Friday, November 4, 2011

Mail voting, Oregon, and coercion in the family [Voting 02, PhD Ideas 01]

In my post on Mail/internet voting v. the secret ballot [Voting 01], I discussed my concern about replacing voting booths with mail/internet voting.

What worries me is a species of what might be termed false completion.  I am thinking specifically about a subspecies of such fraud, the kind that can occur within a household when people vote by mail (VBM).  When voters go to physical polling places and vote secretly at private voting machines, this kind of fraud is rarely possible.  With VBM it becomes much easier.

There are two obvious varieties of false completion within a household: coercion and usurpation. 
  • False completion by coercion occurs when voter X is intimidated by person Y in the household to mark the ballot in accordance with the preferences of Y not shared by X.
  • False completion by usurpation occurs when voter X's rights are nullified because person Y in the household takes the ballot away from X and completes it according to Y's choices,  choices different from X's.

I know that for the last several years the state of Oregon has required all ballots to be mailed.  I wondered whether Oregon had been concerned at all with false completion within households.  Or whether scholars generally have considered this problem.  What I found suggested that the problem of household coercion/usurpation is given lip service at best and has never been studied.

In a 2005 report on the Oregon experience by The Early Voting Center at Reed College, neither the word "secret" nor the phrase "secret ballot"is mentioned. The report makes passing reference to coercion as the third of three security issues, but does not follow up.  The Center's website, visited today, is full of interesting information and thoughtful analysis. Neither "coercion" nor "family," however, net any hits.

The Wikipedia entry on postal voting, viewed today, also refers to coercion as a possible issue but then ignores it.

The security issues that do receive attention are various kinds of fraud, in particular false ballot completion and false registration.  Nobody seems to worry about false completion within a household.  Rather, the imagined situation seems to be wide-scale voter tampering: fraud by strangers to the particular voters. For example, party X operatives might intentionally intercept ballots addressed to people registered with party Y and would then vote those ballots for party X's line.  That is something to protect against, and there are ways to detect it, at least after the fact For example, some of the Ys will complain that they did not get a ballot and will report the theft when they discover that their ballot has already been submitted. False completion within a household is much more insidious but could equally result in disenfranchisement and could skew election results.

I agree that it would be difficult to collect data on the percentage of citizens who did not vote their true choice because someone in their family was standing over them or grabbed the ballot out of their hands.  That information, unlike turn-out and other data that more readily pop out of computers, might have to be gathered by personal surveys undertaken in a setting where the possibly-intimidated voter was not intimidated -- so not door-to-door.  That would not be a cheap or easy study.

But maybe there is a way to learn about false completion within households that could be done with what I jokingly call  RISC architecture  (a play on the real computer science term RISC architecture).  Here are some suggestions, in case anyone is working on a PhD on voting rights.  Or family dynamics.

Ways to Use Data-mining to Study False Completion in Households

Select a past election where an age or gender gap was predicted by pre-election polling.  The vote can be for a candidate or a ballot measure, as long as some kind of demographic gap, one that would occur within households, was expected.

Gaps along the lines of race, ethnicity, or income, where all members of a household would likely be in the same group, could be a good control, but would not themselves reveal anything about false completion within households.

 In hybrid-voting states, where there are physical polls as well as the option to vote by mail, the study would look at whether the gap among voters who cast their ballots secretly at physical polls was statistically different from the gap among VBM voters.

In mail-only (or mail/internet-only) elections, the study could look at the whether the gap was statistically different for single-person households compared to multiple-person households. For example, if an age gap was predicted, the study might look at seniors living alone v. seniors living with relatives.  (Seniors living in assisted living facilities might also be subject to more widescale coercion/usurpation, but that would be a form of false completion by strangers rather than within households.)

Another study might compare measures/candidates on a single ballot, some for which a gap was expected, and some not, and then compare VBM voters and true secret ballot (private voting machine) voters.

Yet another might look at household solidarity (all voters in one household voting alike on every ballot choice) of VBM households versus physical poll households.  That could be done in Oregon, comparing patterns before the switch to mail-only with those after the switch.  It would, however, mean that the researchers would have to see voting data associated with addresses.  That might be impossible due to privacy concerns, even for historical data.  But maybe there would be a way to safeguard privacy by having the software do the seeing and having the humans only learn aggregate results?  Anyway, it's a thought.
If such studies already exist, please let me know.  If you embark on such a study, please let me know that, too!
11/4/11 rjm
rev 11/6/11, typo corrected 6/24/12

No comments:

Post a Comment