Information for EDMS Fall 2006, "Bayesian Inference in Measurement"
Variables Doodle file for exploring distributions.
CoinDoodle Doodle file for two unreliable reports
of a coin flip.
Coin10 Doodle file for ten unreliable reports of
a coin flip, illustrating the use of a plate.
BetaBinomial Example from class on updating
with conjugate distributions--specifically, updating a beta prior with a
binomial likelihood, to get a beta posterior.
Simple regression Linear regression.
CTT Code for in-class CTT example.
CTTshadow Code for PPMC with CTT example.
Rasch example Code for in-class Rasch model
example.
3PL Code for in-class three-parameter logistic IRT
example.
PPMC Posterior predictive distributions
for model checking in IRT. Thanks to Roy Levy.
TwoSkillsDoodle Doodle for estimating
population and conditional probabilities from data, in Two Skills problem
from Assignment 3.
TwoSkillsCode WinBUGS code and data for
estimating population and conditional probabilities in Two Skills problem
from Assignment 3.
Cheating: Manifest Univariate Model
Cheating: Manifest Multivariate
Model
Cheating: Latent Class Model with Uniform(0,.5)
prior on class membership
Cheating: Latent Class Model with Uniform(0,1)
prior on class membership
Multivariate standardized regression
One-dimensional factor analysis
Balance Beam -- five attempts WinBUGS code.
Balance Beam -- two attempts data file.
MovieData Code for example used in the WinBUGS
movie example.
Proficiency1: 100 cases, Proficiency observed directly
Proficiency2: 100 cases, Proficiency observed directly + three History-taking observations
Proficiency3: 100 cases, Proficiency not observed + three History-taking observations
Graded response data : 165 cases, three items with {1,2,3} responses
Graded response spreadsheet : Calculating cumulative and category probabilities for a three response category item.
In-class graded response model. Constructed during the November 13 class period. Warning: Offerred "as is"--has not been tested.
CTT-DIC1, CTT-DIC2, CTT-DIC3, CTT-DIC4.WinBUGS setups for classical test theory model comparisons in presentation on DIC.