Point symmetry adjustment of phi-coefficients in the factor analysis of psychometric test items

Existing computer packages for data analysis are not generally sensitive to the underlying assumptions of factor analysis and issues concerning the appropriateness of data. One serious problem in the factor analysis of test items is that of response skew. This paper discusses some strategies for mit...

Full description

Saved in:
Bibliographic Details
Published in:Personality and individual differences 1992-02, Vol.13 (2), p.211-219
Main Authors: Hammond, Sean M., Lienert, Gustav A.
Format: Article
Language:English
Subjects:
Biological and medical sciences
Fundamental and applied biological sciences. Psychology
Psychology. Psychoanalysis. Psychiatry
Psychology. Psychophysiology
Psychometrics. Statistics. Methodology
Statistics. Mathematics
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Existing computer packages for data analysis are not generally sensitive to the underlying assumptions of factor analysis and issues concerning the appropriateness of data. One serious problem in the factor analysis of test items is that of response skew. This paper discusses some strategies for mitigating this problem and emphasizes the use of alternatives to the Pearson-Bravais inter-item correlation coefficient. In particular, it is shown that the Holley-Guilford G-index of association may be interpreted logically as a coefficient of correlation between two binary items, dichotomised at the median of their underlying continua, following a point symmetry adjustment. The G-index is a point symmetry adjusted phi-correlation coefficient. It is argued that principal components analysis of the inter-item matrix of G-indices is less distorted than that of phi-coefficients where parametric assumptions are in doubt.
ISSN:0191-8869
1873-3549
DOI:10.1016/0191-8869(92)90044-P