Reanalyzing 2020 Reapportionment

Sean Trende’s analysis in “The 2020 Reapportionment and The Voting Rights Act” is helpful, but I would like to supplement his analysis or present a slightly different take. His overall conclusion that it is becoming more difficult to maintain voting rights districts is accurate, although the problem is with the U.S. Supreme Court’s interpretation of the Voting Rights Act (VRA) in combination with the demographic changes he cites.

My reanalysis involves his descriptions of the three “Gingles prongs” or “preconditions” (Thornburg v. Gingles, 478 U.S. 30 (1986)). The first prong states that the minority group must be a majority in a single-member district. The exact legal definition of “majority” was unclear until the ruling in Bartlett v. Strickland, 556 U.S. 1 (2009). In that decision a plurality of the U.S. Supreme Court ruled that the minority group must be a majority of the citizens of voting age. Previously experts, such as myself, had interpreted “majority” to mean that the number of minority voters were sufficient for them to usually elect a representative of their choice, even if their choice was a member of their minority group. Using this definition, federal courts had approved many districts that did not have an absolute “majority of minority citizen of voting age” as appropriate to satisfy Gingles, including some districts in North Carolina (Gingles was a North Carolina case).

There are two problems with the way the Supreme Court has “clarified” this prong. Measures of citizenship are not reliable. The decennial census does not ask a citizenship question. The American Community Survey (ACS) does measure citizenship, but there are problems with applying ACS data to districting. First, the data is not reliable for small districts. Second it is only reliable when you combine five years of ACS data. Thus, even for congressional districts the data is old and therefore less reliable. Third, the ACS is a survey, not a census, so there is always a sampling error to deal with. And fourth, there are problems with the accuracy of the answers to citizenship items in such surveys.

More importantly, the 50%-plus-one standard is irrelevant, as this level of minority concentration is sometimes more or less than is required. In some cases, the rule has allowed Republican representatives to “pack” Democratic-leaning minority voters in majority-minority districts in order to the increase the partisan advantage of the GOP in neighboring districts, using the VRA as a justification. For example, in North Carolina, anything over about 42% black voting-age-population could be considered “packed.” However, in other places, especially for Latino districts, 50%-plus-one may not be enough for Hispanics to ensure the election of a representative of their choice. One thing is clear: The ability or opportunity to elect representatives of their choice is the purpose of the VRA. Before Bartlett, this first Gingles Prong made sense because it answered the question of whether a single-member district system or a different arrangement of districts would provide such an opportunity. Now it is just an impediment to providing minority voters with appropriate representation. The requirement that the district be “compact” is also an impediment as Trende points out, especially because there is no possible absolute mathematical standard of “compact” comparable to the one-man-one-vote standards developed by the courts since the 1960s.

Trende overstates the case for reduced cohesion among minority voters — another Gingles precondition. There is no sign that the extraordinary cohesion among black voters has decreased. Trende cites some disputable figures for the Latino vote for Republican congressional candidates in recent elections. We cannot assume that the remarkable cohesion of blacks (often 90% plus) is the “standard.” Most experts have testified that 55% or 60% cohesion among a minority group is sufficient to satisfy Gingles. The 65% cohesion that is typical among Latinos in general elections (as suggested by Trende) is certainly more than sufficient and has not shown signs of decreasing. The logic of this prong should also be clear. The minority group must be cohesive to demonstrate that they are trying to achieve representation and that there are some common points of view of the minority voters that can be represented. Such common points of view are apparent in Latino voting habits, although not at the level of black voters.

Another Gingles prong is that the voting is racially polarized so that the white voters can prevent the minority-approved candidate from winning in spite of minority cohesion. Here Trende also overstates his case. There may be some reduction in the level of polarization (i.e., white bloc voting) in some places in the North (but not the South), but this is nowhere near enough to say that the voting is not polarized. Much of the increase in the ability of minority voters to elect representatives of their choice in districts that are minority concentrated but majority white is in the enormous increase in minority voting rates in such districts. Districts still must have large concentrations of minority voters, not necessarily a majority, for them to elect their choice. This is true even in the North and West. There are always exceptions of course; that is the nature of politics. But the exceptions are few. In my view, the question is not whether a majority-minority district is necessary, but whether a large concentration of minority voters is necessary. Some concentration is necessary, and this relates directly back to the first Gingles prong, helping us to understand the harm of the Bartlett decision.

Lastly, the possible election of an African-American Republican in Utah (as mentioned by Trende) is not relevant, as she is unlikely to be the candidate of choice of the few black voters there. The VRA protects minority voter’s choices; it is not intended to protect minority candidates who are not the choice of minority voters. Very few black Republicans have been elected to office in modern times, and almost none to statewide office. Former Sen. Edward Brooke (R-MA) is the lone example since Reconstruction; he served from 1967 to 1979. That list is likely to expand with appointed Sen. Tim Scott’s (R-SC) likely election this November, but the odds are that he will still not be the choice of most black voters. Two black Republican statewide nominees in 2006 illustrate this. In the Maryland contest for U.S. Senate, then-Lt. Gov. Michael Steele (R) lost the black vote 74% to 25% to now-Sen. Ben Cardin (D), based on the exit poll; Cardin won 54% to 44% overall. Meanwhile, Ohio Secretary of State Ken Blackwell (R) lost badly in the Buckeye State gubernatorial race against then-Rep. Ted Strickland (D), 61% to 37%. According to the exit poll, Blackwell only garnered 20% of the black vote. While both lost in what was a very Democratic year, neither got much of the black vote. Even in a more optimal environment for a black GOP nominee (such as 2010), it’s hard to imagine that individual holding a Democrat under 60% of the black vote (roughly the deciding line of “cohesion”), even in an uncompetitive race. The situation with Latinos is not as clear, and occasionally a Latino Republican may receive a bare majority of the Latino vote over a white Democrat. But again, these cases are rare.