IFSV > Afdelinger > Afdelinger > Biostatistisk Afdeling > Lokale sider > rasch

SAS macros for Rasch based latent variable modelling

The following SAS macros can be used to estimate item parameters using conditional maximum likelihood estimation and test the assumption of unidimensionality in Rasch models, and to fit regression models where either outcome variables or covariates are latent variables measured using (log linear) Rasch models.

Please feel free to send question and comments to
Karl Bang Christensen

• latreg.sas
  macro to fit a latent regression for a latent variable measured by a loglinear rasch model
• latreg_2d.sas
  macro to estimate the parameters of a twodimensional regression model combining a GLMM (poisson or logistic regression with residual variation) and a latent regression based on a (loglinear) rasch model with known item parameter
• latreg_st.sas
  macro to estimate a poisson or logistic regression model with a latent variable measured by a (loglinear) rasch model with known item parameters as a covariate
• llrasch.sas
  macro to compute CML estimates, gamma polynomials, and maximum of the log likelihood function for a loglinear rasch model
• pml.sas
  macro to compute Martin-L test statistic for test of unidimensionality
• rasch.sas
  macro to compute CML estimates, gamma polynomials, and maximum of the conditional log likelihood function in a Rasch model