As the debate over requiring photo IDs at the polls continues, so does the conversation over how to use data to measure the impact of such policies. In fact, sometimes people using the same data can reach very different conclusions.
For instance, Hans von Spakovsky, a senior legal fellow at the Heritage Foundation and a proponent of voter ID, recently wrote about increases in the total number of Hispanic voters in Georgia. He observed that in 2008, some 43,000 Hispanic voters cast ballots in the presidential election, up from approximately 18,000 in 2004. Similarly, in the 2010 midterm elections, 19,000 Hispanics voted compared with 11,600 in 2006. These increases occurred after the state started requiring photo ID at the polls in 2007, and von Spakovsky cites them as evidence that voter ID had no disproportionate impact on minority voters.
By contrast, Keesha Gaksins with the Brennan Center for Justice and opponent of photo voter ID laws used the same data from Georgia for Hispanics and reached a different conclusion – that participation among Hispanic voters in 2008 and 2010 actually decreased relative to 2004 and 2006, respectively.
While von Spakovsky cites the increase in raw numbers, Gaskins examines turnout — the number of ballots cast by Hispanic voters as a percentage of all registered Hispanic voters in the state. In 2004 the 18,000 Hispanic voters were 60.5 percent of registered Hispanic voters, while the 43,000 in 2008 were 59.6 percent of registered Hispanic voters. Hispanic turnout in 2006 was 26.7 percent and then declined to 25.5 percent in 2010.
Of course many factors influence voter participation beyond election administration policies – who and what is on the ballot, enthusiasm for a particular candidate, the closeness of the race, etc. Similarly, whether a given election includes a presidential race is typically a significant driver of voter participation as illustrated by the consistently lower figures for both raw participation and turnout among Hispanic voters during the midterms versus presidential years.
In the end, however, examining raw numbers or even simple turnout numbers from just one state over only two election cycles provides limited insight about the impact of any policy.