“Look, you had a lot going on. You had people who were nervous to come forward, right? You have undocumented people who were nervous to come forward. I do believe the federal government had a chilling effect. It was down to like 89 – is the differential. Do I think it was accurate to within 89? No. And we’re looking a legal options because when you’re talking about 89, I mean that could be a minor mistake in counting, right?”
Last week, the US Census Bureau released preliminary population and apportionment data for each state (the number of seats each state will now get in the House of Representatives). While many people were surprised that Texas and Florida did not each pick up one more seat, what’s even more surprising is that New York almost didn’t lose any seats. In the closest loss of a seat in Census history, New York lost the final House seat to Minnesota by 89 residents. To make sense out of this we need to understand the apportionment formula.
The challenge in political apportionment is one of fractions. In apportioning representatives, we will run into the problem of remainders – that is, how to decide what to do with a state that should get, say, 10.5 representatives. Since we are talking about non-divisible human beings, rather than money or land, we have to deal in whole numbers. However, any formula to decide representatives will have remainders unless the total population can be neatly divided by the ratio of voters to representatives. However, there are two Constitutional rules that ensure the US will always have to solve the problem of fractions: Constitutional rule (1) limits constituencies to having no more than one representative for every 30,000 people and Constitutional rule (2) requires that every state, notwithstanding rule (1), get at least one representative. From the very first apportionment in 1791, these two rules made deciding the number of representatives for each state much more complicated than one might assume (and more complicated than the framers assumed when drafting the Constitution).*
The US has 331 million people according to the 2020 census. If not for the rule that every state gets one representative, the math would be easier (although the districting problem would be more challenging). Because of the rule requiring no more than one person representing 30,000 the maximum possible size of the House is 11,033. Well, not exactly since the actual calculation is 11,033.34. If we eliminate the remainder and take 11,033 as the maximum size and divide the total population by it, we get another fractional problem. In this case, the size of a constituency would be 30,000.91. (Both calculations so far have had rounding issues within the remainder itself – so, as you can tell, this problem pops up everywhere we need to do the math.) If we multiply the remainder by 11,033, we find 10,040 persons unrepresented (it’s actually 10,040.03). We cannot give these 10,000 folks their own representative because the Constitution requires a representative to represent at least 30,000 people. (It would also run afoul of the “one man, one vote” principle in Reynolds v Sims and Westberry v Sanders). If we eliminate the fraction and redistribute the remainder equally to each of the 11,033 constituencies, each district would add 1 more person. So that would seem to make sense given 0.91 rounds up to 1.0, but the actual number of persons that should be added is 1.098. At this point, you probably understand the problem.
But wait, it’s worse! There are 50 states that must get at least one representative.** Now we have a real problem. Without this rule, the problem of fractions can be solved fairly easily by making sure that there is a rough equality among constituencies even though there will not be absolute equality. Once we factor in this rule we not only have to figure out how many representatives each state gets, but the problem of fractions becomes difficult to resolve. Even if we allowed for the maximum size of the House at 11,033 – because the interpretation of the Constitution is that districts end at state lines – we would still have remainders.
Let’s say State X has 500,000 people and we’ve set our constituency size as 30,001. State X would get 16.66 representatives. State Y, having a population of 300,000, gets 9.99 representatives. State Z has 725,000 and gets 24.16 representatives. So, how do we resolve that? Should we do a standard rounding, round up, or simply eliminate the remainder from consideration (essentially, rounding down)? If we round up, we may run afoul of Constitutional rule (1). If we round down, we run into an equity problem since State Y, which just misses getting 10 representatives, ends up with only nine. State Z’s apportionment, which is slightly over 24, will still get 24 seats. And while the districts in State Z can now be fairly equalized, State Y will have much bigger districts, diluting the power of State Y’s voters relative to State Z’s. The fairest thing to do might seem to be standard rounding, where anything 0.5 and above is rounded up and anything under 0.5 is rounded down. In this case, State Y gets 10 representatives and State Z gets 24, but now State X gets 17. How we deal with the problem of fractions has real-life consequences for the distribution of political power throughout the country. This is the problem that Congress faced in the first apportionment in 1791.
The early Congresses did not set a ceiling on the number of representatives. The final apportionment in 1791, after much haggling over remainders and differing methods to resolve them (including the first-ever presidential veto), gave each district a constituency of 34,000 people. But since 1940, Congress has set the number of seats in the House at 435 rather than the Constitutional maximum (which now would be 11,033 seats).*** This should give us an average House constituency of about 761,000 people. Congress has changed the method of apportioning seats in the House several times, but the current law has been in place since 1940.
