Last week, in our series analyzing the results of the 2020 election, I examined why the model struggled, and the specific areas where it performed poorly. This week, I want to talk about one of the less discussed parts of the election: Trump’s surge with Hispanics.
Once a large voting block for Democrats, this group showed some signs of reversion to the GOP in 2018, but nothing like what we saw on Tuesday night. If you had shown any political analyst the results in Miami Dade and nothing else in Florida, that analyst would have thought that the Republican in the race would be winning Nationally in a Reagan-esq landslide. Somehow, getting 53.3% of the vote in Miami-Dade “only” lead to a 3.4% loss in the state, which is actually a minor miracle. Even stranger, is the strong possibility that Biden did so poorly in the county while not actually losing many voters. In 2016, Clinton received 624k votes to Trump’s 334k, while in 2020 Biden received 617k votes in Miami-Dade. However, the big change was the support for Trump, going from 334k to 532k, an increase of just shy of 200,000 voters. We see the same thing in the Rio Grande Valley in Texas. In Hidalgo county, a border county with Mexico, Clinton received 118k votes, and Biden increased his raw vote count to 127k. Yet again, the same story as Miami-Dade, as Trump increased his vote tally from 48k to 90k, almost doubling his raw vote count in this county alone. We see this same pattern throughout Texas and Florida. Could some of those Hispanic voters have crossed over from Clinton to Trump? Sure, and it is likely a few did. However, there has to be a reckoning with the fact that millions of Hispanics that didn’t vote in 2016 came out in droves and voted for the incumbent President in 2020.
Most people thought that Hispanic engagement at that level was almost impossible. While we at LeanTossup did see a small shift in the high-quality Hispanic polling, we did downplay it, as pollsters are notoriously bad at polling Hispanics, especially in Texas. However, there definitely seems to be different swings with Hispanics in different parts of the country. If you had seen the results in Florida and Texas, you would think Trump would be winning Arizona (and perhaps Nevada) in a landslide, which did not happen. Additionally, at this moment, there does not seem to be much losses for Biden with Northern Hispanics, but we’ll have to wait on full results from New York and Chicago to know for sure. At this point, the only coherent story with this block seems to be a sharp split along educational lines. While there is evidence of Cubans with a college degree in parts of Florida trending towards Trump, overall, the split seems to follow very closely along educational lines. Hispanics along the Rio Grand Valley are mostly low income, and typically do not have college degrees. Although Hispanics in Miami Dade are more likely to have a college degree than others in different parts of the country, there are still a lot of Hispanics in Miami-Dade that do not have a College degree. Take this chart, from the Demographic Statistical Atlas of the US, showing educational attainment in Miami-Dade County. The percentages indicate the percent of the population (male percentage, and female percentage) of the population that has a college degree.
From this data, Hispanics have the second lowest educational attainment rate in the county, second only to Black voters. Additionally, the graph shows the level for each group statewide, which shows us that loses in Miami-Dade might actually have been limited, as Miami-Dade Hispanics are slightly more likely to have a college degree than Hispanic voters in the rest of the state. Although Hispanics in Miami Dade are heavily Cuban, which definitely impacted their voting choice, the fact that many of those Cubans did not have a college degree and are simply following white non-college voters in their rightward swing cannot be ignored. Moving over to Texas, we can see the same issue. Let us see the same graph, but for Hidalgo county:
Immediately we see that Hispanic college degree attainment is in the mid-teens, slightly higher than the state, but significantly lower than Florida, which is likely why Texas had a larger Hispanic shift to Trump than Florida. Although I have decided to highlight Hidalgo county here, the same story can be told throughout the entire Rio Grande Valley, especially in other border counties where the educational attainment rate is even lower than Hidalgo (Starr, the county right beside Hidalgo, is only ~10%).
As we move North of the Rio Grande in Texas, we see that Biden did manage to make gains with white voters in the large urban areas of the state (Dallas/Fort Worth, Austin, Houston, etc). However, when looking at educational attainment of Hispanics in those same areas, we see the same story as Hidalgo, large Hispanic populations without a college degree. Let us look at two counties in the Dallas suburbs that shifted toward Biden; Tarrant county (which contains Fort Worth) and Collin county (North Dallas, TX-03). Tarrant County went from Trump+8.6, to D+0.2. Tarrant County is 27.5% Hispanic according to the Demographic Statistical Atlas, with a Hispanic educational attainment rate of ~12.5%. Collin County went from Trump+16.7 to Trump+4.4. However, unlike Tarrant County, Collin County is much whiter, only 14.8% Hispanic according to the Atlas, with a much higher educational rate of ~23.5%. This led to a ~4% larger shift in Collin County than Tarrant County, in counties that are right beside each other. Additionally, Collin County has a longer history of voting Republican, as Beto won Tarrant County in the 2018 Senate race, while losing Collin County by more than Biden did. When considering the larger Hispanic population in Tarrant County, and the higher educational attainment for Hispanics in Collin County, this differential swing suddenly does not seem so crazy.
