Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
bgsu1182784851.pdf (775.54 KB)
ETD Abstract Container
Abstract Header
Analogy Between Two Approaches to Separately Identify Specific Factors in Factor Analysis
Author Info
Wang, Jing
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=bgsu1182784851
Abstract Details
Year and Degree
2007, Doctor of Philosophy (Ph.D.), Bowling Green State University, Psychology/Quantitative.
Abstract
An observed variable may be decomposed into three components: one or several common factors, a specific factor, and a measurement error. Unfortunately, the specific factor is usually combined with the measurement error and treated as the unique factor in a factor analytic model (FAM) (Harman, 1976; Muliak, 1972). The confounding of parameters leads to several potential problems, such as underestimated reliability, biased estimation of the intercepts, slopes, and measurement error variances. There are two approaches to separately identify the specific factors, the first- and second-order FAM approach and the stationary longitudinal FAM approach, both by including several replicates for each variable from a classical FAM. The purpose of the dissertation is to fully discuss the analogy between the two approaches on separately identifying the specific factors and highlight the role played by the specific factors. First, the first- and second-order FAM approach is extended by including the mean structure and explicitly including the specific factor means and variances. Second, a special condition, i.e., equality of specific factor variances and covariances for the same variable, in the stationery longitudinal FAM approach is presented. To show the analogy between the two approaches on the model identification, all observed and latent variables are permutated to be variable ordered for the longitudinal model. After the permutations, we can easily see that the special condition allows us to build a second-order model on the specific factors, which shows the similarity between the two approaches. However, the first approach requires equal specific factor means for the replicates, while the second approach allows the specific factor means to change over time. From one illustration in a longitudinal FAM, the presence of the specific factors, exhibits themselves both in the mean structure, and in the variance/covariances structure. If the specific factors were omitted, the desired stationarity of the factor structure, i.e., equal intercepts and slopes, would be biased. Furthermore, neglecting the specific factors would lead to negatively biased reliabilities.
Committee
John Tisak (Advisor)
Pages
89 p.
Subject Headings
Psychology, Psychometrics
Keywords
SPECIFIC FACTOR
;
LONGITUDINAL FACTOR ANALYSIS
;
FACTORIAL INVARIANCE
;
SECOND-ORDER FACTOR ANALYSIS
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Wang, J. (2007).
Analogy Between Two Approaches to Separately Identify Specific Factors in Factor Analysis
[Doctoral dissertation, Bowling Green State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=bgsu1182784851
APA Style (7th edition)
Wang, Jing.
Analogy Between Two Approaches to Separately Identify Specific Factors in Factor Analysis.
2007. Bowling Green State University, Doctoral dissertation.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=bgsu1182784851.
MLA Style (8th edition)
Wang, Jing. "Analogy Between Two Approaches to Separately Identify Specific Factors in Factor Analysis." Doctoral dissertation, Bowling Green State University, 2007. http://rave.ohiolink.edu/etdc/view?acc_num=bgsu1182784851
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
bgsu1182784851
Download Count:
645
Copyright Info
© 2007, all rights reserved.
This open access ETD is published by Bowling Green State University and OhioLINK.