Skip to content

Interpreting fa.diagram from package psych

8 messages · David Winsemius, Luigi Marongiu, Ebert,Timothy Aaron

Luigi Marongiu
# 
Hello,
I have plotted data from exploratory factor analysis, and I got a
graph similar to FIGURE 11 (PAGE 36) of this link
file:///home/gigiux/Downloads/An_overview_of_the_psych_package.pdf
How do I interpret the figure? In particular, how do I know what the
colors represent?
Thank you
David Winsemius
#
On 9/10/22 14:08, Luigi Marongiu wrote:
This appears to be a link you a file on your personal device rather than 
an attachment.
Luigi Marongiu
# 
Sorry, the file was automatically downloaded and opened with the
browser instead of pointing to the webpage.
Here is a better link:
https://cran.r-project.org/web/packages/psychTools/vignettes/factor.pdf
The figure is on page 22.
The question is: The dots have different colors; how do I know what
they represent?
Is there a way to show an auto-legend?
Thank you
On Sat, Sep 10, 2022 at 11:33 PM David Winsemius <dwinsemius at comcast.net> wrote:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Ebert,Timothy Aaron
    

    

      
      #
    

  


  

      
It is a bad graphic as the legend that should explain the color coding is missing. The next option is to copy the data and code and see if you can reproduce the figure. You can then play with the code and read a bit about the procedures to figure out what is going on. It should not be too hard. My guess is that there is some additional variable with three states that is being used. If this were the iris data set I would guess it was the three species: setosa, versicolor, and virginica.

Tim

-----Original Message-----
From: R-help <r-help-bounces at r-project.org> On Behalf Of Luigi Marongiu
Sent: Sunday, September 11, 2022 3:02 AM
To: David Winsemius <dwinsemius at comcast.net>
Cc: r-help <r-help at r-project.org>
Subject: Re: [R] Interpreting fa.diagram from package psych

[External Email]

Sorry, the file was automatically downloaded and opened with the browser instead of pointing to the webpage.
Here is a better link:
https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fcran.r-project.org%2Fweb%2Fpackages%2FpsychTools%2Fvignettes%2Ffactor.pdf&amp;data=05%7C01%7Ctebert%40ufl.edu%7C9d0e49ff1aab4d5db2d308da93c3ac01%7C0d4da0f84a314d76ace60a62331e1b84%7C0%7C0%7C637984765998607455%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=2OxukCvxcx9xFJQiAInt2ulqC23qiBjQRPU128O%2Bjn8%3D&amp;reserved=0
The figure is on page 22.
The question is: The dots have different colors; how do I know what they represent?
Is there a way to show an auto-legend?
Thank you

        
On Sat, Sep 10, 2022 at 11:33 PM David Winsemius <dwinsemius at comcast.net> wrote:

            

      
      
      
    

        
--
Best regards,
Luigi

______________________________________________
R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstat.ethz.ch%2Fmailman%2Flistinfo%2Fr-help&amp;data=05%7C01%7Ctebert%40ufl.edu%7C9d0e49ff1aab4d5db2d308da93c3ac01%7C0d4da0f84a314d76ace60a62331e1b84%7C0%7C0%7C637984765998607455%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=KfsnfNKp3klZMJcvvuPD4gqI6fffn95FkpaPm8KyEnA%3D&amp;reserved=0
PLEASE do read the posting guide https://nam10.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.r-project.org%2Fposting-guide.html&amp;data=05%7C01%7Ctebert%40ufl.edu%7C9d0e49ff1aab4d5db2d308da93c3ac01%7C0d4da0f84a314d76ace60a62331e1b84%7C0%7C0%7C637984765998607455%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=7CZst6A76lzHADI4iRNJ%2FVI2%2FMOGLCAVnTL8aRLNjVA%3D&amp;reserved=0
and provide commented, minimal, self-contained, reproducible code.

  
  


  


      

    

    

      
      

        


  

        

      David Winsemius
    

    

      
      #
    

  


  

      
On 9/11/22 07:17, Ebert,Timothy Aaron wrote:

            

      
      
      
    

        
Pretty sure that guess is incorrect.

