in factor analysis when should i use the principal axis factoring method and when the principal components method.what's the diffrence between these two .how can i decide which one i have to use.
Principal components analysis (PCA) differs from the many versions of common factor analysis (CFA)--including principal axis extraction--in several important ways. PCA assumes that all variance is common variance; measurement error (error variance) and variance that is specific to an individual variable are presumed not to exist. CFA allows for all three sources: common variance, error variance, and specific variance, but only uses the common variance to help describe the factor structure. If there is an appreciable amount of error and/or specific variance, PCA results will not match those of CFA.
In general, when the number of variables being factored is low-moderate, PCA tends to yield higher estimates of variable-component relationships ("loadings") than does CFA. Thus, these would be perceived as biased from a CFA perspective. More important, the interpretation of salient variables and factors/components could vary.
PCA is frequently characterized as a data reduction framework, in which the goal is to capture as much of the total variance in a smaller number of independent components, thereby reducing the total number of variables and eliminating any concern over multicollinearity.
Here are two resources that help elaborate these distinctions:
Richard L. Gorsuch (1983). Factor analysis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
Bruce Thompson (2004). Exploratory and confirmatory factor analysis: Understanding concepts and applications. Washington, DC: American Psychological Association.