County by County, Youth of Color Key to Democrats in 2018

Across the country, young people had a profound impact on the 2018 midterm elections. CIRCLE’s exclusive day-after estimate found that 31% of young people voted in Tuesday’s midterms, the highest rate in over two decades. Furthermore, according to national exit polls, young people backed Democrats at higher margins than any other midterm or presidential year since 1992.

To get a better sense of how young voters made a difference, we wanted to dive deeper into election results in key battleground races. In several states with projected toss-up elections, we examined youth vote choice for each candidate by aggregating county-level data. Using a similar method as for our analysis of Ayanna Pressley’s Democratic primary victory in Massachusetts’ 7th District, we wanted to identify counties in each state with high concentrations of youth to see whether these areas broke heavily toward either party.

Indeed, we found that in the Montana Senate race, in the Georgia gubernatorial race, and in the Texas Senate race, counties with a high proportion of youth more strongly supported the Democratic candidate. We also find that, in the latter two races, that effect was further magnified in counties with high proportions of people of color (In Texas, Latinos; in Georgia, African-Americans), which suggests that youth of color may have played an especially key role in supporting Democrats in those races.

For this analysis, we divided counties into thirds of about equal size based on Census data on the percentage of young people among the county’s total population. That allowed us to categorize each county as having a “Low”, “Medium”, or “High” youth population relative to other counties within each state. From there, using countywide candidate support data, we were able to determine that Democratic candidates fared significantly better in counties with high proportions of young people.

In Montana, youth were responsible for propelling incumbent Democratic Senator Jon Tester to a close victory (50.2% to 46.9%—a difference of less than 17,000 votes) despite his lackluster performance in counties with older voters. While Tester won just 31.8% in counties with a low youth population, and 35.8% in “medium” youth counties, he garnered 52% of the vote in high youth counties.

Meanwhile, in the Texas Senate race, young people—and young Latinos in particular—helped Democratic challenger Beto O’Rourke close the gap in his narrow defeat to the heavy favorite, Republican incumbent Ted Cruz.  As in Montana, counties with a higher proportion of youth were significantly more likely to vote for O’Rourke. O’Rourke’s average share of the vote across all Texas counties was 27%, but in high-youth counties it was 38%.

In Texas, we also used the same methodology to divide counties by their proportion of Latinos, and we found that those with higher populations of Latinos were also significantly more likely to vote for O’Rourke. The effect was highest in the 50 Texas counties that fit in both the high-youth and high-Latino categories, where O’Rourke’s average vote share jumped to almost 44%, compared to just 17% in low-youth, low-Latino counties.

In Georgia, we saw a similar pattern with support for Democratic gubernatorial candidate Stacey Abrams. Compared to her overall average of 34% of the vote across all counties, in high-youth counties, Abrams averaged 41% percent whereas in low-youth counties she earned just 29% of votes. As in Texas, another significant driver of these county vote-choice gaps seems to have been young people of color—in this case, African-Americans. A clear pattern emerges when we categorize counties by its African American population in addition to its youth population: 56% of voters in high-youth, high-African American counties backed Abrams—22 percentage points above average. On the flip side, just 19% of voters in counties with low percentages of both youth and Black residents voted for Abrams—14 points below average.

It is important to note that average candidate support within a group of counties is different from  overall candidate support because the number of people in each county varies widely. In general, this method of aggregating county-level data will underrepresent the overall support for Democratic candidates because they won relatively few counties across all three states, but the counties they won had larger populations. Winning counties with smaller populations adds very little to the overall vote total at the state level. As an example, O’Rourke winning 66% vote in Dallas County (pop. 2.6 million) compensates for his losses in many smaller counties, such as his 28% showing in Dawson County (pop. 12,500).

In addition, our method of splitting the counties into thirds of equal size allows for some variation within groups as well as between them. For instance, in Texas, the 83 counties with a “High” proportion of youth range from Hunt County (where youth make up 16% of the population) to Brazos County (39% youth). When we analyzed an even smaller subset of counties with the most Latino youth—the 12 counties in Texas with youth populations greater than 20% and Latino percentages greater than one-third—we found that O’Rourke’s average vote share skyrocketed to 21 percentage points above his average support overall in the state.