By Adam Pagnucco.
Most of us get presents in December. Maybe they are electronics, booze, food, toys… you name it, someone is giving it and someone is getting it. Well, I recently got my favorite kind of present from Jonathan Robinson of Catalist LLC: DATA. And boy is it awesome, so let’s share the wealth!
This specific kind of data is cast vote records for the 2022 primary and general elections in Maryland. Conventional voter file records list the name, address, date of birth, party, precinct, districts and voting history of each voter. You can tell who these voters are and in which elections they vote but you can’t tell for whom they are voting. Of course that’s the case since your decisions in the voting booth are confidential and can’t be released in a way that identifies you. Cast vote records contain a voter’s precinct and ballot style but that’s it. All other identifiers are stripped out so there is no way to know the identity of the voter. But then the cast vote records identify each candidate for whom that voter voted.
Here is an example of an actual cast vote record from the Montgomery County primary.
Precinct: 1-1
Ballot Style: Dem 1-1
Vote for Governor: Franchot
Vote for Comptroller: Lierman
Vote for Attorney General: Curran-O’Malley
Vote for U.S. Senate: Van Hollen
Vote for U.S. House 8: Raskin
Vote for District 14 Senate: Zucker
Votes for District 14 House: Kaiser, Luedtke, undervote
Vote for Executive: Blair
Votes for Council At-Large: Jawando, Gassaway, undervote, undervote
Vote for Council 7: Luedtke
Votes for Circuit Court Judge: Johnson, McGuckian, Pierre, undervote
Vote for State’s Attorney: McCarthy
Vote for Clerk of Court: Bushell
Vote for Register of Wills: undervote
Vote for Sheriff: Bass
Votes for Democratic Central Committee At-Large: 8 undervotes
Votes for Democratic Central Committee 14: Edmunds, Smith
Vote for Board of Education At-Large: undervote
Vote for Board of Education 1: undervote
Vote for Board of Education 3: undervote
Vote for Board of Education 5: undervote
There is no way to identify this individual; all we have is a precinct number and party. But by aggregating these records, we can analyze correlations of voting between candidates. And because the records show undervotes, we can identify bullet voting in races with more than one seat. That’s really valuable for understanding the intensity of support for candidates in delegate and council at-large races.
Do you see why I am so excited about this present??
Now let’s not get carried away with this data, folks. Many insiders are going to over-interpret these patterns as being primarily a product of message. Yes, we know that the candidates and mail firms (at least the good ones) obsess over every single word in their mailers. But most voters don’t take the time to read all of those words and many can’t tell your messages apart. The comment I hear most from political civilians at primary election time is, “They all seem the same to me.”
Instead, three factors seem to account for most correlations between candidates.
Geography – Candidates tend to perform better where they live. That applies to the precinct and regional levels. If a council at-large candidate comes from a particular place – say, inner Silver Spring – that candidate will run better there and voters for the district-based candidates there will also vote for the at-large candidate. That creates a correlation based on raw geography, not necessarily message. (Although it’s certainly possible that candidates from the same place will often have similar messages.)
Race – Race is always a factor in voting whether people admit it or not. And in this county, race overlaps with geography.
Major endorsements – Candidates with the Washington Post endorsement tend to cluster together in voting. The same goes for candidates endorsed by the Apple Ballot. This makes sense – if I value one of those endorsements for one candidate, chances are that I will value the same endorsement for other candidates who receive it too. Messaging plays here indirectly as the Post editorial board and MCEA representative assembly members sometimes look for different things in the candidates they pick.
What about gender? I am not dismissing it entirely but it seems to be less overt than the three factors above. For example, among Brandy Brooks voters, Laurie-Anne Sayles, Lorna Phillips Forde, Tiquia Bennett and Fatmata Barrie over-performed while Marilyn Balcombe, Natali Fani-Gonzalez and Dawn Luedtke under-performed. That suggests a correlation among Black female candidates (Brooks, Sayles, Forde, Bennett and Barrie) but not necessarily all female candidates.
Now this kind of data is really valuable in a crunched format to politicians. I really should be selling it to them! But Montgomery Perspective readers are a special lot so let’s make a deal. I will share it with you so long as you promise not to share it with any politicians who will fork up millions of dollars to get it. It will be our secret. Deal? Good! I will steadily roll it out along with all of the other content on deck.