I'm not sure there should be a legend. The colors just indicate group 
membership derived from a mathematical process that has attempted to 
separate case into distinct groups that maximize the correlations within 
individual groupings. And therefore maximizes the distance separating 
the groups. The number of groups is specified in the function call. You 
should go to the earlier results and see if you can construct the 
groupings to maximize internal correlations. Psychometricians do this 
when they don't really have a theoretical basis for doing classification 
and are asking the data do it for them. If they are doing this on a 
questionnaire dataset, they often go back to the specific 
questions/answer pairings within groupings and try to assign meaning to 
them.? They then build post-hoc explanations and often do further 
studies to see if they can replicate the results and achieve some sort 
of stable synthetic construct.? It's a rather theory-free strategy and 
so trying to assign labels automatically would be difficult.

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Ebert,Timothy Aaron
    

    

      
      #
    

  


  

      
Ok, in looking at the code that makes more sense. The code specifies three groups, so there will be three colors. As the groups do not have meaning (hopefully supplied by the user at a later date) there is no legend. They are there to help the user see overlap between groups (none in this case).

-----Original Message-----
From: David Winsemius <dwinsemius at comcast.net> 
Sent: Sunday, September 11, 2022 11:50 PM
To: Ebert,Timothy Aaron <tebert at ufl.edu>; Luigi Marongiu <marongiu.luigi at gmail.com>
Cc: r-help <r-help at r-project.org>
Subject: Re: [R] Interpreting fa.diagram from package psych

[External Email]

        
On 9/11/22 07:17, Ebert,Timothy Aaron wrote:

            

      
      
      
    

        
Pretty sure that guess is incorrect.

I'm not sure there should be a legend. The colors just indicate group membership derived from a mathematical process that has attempted to separate case into distinct groups that maximize the correlations within individual groupings. And therefore maximizes the distance separating the groups. The number of groups is specified in the function call. You should go to the earlier results and see if you can construct the groupings to maximize internal correlations. Psychometricians do this when they don't really have a theoretical basis for doing classification and are asking the data do it for them. If they are doing this on a questionnaire dataset, they often go back to the specific questions/answer pairings within groupings and try to assign meaning to them.  They then build post-hoc explanations and often do further studies to see if they can replicate the results and achieve some sort of stable synthetic construct.  It's a rather theory-free strategy and so trying to assign labels automatically would be difficult.

--

David

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Luigi Marongiu
    

    

      
      #
    

  


  

      
I see, so the plot in itself is rather NOT informative -- I need to
make one myself after knowing what groups are there... Fair enough.
Thank you.

        
On Mon, Sep 12, 2022 at 1:42 PM Ebert,Timothy Aaron <tebert at ufl.edu> wrote:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Ebert,Timothy Aaron
    

    

      
      #
    

  


  

      
Not quite. What you see in the plot are three distinct groups. The outcome in your application may differ: the groups might be closer together with less clear group boundaries. If the method is successful, you should always ask why? What is the data trying to tell me, and with that information can I see the new pattern in the real world? Here is an example with less clear group boundaries: https://www.qualtrics.com/experience-management/research/factor-analysis/

Also, this is unsupervised learning. At the start of factor analysis you will not know that there are three groups. If you do know and the goal is to predict membership in a group then maybe something like discriminant analysis would be a better approach.

Tim

-----Original Message-----
From: Luigi Marongiu <marongiu.luigi at gmail.com> 
Sent: Monday, September 12, 2022 9:32 AM
To: Ebert,Timothy Aaron <tebert at ufl.edu>
Cc: David Winsemius <dwinsemius at comcast.net>; r-help <r-help at r-project.org>
Subject: Re: [R] Interpreting fa.diagram from package psych

[External Email]

I see, so the plot in itself is rather NOT informative -- I need to make one myself after knowing what groups are there... Fair enough.
Thank you.

        
On Mon, Sep 12, 2022 at 1:42 PM Ebert,Timothy Aaron <tebert at ufl.edu> wrote:

            

      
      
      
    

        

      
      
    

        
--
Best regards,
Luigi