American Elections Have a Literacy Problem
Low literacy is crushing turnout and enabling populists like Trump.
When chasing an answer to the question of why Trump works so well in America, while the rest of the world doesn’t take him seriously, the words education and literacy eventually pop up. 21% of Americans are functionally illiterate, meaning they have difficulties comprehending print material and only understand basic vocabulary or very short texts. “That’s what I always reference”, a friend of mine from the US once told me when we talked about Trump.
I’m skeptical that you can explain Trump’s rise to power just by pointing to America’s education system, because countless factors contributed to his rise: A broken media landscape dominated by Fox News, a celebrity cult, the Republican Party’s missing spine, and so on. I do, however, believe that education and literacy are important factors, but what does the data say. There are two sides to this. We can look at the people who actually voted for Trump, and we can look at those who stayed home. We will start with the latter.
Half of Americans never voted for Trump. He only managed to get between 45-50% in his three presidential elections. But considering turnout, his performance becomes even less impressive. In 2024, only 31% of adult American citizens voted for Donald Trump. Who are the people who didn’t vote? For a long time, less reliable voters were considered more Democratic (young voters, minority voters). This changed with Trump’s appeal to less-informed and spontaneous voters, and in 2024, those who stayed home might even have broken for Trump on Election Day, because they belonged to the one decisive group that backed Trump by a large margin (more on this later).
America’s election turnout is low compared to most European nations and it fluctuates drastically depending on which US state or county we are talking about. There are a few reasons that collide and accumulate to create “turnout deserts” in America and literacy is one of them. A look at county-level data from 2024 shows a clear picture: US counties with a higher adult literacy score had higher turnout rates.
The R2 value of 0.34 shows that literacy played an important role in determining turnout in 2024. Literacy explains about a third of the variance in county-level turnout. A decent result, considering that other variables like education or race likely also affected turnout. (Counties from Alaska and Connecticut are missing due to data collection problems).
Researchers have found that these turnout deserts are most prominent in southern areas with a large African American population, like parts of the deep south - a result of decades of structural racism that has led literacy rates in these counties to be among the lowest in the nation. To increase turnout in these turnout deserts, the states would first have to overcome these decades-old structures so that literacy rates and political education can catch up.
I did not run the same analysis for the 2016 and 2020 elections, but my educated guess is that the effect was even more prominent in 2016, a low-turnout presidential election. 2024 was a high-turnout presidential election, though turnout slightly underperformed the 2020 record-breaking race. So now we know that turnout deserts are linked to literacy rates. In presidential races with record turnout rates and a candidate like Trump who brings low-propensity voters to the polls, another question comes to mind: What about the people who did vote?
First of all, I couldn’t find any meaningful correlation between Trump’s vote share in a county and that county’s literacy rate, at least not on the national level. The reason is pretty simple: While Trump did perform better in rural counties and states with low literacy, Harris performed better in low-literacy, majority-black counties in the deep south that, as mentioned before, have been suffering from decades of structural racism. So, in the end, we get a pretty even picture.
But if we extend our horizon and look at political literacy, a striking difference emerges. G Elliot Morris found that voters who were unable to identify which party controls Congress voted for Trump by a margin of 11 pts. Those who correctly identified the parties voted for Harris by a slim margin. Similarly, polls found that those who do not follow political news “at all” voted for Trump by an overwhelming margin.
Trump’s victory was powered by people who consumed no political news at all and had little knowledge about political matters. That is why higher turnout might even have benefited him, or at least it might not have hurt him, because non-voters tend to be low-information voters as well and therefore would have been more likely to vote for Trump. This is also where basic literacy skills come into play again. I often hear the phrase that regretful Trump voters failed an open-book test, as it was fairly easy to demask Trump’s lies. For example, it would have taken just one quick Google search to learn that Trump is lying about tariffs. So, yes, it was an open-book test, but it’s difficult to pass an open-book test if you can’t read properly. That may sound harsh, but with one-fifth of American adults unable to read and comprehend basic political information, it sadly is the reality we all live in.
So what drives low-information voters’ choices at the ballot box? Popularity could be one factor, but in 2024, high prices are the one thing that almost all Americans felt in their everyday lives, so they punished the incumbent party. With prices remaining high, it is exactly this group of low-information voters who have soured on Trump. Skyrocketing oil prices in the wake of Trump’s war against Iran certainly won’t be helping the President.
For the future, raising literacy levels might give American Democracy a quality boost. Turnout deserts would become increasingly rare, and the outcome of elections would be more information- than vibes-based. After all, consuming news only becomes possible when one can comprehend it. And, as of today, every fifth American adult can’t.
This article was updated on March 12th to fix two errors.