A Brief Comment on Evenwel v. Abbott

The plaintiffs in Evenwel v. Abbott demanded that apportionment in Texas be done on the basis of actual voters or those eligible to vote rather than on the basis of total population. The district court opinion rejected this view. It held that, while there might be more than one permissible basis of apportionment, equal population was clearly one such, and that the choice of Texas to use that basis of apportionment was certainly permissible. It was pretty much inconceivable that a majority of the Court would reverse the district court’s holding, since the Court has repeatedly upheld population based apportionment, and had they reversed the trial court the apportionment method in use in jurisdictions at all levels of government would have been overturned. Moreover, as Nate Persily made clear in his Amicus brief in this case (one which I participated in as a junior author, along with several other social science scholars), the idea of using citizenship has major measurement problems. Similarly, the idea of using voter registration as the basis of apportionment opens up the potential need for frequent reapportionment. And, from a democratic theory standpoint, using a different apportionment basis for federal than for state or local elections seems like a bad idea. Nonetheless, it was conceivable, in my view, that some justices could have sided with the plaintiffs. In fact none did. The decision was unanimous, with five justices writing in the majority in an opinion that is quite dismissive of the arguments raised by plaintiffs, and with three justices making use of concurrences.

However, the unanimity in outcome in Evenwel does not mean that we have seen the last of the issue of “apportionment basis.” The two concurring opinions, representing the views of three Justices, emphasize that there is no magic basis of apportionment that all states must use, and expressed grave reservations about the logical coherence and constitutional grounding of the Supreme Court’s “one person, one vote” jurisprudence, and that in related areas of voting rights. In particular, the concurring opinions made it clear that, in the view of these Justices, the basis of apportionment for state legislatures was essentially a matter for decisions made by each state.

Of course, Evenwel, like many voting rights cases, is actually about partisanship, not about voting equality. Down the road, if Republicans are more concerned about their own power base than about antagonizing Hispanic voters, we may expect that a state under Republican control that is faced with a growing Hispanic population, may choose to adopt a basis of apportionment that would help them reduce the number of districts that might elect Democrats. They can do so by opting for an apportionment basis that potentially minimizes the impact of the Hispanic population by eliminating from the count those not of voting age, and/or those not citizens, and/or those not registered.

However, the consequences of changing the apportionment base may not be as large as one might first think, because changing the apportionment base changes the denominator and not just the numerator, and it is always only voters who vote, regardless of what basis of apportionment is chosen. We can see why the effects of a change in apportionment base might not matter that much with a simple example. Imagine nine districts and an Hispanic population that is 33.33 percent of the total population. If the nine districts are each 33.33% Hispanic and 33.33% of the Hispanics are voters, 90% of whom vote Democratic; and the remaining non-Hispanic population have 66.66% of the population and 66.6% of that population are voters, and these voters vote Republican by 59% to 41%, then dividing this population identically into nine districts we get 9 districts that are won by Democrats by very thin margins (50.8% to 49.2%). Now, imagine that have to apportion on the basis of voters. In that case, Hispanics are 20 percent of the apportionment base, not 33 percent, and so we might think that Democrats would be severely disadvantaged by the change in apportionment basis. But that depends on exactly how districts get drawn.

If we draw the same nine districts, nothing has actually changed as a result of the change in apportionment basis, since each is still perfectly apportioned on the basis of voters, even though the apportionment basis is only 5/9ths of what it had been. Each district has 20 percent of its apportionment basis (actual voters) who are Hispanic and 80 percent who are non-Hispanic, but the Democratic vote share in each district is still 50.8% = (.20*.9 + .8*.41).

Moving from the above toy example to the real world, we observe that, in each of the above categories (voting age population, citizens, registered voters), the expected proportion of eligibles among Hispanics is lower than the corresponding proportion among non-Hispanics. So, once we allow for the possibility of geographic concentrations of voters of a given partisan predisposition, and for the possibility of deliberate partisan gerrymandering, because the effect on the numerator is greater for Hispanics than for non-Hispanics, there might be negative consequences for Democrats of a change in apportionment basis—at least if Republicans are drawing the lines. In particular, if we change the apportionment basis, it is no longer as easy to draw a district that will elect a Democrat by drawing a district with a large number of Hispanic non-voters in it, but still enough Hispanic and non-Hispanic voters to outnumber the Republican voters. These non-voters are no longer available as “filler.” The recognition of that fact is exactly why it is Republicans who have pressed for a change in apportionment base. It is also why the issue of apportionment basis is not yet dead.