A Multilevel CFA-MTMM Model for Nested Structurally Different Methods
Research output: Journal contributions › Journal articles › Research › peer-review
Standard
In: Journal of Educational and Behavioral Statistics, Vol. 40, No. 5, 01.10.2015, p. 477-510.
Research output: Journal contributions › Journal articles › Research › peer-review
Harvard
APA
Vancouver
Bibtex
}
RIS
TY - JOUR
T1 - A Multilevel CFA-MTMM Model for Nested Structurally Different Methods
AU - Koch, Tobias
AU - Schultze, Martin
AU - Burrus, Jeremy
AU - Roberts, Richard D.
AU - Eid, Michael
PY - 2015/10/1
Y1 - 2015/10/1
N2 - The numerous advantages of structural equation modeling (SEM) for the analysis of multitrait–multimethod (MTMM) data are well known. MTMM-SEMs allow researchers to explicitly model the measurement error, to examine the true convergent and discriminant validity of the given measures, and to relate external variables to the latent trait as well as the latent method factors in the model. According to Eid et al. (2008) different MTMM measurement designs require different types of MTMM-SEMs. Eid et al. (2008) proposed three different MTMM-SEMs for measurement designs with (a) structurally different methods, (b) interchangeable methods, and (c) a combination of both types of methods. In the present work, we extend this taxonomy to a multilevel correlated traits–correlated methods minus one [CTC(M−1)] model for nested structurally different methods. The new model enables researchers to study method effects on both measurement levels (i.e., within and between clusters, classes, schools, etc.) and evaluate the convergent and discriminant validity of the measures. The statistical performance of the model is examined by a simulation study, and recommendations for the application of the model are given.
AB - The numerous advantages of structural equation modeling (SEM) for the analysis of multitrait–multimethod (MTMM) data are well known. MTMM-SEMs allow researchers to explicitly model the measurement error, to examine the true convergent and discriminant validity of the given measures, and to relate external variables to the latent trait as well as the latent method factors in the model. According to Eid et al. (2008) different MTMM measurement designs require different types of MTMM-SEMs. Eid et al. (2008) proposed three different MTMM-SEMs for measurement designs with (a) structurally different methods, (b) interchangeable methods, and (c) a combination of both types of methods. In the present work, we extend this taxonomy to a multilevel correlated traits–correlated methods minus one [CTC(M−1)] model for nested structurally different methods. The new model enables researchers to study method effects on both measurement levels (i.e., within and between clusters, classes, schools, etc.) and evaluate the convergent and discriminant validity of the measures. The statistical performance of the model is examined by a simulation study, and recommendations for the application of the model are given.
KW - Sociology
KW - MTMM analysis
KW - multilevel structural equation modeling
KW - structurally different and interchangeable methods
UR - http://www.scopus.com/inward/record.url?scp=84945303884&partnerID=8YFLogxK
U2 - 10.3102/1076998615606109
DO - 10.3102/1076998615606109
M3 - Journal articles
AN - SCOPUS:84945303884
VL - 40
SP - 477
EP - 510
JO - Journal of Educational and Behavioral Statistics
JF - Journal of Educational and Behavioral Statistics
SN - 1076-9986
IS - 5
ER -