Comparison with SPSS

[loukasilias]

Hello,

I am comparing the results of this library with SPSS and I observe that they are different.
Is it logical or is it a bug?

Thank you!

[desilinguist]

Hi @loukasilias, this library is mostly a direct port of functionality from R's psych library. There's no expectation of parity with SPSS.

[nachomaiz]

@loukasilias maybe your issue is related to #89 or #90 (I had a similar question in the past). There are also some Important Notes in the docs which might be relevant.

[JoelPasapera]

I have a similar problem and it is that when performing the factor analysis the factorial matrix of this library differs with the factorial matrix of SPSS.

Specifically, the library assigns factor loadings differently. This influences:

1. how the items are grouped,
2. how many items each dimension or construct has,
3. how the path diagram is structured.

could you put the default spss configuration in the library calculations ? or in any case specify which methods are used in spss and r studio to choose from ?

I think it would be a good idea to replicate the SPSS functionality as a standard for calculations. But in case you want to keep the R studio approach you could give the option to the user to choose to run the factor analysis with the default values instead of having to understand all the mathematics behind the factor analysis and do it manually.