Tuesday, July 6, 2010

Harvesting Relevant Cases on Lexis and Westlaw



LexisNexis and Westlaw are both very powerful research systems. Their coverage has a lot of overlap -- they both have state and federal cases, statutes, and regulations, they both have lots of law reviews, and so on. And their features are generally comparable -- you can search with connectors or with natural language, you can check the status of cases with a citator (KeyCite or Shepard's), and so on.

Focusing on the the similarities and parallels, I've sometimes underplayed the differences, telling students it's largely a "Coke-Pepsi" or "Ford-Chevy" thing.

But there are some differences, and it's worth conducting a taste test or taking a look under the hood (depending on whether you favor the soda or the car metaphor). Susan Mart investigates two parallel techniques for finding cases once you know one relevant case:
  • searching for more cases like a given headnote, and
  • using a citator to find later cases discussing the point of law in a headnote.
Susan Develow Mart, The Relevance of Results Generated by Human Indexing and Computer Algorithms, 102 Law Libr. J. 221, 2010 Law Libr. J. 2010-13.

Mart's methodology required a lot of painstaking work -- more than most of us would be willing to undertake, which is why it's so helpful that she published her results. She selected 10 prominent cases, then found pairs of corresponding headnotes to use from each (e.g., LexisNexis headnote 7 and Westlaw headnote 8 from Brown v. Board of Education). She ran searches using Westlaw's Custom Digest and two methods in LexisNexis: "More Like This Headnote" and the list of topics. Having written in advance what would count as relevant, she then sifted through all the cases retrieved and scored them as relevant or not. And then she compared the sets.

The sets of cases retrieved using the West headnotes tended to have a higher percentage of relevant cases than did the sets retrieved using either of the approaches based on Lexis headnotes. Does that mean we can just drop Lexis and use only Westlaw? No -- because each time there were a number of relevant cases that were found only using the Lexis headnotes.

What of the citators? How do Shepard's and KeyCite stack up? Again, the results were more divergent than you might expect. Each system turned up lots of cases supposedly discussing a point of law that the other did not turn up, even when the headnotes were comparable.

From these samples, Mart draws the lesson that researchers who need to be thorough should use both systems and conduct multiple searches. Don't assume that following a headnote from one case in one system will yield all the relevant cases. Use many seed cases. And use more tools -- for instance, A.L.R. annotations and law review articles, not just headnote searches or citator scans.

Years ago our Lexis rep used to say in training sessions, "One search is not research." This study underscores the wisdom of that saying. No one search -- in either system -- does it all for you.


Photos from Washington Economic Development Commission and NIH Nat'l Ctr. for Complementary and Alternative Medicine.

1 comment:

Bircheeta said...

Excellent post! I will be sharing it with my ALR students as we discuss key numbers, more like this, keyciting and shepardizing. I will also be sure to read the article you are discussing.