How records are built

Each entry is a court's ruling on a motion to exclude or strike an expert's testimony in a patent case, taken from the public docket. The order is linked from every entry and controls over our summary.

What counts

Rulings on Daubert motions, motions to strike expert reports, and appellate decisions reviewing them, in federal patent cases. An entry records the expert, the side that retained them, the court and judge, the date, the outcome and its scope, the methods the challenge targeted, and the court's own words.

Outcomes

Opinions stood means the challenge was denied. Trimmed means some opinions or paragraphs were struck and the rest stood; most challenges in the Eastern District of Texas end this way. Excluded means the testimony on the challenged subject was excluded. When an appellate court later treats the same testimony differently, the entry says so.

Sources and review

Orders come from the federal courts' public dockets through CourtListener's RECAP archive. Entries are first extracted from the order's text and then checked by a person against the order before they are marked reviewed. Sealed filings are never used.

The Daubert survival score

Each decided motion to exclude or strike an expert's testimony counts 1 if the opinions stood, ½ if they were trimmed, and 0 if they were excluded. The score is the expected share of opinions surviving a challenge, estimated so that an expert with few rulings is pulled toward the average of all experts: two survivals out of two is encouraging, but it says less than seventeen out of twenty. The 80% range shows how much the record supports. Motions that are pending, withdrawn or mooted by settlement count as challenges but are never scored. Busy experts in large cases draw more challenges, so the number of challenges alone is not a mark against anyone.

Adjusting for the circumstances of a challenge

Outcomes depend on more than the expert: some judges exclude more readily than others, and some methods, such as apportionment, draw successful challenges more often. The rankings come from a model fitted to every decided challenge in the record, which estimates at once how much of each outcome is due to the expert, the judge, the method attacked, the court, the retaining side and the role. Each estimate is shrunk toward the average in proportion to how little the record says about it, and each comes with an 80% range. We report whether the model, tested on rulings it wasn't fitted to, predicts better than the overall average; until it clearly does, treat the rankings as a careful description of the record rather than a forecast.

Corrections

If an entry misstates an order, tell us which one. We check it against the order and correct it, whether or not the expert has claimed the profile or pays for placement.

Paid placement

Experts may pay to be featured in categories they choose. Featured profiles are labeled Sponsored, appear above the results they match, and are never counted or ranked differently in the record itself.