Election Updates

Okay, my email has been full of material on the humidity issue and paper ballots.

It turns out that Doug Jones of Iowa has written a bit about this problem relating to optical scanning devices:

t is worth noting that paper ballots change size slightly with changes in humidity! Commercial printers use the rule of thumb that a 10% increase in relative humidity causes paper to expand by about by one part in 1000; as a result, the size of a piece of bone-dry paper could expand by as much as 1% as it picks up moisture in an extremely humid environment. This comes to 1/10 inch in 10 inches, while many mark-sense voting targets are about 1/8 inch in their short dimension!

The early mark-sense scanners used for scoring tests solved the problems caused by humidity changes by storing all test papers in a controlled atmosphere for some time prior to tabulation, but on modern mark-sense vote tabulators, the design accommodates humidity changes by spacing the sensors to the midrange of variation in paper size and using voting targets that are wider than the sensors.
...

If one or more of the top or bottom index marks are not seen on a pass through the scanner, the scanner typically stops with an error message indicating a misfeed. This could be due to an extreme dimensional change in a ballot, for example, because the humidity is too high for reliable scanning or because the paper has been crumpled and then flattened, or it could be because the paper passed through the scanner on a diagonal.

The fact that sensing tracks are narrower than the voting target means that a mark may be seen on one pass through the scanner and not on a later pass, perhaps because of a humidity change from one pass to the next. The fact that a ballot may be on a slight diagonal as it is scanned introduces the possibility that identical marks in two different places on the ballot might be counted differently, even when they are seen by the exact same photosensor.

And here is a report from 2004 in Lawrence, Kansas:

Today's humidity did more than make voters uncomfortable -- it also helped gum up the works during vote counting at the county courthouse, election officials said.

Marni Penrod, deputy Douglas County Clerk, said Tuesday night that one of the county's two vote counting machines typically was susceptible to minor problems related to the counting of paper ballots. But today, in particular, the high humidity made some ballots stick together and the machine had trouble counting some ballots.

Ballots from two precincts were particularly troublesome: Those from the North Wakarusa precinct, which votes at Wakarusa North Fire Station and those from the Coffin Sports complex at Haskell Indian Nations University. Neither of those polling stations have air conditioning.

"There were a lot of problems with those ballots, Penrod said. "They've gathered a lot of humidity all day."

And there were many other similar reports ...

Thanks to those of you who mailed me these materials and examples today.

I've written and spoken recently about the need for election officials to think through potential problems with their voting systems and to produce contingency plans for when things go wrong (for example, security threats, large-scale problems, or natural disasters). I talked about this again a few weeks ago at the AEI/Brookings election reform workshop.

But as I was reading through the latest edition of Newsweek this evening, I saw a story that didn't at first seem very related to voting technology. But I later scrambled to grab the issue of Newsweek out of the recycling, because I realized that the story may have some very important ramifications for optically scanned paper ballots. And I then thought that election officials using optical scanned ballots might have an enemy that I never seriously though about --- humidity.

Here's the basic problem. As recounted by the Newsweek story, recently at least four thousand high school students throughout the nation received word from the College Board, the entity that administers the dread SAT, that their SAT scores were scored inaccurately --- their actual score was greater than what they were initially told. Amazingly, the number of incorrectly scored tests is somewhat greater than 4000, as there were some unknown number of scores (the College Board is quoted in the story as saying "substantially fewer than 4000") that were incorrectly scored higher than they should have been, but these students are not being contacted and their scores are not being changed (why the error has such a skewed distribution, that is, producing a greater number of incorrectly lower scores than incorrectly greater scores, is not clearly addressed in the article). So the number of incorrectly scored tests is greater than 4000, but apparently less than 8000.

In some cases, the errors look relatively large. It seems that forensic analysis by the College Board traced the errors to one batch of tests, those conducted last October. But the Newsweek story noted that the distribution of errors was potentially problematic for the students taking the October SAT:

A total of 495,000 students around the country took the SAT in that sitting, meaning that fewer than 1 percent actually saw their scores change. And in most cases, it might not have made a difference: 83 percent of those affected lost only 10 to 40 points. But some 200 students missed out on more than 100 points, and the scores of 16 were off by 200 to 400 points.

But what does this have to do with voting technology?

A contractor for Pearson Educational Measurement is quoted in the story as blaming the scanning technology used to assess SAT tests. He said:

"On the day of the test administration, in the Northeast part of the country, there was heavy, heavy rainfall. We're thinking the test documents expanded because of the moisture in the air."

Then the theory goes on:

At the Austin, Texas, scanning facility to which they were shipped, they may have sat in a warehouse for just two to three hours --- not long enough to dry before they were sent through the machines, and some answers on the expanded paper didn't align with the scanners. Tests from other parts of the country may have picked up extra points because they had been filled in too lightly or contained stray pencil marks.

So the theory is not blaming the heavy rain for actually getting the test forms wet, but for just providing enough humidity to distort the paper sufficiently so as to render their alignment with the scanning devices problematic.

Now we have a new problem to wonder about on rainy election days. In the past we tended to just be concerned about how rain influenced voter turnout. But now we might need to worry about humidity levels, especially as the most recent estimates have at least four of every ten votes being cast on an optically scanned ballot this year. Cross your fingers and hope for dry weather on election day! Or make sure that you have procedures in place for drying out paper ballots if for some reason the humidity levels are high in your jurisdiction on election day (and be prepared to tell candidates, the media, and the public that they won't get immediate gratification in the form in real-time election returns immediatedly after the polls close).

One other minor point to make for my colleagues who study residual votes and voting technology; does this story suggest that they need to control for rain or humidity levels when estimating the residual vote associated with optical scanning devices? I'm only joking, but I do wonder.

