Which specific House seats would have changed hands in the 2020 apportionment if only citizens were counted?
Executive summary
A definitive, government-sanctioned apportionment based on "citizens only" does not exist for 2020; however, independent analyses that re-run the Huntington–Hill/math-of-equal-proportions apportionment using citizen-population estimates conclude that several seats would have shifted compared with the official resident-population apportionment (which gave six states gains and seven states losses) [1] [2]. The most detailed citizen-based estimate assembled by the American Redistricting Project (ARP) finds that California would have lost four seats, New York two, and New Jersey and Illinois one each — a net pattern of seat movement away from large immigrant-rich states — but other credible analyses produce much smaller changes, and all rest on estimates and sampling data rather than an alternate census enumeration [3] [4] [5].
1. How apportionment actually worked in 2020 and the baseline results
The law requires apportionment be calculated from the Census Bureau’s resident population counts and uses the method of equal proportions (Huntington–Hill) to assign the 435 seats; the official 2020 apportionment — using resident population — resulted in six states gaining seats (Colorado, Florida, Montana, North Carolina, Oregon, Texas) and seven states losing seats (California, Illinois, Michigan, New York, Ohio, Pennsylvania, West Virginia) [6] [1] [2]. Those official numbers are the starting point against which any hypothetical “citizen-only” reallocation is compared [1] [2].
2. The ARP “citizen population” re-run: which seats would flip in that analysis
The American Redistricting Project re-estimated 2020 apportionment using American Community Survey–based citizen population estimates and projects that, under a citizen-only apportionment, California would lose four seats, New York two seats, and New Jersey and Illinois would each lose one seat [3]. ARP’s headline—large losses concentrated in a handful of states with relatively large noncitizen populations—flows from substituting ACS-derived citizen totals for the Census Bureau’s resident totals and then applying the statutory apportionment formula [3].
3. Smaller-seat estimates from other analysts and the “unauthorized immigrant” angle
Other reputable analyses reach more modest conclusions: Pew’s modeling focused on removing unauthorized immigrants (not all noncitizens) and concluded that excluding unauthorized immigrants from the apportionment count could produce changes in three states losing seats and three different states each gaining a seat — a far smaller net reshuffle than ARP’s citizen-only scenario [4]. Conversely, the Center for Immigration Studies (CIS), using its own projections, argued that immigration (including lawful and unauthorized immigrants plus U.S.-born children) would be responsible for a shift of 26 seats by 2020 — an outlier estimate that reflects different assumptions about which populations to exclude and how to model them [5].
4. Why these estimates diverge: data, definitions, and methods
All estimates that replace resident counts with “citizens only” must rely on the Census Bureau’s American Community Survey (ACS) or other samples to estimate citizenship because the decennial short-form census does not enumerate citizenship, and CRS has warned that the ACS is the only available source for citizenship status in the absence of a specially designed census question [7]. Differences in which noncitizen groups are excluded (unauthorized immigrants only versus all noncitizens), how ACS sampling error and undercounts are adjusted, and how population projections are weighted produce divergent seat-change tallies [7] [4] [3].
5. Legal and practical constraints on creating a “citizen-only” apportionment
Federal statute and long-standing Census practice use resident population for apportionment, and implementing a citizenship-based apportionment would require either a constitutional amendment or a legally contentious change in executive or bureaucratic policy — a reality that analysts note when describing hypothetical alternative tallies [7] [6]. Moreover, official apportionment must be delivered on statutory timelines using the bureau’s counts, so no alternative official seat list based on citizens was produced for 2020 [6] [1].
6. Bottom line — which specific seats would have changed hands?
There is no single, authoritative “citizen-only” 2020 apportionment from the federal government; using published independent reconstructions, the clearest and most-cited citizen-based result (ARP) identifies specific state-seat losses: California −4, New York −2, New Jersey −1, Illinois −1 [3]. Other credible studies give much smaller effects (Pew: three states could lose seats and three could gain if unauthorized immigrants were excluded) or larger, model-driven effects (CIS: up to 26-seat shift depending on definitions) — the precise list of district-level seats that “change hands” therefore depends on the analyst’s definitions and data choices and cannot be pinned down as an official alternative to the resident-count apportionment [4] [5] [7].