data for politics #19: Young Voters, Not Gentrification, Drove Ocasio-Cortez’s Victory

Data for politics logo long-01.png

By: Kevin Morris (@ktnmorris)

A couple weeks ago on the blog, I took a look at who voted in the June 26th federal primaries in New York City. The trend was clear: younger voters supported challengers to incumbents across the board, and helped propel Alexandria Ocasio-Cortez to a long-shot win over Representative Joseph Crowley. We talked about how voting laws that fall heavily on younger voters are deeply problematic not only because they limit access to the franchise, but also because they may be having electoral impacts.

This week, I’m taking my voting rights hat off and putting my urban planning hat on. How communities choose their representatives and how representation can change due to population shifts in urban environments is of extreme importance as young, educated Americans flock to major cities.

In the race to say something fast about Ocasio-Cortez’s victory before real data were available, many pundits went straight for the 'G’ word  - gentrification. A piece over at the Intercept, for instance, attributed her win changing demographics and an influx of new residents. It's a story that fits a broader narrative: young, college educated kids move into working-class neighborhoods, bringing their campus idealism with them.

Despite the overblown importance placed on gentrification in discussions about urban politics today, we do need to pay attention to how the folks choosing a neighborhood's elected officials might be changing. If newcomers are able to elect representatives unable or unwilling to understand the needs of longer-term residents, the harms of urban displacement will be magnified. This is no idle concern: earlier this month, a white woman called the police on Crown Heights state senator Jesse Hamilton for the crime of shaking hands with voters in his own district. The woman was not happy to just disagree with Senator Hamilton. She instead tried to enlist police power against the community’s elected official. Although Hamilton is facing a well-deserved challenge next week from Zellnor Myrie, Crown Heights and Bedford-Stuyvesant are right to be concerned and outraged by this treatment of Senator Hamilton both because of its racist predicate and its potential political ramifications.

In the case of the NY-14 congressional primary, this doesn't seem to be the case. Of course, everyone who uses the term gentrification means something different. To some, it's an influx of white residents. To others, it's evidenced by rising rents, or the sudden arrival of college-educated people. We looked at all the classic signs of gentrification, and found that none of them were associated with higher vote share for Ocasio-Cortez.

Methodology

To test the claim of marauding socialist gentrifiers, we needed to get a better picture of the backgrounds of these voters than we can get from the voter file alone. Unfortunately, the Census Bureau doesn't publish statistics at the precinct level. To get around this limitation, we mapped each actively registered Democrat in the New York 14th to her home census block group (the lowest-level of geography for which the Census Bureau publishes data each year). We then assigned each voter the average characteristics for her block group. Doing so is a bit of a leap - we had to assume that the average income, for instance, for all voters is the same as the average for the block group, but it's the best we can do.

After assigning demographics to each voter based on his home block group, we took the average for all the voters who cast a ballot in each election district. This allows us to get a general idea of the characteristics of the electorate in each precinct. (An important note before we jump in: these data aren't perfect. The Census Bureau won't publish 2017 data until December of 2018, so we can't see changes from the past year and a half. Moreover, these are five year averages, potentially masking big changes from year-to-year.)

What We Found

Our results refute the notion that voters new to the congressional district supported Ocasio-Cortez at higher rates than longer-term residents. Let’s start with a reminder of what we saw a couple of weeks ago when we looked at the average age of a precinct and its support for Ocasio-Cortez. The relationship is striking:

Some observers have claimed that this age trend was driven by younger voters in white districts alone. It wasn’t. Younger districts were more likely to support Ocasio-Cortez regardless of whether they were plurality-white, as the charts below show:

age_races_14.png

While it's true that precincts where white residents made up the preponderance of the population were younger, on average, than others, it's clear that youth support wasn't limited to the white community. In fact, the slope of the lines in the two charts above are nearly identical, and the both are statistically significant at the 99 percent level.

As we continued to play with the data, we found another interesting predictor of a precinct’s support for Ocasio-Cortez that doesn’t have anything to do with gentrification, but does bolster a lot of the reporting we heard about the race. Precincts with higher turnout, we found, really did support Ocasio-Cortez more than those with lower turnout. This maybe isn’t surprising - voters were inspired by Ocasio-Cortez’s message. We’ve all seen the picture of her shoes from after the election. Turnout is a good indicator of enthusiasm and excitement surrounding a race; when people don’t care, they don’t vote. And it looks like where New Yorkers were excited to participate in a race, where they found a candidate worth trekking to the polls for, they showed up for Ocasio-Cortez:

 

There’s an important clarification around this issue of turnout that needs to be addressed, though. Shortly after the election, many commentators claimed that Ocasio-Cortez won because she was able to mobilize new voters. This isn’t exactly true. In fact, just 921 voters cast their first ballot in this election, and fewer than 4 percent of participants registered after the 2016 presidential election - a share smaller than in the NY12 and in the NY9. There’s some evidence that voters who did show up for the first time lived in neighborhoods more likely to support Ocasio-Cortez, but there’s a dangerous narrative that needs to be corrected here. The progressive policies of Ocasio-Cortez didn’t just win the support of new voters. She won over regular, rank-and-file Democrats.

