Job Statistics: What about Unknowns?

In Placement and Accreditation, I discussed the placement benchmarks of suggested Interpretation 301-7 (supposedly are no longer being considered).  Suggested Interpretation 301-7 would have treated "unknowns"--graduates who have not reported their job status to their law school--as unemployed. A law school's unemployment rate would include both graduates reporting themselves as unemployed, and seeking employment, but also graduates that did not respond.

I see two possible rationales for that approach.  First, it encourages law schools to try harder to get responses from its graduates. Second, it assumes that non-responders are probably unemployed, but are ashamed to admit it.

How are unknowns distributed across law schools?  As the following chart shows, there is a general connection between a law school's proportion of unknowns and its relative eliteness, as measured by the 75th percentile of the LSAT scores of its entering classes.

Only three law schools have proportions of unknown" above 20%:  University of Puerto Rico (148, 83.4%); North Carolina Central (153, 32.0%); and Capital (156, 25.5%). Fifteen other law schools have proportions in the range of 10% to 20%. All 18 law schools (9.1% of 198) had LSAT 75th percentiles of 157 or below.

Almost 40% (38.5%) of law schools had 2006 entering LSAT 75th percentiles of 157 or lower ("low-LSAT schools"). As the following chart shows, low LSAT schools were much more likely to have higher proportions of unknowns for their Classes of 2009.

Most strikingly, although 19.2% of all law schools had proportions of unknowns of at least 5.0%, 44.6% of low-LSAT schools had proportions in that range, as opposed to 3.4% of higher-LSAT schools.

Comments

[Cheap Brochure Printing]

I guess this is an unbiased review and an accurate one! Thanks for sharing this statistics. I agree with you, some Law graduates are just shy to admit that they are jobless. -Sarah-

[Erik Girvan]

The data fit a story that makes sense. Even so, the assumption that 100% of non-responses are unemployed seems rather strong without an estimate of the actual proportion. In particular, I worry that it might be systematically misleading (and burdensome) due to third variables like school resources and general alumni satisfaction, which may be more strongly related to non-response rate than unemployment. How predictive is the rate of unemployment calculated from alumni responses for a school of the school's non-response rate (ideally controlling for resources, e.g., student-faculty ratio, and alumni satisfaction, e.g., giving rates)? It is not entirely clear that the tendency not to respond because one is unemployed and embarrassed about it is unrelated to "elite status"--given that relative deprivation is often more influential than absolute deprivation, graduates of high status schools might be more embarrassed by unemployment and thus have a greater tendency to not respond or lie. Even so, it seems like a more direct test of the hypothesis that non-response is due to alumni unemployment.
-Erik

[Scott Boone]

Correlation is a dangerous game.

Other commonalities between "low-LSAT" schools, such as resources and satisfaction that EriK points out, can contribute to higher percentages of unknowns. I think many people familiar only with well funded elite schools overestimate the resources available at some or many "low-LSAT" schools. Keep in mind that any increased resources put to this will have to either come out of some other function of the school (career placement, teaching position, etc) or out of student pockets in the form of higher tuition. I also think some people underestimate the extent to which some alumni go to hide themselves from their schools (primarily out of a mistaken fear that they are going to be pestered for donations). Yes, this is real.

Another factor that contributes to a high level of unknowns is how dispersed the alumni are geographically. The more concentrated the alumni are geographically, the easier it is to discover employment status and contact information through tighter networks.