Factor Analysis with Restrictions on Weights and Loadings

[Philip Baker]

Dear All,

I am trying to replicate a paper to learn their methodology. In one specification, they require that the sum of factor loadings be equal to one. In another one, they require the sum of factor weights to be equal to one. I don't see any option in "factor" command to impose restrictions on the factor weight or loadings. I was wondering if there is any way I can impose such restrictions. I could also try to manually write a loop to do the optimization and find the factors under such restrictions, but not sure where to start. Any guidance would be highly appreciated.

Thanks

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

Are you doing exploratory or confirmatory factor analysis? This is easy in confirmatory factor analysis. I guess in exploratory you could change the variance of the unobserved factor to change the sum of the loadings, but I'm not sure why this makes sense or how to implement it. If you had the factor structure from exploratory, you might be able to use confirmatory with that structure, and constrain the sums of the weights etc., while allowing the variance of the unobserved factors to be free.