Keywords

dimensionality of test data, divisive latent class reliability coefficient, psychometrics, reliability, standard error for reliability estimate, test score reliability

 

Authors

  1. Sijtsma, Klaas
  2. van der Ark, L. Andries

Abstract

We respond to three commentaries on our discussion article on different conceptions of test score reliability. First, we discuss the use of standard errors for reliability estimates. Second, we discuss the desirability not to confuse issues pertaining to the dimensionality of the test data (closely related to construct validity) and the degree to which measurement values are repeatable under the same circumstances (i.e., the reliability issue). Third, we discuss a new reliability estimation method that is almost unbiased irrespective of the dimensionality of the test data.