Utah v. Evans
The Supreme Court ruled that the Census Bureau's use of "hot-deck imputation" — filling in missing address data by copying information from a similar nearby address — did not violate the federal law banning census "sampling," nor did it violate the Constitution's requirement of an "actual Enumeration."
The decision let stand a population count that gave North Carolina one more House seat and Utah one fewer, and it clarified how far the Census Bureau can go in filling gaps in its data without crossing into the kind of statistical extrapolation Congress and the Constitution forbid for apportionment.
“We do not believe the Constitution makes the distinction that Utah seeks to draw.”
The Court's rejection of Utah's argument that only direct enumeration methods satisfy the Constitution.
How it got here: A three-judge federal district court ruled for the Census Bureau; Utah appealed directly to the Supreme Court, which postponed the standing question until argument on the merits.
The Case in Depth
What happened
After the 2000 census, Utah discovered that the Census Bureau had used "hot-deck imputation" to fill gaps where it could not determine whether an address was occupied or how many people lived there, copying data from a similar nearby address. This process added about 1.2 million people nationwide and shifted one congressional seat from Utah to North Carolina. Utah sued the Secretary of Commerce and Census Bureau, and North Carolina intervened to defend the count.
The question before the Court
Could the Census Bureau use a data-filling technique called "hot-deck imputation" in the 2000 census, even though it cost Utah a House seat and gave North Carolina an extra one?
The Court's answer
Yes — the Court held that hot-deck imputation was lawful under both the statute and the Constitution. It found the Census Bureau's practice of filling gaps by copying data from a similar nearby address was fundamentally different from the "sampling" Congress banned for apportionment: sampling deliberately surveys only part of the population to extrapolate to the whole, while imputation is used only as a narrow, last-resort fix after the Bureau already tried to count every address directly.
On the constitutional side, the Court read "actual Enumeration" as leaving Congress broad authority over census methodology rather than locking in one specific counting technique, and found nothing in the Constitution's text or history barring a small-scale data-filling tool used only when other counting methods failed. Because imputation affected a tiny share of the population and was aimed at achieving greater accuracy, not manipulating results, the Court let the 2000 census count — and North Carolina's extra seat — stand.
Curious how the Court got there? See the step-by-step legal reasoning →
Why it matters
The ruling let North Carolina keep the additional congressional seat it gained over Utah because of imputation, directly shaping which state's voters got an extra vote in the House. More broadly, it gave the Census Bureau continued authority to use imputation as a data-filling tool of last resort in future censuses, without triggering the same legal bar that applies to broader statistical sampling.
What changes now
This is a final merits decision, so the 2000 census apportionment stands: North Carolina keeps the additional House seat, and Utah does not receive it back. The ruling settles, for now, that the Census Bureau may keep using hot-deck imputation as a last-resort data-filling tool in future censuses, though the Court left open how far a much larger reliance on statistical estimation could go before it would cross constitutional or statutory lines.
What this does not decide
The Court expressly said it was not deciding the outer constitutional limits on census methodology or whether larger-scale statistical substitutes for actually reaching households would be permissible. It limited its holding to imputation used sparingly, as a last resort, affecting only a tiny share of the population.
Concurrences and dissents
Dissent in part — Justice O'Connor
Justice O'Connor agreed with the majority on standing and jurisdiction (Parts I and II) but disagreed that imputation is different from sampling. She argued the Bureau selected a subset of the population (donor addresses) and used it to estimate characteristics of the rest, which fits the statute's own definition of sampling, so she would have found imputation unlawful under the statute and never reached the constitutional question.
Dissent in part — Justice Thomas
Justice Thomas agreed that the statute does not bar imputation, but argued the Constitution's "actual Enumeration" requirement demands an actual count of real people, not estimation. Drawing on founding-era dictionaries, records of the Constitutional Convention, and early Census Acts, he argued the Framers deliberately chose precise language to prevent exactly this kind of statistical estimation, out of concern for political manipulation, and would have held hot-deck imputation unconstitutional.
Dissent — Justice Scalia
“This is an astonishing exercise of raw judicial power.”Scalia's objection to the majority reading a judicial correction power into the reapportionment statute.
Justice Scalia would not have reached the merits at all, arguing Utah lacked standing because any real relief depended on discretionary acts by the President and Congress that courts cannot control or predict. He accused the majority of rewriting the reapportionment statute to invent a judicial-correction option Congress never provided, calling it a violation of separation of powers, and would have vacated for lack of jurisdiction.
How the Court got there
The legal reasoning, step by step
- The Court first addressed whether Utah had standing — the right to sue — by asking whether a court order requiring a corrected census report would actually be likely to lead to a new, more favorable apportionment; relying on its earlier decision in Franklin v. Massachusetts, it found the relief realistically likely enough to satisfy this test.
- Turning to the statutory question, the Court distinguished imputation from the banned "statistical method known as ‘sampling’" by comparing their nature, methodology, and immediate goal: sampling deliberately surveys only part of a population to extrapolate to the whole, while imputation only fills in isolated missing data points after the Bureau already tried to survey every address.
- The Court looked to technical statistical definitions, the 1958 legislative history of the sampling ban, and the Bureau's own long-standing, unchallenged interpretation of the statute, concluding that Congress had sampling in the traditional survey sense in mind, not gap-filling imputation.
- On the constitutional claim, the Court read the phrase "actual Enumeration" as a general instruction to count people, not as dictating a specific counting methodology, noting that the Constitution leaves the "Manner" of counting to Congress.
- Examining founding-era dictionaries, drafting history, and early Census Acts, the Court found no evidence the Framers meant to bar every method short of literal person-by-person contact, especially where imputation is used only as a last resort after other counting efforts failed and involves a tiny share of the population.
- Because imputation fit neither the statutory definition of sampling nor the constitutional limits on enumeration methods, the Court concluded the 2000 census count was both lawful and permissible to use for apportionment.
Doctrinal impact
Cases affected by this decision
Reaffirms Franklin v. Massachusetts (505 U.S. 788)
The Court relied on Franklin's standing analysis to find Utah could sue over the census count.
Distinguishes Department of Commerce v. United States House of Representatives (525 U.S. 316)
The Court said the sampling methods barred in that case differed enough from imputation to reach a different result here.