Partial vs marginal slopes

• Administrivia:
  • Handout sample exam from last year
  • It covers about up to here. We will have some multiple regression on the exam also.
  • Exam a week from Thursday
  • 2nd Homework due 2 weeks from Today

• Basic idea
  • X --> Z, Z --> Y, Z --> Y looks like X --> Y
  • Example: breast feeding --> bad water --> infant mortality
  • Correct analysis finds this: regress Y on both X and Z
  • Relationship between X and Y is marginal slope
  • Relationship between X and Y HOLDING Z FIXED is partial slope
  • Partial slopes are better!
• Example:
  • Y = beta₀ + beta₁ X₁ + beta₂ X₂ + error
  • GP1000M = 11.7 + .0089 Weight + .088 Horsepower + error
  • GP1000M = 9.43 + .0136 Weight + error
  • HP = Weight / 20 + error
  • explination: GP1000M = 9.43 + .0089 weight + .088 * (weight/20)

• Handout on Effort and Reward in College

Last modified: Tue Feb 22 13:13:26 2000