Home » Voting » A first look at November 2018 CPS data on registration and voting

A first look at November 2018 CPS data on registration and voting

To begin this project on voting, I decided to look at publicly available data from the November 2018 Current Population Survey (CPS). The CPS is a monthly household survey taken by the Census Bureau in the Commerce Department with collaboration from the Bureau of Labor Statistics in the Labor Department. People from around the United States are interviewed about basic aspects of their household and family units, with supplementary questions in some surveys to focus on specific topics. In November of federal election years – all the even-numbered years – the supplement is on elections.

The November 2018 CPS involved interviews with close to 150,000 people living in the United States. Each interviewee’s responses were given a weight, essentially an estimate of the number of people with similar demographics living in the same geographic area. Summing the weights gives the total resident population of the United States, or something close to it.

The supplementary questions in November 2018 asked whether the interviewees were registered to vote (assuming they were old enough and United States citizens) and whether they voted. If the response to either question was negative, the interviewees were asked why they had not registered or voted. Each question offered a menu of possible responses. Additional questions asked how the interviewee registered, whether the interviewee voted in person or by mail, and whether the voting was early or on Election Day.

These data should allow me to figure out where registration and voting by the Black community is low relative to the same actions by people who classify themselves as White (with no other race). I’ve chosen White people as a comparison group, since my assumption is that they face no unusual barriers to voting. I’ve defined Black people as any interviewees who said they were Black or Black mixed another race. I did this because Black people in the United States have faced prejudice regardless of their exact racial mix.

Because legislation on voting can differ from state to state – and compliance can differ from county to county – I started by separating the survey responses geographically. My challenge was to ensure sufficient data in each geographic subset for robust statistics. Even though a single response to the CPS represents, on average, more than 2,000 Americans, it is still just one data point.

The CPS classifies responses by state and core based statistical areas (CBSAs). These areas generally represent metropolitan areas and can cross state lines; you can see maps of each state by CBSA here. Since each state can set different regulations for voting, I split up the CBSAs by state. Then I looked for CBSA fragments that had at least 100 responses from Black interviewees. I called these “big” areas. Then I classified CBSA fragments with responses from 50 to 99 Black interviewees as “medium” areas. I grouped together the remaining areas by state, so each state has a catch-all small town and rural subset.

The “big” areas are centered around the following cities: Atlanta, Baltimore, Baton Rouge, Birmingham, Chicago, Dallas, Detroit, Houston, Jackson, Los Angeles, Miami, New Orleans, New York (separate entries for the State of New York and New Jersey), Philadelphia (separate entries for Pennsylvania and Delaware), Richmond, Virginia Beach, and Washington (separate entries for the District of Columbia and Maryland).

The “medium” areas are centered around these cities: Boston, Charleston, Charlotte, Cleveland, Columbia, Dover, Lafayette, Las Vegas, Little Rock, Louisville, Memphis (separate entries for Mississippi and Tennessee), Montgomery, Nashville, Orlando, Phoenix, Pine Bluff, Providence, Raleigh, Riverside, San Francisco, Shreveport, St. Louis, and Washington (for Virginia).

A caveat: Any of these subsets could cross county lines, so the results I obtain should be viewed as indicative rather than definitive. As I like to say, the data help us to ask the right questions. If we find something strange in a CBSA fragment, we’ll want to look more closely at all of the counties involved.

Well, that’s already a lot of information, and I haven’t even started writing about the analysis. But it’s important to have a good grip on the data before presenting any results, even preliminary ones. That’s what I’ll do in the next installment.

[Image: U.S. Department of Commerce]