You Have to Prove Something’s Missing Before You Can Look for It: The Administrative Record

In March 2018, Commerce Secretary Wilbur Ross announced that the 2020 census would include a citizenship question. The official story was straightforward: the Justice Department asked for citizenship data to enforce the Voting Rights Act, the Census Bureau studied how to collect it, and Ross decided to put the question on the form. The administrative record the government filed contained exactly that. The Justice request, the Bureau’s analyses, the Secretary’s explanation.

Then, after the litigation began, Ross filed a supplemental memorandum acknowledging he had been considering the question since early 2017, well before Justice asked, and that Commerce had asked whether Justice would make the request. That gave the plaintiffs something concrete: a reason to believe the record wasn’t complete. The parties eventually added more than twelve thousand pages of internal documents.

Gwen and Marc use that case to work through what the administrative record is, who assembles it, and what happens when someone thinks the court isn’t seeing everything. The agency whose decision is under review is also the party that builds the file the court will use to review it, and that arrangement is harder to replace than it first appears.

In this episode:

• What belongs in the record: the proposed rule, the studies, the comments, the responses, and everything the agency considered directly or indirectly, not just what the administrator personally read
• Three things worth keeping separate: the agency’s files, the public docket, and the record filed with the court
• APA section 706 and the whole record, which includes material cutting against the agency’s conclusion
• Overton Park, and why post-hoc affidavits can clarify an agency’s reasoning but can’t replace it
• Why building a record that can withstand review isn’t evidence of bad faith, any more than drafting a contract a court might read someday
• The presumption of regularity, and what a challenger needs before a court will test it
• Completion versus supplementation, and why the distinction matters even though courts are inconsistent about the labels
• The four circumstances in Lands Council v. Powell, including technical explanation and the strong showing required for bad faith
• The chicken-and-egg problem: making a bad-faith showing without the documents that would prove it
• What the Supreme Court actually held about the census, which wasn’t that Ross lacked authority or that a secretary can’t have policy preferences
• The deliberative process privilege, and why protecting candid internal disagreement is a feature rather than a dodge
• Blue Mountains Biodiversity Project, where pre-decisional material falls outside the record entirely, so there may be no privilege log and no way to know what was left out
• The same questions in an individual adjudication: a disability claimant’s record, the ALJ’s duty to develop it fully, and the narrow route for new and material evidence

The record does two jobs at once. It gives the court what it needs to review the decision, and it sets the outer limit of that review. Those boundaries are what keep a court from taking over the agency’s job, and they are also what can make a real problem hard to find.

Next week: questions of law, questions of fact, and mixed questions, and why the same record produces very different review depending on which one the court is answering.

Full written explainer with cases, context, and related guides: remediespodcast.com/learn-administrative-record-explained