The current method of apportionment is called the Method of Equal Proportions. According to the US Census, “the goal of the Method of Equal Proportions is to minimize the relative (or percentage) differences in representation (the number of people per representative) among the states.” First, every state gets one seat. Second, the remaining 385 seats are distributed according to a “priority value” based on a state’s apportionment population. The priority value is equal to its apportionment population divided by the square root of N(N-1), with N representing the number of the seat. So in the above example, State Z would have a priority value of 530330.085 for seat two (750,000 divided by the square root of 2(2-1). If that priority value is the highest after all other states get their first seat, then state Z would get the first second seat. This calculation continues until all seats are distributed. (It sounds confusing, but the math is straightforward). Take a look at the Excel spreadsheet for the 2020 priority values and it will help you visualize the process better.
Back to New York’s dilemma. If you look at the spreadsheet you’ll see that Minnesota barely beat the Empire State for the 435th and final seat. (The Census continues to calculate past the 435th seat for a number of reasons, including if Congress decides to increase the number of seats in the House.) Also interesting to note is that Montana barely got their second seat too; it’s right ahead of Minnesota in seat 434. New York has not lost population in the last ten years. In fact, the Empire State has gained almost a million more people (837,649) in the past decade. So, how can it lose a seat?
It seems counter-intuitive that a state can lose a seat when its population has increased.**** In fact, that would not happen if we did not have a cap on the number of seats in the House. Even if Congress did not set a cap, Constitutional rule (2) creates a moving cap based on total population every ten years. According to Minnesota state demographer Susan Brower "What I tell people is that it is not only what our population turns out to be, but it also relies on what every other state's population is. There is not any one population target or threshold that we're trying to meet. It really depends on where we fall relative to other states' populations."
This is why states attempt to make sure that their count is as accurate as possible, and some even have efforts to influence the Census estimates that are produced in the years between the decennial whole counts. It’s not just apportionment that is impacted by the Census, but the distribution of federal aid and loans. (Also, states use Census data for their own apportionment and state aid formulas.)
How is Cuomo’s argument related to the unexpected failure of Texas to pick up three seats and Florida to pick up two? It’s in his reference to the federal government’s “chilling effect” that discouraged undocumented – and perhaps even documented – immigrants from reporting to the Census. With an incompetent president and administration, as was true last year during the Census count, it is not surprising to see ironic outcomes.
The Trump Administration’s campaign to add a citizenship question to the Census faced intense backlash because the question would likely – and as we eventually found out was intended to – scare immigrants into refusing to participate. Trump and other Republicans hoped this would depress the population counts in Democratic states like California and New York, and thus lose them additional seats in apportionment. What happened instead is that the question was not added, but the legal and political battles over it further delayed the work of an already COVID-19 challenged count and arguably intimidated some immigrants from responding to the Census. In the end, it hurt Texas and Florida which each gained one fewer seat than expected while California lost just one seat (expected) and New York only lost one (also expected, but many thought the state would lose two).
Cuomo’s other argument appears to be that such a small number separating New York and Minnesota is likely an error. He’s wrong. While there may be an error, just because the number is small - even if it is the smallest in history - does not mean it is an error. Sure, New York might be able to find 89 more residents. Minnesota can probably find a few dozen uncounted residents too. The difference is perhaps so close that it warrants an extra review, but the likelihood of New York prevailing is much lower than Cuomo publicly assumes it is.
We’ll take a deeper dive into New York’s challenge to the results as it develops. Likewise, as we get more detailed demographic data we’ll revisit whether Texas and Florida failed to pick up more seats because of the Trump Administration’s efforts to discourage immigrants from participating in the Census. In the next installment of this series, we’ll look at the apportionment itself and examine the redistricting methods in the states that are losing and gaining seats.
————
* Until the passage of the 14th Amendment in 1868, there was one other rule that made apportionment difficult: the 3/5 clause. Here, fractions of persons were required. The unintended consequence of this is that we have census information – not detailed, but aggregate – on the US slave population during the first 75 years of the Republic that we likely would not have had if the framers had decided not to include slaves in the apportionment formula.
** This has been generally accepted as meaning one full congressional district, not one that also includes residents in a neighboring state.
*** The original First Amendment approved by Congress and sent to the states with the rest of the Bill of Rights would have required there to be a representative for every 30,000, which would force Congress to raise the membership of the House to the Constitutional maximum. It remains the only one of the twelve articles of the Bill of Rights not to be ratified by the states.
**** Only three states - West Virginia, Illinois, and Mississippi - lost population between 2010 and 2020. While population growth was slow overall, the other 47 states gained population in absolute terms.