The 200-plus-page report looks at several issues related to laws aimed at tightening rules around voting. The GAO compiled detailed data on various demographic groups in states that changed their laws, reviewed past studies on the effects of new laws on turnout, and attempted to gather data on instances of voter fraud, the rationale usually provided for changing voting rules. Democrats counter that the laws are thinly veiled efforts to reduce the number of their supporters that vote, by adding additional obstacles to black and young voters.
The GAO report suggests that, intentional or not, that's what happened in Kansas and Tennessee. This chart summarizes what it found.
Looking at three sets of data -- numbers from the United States Election Project, data from the states themselves, and Census data -- the GAO compared Kansas and Tennessee with Alabama, Arkansas, Delaware and Maine. The change in turnout in the latter four states is indicated with the black bars for each of the three datasets. The data for Kansas and Tennessee are lighter.
In short: The GAO found that turnout was at least 1.9 percentage points lower in 2012 in Kansas vs. 2008, and 2.2 percentage points lower in Tennessee thanks to their newly implemented voter ID laws. (The report, as you'd expect, has a full articulation of its methodology starting on p. 106, if you want to dive deep.)
According to data from the states (here and here), turnout dropped 5.5 percentage points overall in Kansas and 4.5 percent in Tennessee. With registered voter pools of about 1.77 million and 4 million, respectively, that means that 34,000 Kansans and 88,000 Tennesseans likely would have voted if the new laws weren't in place.
The effects of the change weren't evenly distributed. Broken down by demographic:
Young people, black people, and newly registered voters were the groups that were more likely to see bigger drops in turnout. Sixteen percent of voters in Kansas in 2012 were under the age of 30, according to exit polls. In 2008, the group comprised 19 percent of the vote. That change wasn't entirely due to voter ID, of course, but the GAO report suggests it played a part.
The report is very clear in stating that its findings apply only to the two states under consideration. The GAO invited those states to respond to the findings. Tennessee's secretary of state noted that other states may have had more compelling issues on the ballot and that the GAO used data from Catalist, which it labeled a progressive firm. Kansas's secretary of state echoed the first point, and noted the small population of African Americans in the state, which raises questions about the second graph, above. (Our colleagues at GovBeat explored the issue of race in more detail.)
As for instances of fraud, the GAO largely agreed with existing compilations of data: There exist a few scattered examples, but it was "difficult to determine a complete picture of such fraud."
And the GAO did not come up with the 122,000-voter figure. But if its analysis is correct, and the turnout drop was attributable to voter ID laws, that figure necessarily follows.