[EL] new ID data

Tue Jul 23 12:55:27 PDT 2013

I agree with Justin.  The typical analysis of the effect of voter ID laws assumes we are only interested in finding people who (a) lack a valid ID under the new law, AND (b) voted last time under the old law.  That is the group the North Carolina Secretary of State's office has essentially tried to measure. I  appreciate their efforts.  But when you think about it, this is a relatively narrow slice of the phenomenon we ought to be interested in -- even if we were only concerned with ID laws' effects on turnout.

To see all the sides of this problem, just relax (a), (b) or both.  As Justin points out, a problem with (b) -- only looking at those who turned out last time -- is that every cycle, some number of people will attempt to vote who didn't last time.  Indeed, every cycle, some number of eligible but unregistered people will try to register and vote -- or would have, absent new restrictions.  Meanwhile, (a) doesn't capture all the effects of the laws either.  Voter ID laws can deter even some people with valid IDs from voting.  I give the Kyle Dropp paper some credit for considering that effect.  How would this happen?  Well, as that paper briefly notes, even some people with IDs that are actually valid under a new law may be turned away -- or deterred from attempting to vote at all -- by erroneous poll worker demands (e.g. saying we don't accept anything other than a driver's license when the law allows other forms of ID, requiring an address match when the law does not, etc.) or confusing or misleading messages from either signs at polling places, poll workers, campaigns, or even news stories.  

The nice thing about difference-in-differences (as in the Dropp paper) is that it measures what difference the laws actually make.  In theory it gets us beyond (a) and (b), by just measuring the overall effects on turnout of enacting an ID law, versus not enacting an ID law.  But even here, a note of caution is in order.  In a world where poll workers so often depart from the actual requirements of state law, and in which many voters form erroneous beliefs about what is required, the enactment of a large wave of voter ID laws may have had interstate spillover effects.  That is, even in states that did not enact such laws, I would be surprised if there were not some tightening up of the informal demands of poll workers, and the perceived requirements on the part of at least some voters.  Any such spillover effects would obviously cause a difference-in-differences methodology to underestimate the size of the effects of the new laws.

At the end of the day, we need to get beyond the focus on Voter ID laws' effects on election outcomes -- and even on turnout.  I understand that most politicians focus mostly on outcome effects.  Their political lives depend on it.  Turnout effects are a little broader than outcome effects; hopefully most people who are not politicians are capable of understanding that even if the outcomes were totally unaffected, participation matters.  But beyond either of those effects, it seems to me that we should care about whether some people are kept out of the political process and told they are not full citizens like everyone else -- especially if they're the same people, such as people with disabilities, poor people, etc., who often get that same message from many other parts of our economic and political system.

Joey

Joseph Fishkin
Assistant Professor
University of Texas School of Law
727 E. Dean Keeton St., Austin, TX 78705
jfishkin at law.utexas.edu

On Jul 23, 2013, at 1:37 PM, Justin Levitt wrote:

> With respect, the study that Jonathan linked to may well be the most thorough empirical analysis of the impact of voter ID laws on turnout thus far (and I look forward to reading it more thoroughly, and seeing whether it solves some of the problems with modeling turnout effects of single election laws that Erickson and Minnite have identified).  
> 
> But as I've written, laws increasing barriers to entry have an impact beyond the marginal person who voted last time.  A law permanently prohibiting any person who didn't vote in 2008 from voting in the future would have shown only a modest effect on aggregate turnout, at least in the short run -- the vast majority of 2012 voters were       also 2008 voters.  But I would not say that such a law has only a minimal impact.
> 
> If you're only looking at turnout, that's better than just looking at the outcome of a Presidential race ... but it's still only part of the impact.  
> 
> Justin
> 
> 
> On 7/23/2013 10:33 AM, Jonathan Rodden Stanford wrote:
>> I haven’t seen it posted on this list yet, so here is a link to what is by far the most thorough empirical analysis of the impact of voter id laws:
>> 
>> http://kyledropp.weebly.com/uploads/1/2/0/9/12094568/dropp_voter_id.pdf
>> 
>> Best,
>> Jonathan
>> 
>> 
>> 
>> On 7/23/13 10:10 AM, "Rick Hasen" <rhasen at law.uci.edu> wrote:
>> 
>>   I'd add that a .3 swing is a pretty significant risk of swinging a swing state even if one were concerned only about presidential elections.  
>>  
>> On 7/23/13 10:06 AM, Justin Levitt wrote:
>>  
>>  
>>  
>> What's also missing in this analysis is concern about anything other than the final outcome of a Presidential race.  
>>  
>>  Yes, the  piece <http://www.newrepublic.com/article/113986/voter-id-north-carolina-law-hurts-democrats>  finds that "the electoral consequences of voter ID seem relatively marginal," by noting that with ID, Obama's final share of the North Carolina vote might have dropped from 48.3 to 48%.
>>  
>>  But the piece also notes that this latest data reveals that there are somewhere around 319,000 registered voters currently without a state-issued photo ID, "just" (just!) 138,425 of whom participated in the 2012 general election.  There is no estimate of the number of currently unregistered but eligible voters who don't now have a state-issued photo ID, but it's got to add to the pile.  
>>  
>>  For those who think the most important measure of the impact of an electoral policy is the outcome of a Presidential race, why have a national election at all?  Polling science is pretty good: we could just declare the winner of every state where the margin of victory is larger than the margin of error in several consecutive polls in the last week of October, and only bother with actually letting people vote in the very few states where polls don't deliver a clear answer.  Holding an election seems like a really expensive way to confirm the pretty-much-guaranteed winner.  Or, put differently, if you're just focused on Presidential outcome, "the electoral consequences of holding an election seem relatively marginal."
>>  
>>  Justin 
>>  
> 
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> http://department-lists.uci.edu/mailman/listinfo/law-election

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20130723/ad35f32b/attachment.html>