Please use this identifier to cite or link to this item:
http://hdl.handle.net/1942/16178
Title: | Reliability Measures In Item Response Theory: Manifest Versus Latent Correlation Functions | Authors: | MILANZI, Elasma MOLENBERGHS, Geert ALONSO ABAD, Ariel VERBEKE, Geert De Boeck, Paul |
Issue Date: | 2015 | Source: | British Journal of Mathematical & Statistical Psychology, 68 (1), p. 43-64 | Abstract: | For item response theory (IRT) models, which belong to the class of generalized linear or non-linear mixed models, reliability at the scale of observed scores, i.e., manifest correlation, is usually of greater scientific interest, while its calculation is more difficult than latent correlation based reliability. This is not in the least because it cannot be calculated explicitly when the logit link is used in conjunction with normal random effects. As such, approximations like Fisher’s information coefficient, Cronbach’s , or the latent correlation are calculated, allegedly because it is easy to do so. Cronbach’s has well-known and serious drawbacks, Fisher’s information is not meaningful under certain circumstances and there is an important but often overlooked difference between latent and manifest correlations. Here, manifest correlation refers to correlation between observed scores, while latent correlation refers to correlation between scores at the latent scale, e.g., logit or probit scale. Thus, using one in place of the other can lead to erroneous conclusions. Taylor series based reliability measures, which are based on manifest correlation functions, are derived and a careful comparison of reliability measures based on latent correlations, Fisher’s information, and exact reliability is effectuated. The latent correlations are virtually always considerably higher than their manifest counterparts, Fisher’s information measure shows no coherent behavior (it is even negative in some cases), while the newly introduced Taylor series based approximations reflect the exact reliability very closely. Comparisons among the various types of correlations, for various IRT models, are made using algebraic expressions, Monte Carlo simulations, and data analysis. Considering the light computational burden and the performance of Taylor series based reliability measures, their use is recommended. | Notes: | Molenberghs, G (reprint author), Hasselt Univ, Ctr Stat, B-3590 Diepenbeek, Belgium geert.molenberghs@uhasselt.be | Keywords: | 1PL model; 2PL model; logit link; probit link; Rasch model | Document URI: | http://hdl.handle.net/1942/16178 | ISSN: | 0007-1102 | e-ISSN: | 2044-8317 | DOI: | 10.1111/bmsp.12033 | ISI #: | 000347892300003 | Rights: | © 2014 The British Psychological Society | Category: | A1 | Type: | Journal Contribution | Validations: | ecoom 2016 |
Appears in Collections: | Research publications |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
explicit15.pdf | Peer-reviewed author version | 210.05 kB | Adobe PDF | View/Open |
Milanzi_et_al-2015-British_Journal_of_Mathematical_and_Statistical_Psychology.pdf Restricted Access | Published version | 350.83 kB | Adobe PDF | View/Open Request a copy |
SCOPUSTM
Citations
8
checked on Sep 2, 2020
WEB OF SCIENCETM
Citations
14
checked on Oct 13, 2024
Page view(s)
62
checked on Sep 7, 2022
Download(s)
156
checked on Sep 7, 2022
Google ScholarTM
Check
Altmetric
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.