In The Use and Reliability of Federal Nature of Suit Codes, Christina Boyd (Georgia--Poli Sci) and David Hoffman (Penn) identify a lingering and subtle coding problem involving the structure of the "Nature of the Suit" ("NOS") variable. The source of this problem is clear: while the NOS coding protocol requires a single "case type" identification for every federal civil lawsuit, a single lawsuit typically joins multiple types of legal issues. While this coding problem might not catch the eye of many it absolutely informs legal empirical folk who use the array of federal civil datasets that rely on NOS codes. As the paper notes, "[NOS] codes ... are the foundation for almost all quantitative analysis of the federal litigation system. Scholars seeking to understand the relationship between substance and procedure use NOS codes to select their datasets. So do court administrators looking to equalize workload between and among districts and justify new judgeship requests." The paper's take-away includes good and bad news. "The bad news is that NOS codes in their current form are much too noisy to be reliable. The good news is that a complete and almost free solution is easily at hand." The paper's abstract follows.
“When filing a civil case in a federal district court, attorneys must identify one, and only one, of ninety issue area nature of suit (NOS) codes that best describes their case. While this may seem like a trivial moment in litigation, the selection of this single descriptor has significant implications for court statistics, empirical research findings, and the allocation of resources to federal courts, including judgeships. Despite the import of NOS codes, there is little within the process of choosing them to guarantee reliability in the selected NOS codes. To assess how reliable NOS codes are, we examine a database of nearly 2,500 federal civil complaints and the individual causes of action within those complaints. Our data reveal that for lawsuits like employment discrimination and intellectual property cases, the selected NOS codes do a very good job of summarizing the legal content of the complaint. However, in other types of civil lawsuits, including many contract, tort, and real property cases, there is a great deal of inconsistency between the NOS codes and the complaint contents. The difficulty in reliably selecting an NOS code is particularly high as the number and variability of underlying legal claims rise. We conclude by recommending that federal courts adopt a modest revision to the NOS code selection strategy. Rather than relying on attorneys to summarize their frequently complex lawsuits into a single NOS code, filing attorneys should instead classify their individual causes of action. From there, the courts can automate the grouping of content-similar cases. The result will be NOS codes that much more accurately and reliably capture the nature of the suits.”