« The Power of Arsy | Main | By the Book »

August 24, 2008

Pulp Non-Fiction

Any competent statistician will tell you that sample size is telling. So why do Andrew S. Baluch and Stephen B. Maebius just look at the first 30 of 308 inter partes reexaminations since late 1999, when the practice began? From the smattering, the two Foley & Lardner attorneys find reexamed claims downed a whopping 73% of the time. 43% of the time, the patent holder didn't even bother to reply. The authors term it "surprising efficacy." A sense of perspective would call it numbers that tell you nothing noteworthy.

Old reexams, literally the first batch, that didn't fare well. Another exercise in rather meaningless patent stats by lawyers with, at best, facile statistical acumen. If they knew better, they would have bothered to provide a more complete, and interesting, picture.

A hat-tip of thanks to Professor Crouch at Patently-O.

Posted by Patent Hawk at August 24, 2008 10:28 AM | Prosecution


The problem is that of the remaining 308 reexams, most are still pending. Since they haven't been finally resolved, it's rather hard to evaluate their resolution.

Posted by: Judith at August 25, 2008 11:27 AM

Judith points out exactly the sampling bias - the sample only includes the reexams that were concluded most quickly, which of course will have a very high sample bias in favor of those for which "the patent holder didn't even bother to reply"

Posted by: David Boundy at August 26, 2008 8:31 AM