Beyond Assessment of Quality of Life in Schizophrenia by A. George Awad & Lakshmi N.P. Voruganti

Author:A. George Awad & Lakshmi N.P. Voruganti
Language: eng
Format: epub
Publisher: Springer International Publishing, Cham

If the kernel smoothing model in the left column in Fig. 8.5 is taken as the truth, the other IRT models have various degrees of success in modeling the option characteristic curves. In particular, the rating scale model seems to provide a poor fit for MH5 – not capturing the broad range over which the response “Most of the time” is the most probable. A likely explanation for this misfit is that the response choices for this item have been “reverse coded” compared to the other items due to the positive item formulation (“happy person”). Thus, it is not reasonable to assume that the same rating scale model would apply to this item. A single rating scale model is usually only applied in situations where all the items share the same response choices (and direction of scoring). For an improved rating scale analysis of the current items, one rating scale model would be specified for MH1 and MH4 and another for MH5.

When comparing the option characteristic curves of the normal-ogive graded response model and the generalized partial credit model, the latter seems to provide a better fit. A possible reason for this is that the normal-ogive graded response model for illustration purposes was calculated based on the results of a categorical data factor analysis (Muthén and Muthén 2014). While this approach is correct in principle, a direct estimation of the normal-ogive graded response model would probably have resulted in a better fit.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.