Wednesday, August 28, 2019

A Modest Proposal

I saw that Stanford was making physics inclusive and creating an additional physics course that is about experiences and not physics. And it isn't, apparently, a mentoring program.

You can probably gather from an earlier post that I think of this sort of thing misses the point.

But I want to be positive for a change.

I have a modest proposal.

Most of the claims about racism and toxic environments are blissfully innocent of any data. It would not be expensive to accumulate some.

Every incoming student's records are already associated with their 1) age and 2) college board scores. Ask 3 questions during onboarding, that also go in their record:

  1. Sex: male, female, "whatever"
  2. Ethnicity: (include "prefer not to answer")
  3. First and second choices for major ("undecided" is OK)

Five years later, look up the all those students' records and determine

  1. Did the student graduate?
  2. If so, with what major?

Any such study has to preserve anonymity, so the only things included in the database for research are the 7 items.

A lot of students change majors, a number drop out, some get sick or even die. But even so, for a state university you'll start to get good statistics with the first or second cohort. In 10 years you might even have enough statistics about minority groups and about uncommon majors to begin to learn something useful, but even after 6 years (the second cohort) you'd have enough to start saying useful things about men and women in college.

After you learn the patterns ("18-year olds who want to do chemistry but have combined SAT less than 1150 generally don't complete the degree, but 35+ year olds do") you could, for the first time, actually use statistics to say whether some department is less friendly to women than others, or than the same department in another university. Right now we have subjective anecdotes. Sometimes those are revealing, and sometimes they are less than fully accurate.

Easy, cheap, high statistics--what's not to like? I know the diversity industry might fear it, but they aren't the only players in town.


J Melcher said...

Demographers do, routinely, collect this sort of data.

They don't collect other interesting attributes of the population. Among students, they (we) don't collect data on home many school districts or jurisdictions a student has attended during K-12. Data on religious affiliation, often collected for political polls and sampling, is rarely obtained for college populations. (Would it not be interesting to learn, for hypothetical example, that the poorly-defined voting faction "Evangelicals" tend to finish in fewer years with less debt? That Catholics tend toward "pre-law"? Or whatever?)

Freshman orientation ceremonies often force students thru the (IMO, bogus) Myers-Briggs personality survey, but as far as I know never accumulate results into useful data or trends.

I'd love to see an investigation done to see which STEM-track college students remember what publishers' math textbook series they used in sixth-thru-eighth grades. Can our education system EVER grade or scorecard the educational materials we spend and waste so much money on?

Can we use "family visits to Disney theme parks" as a proxy for a variety of factors -- family wealth, thrift (or lack of same), parental involvement, assimilation, concern for the environment -- to help make admission decisions?

Or simply request every incoming student to take the military ASVAB assessments and track forward from there. NOT an admission test, but simply a way to match raw young "aptitudes" to graduation rates, majors, etc.

Lots of info out there we aren't even trying to collect.

james said...

Those would all be interesting to know, indeed.

I think freshman orientation here was more lectures and talk sessions than testing, but I didn't go through them myself--just my daughters.

I was looking for something easy with high statistics that any university could create with no privacy issues. If you have a freshman class of 300, too many questions would make it easier to associate the records to a particular student.

I wonder if the Dept of Psychology is doing anything here.