‘Roughly’ Equal Population
You may wonder where the equal population thresholds in DRA 2020 come from. The short answer is from case law. The details are below.
Population Deviation
A set of districts is said to have roughly equal populations, when the population deviation is sufficiently small, where the formula is:
Deviation = [(Maximum — Minimum) / Average] * 100
where the target district size is the average (i.e., they’re the same thing).
Congressional Districts
The screenshot below is part of P. 11 from Benisek v. Lamone complaint (Maryland). There the plaintiffs argued that courts have said that 0.70% deviation can be found to be too small a population deviation for congressional districts but 0.79% can be found to be acceptable, both depending on whether the state has some legitimate interest in the deviation from more equal populations (such as keeping counties intact).
The 0.75% population deviation threshold that we use for congressional districts in DRA 2020 splits that difference and is really a caution that there’s little precedent for congressional plans with bigger deviations being allowed. In other words, a deviation less than 0.75% is equal enough, but bigger than that is not.
Some would argue that no deviation is too small, and in practice many states try to get population deviations effectively zero — e.g., one person deviations — probably to avoid judicial scrutiny of gerrymandered plans. But courts have shown some flexibility when states have legitimate interests in the deviations.
On the other end of the spectrum, the average size of a census blockgroup is ~0.20% of a state’s population, and that seems like a pretty reasonable “good enough” target. In other words, try to get get your congressional maps somewhere in the range 0.20–0.75%.
State Legislative Districts
The equal population requirements for state legislative districts are more relaxed and less concrete. It seems that 10% is generally accepted as the acceptable threshold for state legislative redistricting, again when states have some legitimate countervailing interest.