A very interesting analysis would be if those gains in the majority white precincts would have been enough for Biden to win the state, assuming no major loss of Hispanic support in the rest of the state (“Election Twitter, if you’re listening…”). Although the end result for Texas Democrats was very disappointing in terms of the statewide margin, it’s very possible that the suburban gains with white voters are much deeper than a 5 point loss would suggest, but are being papered over by large losses with Hispanic voters. It would be very interesting to compare the size of white suburban gains in Texas to Georgia, where Democrats made enough suburban gains to win the state. Possibly, with a much smaller Hispanic population in Georgia, Texas and Georgia might be moving in a more similar direction than we currently think. In terms of Texas long term, Democrats might actually have a better chance at winning Texas in midterm environments, where the Hispanic share of the electorate is lower at the expense of white college voters. While Texas is still trending to the Democrats, their eventual flipping of the state will have to wait even longer if their support among non-college Hispanics continues to collapse, in a mirror image of their problems with Northern white non-college voters. Eventually, their suburban gains will become insurmountable, as a larger percentage of the electorate will have a college degree, and live in a suburb, that the GOP nominee outright winning the Rio Grande will not be enough to overcome the Democratic margins with those college educated voters.
Moving further west, we see that this trend continues in other heavily Hispanic counties. Further along the US-Mexico border, are other counties that have similar demographic profiles to those Rio Grand counties. Let’s look at two specific counties in New Mexico, Luna County, and Dona Ana County. According to the US census, Luna County is 64.9% Hispanic, while Dona Ana County is 67%. In 2016, Trump won Luna County 47.4-43.8, but lost Dona Ana County 53.7-35.9. According to the NYT result map, in 2020, Trump won Luna County by an expanded 54-44, but lost Dona Ana County by a worse 58-39.7 margin. These counties are right beside each other, with Dona Ana county bordering El Paso. How can they swing differently? Education attainment. Only 18.7% of residents have a college degree in Luna County, while 35.4% of residents have a college degree in Dona Ana county. While the atlas has a bad link on the page for Dona Ana county (likely because of the unusual character in it’s name) we know that specifically Hispanic degree attainment in Luna county is roughly 7%, less than half of the county total. Assuming a similar same ratio in Dona Ana county, the differential swing makes sense, even in bordering counties. There are further examples of this, in border counties in Arizona and Mexico, but the same point still stands: voters without a college degree continue to trend away from the Democrats, no matter their race.
What does this mean for Democrats going forward? It means they have a huge problem. While in previous elections, Hispanic voters seemed to vote as a block, college educated or not, this appears to be changing rapidly. To reverse these losses, Democrats will have to convince Hispanic voters without a college degree (which is the overwhelming majority) to support them again, and hope that they do not follow the same path as white non-college voters, a group which has never looked further away from Democrats than right now. While Democrats (and pundits) believed that eventually demographic change, an ever increasing population of Hispanic and other non-white voters in all parts of the country, would eventually deliver them an unbreakable electoral stranglehold, that might no longer be true.
Somehow, Democrats are only able to appeal to voters who have a college degree, and while that used to continue if a voter was either Black or Hispanic, this appears to no longer the case, as Trump was able to gain with Black and Hispanic voters in 2020, neither of which approach the same level of educational attainment as white voters. This might reflect Republicans repeated messaging that the Democrats are the party of the coastal elites. While white college educated voters in the suburbs are more likely to identify with those “cultural coastal elite” values, the assumption was that non-white voters would also be able to identify with them, and to be supportive of policies that their creators believed would benefit them. However, much to the surprise of everyone, non-white voters without a degree are now joining white non-college voters in their displeasure of these leftwing goals.
What is the solution? I honestly don’t know. Much of it will involve directly appealing to these voters, instead of taking them for granted. For years, Democrats have been terrified of explicit appeals to non-white voters, in fear of a backlash against them from white voters. One can only imagine the response from Republicans that signs with a “Blacks for Biden”, or “Latinos for Biden” slogan would generate, while somehow, “Blacks for Trump” and “Latinos for Trump” were only met with a collective eye-roll from the media. While it is obviously important to question the impact of simple slogans, the implication of them is clear: Latinos had a place in the Trump coalition, while the Biden campaign dared Hispanic voters to go online and find their campaign platform to learn if they would benefit from a Biden presidency or not. If Democrats cannot find some way to appeal to non-white voters without a college degree, the next decade risks the Rio Grande Valley (and South Florida) looking like rural Wisconsin on election night.