NotesFAQContact Us
Search Tips
ERIC Number: ED454245
Record Type: Non-Journal
Publication Date: 2001-Feb
Pages: 30
Abstractor: N/A
Reference Count: N/A
Does Method of Handling Missing Data Affect Results of a Structural Equation Model?
Witta, E. Lea
The influence of method of handling missing data on estimates produced by a structural equation model of the effects of part-time work on high-school student achievement was investigated. Missing data methods studied were listwise deletion, pairwise deletion, the expectation maximization (EM) algorithm, regression, and response pattern. The 26 variables selected from the National Educational Longitudinal Survey of 1988 database were those previously used by K. Singh and M. Ozturk (1999) in an analysis of part-time work. Results indicate the data was not missing completely at random, and although the covariance matrices, measurement models, and structural models using the five missing data methods were not significantly different statistically, the individual best fitting structural model for each missing data method differed substantively. Results are discussed. (Contains 4 figures, 3 tables, and 20 references.) (Author/SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A