Michael Pitts has posted this draft on SSRN (University of Richmond Law Review).  Here is the abstract:

This article represents the continuation of a series of studies that measure the impact of photo identification on the electorate by examining provisional ballots cast and not counted because prospective voters lacked photo identification. Prior studies examined provisional balloting at Indiana’s 2008 primary and 2008 general elections. This article presents results from the 2012 primary and proceeds in two parts. Part I briefly discusses the details of Indiana’s photo identification law, the various methods used to measure the impact of photo identification laws generally, and the research methods employed for this study. Part II presents and analyzes the empirical data generated in this study and, notably, compares the impact of the photo identification law at Indiana’s 2008 primary election with the impact of the photo identification law at Indiana’s 2012 primary election.

As with Pitts’ earlier study of Indiana’s implementation of its voter id law, this is a must read.  He finds much less disenfranchisement likely caused by Indiana’s i.d. law than opponents of the law usually claim.