Tuesday, February 27, 2007

Double Whammy for Wisconsin Voter ID Bill

Voter ID is in the news once again in Wisconsin. Assembly Republicans are trying to push it through over the objections of Governor Doyle and the Democratic Senate. But their job just got a heck of a lot tougher because the first scientific data on the effects of voter ID was officially released, and it's a double whammy for the Republicans. Not only does the study show that voter ID very significantly reduces overall turnout, but it also shows that the Wisconsin bill is actually more likely to cut white turnout than minority turnout.

Now, I would never accuse my good Republican friends of pushing voter ID for political gain, but somehow I suspect that they won't be quite as enthusiastic when they realize that the political party most likely to be hurt by the law is their own.

The study, conducted by researchers at Rutgers and Ohio State for the US Election Assistance Commission, was completed last June but not publicly discussed by the Commission until earlier this month. It showed that tough voter ID requirements cut overall voter turnout by nearly three percent, black turnout by nearly six percent, and Hispanic turnout by ten percent. But requiring a photo ID, as opposed to other methods such as non-photo ID's or signature matching, cut white turnout far more than minority turnout. Bummer, if you're a Wisconsin Republican.

The full study was actually released last summer but nobody noticed, perhaps because it's kind of tough reading unless you're a masochist. Within the last couple of weeks both USA Today and The New York Times have done articles about the study, although the most interesting write-up I've seen is on this blog. But all of these reports skip over a lot of the details important in Wisconsin, such as the relationship between turnout reduction and the specific type of voter ID. You really need to push through the study if you want to find the good stuff.

This study also didn't examine the rate of fraudulent voting, so it wasn't able to compare the number of valid votes lost due to stricter voter ID requirements to the number of fraudulent votes prevented. But based on the overall findings, the proposed Wisconsin voter ID bill would reduce turnout by perhaps fifty-thousand votes, and it's kind of hard to imagine that there could be anywhere near that many fraudulent votes here.

I've always thought that the basic question voter ID supporters must be willing to answer is how many good votes they're willing to sacrifice to prevent one bad vote. Compared to most other democracies we already have a very low voter turnout, due at least in part to long-standing efforts to disenfranchise one group of voters or another, yet democracies can only work when government has the informed consent of its citizens. Seems to me that we should be doing everything possible to get as many people as possible to the polls, not the other way around.

Yes, we must always be watchful for fraud, but every fraud prevention technique should be judged solely by how it effects the integrity of our democratic process. If a stricter voter ID requirement eliminates more good votes than fraudulent votes, or if it has disproportionate effects on particular groups of voters, then it's a failure. Based on the data now starting to come out, the proposed Wisconsin voter ID law is not likely to meet this standard and should be reconsidered. Here's how the authors of the study put it:
"This research also is unable to answer the question of whether stricter voter identification requirements succeed at preventing vote fraud. The results, instead, tell part of the story. It appears that stringent requirements can reduce turnout. But it remains to be seen whether the reduction in turnout is the price to pay for greater ballot security. That may, indeed, be the case. But it is also possible that strict voter identification requirements, designed to promote legitimate election results, could actually undermine that legitimacy instead."
----------------------------------------

Speaking of voting issues that have a disproportionate effect on particular groups, there's some very interesting data just out from New Mexico. In 2006 New Mexico scrapped all their touchscreen voting machines and switched to using only hand marked ballots with optical scanners, so the good folks at Voters Unite thought it would be interesting to retrospectively examine the change and see it had actually made any detectable difference in voting patterns.

The results were rather shocking. They found that DRE (Direct Record Electronic) touchscreen machines had a top-ticket (president of governor depending on the year) undervote rate that was typically three to four times higher than paper ballots. But strangely enough, that huge undervote discrepancy only occurred in precincts that were predominantly Hispanic or Native American. There was little difference in white areas. Check out this chart from the report:



Voters Unite didn't even speculate about why this occurred, but it's pretty clear from this and many other cases that weird things happen with touchscreen machines, and they need a heck of a lot closer scrutiny than they've received so far. Perhaps it's just a coincidence that Republicans love both tough voter ID laws and touchscreen voting machines, but given that in almost every case the resulting changes in voting patterns tend to directly benefit them, it kind of makes you wonder...

Read more about this on Brad Blog.

6 comments:

Anonymous said...

Doesn't this guy come to the opposite conclusion? http://www.madison.com/tct/opinion/column/index.php?ntid=120912&ntpid=0

Russell said...

"Doesn't this guy come to the opposite conclusion?"

He does, but he probably read only the NYT article the not report itself. As I mentioned, none of the MSM articles picked up on the fact that while voter ID in general significantly reduces minority turnout, the study concluded that isn't the case when the law requires a photo ID as opposed to other forms of identification.

One of the big problems with the MSM is that they rarely provide links to their original sources. I realize that there can be copyright and other legal issues involved, but I believe the big reasons are time pressure and simple laziness. It's far easier to just read the NYT and whip out a "me too" article than to spend the time doing the background research yourself.

I put five or six hours into researching and writing this post, and I'll bet that's ten times more than Zweifel put into his article. Besides, if their readers could look at the original sources they would realize how often the MSM is just plain wrong, and that wouldn't be good for business, now would it.

Anonymous said...

Never understood why you think you have the merit to comment on these subjects. You are very good and spewing inconsistencies that harm the merit of academics. BLAH

Russell said...

I'm harming the merit of academics? I don't understand what you're trying to say.

As far as my merit to comment, I actually read the report, which appears to be more than the authors of the MSM stories did, with the possible exception of the NYT. Seems to me that gives me at least as much right to express an opinion as they have.

--------------------

Proletariat, could you clarify your population growth comment? It doesn't make any sense to me. The researchers didn't just use raw vote totals, they created a statistical model starting with county-level voting data to correlate type of voter ID (and a lot of other stuff) with voter turnout. Population growth is not an issue given the way they structured the study.

Russell said...

Proletariat, the paragraph you quoted has basically nothing to do with population growth. The researchers are just justifying using a correction factor derived from 2000 census data to adjust their 2004 numbers. This is only valid if the correction factor hasn't changed much between 2000 and 2004, which is what the authors are arguing.

The correction is based on the ratio of citizens (who can vote) to non-citizens (who can't). That ratio remains fairly constant over the space of a few years unless there are large changes in immigration patterns during that time.

If you read the entire paragraph which you excerpted you'll see that the authors state that the US Census Bureau provides annual estimates of population changes, but that this doesn't include citizenship data, which is why they had to use the numbers from the 2000 census.

Russell said...

You're right that for some counties immigration could distort the results, but although the study started with county level data, it was used essentially to calculate a composite turnout for all areas using a particular voter ID method. So individual differences in counties average out.

Dave who?