Brooklyn Borough president Eric Adams secured a victory in the city’s first ranked-choice voting primary election—and for the first time new data mapped by the Center for Urban Research at The CUNY Graduate Center shows where and how he won, plus how leading candidates saw their votes transfer through eight rounds of tabulation. (You may recall we did this after Mayor Lenny's victory in our practice election.)
The interactive maps are based on an analysis of the cast vote record, released last week by the New York City Board of Elections after they finished certifying all the primary election results.
To understand a few of the most interesting trends these maps show—and what a cast vote record actually is—we turned to Steven Romalewski, Director of the CUNY Mapping Service at CUNY. Here is a lightly edited version of that conversation and, of course, those maps!
How are these results different from previous election results?
Before ranked-choice voting, the New York City Board of Elections would just publish the vote count for each candidate by election district and assembly district. People who wanted to analyze the results data could simply map that information against election districts and find relatively straightforward patterns.
Now that voters have the option of ranking as many as five candidates, it’s more complicated to show those results. The scanners that the city uses are able to keep an anonymous, digital record of each ballot that's scanned in. The final file that is produced of all those ballots is called the cast vote record.
Was the data easy to analyze?
On the one hand, the city BOE did a great thing by publishing that data online for anyone to access; it makes it more transparent, especially given the error the BOE made in the initial tally of ranked-choice results. It's really good for people to be able to take this information and make sure that everything is accurate.
However, the format was really complex and the BOE provided no documentation about what the format was. We consulted with the Ranked Choice Voting Resource Center, which developed the software that the BOE uses, to process the cast vote record and turn that into tallies round-by-round to determine how each candidate did in each round and how the ultimate winner did.
What can we learn from the cast vote record that you can’t discern from results alone?
The cast vote record data shines a really powerful light into what the patterns are for this first ranked choice voting election and we can really drill down and not only see the patterns by election district of how the votes were transferred round by round from candidate to candidate, but we can also explore more fully how much voters in each neighborhood really used the new system to rank all five or fewer candidates.
These maps talk about the “vote share” - what is that?
That’s the number of votes that each candidate received in each election district. The candidates are color-coded. For Eric Adams, we used a blue color scheme. The districts that are shaded in light blue are where he got relatively few votes and the districts that are shaded in dark blue are where he received a much greater share of votes by election district.
How do you read these maps?
For each round of rank choice voting, the Board of Elections looks at the vote totals and if no one candidate has 50% or more of the result, then the lowest vote getter is eliminated. Those votes are then transferred to their next selected candidate that is still in the race. There is a tool on the map that lets you select a candidate and another that allows you to move through each of the rounds of tabulation for the mayoral and city comptroller primaries. As you do that, in the mayoral primary, for example, you can start to see where votes get transferred to Adams or any of the other candidates still in the race. Some of the changes are not very perceptible because Adams only gained a couple of votes. But in other cases there are big jumps, where the shading goes from light blue to dark blue.
Where is there an example of that?
In Borough Park, (Brooklyn - AD 48) voters who picked Andrew Yang number one on their ballot also appeared to choose Adams as their number two. When Yang was eliminated, most of those votes in many of those election districts went to Adams. All of a sudden the color pattern on the map darkens and Adams’ vote share in that Assembly district doubles, from 40% to 80%.
ADAMS' VOTE SHARE BEFORE & AFTER YANG WAS ELIMINATED
(Adjust the slider tool from round 6 to 7 to see the change)
The software behind the map is really powerful because you can move your cursor over the map and the diagram will instantly change for each of the 5,600 election districts or 65 Assembly districts.
Are there other trends that jumped out at you?
In those districts, like AD 53 in Brooklyn, where Maya Wiley did very well, you can see that for the most part there's a big swath of votes that went from her to Kathryn Garcia when Wiley got eliminated in round seven. At the same time, if you look at some other districts where Adams did very well, and you look at the vote flow, for the voters who supported Wiley in those areas, most of her votes in those areas went to Adams when she was eliminated.
You can really see that trend in southeast Queens
That’s why this localized, very detailed data is really important because it enables us to see the nuance and the subtleties and the variation in how voters behave across the city. And by mapping it like this, we really get a sense of what the neighborhood by neighborhood almost block by block pattern.
There have been some really heated debates about ranked-choice voting and whether it will lead to disenfranchising voters in neighborhoods with larger populations of Black and Latino voters. Does this data tell us anything about whether that happened?
The cast vote record data enables us to look at those kinds of questions in great detail. In addition to showing the vote flow candidate by candidate, we also show the number of ballots that become inactive, where voters chose someone who got eliminated and did not pick one of the two candidates who were in the final round of tabulation.
The concern is that if you don't communicate effectively enough and educate people on how to completely rank choices, then your ballots will be exhausted earlier and your votes in effect won't count. The map shows very clearly that number one in areas where Eric Adams did very well, for example, all of those ballots counted, because he was one of the finalists in the last round of voting and the ultimate winner. He did very well in predominantly Black and Latino communities and most ballots in those communities counted in the final round.
Did you find any interesting trends among the inactive or exhausted ballots?
You can see very local patterns where ballots became inactive. For example, in northeastern Queens, where Andrew Yang tended to do very well, in a lot of the election districts voters either chose Yang as their first choice candidate, and then did not pick Garcia or Adams, or picked maybe no one else. Those ballots became inactive and you can see that in the diagram, on the map, the color scheme goes to gray. Where we use big columns of gray, that's where there were a lot more inactive ballots. That's interesting because Yang was promoting ranked-choice voting and also promoting an alliance with Kathryn Garcia, urging people to pick her as their number two choice. That seems to have worked in some places, but not in a lot of places.
What about people who only picked one candidate or people who made mistakes on their ballots - do your maps tell us more about that?
Not yet. This provides a first cut at this data and understanding these types of patterns. We're going to be working in partnership with Common Cause to go much more in-depth into this analysis to understand all of the different ways people made use of ranked-choice voting or didn't fully make use of it, and how those patterns compare with local demographics and local vote patterns.
