Whether these tests are necessary to report in case of Confirmatory Factor Analysis.What is the difference between these two and should we report both in case of CFA while testing Common method bias.
Ramendra,
Harman’s single factor test is one technique to identify common method variance. In EFA one examines the unrotated factor solution to determine the number of factors that are necessary to account for the variance in the variables. If a single factor emerges or one general factor will account for the majority of the covariance among the measures then it is concluded that a substantial amount of common method variance is present. However, this is an exploratory method and not a statistical test, therefore is should not be used.
A better method for testing whether a single (global) factor accounts for the majority of the variance in the variables or not is CFA. This method provides a chi-square test so that it is possible to judge whether the model fits the data or not.
The single factor test (EFA and CFA) is actually not a good test for common method variance. If by means of a CFA model a single factor emerges, then one cannot be sure that this factor comprises actually method variance. Instead, it could be a factor measuring a single trait. Therefore, it would be necessary to include in your model at least two different methods, for example, self-report and peer-report. Then, these methods can be modeled as latent factors. The advantage of allowing the indicators of the constructs of interest to load onto the trait factor as well as the method factor is that the impact of method factors can be estimated.
This is very nicely explained in this article:
psych.colorado.edu/~willcutt/pdfs/Podsakoff_2003.pdf
HTH, Karin
Very clear explanation. If you really want to test the effect of method factors, you need to include different method factors in addition to latent variables. See the attached articles for more information.
Article Multitrait–multimethod matrices in consumer research: Critiq...
Article Assessing Construct Validity in Organization Research
to run it on spss, watch this video
https://www.youtube.com/watch?v=bIw6DbvOlUs
Hi,
If there is an issue of common method bias (single factor extracted is more than 50%), then what is the solution to deal with it?
There are some remedies presented in this paper:
http://www.jwalkonline.org/docs/Grad%20Classes/Survey/articles/required/Podsakoff%20-%20common%20method%20biases.pdf
where do we need to use Harman's method? is it in CFA or in EFA
Mainly I used it in EFA, put what I understand from Karin's response that we should not used it.
can you explain more
Recommend reading this:
Kock, N., 2015. Common method bias in PLS-SEM: A full collinearity assessment approach. International Journal of e-Collaboration (ijec), 11(4), pp.1-10.