PolitiFact: Roberts too selective in choosing voting data

This article was edited for length. To see a complete version and its sources, go to www.politifact.com/truth-o-meter/statements/2013/mar/05/john-roberts/was-chief-justice-john-roberts-right-about-voting-/.

“Do you know which state has the worst ratio of white voter turnout to African-American voter turnout? Massachusetts. Do you know what has the best, where African-American turnout actually exceeds white turnout? Mississippi.”

Chief Justice John Roberts during oral arguments Feb. 27 at the Supreme Court


Chief Justice John Roberts quizzed — and stumped — U.S. Solicitor General Donald Verrilli Jr. on some deep-in-the-weeds voting statistics during oral argument over a key provision of the Voting Rights Act.

Shelby County, Ala., has challenged how the landmark 1965 law — which ensured racial minorities the right to vote — determines whether a state or locality needs advance approval from the U.S. Justice Department before it changes any voting laws, regulations or procedures.

The law requires advance approval for Alabama, as well as five other Southern states (Georgia, Louisiana, Mississippi, South Carolina and Virginia) that demonstrated a poor historical record of allowing minorities to vote. Alaska, Arizona and Texas, plus a variety of local jurisdictions, are also subject to “preclearance.” Shelby County says these requirements are outdated.

At one point in the oral argument held Feb. 27, Roberts asked Verrilli, “Do you know which state has the worst ratio of white voter turnout to African-American voter turnout?”

Verrilli, who was defending the law as currently written, responded, “I do not.”

Roberts said: “Massachusetts. Do you know what has the best, where African-American turnout actually exceeds white turnout? Mississippi.”

Roberts was pointing out a striking irony — that Massachusetts, a state well-known for its liberalism, was seemingly a poor performer in protecting minority voting rights, yet Mississippi, a state for which the Voting Rights Act demands intensive federal scrutiny, seems to be a model of voting equality.

We wondered, does the chief justice have his numbers right?

We should note that the Supreme Court has declined to elaborate on Roberts’ data selection, but there is wide agreement among voting experts that he was using data from the Current Population Survey, a widely used U.S. Census Bureau survey.

Since Roberts used the present tense, we think a reasonable listener would assume that the chief justice is referring to the most recent data available. So we will focus on data for 2008 (the last presidential election year available) and for 2010 (the last midterm election year available).

We’ll begin by looking at some of the problems with using Current Population Survey data to back up Roberts’ claim. After that, we’ll grant for the sake of argument that the Current Population Survey is valid for this purpose and look instead at why Roberts’ interpretation of the data is less than ideal. Finally, we’ll discuss why, despite these problems, Roberts has a reasonable point to make.

A chief problem with using the Current Population Survey is that in collecting data on Massachusetts, it is using a relatively small group of respondents — blacks make up only 6 percent of the state’s population — so there’s a large margin of error. That increases the chance that a few bits of off-base data will skew the whole sample.

The survey found that in Massachusetts, 52 percent of all adult citizens voted in 2010, with a margin of error of about 2 percentage points. But the same survey found that 39 percent of black voters in Massachusetts voted in the same year, with a margin of error of 11 points. This means the black voting rate actually may have been as high as 50 percent — pretty close to the percentage for the state as a whole.

The Current Population Survey data for voting registration among Massachusetts residents also fluctuated wildly between 2008 and 2010. The registration gap between whites and blacks ballooned from 13.8 percentage points in 2008 to 36.5 percent in 2010, an almost threefold increase in just two years. Such a large change casts doubt on the reliability of the Massachusetts sample.

Even if you assume the Current Population Survey is a good source of data, Roberts has cherry-picked by citing Massachusetts

We analyzed the registration and voting numbers for 2008 and 2010 and found that Massachusetts was well out of the national mainstream in both years. This becomes clear when you compare Massachusetts to other large, relatively diverse Northeastern states. We looked at four other states for comparison — Connecticut, New Jersey, New York and Pennsylvania.

In 2010, when the white-black registration gap in Massachusetts was a whopping 36.5 percentage points, the average for the other four states was a much more modest 9.8 percentage points. Two years earlier, Massachusetts voting rates showed a 25.8 percentage point white-black gap, compared with just 6.3 percentage points for the other states.

If Roberts wanted to make his point by looking at the voting gaps in large, Northeastern, “liberal” states, he could have chosen four that had relatively modest racial gaps. Instead, he chose the one outlier, Massachusetts, and that amounts to cherry-picking.

Where Roberts has a point

Roberts is on more solid ground with his discussion of Mississippi. Unlike the comparison of Massachusetts with the other Northeastern states, Mississippi’s data in the Current Population Survey fit in closely with that of its neighbors in the Deep South, which we’ll count as Georgia, Alabama, Louisiana, Mississippi and South Carolina. And because the black population in these states is proportionally much larger than it is in Massachusetts, the margins of error for these states in the Current Population Survey are much less problematic.

In 2010, three of these states had black registration rates higher than for whites, and four had black voting rates higher than for whites. The remaining Deep South states were roughly at parity between the races. And this pattern was much the same in 2008.

The high rates of participation suggest that Roberts has a point: Voting rates in Mississippi, and the Deep South generally, have improved over the years.

Our ruling

Roberts said, “Do you know which state has the worst ratio of white voter turnout to African-American voter turnout? Massachusetts. Do you know what has the best, where African-American turnout actually exceeds white turnout? Mississippi.”

For several reasons — including unacceptably wide margins of error and strange year-to-year inconsistencies — the survey data Roberts likely used to back up his point on Massachusetts is questionable. And even if this data source had been reliable for his purposes, Roberts’ decision to hold up Massachusetts as his primary example amounts to cherry-picking, since its data diverge significantly from that of its regional neighbors.

Roberts makes a valid point, however, when he notes that black voter turnout in Mississippi and other Deep South states is high, and indeed often exceeds white turnout. So while Roberts is on shaky ground with Massachusetts, he’s on more solid turf when it comes to Mississippi. On balance, we’ll call it Half True.

About the Author