Income, Race, Age, and Gender: A Data-Driven Look at Voter Turnout in Oakland County

(Kurt Metzger, March 13, 2024)

Oakland County, MI – Another Presidential election year is upon us and voter turnout is once again an important topic. Election pundits will be discussing a wide range of factors that will influence turnout, including internal and external factors.

Internally, individual voters are affected by motivation, eagerness and knowledge. Externally there are issues of eligibility, mobilization, and voter suppression.

While Oakland County voters in all communities faced these factors in both 2020 and 2022 (along with COVID and successful voter-initiated proposals), variations in voter turnout – 46.8 to 89.1 percent – require different factors for explanation.

The demographic characteristics of a community’s population are often used to predict turnout. The standard demographic factors that have been shown to affect voting are:

Women vote at higher rates than men.
White, non-Hispanics vote at higher rates than persons of color.
Persons with college degrees vote at higher rates than those without.
Persons with higher incomes vote at higher rates than those on the lower end.
Persons 65 years and over vote at higher rates than any other group.

In order to test the impact of each of these, I pulled the latest data on gender, age, race/ethnicity, income and educational attainment for the 51 communities for which voter turnout data are available. I then ranked the communities 1-51 on each factor, and ran correlations between each factor and community rank on voter turnout in both 2020 and 2022 November elections.

Correlation is a statistical measure that expresses the extent to which two variables are linearly related (meaning they change together at a constant rate). It’s a common tool for describing simple relationships without making a statement about cause and effect. Correlations run from a low of 0.0 to a high of 1.0. The closer the value is to 1.0 – the closer the two rankings correspond across the communities. In other words, as the value rises, the ranking across variables becomes more consistent (1=1, 5=5, 51=51, etc.).

Comparing communities based on the percent of women in the population yielded a small negative correlation (-.20). In other words, as the percentage of women increased, voter turnout tended to decrease Huntington Woods provides a perfect example of this fact in that, while it ranked 3^rd for 2020 turnout, it ranked 40^th on percentage of women, while Holly township ranked 8^th for percent women but 45^th in turnout.

Looking at the percent of the population 65 and over yielded a relatively small positive correlation (.40). A stronger, though still moderate, correlation resulted when looking at the percent of the population that is white, non-Hispanic (.59).

Once I moved on to the socioeconomic characteristics of education and income, I found very strong correlations. When percent Bachelor’s degree or more was the factor under consideration, the correlation coefficient rose to a strong .73. However, it was income that told the tale. I first looked at median household income and found a similar strong correlation of .74. The second income variable, percent of households with incomes of $100,000 or more, the coefficient jumped to a very strong .84, meaning that the rankings across both variables were highly consistent.

I have provided 2 illustrations of the data mentioned. The first provides the entire list of 51 communities with their rankings on 2020 and 2022 turnout, percent 100K+, median household income and percent Bachelor’s+. The second illustrates the high correlation between 2020 turnout and percent 100K+. As one can see, there is a consistent pattern of both ranks moving together.

A perfect correlation between income and turnout would have yielded a coefficient of 1.0. Obviously, there were some overperformers – turnout higher than income would predict, as well as underperformers – turnout lower than would be expected. While you can pull these out by looking at the ranking chart, I will give you a few of each to mull over.

Over Performers – Sylvan Lake, Lathrup Village, Clawson and Ferndale

Under Performers – Birmingham, Bloomfield Hills, Troy, Groveland Township and Rochester

While I understand that this article is rather ‘wonky,’ it is important to know how such community demographics affect campaigns. While politicians are obviously interested in past turnout and party preferences at the community level, their campaign folks will use demographics to identify where they will get the most payoff and need to concentrate their efforts. It is understood that persons of low socioeconomic status do not see politics as the answer to their issues, but their resulting low turnout just reinforces the lack of attention. We must work – both residents and politicians – to change the playbook.

Check out our previous stories: