The Way the Census Counts Prison Populations Seriously Distorts Redistricting

Sing Sing prison as seen from Rockland Lake State Park.
Sing Sing prison. Wikimedia Commons

The Trump administration’s failed efforts to add a citizenship question to the 2020 census have rightfully dominated Supreme Court coverage in the days since the end of the 2019 term. The question was blocked from appearing on the census after the court ruled that the reason for it had been “contrived,” with ballooning evidence that the actual reason was to redistrict maps that would present “a disadvantage to the Democrats” and be “advantageous to Republicans and non-Hispanic Whites.” The defeat of the census question was assuredly a voting rights victory, even as a separate ruling by the Supreme Court that deciding partisan gerrymandering cases was beyond the justices’ reach poses future dangers to majority rule.

Less noticed, though, than these two cases was a galvanic blow delivered to democracy by the Census Bureau just last year—when the bureau declined to change long-standing policy and count incarcerated persons where they previously lived, rather than in the place of their imprisonment. Because prisons tend to be located in white, rural districts, counting prison populations by prison location artificially inflates these districts’ political power and diminishes political representation for urban and suburban areas.

The Census Bureau’s decision to preserve its stale policy on counting people in prison did not invite the saber-rattling condemnations that met Donald Trump’s failed census maneuver, but it deserves far greater criticism. The bureau invited public comment on its 2020 census residence criteria in 2016 as part of a decennial review to assess compliance with the Census Act of 1970. According to the bureau, of the hefty 77,887 public comments it received addressing the prison population issue, nearly all the comments urged the bureau to count people in prisons at their home or pre-incarceration address. The bureau demurred, however, citing the distinction between legislative redistricting and the broader purposes of the census count.

But the bureau failed to understand the nature of its role. The current census policy leads state administrators to shuttle incarcerated people across states and deposit them in districts where atrophying population counts are bloated by the presence of prisons. One study analyzing state senate districts found that prison populations were often shifted in tandem with the life cycle of legislatures or governorships, conferring partisan advantage after switches in party control over the redistricting process. The bulk of the prison population—which numbers in the millionsis planted in small towns and rural areas, though many individuals hail from cities and surrounding suburbs. The result is that census data—both in terms of redistricting and in terms of an accurate population count—is soiled by the bureau’s own rules on counting people in prison.

The Prison Policy Initiative has observed that a substantial percentage of the prison population is in jail awaiting trial or serving short sentences, according to county data. Most people held in state prison return home after their release, which means that their actual place of residence is not being accurately reflected in the census. Prison-based gerrymandering additionally midwifes perverse incentive structures, nudging state legislators to keep on the books punitive sentencing laws, such as inflexible mandatory minimums, that entrench stark racial disparities in the criminal justice system.

Some states have worked to correct this blight on redistricting. New York, Maryland, California, Delaware, and Washington have passed laws to end prison-based gerrymandering. In 2016, a federal judge held that commission districts in Jefferson County, Florida, violated the touchstone constitutional principle of “one person, one vote” in employing prison-based gerrymandering, though the court’s reasoning was seemingly narrowly drawn to apply to the facts at hand. A similar case brought by the NAACP in Connecticut was green-lighted to go forward in federal court earlier this year. The legal arguments in these cases complement universal redistricting practices based on total population counts, which include noncitizens and children, a policy blessed by the Supreme Court in Evenwel v. Abbott.

State representatives, meanwhile, have limited ability to direct resources to people incarcerated in their districts (unlike noncitizens and children), defeating another purpose of the census count. And in terms of voting patterns, people in prison must vote by absentee ballot from their pre-incarceration address if they are allowed to vote at all (which they most often are not). Still, they are typically jettisoned from district to district, secreted away in new prison facilities at the whim of state administrators.

With the Roberts court ill-disposed to opine on political redistricting that confers partisan advantage, a state-by-state piecemeal approach to challenging the distorting effects of prison-based gerrymandering pales against what would have been a simple change in federal census residency rules. The Census Bureau’s decision to reject that change reflected legalistic swaddling and poorly conceived policy. The next administration must instruct the bureau to kick-start a rulemaking process that would produce fairer legislative districts. In the meantime, states should not wait another decade to issue a legislative fix.