Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Structural Model Evaluation and Modification: An Interval Estimation Approach

Structural Model Evaluation and Modification: An Interval Estimation Approach Multivariate Behavioral Research, 25 (2), 173-180 Copyright O 1990, Lawrence Erlbaum Associates, Inc. Structural Model Evaluation and Modification: An Interval Estimation Approach James H. Steiger University of British Columbia Procedures for evaluation and sequential modification of structural models have attracted much interest in the recent psychometric literature. Kaplan (1990) proposes to extend a procedure (post hoc model modification, or PMM) popularized by Joreskog and Sorbom (1984). PMM is designed for cases where one's best attempt at an a priori theoretical model has been found to have poor or marginal fit to the sample data. The researcher, in desperation, may wonder if there is any model which fits the data. The PMM approach uses modification indices to predict which path, if added to a structural diagram, would decrease the chi-square fit statistic the most. One frees the parameter associated with that path to obtain an improved model. Kaplan (1990) recognizes some technical problems with the procedure. In particular, aparameter when freed may not change much from zero, even though the chi-square changes a lot. This can occur for at least two reasons. First, sample size may be huge, and so the big change in chi-square was really associated with a http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Multivariate Behavioral Research Taylor & Francis

Structural Model Evaluation and Modification: An Interval Estimation Approach

Multivariate Behavioral Research , Volume 25 (2): 8 – Apr 1, 1990

Structural Model Evaluation and Modification: An Interval Estimation Approach

Multivariate Behavioral Research , Volume 25 (2): 8 – Apr 1, 1990

Abstract

Multivariate Behavioral Research, 25 (2), 173-180 Copyright O 1990, Lawrence Erlbaum Associates, Inc. Structural Model Evaluation and Modification: An Interval Estimation Approach James H. Steiger University of British Columbia Procedures for evaluation and sequential modification of structural models have attracted much interest in the recent psychometric literature. Kaplan (1990) proposes to extend a procedure (post hoc model modification, or PMM) popularized by Joreskog and Sorbom (1984). PMM is designed for cases where one's best attempt at an a priori theoretical model has been found to have poor or marginal fit to the sample data. The researcher, in desperation, may wonder if there is any model which fits the data. The PMM approach uses modification indices to predict which path, if added to a structural diagram, would decrease the chi-square fit statistic the most. One frees the parameter associated with that path to obtain an improved model. Kaplan (1990) recognizes some technical problems with the procedure. In particular, aparameter when freed may not change much from zero, even though the chi-square changes a lot. This can occur for at least two reasons. First, sample size may be huge, and so the big change in chi-square was really associated with a

Loading next page...
 
/lp/taylor-francis/structural-model-evaluation-and-modification-an-interval-estimation-cm81FdQ3d0

References (8)

Publisher
Taylor & Francis
Copyright
Copyright Taylor & Francis Group, LLC
ISSN
1532-7906
eISSN
0027-3171
DOI
10.1207/s15327906mbr2502_4
Publisher site
See Article on Publisher Site

Abstract

Multivariate Behavioral Research, 25 (2), 173-180 Copyright O 1990, Lawrence Erlbaum Associates, Inc. Structural Model Evaluation and Modification: An Interval Estimation Approach James H. Steiger University of British Columbia Procedures for evaluation and sequential modification of structural models have attracted much interest in the recent psychometric literature. Kaplan (1990) proposes to extend a procedure (post hoc model modification, or PMM) popularized by Joreskog and Sorbom (1984). PMM is designed for cases where one's best attempt at an a priori theoretical model has been found to have poor or marginal fit to the sample data. The researcher, in desperation, may wonder if there is any model which fits the data. The PMM approach uses modification indices to predict which path, if added to a structural diagram, would decrease the chi-square fit statistic the most. One frees the parameter associated with that path to obtain an improved model. Kaplan (1990) recognizes some technical problems with the procedure. In particular, aparameter when freed may not change much from zero, even though the chi-square changes a lot. This can occur for at least two reasons. First, sample size may be huge, and so the big change in chi-square was really associated with a

Journal

Multivariate Behavioral ResearchTaylor & Francis

Published: Apr 1, 1990

There are no references for this article.