Two candidates who narrowly lost primary races in Starr County, Texas, earlier this week have alleged that irregularities in early and absentee voting may have cost them the race, according to a story in the McAllen, Texas, Monitor. One of these candidates (David Porras) lost by 55 votes, while the other (Dave "Chachi" Jones) lost by 300 votes, of the nearly 6,900 votes cast in Tuesday's primary.

Here are the allegations that were reported on in this story.

Porras and Jones had campaigned together for their respective races, as had Maldonado and Garcia. They won based on the number of ballots cast at polling stations in early voting and on Election Day but garnered less than half the number of mail-in ballots Maldonado and Garcia received.

Jones, who works as a private investigator for a bail-bond service, said he was suspicious because he had helped more than 600 people — most of them elderly — request mail-in ballots but the elections department listed just 253 of those ballots in its official tally.

“I know for a fact that I assisted more than 253 people,” he said. “The numbers that I have and the numbers they counted don’t match.”

Interestingly, these losing candidates don't seem to think that the local recount will resolve the election in their favor:

The logistics of the recount are yet to be determined. Both Porras and Jones said they didn’t expect the recount to vindicate them but had asked for it because it as a precursor to challenging the entire election with the Texas Attorney General’s office — their likely next step, they said.

This goes back to something that Thad wrote about on March 4th; he is running a project that is going to collect and analyze information from every state's election code about their rules and regulations for ballot counting and recounting. I do have a volume on my desk that is about three inches thick, that has the entire body of Texas election law in it, and it will be exciting when Thad's project is complete and we will have a more analytic understanding of of these aspects of state election code vary across the states, and how the count/recount procedures may be related to measures of the performance of a state's election system.

In the 2005 special election in California, Orange County (California) implemented a number of new tools that allow folks to monitor the election administration process remotely (over the Internet). Orange County is having an election on April 11, 2006 (the 35th Senate District primary), and in addition to the same remote adminstration viewing tools they let the public use last fall, they are adding some new functionality to their website for this primary election. You can go to to the following link for the same basic tracking and viewing tools that were available in the 2005 special election.

But they have added the ability to view the number of absentee ballots mailed, and returned, on a daily basis. Unfortunately it is a bit difficult to navigate to these two sites, but here is the link for the daily report on absentee ballots mailed, and here is the link for the daily report on the number of absentee ballots counted. This is great functionality for election administrators to add to their public reporting of election administration, especially in states like California where early and absentee voting is quickly becoming one of the primary modes of participation for voters.

Recently the U.S. Election Assistance Commission (EAC) released a report on their 2004 survey of UOCAVA voting. While this report is focused on UOCAVA voter participation, and the number of UOCAVA ballots that were returned, the report is unsatisfying because of the problems the EAC encountered in trying to collect the information needed to account for UOCAVA voter participation in the 2004 presidential election. Due to a variety of issues (which I'll discussion in more detail below), the EAC was unable to compile the necessary information that would tell us the number of UOCAVA voters who requested absentee ballots, the number of those ballots that were returned, and the number that were then counted --- basic facts about election administration that should be reported to the public, but which have not been, even with the EAC efforts to collect and distribute this information.

The difficulty that the EAC faces in trying to collect and distribute this basic data about the democratic process in the United States demonstrates how much work needs to be done on data collection and distribution practices by election officials.

Turning specifically to the report, in it the EAC details the methodology used and the survey that was distributed to state election officials. But:

1. Some states failed to respond to the requests for information, or to requests for additional information.
   
2. Some states altered the information they reported, and it is unclear what the altered information reflects (incorrect original information, or estimates based on new data).
   
3. Some states are unable to break UOCAVA voters from their abstentee voting totals, thus reporting to the EAC not the UOCAVA participation rates, but the total absentee voting rates.
   

The last problem is serious, and unfortunately in the report I cannot find any reference to which states are providing total abstentee voting estimates and not UOCAVA voting estimates.

Based on the information provided in the report, it appears that about 1.1 million UOCAVA abstentee ballots were requested in 2004, and about 840,000 were returned (it is unclear how precise this estimate may be, given the problems of data reporting listed above). Once we eliminate the reported 62,468 UOCAVA absentee ballot requests from the calculation from California (California reported UOCAVA absentee requests but not the returns), we get an estimate of about 81% for UOCAVA absentee ballot returns.

But as we really don't know here which states reported total absentee numbers, nor for those states the UOCAVA absentee ballot request or return rate, we do have to keep in mind that these are estimates, and that they are likely to be biased (though the direction and magnitude of the bias is unknown).

Yet the bottom line here is how difficult it is for the EAC (and by implication researchers and the interested public) to get detailed post-election information about even basic questions, like the participation rates of this special class of voters. Groups like the Caltech/MIT Voting Technology Project have called for better quality data (see the October 2004 report, "Insuring the Integrity of the Electoral Process: Recommendations for Consistent and Complete Reporting of Election Data"), and the EAC has tried to collect and distribute information in response to these calls (see the EAC Election Day Survey, as well as our thoughts about how much further we have to go to get the necessary data to really understand the election process and how it performs in the United States ["EAC Election Day Survey underscores need for better retention, collection and distribution practices"]).

At this point, time is short to work with election officials to improve the information they collect and distribute to the public. Many states will be having complex primary elections very soon, and the November elections are really just months off. Perhaps the EAC can provide some guidance or guidelines for election officials about what information they should collect and distribute; or perhaps some of the other entities that assist election administrators could step up and help provide some guidance and assistance for improving collection, reporting and distribution of detailed election administration data.