Statistical analysis of Discrimination over the past 40 years
- Administrivia:
- If you dont have a partner for your project, stay after
class and see if you can find one.
- Please email me and nuria with your project proposal.
(See web page for email addresses)
Discrimination: the haydays
- Who got Einstiens office? I didn't.
- But I did get John de Cani's office.
- He did "anti-discrimination" in its hayday.
- Drove around the south with the ACLU knocking down
discriminatory jury duty laws
- Example: 25% Blacks in state
- 250 people on juries in past year
- Not ONE black on ANY jury
- He said: "the chance of this occuring by chance alone is
like dealing 5 full houses in a row in poker."
- Judge said: "That would be cheating."
- he said: "Exactly."
Discrimination: modern times
- Prop. 209 (my small part)
- Prop 209 made affimative action illegal
- Continuing in the tradition of the office, I was part of
a ACLU friend of the court filing against it.
- We lost.
- Legacy admissions/children of allumni still allowed in
extra. These forms of discrimination are still legal.
- Income is the battle ground now adays
- 20% difference is considered significant (legal
definition)
- Statistically significant different is only sometimes
needed! (should it be the definition?)
- Covariates now are used: age, profession, years worked,
etc.
- Example AAUP (American Association of University Professors)
- Every year in their salary issue, they point out three
problems with status of women:
- Women are paid less
- Women are lower ranked
- women are younger
- Court cases don't hold up though?
- What is going on?
- Draw picture of age (X) and income (Y)
- positively related
- percentage of women has been increasing (at least in
some fields)
- so women are younger
- thus sex and age are correlated
- thus we need to do a multiple regression
- Draw various pictures of age, income and color coded sex
- easy discrimination (women uniformly lower than men)
- in equlibrium (fall on same line, at same location)
- current reality (same line?, different X distribution)
- Is this discrimination or not?
Last modified: Tue Mar 7 12:47:02 2000