To explore the gentrification question, we start by looking at how changes in the racial demographics of a neighborhood correlated with a precinct’s support for the 14th’s challenger. The chart below plots percent change from 2011 to 2016 in the share of the population that was non-Hispanic white. That best-fit line is flat; areas that saw an influx of white voters didn’t support Ocasio-Cortez any more than other neighborhoods.

How about where rents have risen most quickly? Rapidly rising housing costs can be indicative of neighborhood change that can't be captured by data on race. It’s also become an easy way of looking to see where housing displacement, gentrification’s ugly sister, might be tagging along. We do see that precincts with rising rents were a little more likely to support Ocasio-Cortez; a 10 percent increase in the rate at which housing costs went up was associated with AOC winning a little more than 2 percent more of the vote share. This is a relatively small impact, though, and the causality isn’t clear. Are the people moving into and bidding up rents in these areas voting for Ocasio-Cortez, or are renters concerned by the threat of displacement supporting a more progressive candidate? Both may be true, but it’s clear that areas with quickly rising housing costs didn’t show up in hugely higher numbers for Ocasio-Cortez.

The story gets a little more complicated when we look at the impact of changing levels of education and geographic mobility. Increases in both of these measures are associated with a slight increase in propensity to support the challenger candidate.

The difference, however, is still small: for every one additional percentage-point increase in the share of residents with some college, Ocasio-Cortez won an additional 0.17 percent of votes. Districts where more people moved in the past twelve months were also slightly more likely to support Ocasio-Cortez. For every one percent increase in the share of folks who had moved, she picked up an additional 0.3% of the votes (though the relationship is not statistically significant). Neither of these distributions share anywhere near the same sort of correlation that we saw last week between average age and vote share.

 

What happens when we put this all together? To systematically examine the impact of each of these factors’ relationship with the vote share won by Alexandria Ocasio-Cortez, we built out a standard regression model, presented below. The first column examines basic demographic information on the precinct’s neighborhood, including the average age, as we discussed last week. The second column adds in the gentrification variables. In the final column, we weight the precinct observations by the number of votes cast in each precinct.

 

The regression model corroborates what we’ve seen looking at each of these variables independently. Firstly, voters’ age was the most statistically significant predictor of support for Ocasio-Cortez - a ten-year increase in median age was associated with a decrease of between 15 and 17 percentage points in a precinct’s support for the candidate. Race was also associated with support, but the effect was not particularly large (the Latino share of the electorate has the largest coefficient; a ten percentage point increase in the Latino share was correlated with  2.2 - 2.8 percentage point increase in share of votes going to Ocasio-Cortez.). Precincts where a larger share of registered Democrats turned out also supported Ocasio-Cortez at a higher rate, lending credence to the anecdotes about the campaign’s strong GOTV operation. 

And the gentrification variables? After we control for other precinct demographics, none of them are significant. In fact, the adjusted R-squared (a measure of a model’s fit that accounts for the number of variables) actually drops when we layer in these variables. And this isn’t just because they all went in at once; it’s not shown in the table above, but I also ran the model by adding gentrification variable by itself to the demographic variables. None of them were significant.

This is really good news for two reasons. To start, it means that despite a hard-fought nomination, Ocasio-Cortez doesn’t just represent a subsection of her electorate. With the exception of age, precincts that are getting white were as likely to support her as were precincts getting less white, as are precincts where rents are going up and going down. There are, of course, individuals disappointed by her win, but we don't find evidence that a preponderance of any one type of neighborhood is unhappy with the primary result.

Secondly, it means that at least in the New York 14th, gentrifiers aren’t usurping the representation of long-time residents. These charts show some surprising, and troubling, statistics - rents rose by 10% in the average precinct, surely putting stress on many households. And nearly ten percent of households moved in the past year in the average precinct, pointing to housing instability. These are real problems that the city and the state need to address. However, one thing is fairly clear - the hot takes proclaiming Ocasio-Cortez the unlikely beneficiary of a gentrifying district don’t pan out. There may well come a time when new arrivals exhibit different political preferences than their neighbors, causing representational imbalances, and we should all watch for that. In the case of the New York 14th, though, we should resist oversimplifying Ocasio-Cortez’ primary victory. Her support wasn’t concentrated just in gentrifying areas - support for her candidacy was spread across many, many types of neighborhoods in her district.

Kevin Morris (@ktnmorris) is an urban planning student at NYU’s Wagner School of Public Service.


Note: Where data was not published at the block-group level, we used tract level data instead.