Three Factors Contribute to ‘Bloated’ Voter Files in Kansas

Do nearly 250,000 voter registrations merit scrutiny?

Mar 04, 2024

Two recent WatchdogLab articles show issues with Kansas voter registrations:

49,667 Kansas voters, who registered to vote in 2012 or earlier, have never voted in any election according to voter history.
77,948 voters, who have voter history for voting in 2012 or earlier, have not voted since.
142,308 voters in Kansas have an “inactive” status in the voter file. This means election offices cannot contact these voters at their last known street addresses by mail. Without a known address, the election office does not know what ballot style applies to an inactive voter.

All these voters can vote if they show up at the polls, or request an advance ballot, but don’t they merit additional scrutiny?

Overlap

Let’s determine how much overlap there is among these three sets of voters using an “UpSet plot.” An Upset plot provides information similar to a Venn diagram, but is often easier to interpret, especially when there are many sets to compare.

The blue horizontal bars at the lower left are the sets to compare, which would be the circles in a Venn diagram. The black vertical bars show the set intersections — how many voters are common to the original sets.

Upset plot showing overlap among three sets of voters.

Interpretation of the five black vertical bars:

120,658 voters are “inactive” and have unknown or invalid mailing addresses. The process to remove “inactive” voters works but is quite s-l-o-w.
65,793 have voter history from 2012 or before; no voter history in recent years. It’s unclear why most of these are not “inactive” voters now. However, “disinterested voters” cannot be removed only for not voting.
40,172 voters were registered in 2012 or before; have no voter history; have never voted. It’s unclear why most of these are not “inactive” voters now. Again, “disinterested voters” cannot be removed only for not voting, but the post office’s National Change of Address does not identify everyone no longer at an address.
12,155 voters are inactive; have voter history from 2012 or before; have not voted in recent years.
9,495 voters are inactive; were registered in 2012 or before; have no voter history; have never voted

The sum of the numbers above the black bars shows how many unique voters are involved in these three issues: 120,658 + 65,793 + 40,172 + 12,155 + 9,495 = 248,273 voters.

Why do these issues matter?

University of Michigan professor J. Alex Halderman teaches in his online Securing Digital Democracy class to approach all aspects of voting and elections with a “security mindset.”

Currently there are few “security mindset” checks and balances applied to inactive voters, or those who have not communicated with their election offices in many years.

In Kansas if you vote in person one must show a valid ID, usually a driver’s license, and often one must answer questions about your name and address on the ID from the poll worker.

If you request an advance ballot one must only give a driver’s license number (or a copy of a photo ID). Someone with access to another’s ID could request a ballot. There’s no poll worker to challenge someone with a mail ballot application about their name and address — fewer safeguards for mail ballots.

It’s unclear how good the signature verification process is when a ballot is returned. There is almost no transparency on how that process works or what its failure rate is. It’s unclear how much better signature checks are than the “honor system.”

So, eliminating or reducing the number of inactive voters, or the number of voters who have last contact with the election office, would reduce the risk of mischief, especially with mail ballots.

With security breakdowns, inactive voter names could be used by corrupt election officials to cast ballots in others’ names with little chance of being caught. I’m not saying this has ever happened in Kansas, but voter “bloat” enables mischief.

Maps: differences by county

Each of these three issues show rate differences in the 105 Kansas counties in the maps below. At what point do these differences become a violation of the Fourteenth amendment for “equal protection under the laws”?

The inactive rates do not correlate with either of the other two factors by county (correlation was only 0.3).

However, “Never Voted Since 2012” has a correlation of about 0.7 with “Voted 2012 or Before.” (A correlation of 1.0 is a perfect positive correlation, and 0.0 indicates no correlation.)