What thresholds should I use for factor loading cut-offs?
Hair et al give rules of thumb for assessing the practical significance of standardised factor loadings as denoted by either the component coefficients in the case of principal components, the factor matrix (in a single factor model or an uncorrelated multiple factor model) or the pattern matrix (in a correlated multiple factor model).
On the other hand Field (2005) advocates the suggestion of Stevens (1992) to use a cut-off of 0.4, irrespective of sample size, for interpretative purposes.
Hair et al. (p112) Table of Loadings for Practical Significance
Factor Loading |
Sample Size needed for significance |
|
0.30 |
350 |
|
0.35 |
250 |
|
0.40 |
200 |
|
0.45 |
150 |
|
0.50 |
120 |
|
0.55 |
100 |
|
0.60 |
85 |
|
0.65 |
70 |
|
0.70 |
60 |
|
0.75 |
50 |
|
References
Field A (2005) Discovering statistics using SPSS. Second edition. Sage.
Stevens JP (1992) Applied multivariate statistics for the social sciences (2nd edition). Hillsdale, NJ:Erlbaum.