Monday, November 05, 2007

The mythical dead voter

Here is a persistent myth that has a lot less support than people think. Every election cycle people trot out the belief that there exists a huge number of dead voters who show up at the polls. This year WPXI has gone a little farther and investigated the phenomenon: Ghost Voters Cast Ballots In Pittsburgh Elections. Investigation Of Voter Records Yields Surprises.

Surprises eh? According to this web article they poured over records to find 9 thousand or so voters already deceased. Is that a lot? Comes out to about 1% of all registered voters and less than the number of deaths in the county each year. They found "16" of them who actually voted. 16 out of just shy of 900K registered voters in the county. Thats an amazingly good record if WPXI claims their research was thorough. The surprise in the headline is how few dead voters there are, not how many. Yet I bet many who don't read past the headline think this is some huge problem now proven by the story title. 16 voters spread throughout the city can affect virtually no election out there even if they are fraudulent. and even the WPXI story suggests the possible explanation of similar names and sheer clerical error. Not that votes were created, but just misreported.

Here is the deal... if people really are not removed from the voter rolls soon after they are deceased, it would be a fairly hard thing to hide. I am sure there are some limited examples of people who are missed. Remember, people die all the time. If even a small fraction of deceased voters were routinely left on the rolls, it would be a very short time before the voter rolls showed a severe imbalance in the number of older registered voters and what you would expect by looking at the population.

I have pointed out this simple graph to people in the past and they still swear every person deceased in the last decade are still on the rolls. Here is a graph of the age distribution of all registered voters in Allegheny County compared to the age distribution of the population. Both are for 2006.




What do I see? At the older age ranges there is a slight difference between the proportion of registered voters compared to the population. Its pretty slight and I tried to quantify how many.. Mostly a guess, but the discrepancy would be consistent with an average delay in getting voters off the rolls of less than 6 months. It really just does not show up.


What is more interesting is the higher registration proportion for the younger voters which may seem counterintuitive, but really isn't. That results because it is a lot harder to purge the local voter rolls of people who just pick up and move away. Thus the apparent discrepancy between the two sets of numbers. Those are the voters who are most likely to be moving in and out all the time, which leads to there being more registered than actually living here. Maybe every time a poll workers sees a rare voter under 40 they should be asked to certify if they still live in their voting district, that is certainly easier than asking people to certify they are still alive.


That problem exists not only for the number of people who move in or out of the county, but for those moving within the county as well. The graph above suggests a far larger problem is not the number of deceased voters on the voter rolls, but people who are still alive but no longer live where they are registered, yet still voting there. Especially in general elections when people want to vote for the top race, things like Prez, Gov of Senator, people who have not had time to re-register will go back to where they used to live to vote. I doubt it is as big an issue for municipal elections, but I have no doubt it happens to some degree. When you add in the fragmentation of the county, you have to believe there are many people voting for municipal officers that are not their own. Given that there is a fairly steady flow of people moving out of the city yet still living in the region, I bet there will be more than a few people voting for mayor tomorrow who do not live in the city and I think it's just as illegal as those mythical dead people voting and as I explain with the graph above, far more likely to actually be happening than votes from the grave.


Why the discrepancy between popular perception and the data. Like I said, I am sure it happens sometimes when someone dies yet does not get removed from the rolls.. that or there is some normal delay for mortality records to make it over to the elections division. So when you go and vote and are looking over the shoulder of the clerk and see some name you know is deceased I bet you over-interpret what you see.. especially given the persistent stories like what was just on WPXI. Its a fun story, but the data just isn't there, as WPXI seems to confirm despite the headline. I bet many of those who are deceased, yet remain on the rolls are people who moved out either seasonally or in recent years and who passed away in some other jurisdiction which makes reconciliation with local voter rolls at least slightly more difficult or delayed.

2 Comments:

Anonymous Anonymous said...

It is ironic that WPXI is running a story on a nonexistent problem when the much bigger problem relates to states over-purging the voter rolls. Much of the problem with Florida in 2000 (and Ohio in 2004) was the fact that tens of thousands of voters had been improperly purged from the rolls. Many people vote only occassionally. So the Help America Vote Act only allows a name to be purged in the case of death, written request of the voter, or failure to vote in two consecutive elections AND a failure of the voter to respond to a written inquiry sent to the voter's registered address.
In Pennsylvania, the Bureau of Records regularly compiles monthly lists of the dearly departed and sends them to the county elections offices for purging. HAVA prevents ANY purging of the rolls within 30 days of a federal election.
Additionally, the NVRA ("Motor Voter Act") automatically updates registration information when you move--unless you fail to report your new address to PennDOT. So I doubt that many persons are regularly voting in a district where they do not live.
Also, in Pennsylvania you can vote one time at your old polling place, even if you have moved.

Monday, November 05, 2007 4:36:00 PM  
Anonymous Anonymous said...

Oops, should have said "two consecutive FEDERAL elections".

Monday, November 05, 2007 4:37:00 PM  

Post a Comment

Links to this post:

Create a Link

<< Home