We usually use multidimensional item response theory (MIRT) model to research the educational longitudinal survey data, also known as longitudinal item response data. That is, sometimes students need to take the test in several different time spots. However, When the dimensions of the latent variable increases, difficulties in the operation of the computer is also increasing. Sometimes latent variable distribution cannot be integrated out, such as the IRT models for binary data. This article based on the pseudo-likelihood theory, using pairwise modeling to estimate item and population parameters in longitudinal item response study. Finally, we use simulatio- n to verify the advantages of this method.
|