Methodology Review: Statistical Approaches for Assessing Measurement Bias

This paper provides an up-to-date review of state-of-the-art methodological developments appearing in the literature since the publication of the Handbook of Nlethods for Detecting Test Bias

Roger E. Millsap; Howard T. Everson

2007

Scholarcy highlights

  • Statistical methods developed over the last decade for detecting measurement bias in psychological and educational tests are reviewed
  • This paper provides an up-to-date review of state-of-the-art methodological developments appearing in the literature since the publication of the Handbook of Nlethods for Detecting Test Bias
  • The standardization procedure uses the total score as a substitute for the unobserved latent trait, and should encounter problems in data generated by multiparameter item response theory models
  • This section reviews developments in bias detection methods that operate within an assumed measurement model relating the observed measure Y to the latent variable ~ The section is divided in three subsections
  • The final section reviews methods that consider YV as a multivariate latent variable. Scored easures These methods assume that a unidimensional IRT model underlies performance on the studied measure
  • Progress has been made in developing detection methods, problems still remain
  • Lord proposed that the null hypothesis of identical IRFs across groups be tested using a test for equality of item parameters under an assumed parametric model

Need more features? Save interactive summary cards to your Scholarcy Library.