Update: Just to clarify for some folks who have read the headline without reading to the end of this post: No, I have not found any evidence that dead people or children are voting in Utah. I have only found that there are unusual birthdates recorded for many voters. This post below was just a brief note on an ongoing research project, and it wasn’t meant as a final statement on the quality of Utah’s public data.
This morning I was browsing through Utah’s voter file, the publicly available index of all of Utah’s registered voters. It has some issues. From what I can tell, there’s a lot of children and dead people registered to vote in Utah.
The voter file doesn’t contain much information. It lists each voter’s name, partisan affiliation, birth date, address, and turnout history. It also lists each voter’s status. Most are marked as “active.” Some are marked as “inactive,” “not eligible,” “removable,” or “suspense” depending on the circumstances. Election officials need this information to administer elections. Candidates can purchase this information, as can researchers or anybody else. It’s a public record.
There were several dozen entries that were so corrupted or incomplete that I couldn’t read them into my database. Setting those aside, I was able to import the data for 1,551,086 registered voters. In 2010, the state reported that there were 1,267,250 registered voters. I’m at a loss to explain why I’ve got 283,836 more registered voters in my database than they reported in November 2010. Utah’s population hasn’t grown nearly that fast. (Update: The 1,267,250 voters reported in Nov 2010 was the number of “active” voters; my copy of the voter file shows that of 1,551,086 total, 1,247,831 are active, which is roughly the same as in Nov 2010.)
Of the 1,551,068 registered voters, 271 are children as of today, having been born after August 15, 1993. Oddly, 43 of those 271 voters are marked as “active,” and most of those 43 have cast at least one ballot since the 2008 elections. In fairness, most of these “children” appear to be adults who carelessly wrote down the current date instead of their birthdate when registering to vote. And to the state’s credit, most of the 271 with an under-18 birthdate are marked as “not eligible” or something other than “active.”
Meanwhile, 9,259 registered voters are over 100 years old, having been born in 1910 or earlier. Of those, 5,966 are marked as “active” voters; most of these 5,966 have cast at least one ballot since 2008.
Now, I’m sure some of these folks really are alive and voting. But I’m suspicious at that large number, seeing as the 2000 Census reports only 7,177 folks who were 85 or older living in Utah at the time. There can’t be more than a couple hundred centenarians in Utah at most, yet 9,259 are registered to vote. Something’s wrong.
Here’s something even stranger. There are 3,830 registered voters with a birthdate in the 1800s or earlier. The overwhelming majority of these voters (3,574) have a birthdate between 1800 and 1810. These are obvious mistakes. Still, a stunning 2,871 of them–75%–are marked as “active” voters, and most have voted at least once since 2008.
Now, I’m not into conspiracy theories. I don’t really believe that I’ve found evidence that children and dead people are voting in Utah. I’m guessing most of these records are for living adults who simply had their birthdates recorded incorrectly.
Update: See which counties have the most errors in their voter registration data.
Update #2: I’ve been working to find out why these errors are in the data. Many of the errors are intentional “errors” by county clerks. If a voter first registered back before the state required that birthdates be collected with registration information, for example, then counties would generally enter an error code like “09/09/1809” as an error code indicating that the voter in question was not required to provide a birthdate back when he or she registered to vote.
Perhaps some of these don’t want to give out the information to the public.
I am sure the birth year attached to the glasses I ordered is not correct, which is good as the company had their records hacked.
It’s very different when a corporation collects ridiculous amounts of personal information. I routinely make stuff up when corporations ask for personal details without a good reason, just as you probably made up the name “George.”
But it’s different when it comes to regulating elections. If people don’t give a valid date, they shouldn’t be on the voter rolls.
Now, whether the state should disclose those birthdates as opposed to just the birth years is a different matter. But my post isn’t about disclosure of information, it’s about collection.
Just FYI, the dates of birth can be a valuable resource for election administrators seeking to make corrections to their lists: a number of states take part in interstate compacts to identify voters who have moved across state lines (such as Washington and Oregon).
In my town of 35,000 voters, for example, using both names and dates of birth, we have identified 700 voters who are registered with us and in the state of New York, and 500 voters registered with us and in the state of Florida. Using only the year (as Ohio, North Carolina, and Vermont do) produces too many false positives — 365 times as many false positives — to be of any use. Likewise, a voter file with dates of birth can be matched against the Social Security Death Index to identify deceased voters en masse.
If you took the opportunity to sit down with a collections agency, you would quickly realize that the volume of data commercially available for each individual dwarfs that contained in the voter file, so I’m not sure how serious the privacy implications are. Dates of birth along with historical address information, credit information, consumer preferences, etc are all available to motivated purchasers. But there are real benefits to the election process in having the information available along with voter records.
Out of curiosity, does the problem occur more frequently in any specific county? After the botching of district boundaries in LD57, I’m sure a lot of folks would like to know if it’s a pattern.
Good question. Give me 2 minutes and I’ll have that.
See here: Which counties have more registration errors?