Tags: Investigative Research · API assisted reporting · Courts
Context
Reporting duration: nearly a year, on and off, alongside coursework and other assignments
Team project: ProPublica, in partnership with Boston University's Justice Media Computational Journalism co-Lab (BU Spark!); co-published with ABC News
My role: researcher on a multi-person team
Supervised by my professor, Brooke Williams and my editor, Noah Pranksy.
This was the first investigation in which I paired manual document sourcing — the kind of judgment call you can only make by reading a disclosure form line by line — with computational search at scale. Running keyword queries through the SEC's API let us test hypotheses across thousands of filings in the time it would've taken to manually check a handful. It taught me that the two approaches aren't a tradeoff: the API told us where to look closer, and the closer look is what turned a data pattern into a documented conflict of interest.
Across all 50 states, spouses, parents, and children of powerful judges sit on corporate boards, run political campaigns, and lobby the very governments whose cases land on the judge's docket.
I spent nearly a year working with a team to trace those hidden threads: cross-referencing campaign contributions, recusal forms, and financial disclosures against thousands of court dockets, then verifying family and business ties through social media sleuthing across platforms. Where the paper trail thinned out, we ran manual keyword searches in PACER and Westlaw to surface cases judges hadn't flagged themselves, and fed disclosure data into the SEC's full-text search API to scale that search computationally across records no team could comb through by hand alone.
We found dozens of judges — a Florida Supreme Court justice whose wife's PAC took a large campaign check from a tribe with billions riding on his ruling, a Minnesota federal judge who dismissed a case against his wife's own PR client, state supreme court justices in Ohio and North Carolina who declined to step aside from cases involving their own parents — who chose not to recuse, protected by ethics rules too ambiguous to enforce.
I learned:
How to build and sustain a long-horizon investigation across a scattered, inconsistent national record system — every state (and the federal judiciary) discloses judicial financial and family information differently, so the methodology had to be adaptable rather than uniform
How to combine manual, judgment-heavy research (reading disclosure forms, verifying social ties) with computational tools (API-based full-text search) to work at a scale no single reporter could manage by hand
The value of triangulating multiple record types — campaign contributions, recusal orders, financial disclosures, court dockets — since no single source told the full story of a conflict
How to work within a hybrid newsroom/classroom structure, coordinating with professional editors at ProPublica and ABC News while producing student-led original research