• Description of variables in the High School and Beyond data set.



• High School & Beyond: measures of association and correspondence analysis.



• 2004 General Social survey Vocabulary Examples. Includes input code for
  • Data file
  • Measures of assocication: Items A & C
  • Logit model of A as a function of rest-score and degree earned.
  • Log-linear models
  • IRT modeling (log-linear by linear and quasi-symmetric---Rasch models)
  • Log-multiplicative models corresponding to the 2pl.
    The MLE estimation of these models using SAS is "new" in the sense that I haven't written about it yet. This uses PROC NLP in the OR (Operations Research) SAS product. Includes input code for
    • Data set-up
    • Using PROC NLP to fit the log-multiplicative assocation model corresponding to the 2pl.
    • Computation of fitted values, goodness-of-fit statistics, and standardized residuals.
    • Comparisons of results with LEM and standard IRT models.
  
  


• Celebirty Choice Data. Includes input code for
  • Data
  • Bradley-Terry-Luce model at a logit model (with some diagnostics)
  • Thrustones comparative choice model as a probit model (with some diagnositics)
  • Bradley-Terry-Luce model as a log-linear model of quasi-symmetry (also quasi-independnece and symmetry, but these are bad models for these data)


• High School & Beyond: Discrete choice model for careers. Includes input code for
  • Data
  • Data manipulation to set up for modeling
  • Use PROC MDC
  (specifically designed to fit conditional discrete choice models
  • Use PROC GENMOD (log-linear/Poisson regression model)
  • Use PROC PHREG (fit as proportional hazard regression model)
  • Comparision of results from different procedures
  • Model simplification

R code for various analysis:

• High School & Beyond data set: measures of association (including correspondence analysis):
  • Description of variables in the High School and Beyond data set.
  • HSB Data (all the data)
  • 2-way table (career X gender)
  • R script



• 2004 General Social survey Vocabulary Examples:
  • Individual Level Data
  • Tabled Data
  • R script


• Celebrity Choice Example (Thurstone's Case V and the Bradley-Terry-Luce Model):
  • The data
  • R script



• High School \amp Beyond: career choices (Conditional Multinomial Choice Model):
  • The HSB data in expanded format.
  • R script.




---

Lem files for item response models

Program input at the top and output follows:
• Log-linear model of independence.
• Marginal maximum likelihood estimation of the Rasch model.
• Log-multiplicative association model corresponding to the 2PL.
• Marginal maximum likelihod of 2PL.

---

Report broken links and send email to cja@uiuc.edu



Last revised June 16, 2008