Saturday, September 20, 2003

Another Self-Indulgent Post About Statistics

The ever fascinating David Horowitz of Front Page appears to have spawned another interesting political project. It's a grass roots campaign for the adoption of an Academic Bill of Rights. Like any effective grass-roots campaign, it's being fought on two fronts; through lobbying of legislators and on-the-ground activism and research by conservative students under the banner of Students for Academic Freedom (SAF). SAF has already put in the hard yards at 32 Elite US Universities and Colleges, producing this damning report on the dominance of lefty-liberalism at these institutions. US students who suspect that their own college or university is similarly tainted with academic bias can find handy tips on how to confirm their suspicions here.

The research protocol is an eminently simple, practical one. First, you make a list of the names of all the faculty and administrative staff of your institution. If possible, you should get a full list of names and home addresses, as this will make it easier to accurately identify individual staff members when you get to the second stage of the process, which is to trot down to wherever your State or county keeps its voter registration records and check the political affiliations of the people on your list. It's possible of course, that your identifying data may be inadequate when you cross-check against the official records; here's how you deal with that problem:

The person being investigated is “Andrew Jones,” and there is no address available.
If your result is one “Andrew Jones,” this is conclusive. Record the party.
If your result is two people named “Andrew Jones” this is not conclusive. Record as TM – too many positive hits.
If your result is “Andrew L. Jones”: this is conclusive. Record the party.
If your results are “Andrew L. Jones” and “Andrew N. Jones”: not conclusive. Record as TM – too many positive hits.
If your results are “Andrew Jones” and “Andrew L. Jones”: still not conclusive. Record as TM – too many positive hits.

It would be disingenuous of me to suggest that there is a simpler solution to this problem, which is to approach "Andrew Jones" (or "Andrew L. Jones") and ask him to his face what his political party affiliations are. After all, he might just ask why you want to know and, if you were foolish enough to tell him what you were about, might form an unfairly low opinion of your character based on the ridiculous prejudice that checking the political affiliations of college and university professors in this way is a bit of a sneaking, underhanded business. Better by far to accept the possibility of a few false positives and get on with the research.

Another interesting omission, which bears on the reported results of the 32 campus survey, is a simple test you really need to perform before you can say, with any authority at all, that the political affilaitions of the staff at your college or university are out of whack with those of the general community. That, of course, is to tally up all the political affiliations recorded in the county or state records, to find out how many registered Democrats (i.e. leftists) and Republicans (i.e. sensible conservatives) there are in the general population.

If this is too tedious, you could at least take a sample, by checking several random pages from the listings. It seems to me that anyone who is prepared to take the trouble to do all the other tedious data gathering reccommended in the SAF protocol ought to have the motivation to take this extra step as well. Its omission is troubling, although it may explain why the summary report on the 32 elite campuses makes no mention of any statistical test which may have been applied to determine whether the academic bias identified at these institutions was genuine, or consistent with random variations in the population. But this is mere nitpicking; the parlous state of the US academic community is well known to people of sense, so it would be unreasonable to insist on too high a standard of proof.

No comments: