The Census Bureau announced last week that their estimates of each state’s population had errors, resulting in six states with undercounts and eight states with overcounts. This may shock, or at the very least, disturb some people, who believe that the process of tabulating the nation’s population should be precise and accurate.
The challenge in conducting a census count is identifying and reaching every person. Moreover, population counts are dynamic, with both daily births and death, as well as people moving, even if a fixed date of record is specified. As such, the best one can hope for is a population estimate, and with all estimates, there is uncertainty.
This has been observed with polling data used to forecast the outcome of elections. When a subset of voters are polled, if this subset is not representative of the population of voters, the forecast could be wrong. This was experienced during the 2016 presidential election.
Census data is used for a variety of purposes, including allocation of federal funds to states and communities, city planning and emergency response. Its most visible role is determining how many congressional seats are allocated to each state, which impacts how each state’s congressional map gets drawn.
The apportionment formula is a function of a state’s population. Every state’s potential seat is assigned a priority value, and all these values are ranked from highest to lowest. Since every state is allocated at least one seat in the House of Representatives, the largest remaining 385 priority values (seats 51 through 435) result in where the seats beyond one-per-state are assigned.
Based on this formula and apportionment process, each states population is critical for gaining or losing congressional seats relative to what it was allocated after the 2010 census. This is where an undercount or an overcount can make the difference between gaining or losing congressional seats.
The states most affected are those near the seat allocation cut line. For example, New York was apportioned one fewer House seat, but was next in line to have not lost it (ranked 436 in priority values). Given that its population was overcounted by 3.44 percent, this overcount had no influence on its final seat allocation.
Ohio is another state in a similar situation, with an overcounted population that was apportioned one fewer House seat (ranked 437 in priority values).
The two states with significant population undercounts that were close to the cut line for adding another seat were Florida and Texas. Florida’s population was undercounted by 3.48 percent. This undercount cost Florida one additional House seat (ranked 439 in priority values). Texas’s population was undercounted by 1.92 percent. This undercount cost them one additional House seat (ranked 438 in priority values).
So, where would these two seats have come from? Minnesota was the last state to be apportioned a seat (ranked 435 in priority values). Since its population was overcounted by 3.84 percent, it would have lost one of its House seats, which was expected by political pundits. The same is true of Rhode Island, whose population was overcounted by 5.05 percent (ranked 428 in priority values).
Given the nation’s contentious political divisions, with partisan gerrymandering working to capture every possible congressional seat, one extra or fewer representative elected can mean the difference between winning or losing control of the House in the 2022 midterm elections. With both Florida and Texas Republican gerrymandered states, those two extra seats would have likely added to their House seats, increasing the likelihood of Republicans regaining control in the upcoming midterms.
Note that of the six states that have undercounted populations, five are Republican leaning (Illinois being the exception). Among the eight overcounted population states, six are Democratic leaning (Ohio and Utah being the exceptions). Such observations are not indicative of any nefarious intent or actions on the part of the Census Bureau, but rather, demonstrate that compiling population counts have an element of uncertainty, although in the upcoming close elections, could have significant legislative consequences.
Nothing can be done to correct the consequences of population undercounts or overcounts on apportionment. Republicans appears to have drawn the short end of the apportionment stick this time, although their history of gerrymandering dominance could more than compensate for it in the November midterms. Whether that plays out at the ballot box remains to be seen.
Sheldon H. Jacobson, Ph.D., is a professor of computer science at the University of Illinois at Urbana-Champaign. A data scientist, his research group on computational redistricting is committed to bringing transparency to the redistricting process using optimization algorithms and artificial intelligence.