This Article Full Text (PDF) References Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to Saved Citations Download to citation manager Request Permissions Request Reprints Add to My Marked Citations Citing Articles Citing Articles via Google Scholar Citing Articles via Scopus Google Scholar Articles by Lee, S.-Y. Articles by Xu, L. Search for Related Content Social Bookmarking                         What's this?

Case-Deletion Diagnostics for Factor Analysis Models With Continuous and Ordinal Categorical Data

Liang Xu,

The Chinese University of Hong Kong

This article proposes a method to detect influential observations in a factor analysis model with continuous and ordinal categorical variables. The key ideas are to treat the latent factors as hypothetical missing data and then develop the diagnostic measures on the basis of the conditional expectation of the complete-data log-likelihood function in the EM algorithm. A one-step approximation is proposed to reduce the computational burden. Building blocks for achieving the diagnostic measures are computed via observations generated by the Gibbs sampler. Results from a simulation study and an illustrative real example are presented.

Key Words: influential observations • EM algorithm • Gibbs sampler • one-step approximation

Sociological Methods & Research, Vol. 31, No. 3, 389-419 (2003)
DOI: 10.1177/0049124102239081


CiteULike    Complore    Connotea    Del.icio.us    Digg    Reddit    Technorati    Twitter    What's this?