more complicated than an electorate becoming more Democratic as it becomes less densely White.