How a democracy data analyst assesses changes in voter turnout.

In November 2022, Alaska saw its lowest voter turnout in any general election since 1980. While it's tempting to try to blame a single flashy reason for that fact, there are many factors that can raise and lower overall voter turnout: voter interest, competitive high-profile contests, media coverage, ballot access, even the weather on Election Day. Another factor that can impact voter turnout: changes in election laws, such as Alaska's 2022 implementation of ranked choice voting (RCV).

Since higher voter turnout is a win for all of us, it's important to find out if ranked choice voting has any impact on voter turnout and how that impact stacks up against the system's other pros and cons. In this article, I'll walk through how I calculate voter turnout, how I explore a data set to investigate any changes in voter turnout, and what the November 2022 election can tell us about ranked choice voting's impact on turnout. (Spoiler: it can't tell us much.)

What's in a denominator?

To start, 267,047 Alaskans cast a ballot in November 2022. Simple! But not super useful without any context. Is that high? Is it low? How does it compare to other elections or other states?

In the 1958 election for Alaska statehood, only about 48,000 people cast a ballot. Down in Washington state, more than 3 million people voted in the 2022 general election. Comparing the raw number of voters in Alaska in 2022 to these two numbers doesn't say much about voting behavior. All it really shows is that way more people live in Alaska now than in 1958, and that way more people live in Washington now than in Alaska.

Since the raw numbers lump together differences in voting behavior with differences in population, I can isolate the impact of voting behavior by dividing the raw vote count by some measure of population. Choosing the right denominator is important for calculating voter turnout.

A few different measures of population are useful for understanding different aspects of voter turnout. Total residents is useful for assessing the impact of large-scale disenfranchisement, but the huge number of legally ineligible voters under 18 will drown out the smaller impacts that I'm concerned with here. Registered voters is good for administrative checks (because each voter is listed individually, unlike aggregate census data) or when assessing partisan turnout differences, but factors such as automatic voter registration can artificially decrease registered voter turnout by increasing the number of registered voters.

The sweet spot for the Alaska election turnout analysis is the voting-eligible population, a metric published by researchers at the University of Florida that counts the number of people who are legally eligible to vote in a given jurisdiction. Many election researchers prefer the voting-eligible population as a gold standard for calculating voter turnout in their analyses.

So! In Alaska's 2022 general election, 267,047 voters cast a ballot. That's 50.27 percent of the state's 531,272 eligible voters.

Next, to figure out if that number is meaningful or unusual, I'll compare it to the state's previous general elections.

Visualizing and contextualizing

Now that I know what data I'm working with, my next step in trying to find any stories that the data is telling is to visualize this data. Turning the data into a plot or graph often makes it easier to spot trends or details that wouldn't be clear from the numbers alone. I typically focus my attention on ranges (the spread of the values), patterns (any trends or repetitions), outliers (unusual points that break patterns), and subgroups and supergroups (comparisons between sections of the data or to other data sources).

Ranges

Looking at the range of the data is a useful start to finding any lessons or highlights from the information. We want to understand all the values our data could take and the values it takes in practice. I'll start with a l...