ideas for analyzing associations between items from two different scales

[pji]

Hello all, hope everyone is safe and healthy.

I am curious about ways to analyze associations between items from two different scales.
We have scale 1 that has 20 items and scale 2 that has 25 items and we were curious to see if certain items were associated across the two scales.
For example, something like:
Scale 1 Scale 2
Item 1 <-----------> Item 1  (item 1 and item 1 on scale 1 tend to correlate on scale 2)
Item 2 <-----------> item 2 (item 2 and item 2 on scale 1 tend to correlate on scale 2)
Item 3 }____________> item 3 (item 3 and item 4 on scale 1 tend to correlate with item 3 on scale 2)
Item 4} item 4 (item 4 on scale 2 doesn't correlate with anything)

A correlation matrix would a bit cumbersome to determine if groups of items also correlated with a single or another group of items.
I tried using a principle components analysis but the rotation just gives me the two scales as factors, so I don't rotate them, but didn't get much in the way of results.
I was thinking of multi-dimensional scaling, but SPSS didn't like the setup indicating that there too few cases.
I was thinking about canonical correlation but I'm not familiar with that analysis.

Ideas? Thoughts?
Much appreciation in advance.


Peter Ji, Ph.D
Associate Professor, Clinical Psychology Department
Adler University
312-662-4354
[hidden email]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Enzmann]

I would use a SEM with two correlated factors of the respective scales
and investigate the modification indices for cross-loadings and/or error
terms between items.

Best,
Dirk

=====================================
Prof. Dr. Dirk Enzmann
Institute of Criminal Sciences
Dep. of Criminology
Rothenbaumchaussee 33
D-20148 Hamburg
Germany

phone: [phone] (office)
        +49-40-42838.4591 (Mrs Billon)
fax:   [phone]
email: [hidden email]
https://www.jura.uni-hamburg.de/die-fakultaet/personenverzeichnis/enzmann-dirk.html
=====================================

On Wed, 25 Mar 2020 03:00:28 +0000, "Ji, Peter" <[hidden email]> wrote:

> Hello all, hope everyone is safe and healthy.
>
> I am curious about ways to analyze associations between items from two different scales.
> We have scale 1 that has 20 items and scale 2 that has 25 items and we were curious to see if certain items were associated across the two scales.
> For example, something like:
> Scale 1 Scale 2
> Item 1 <-----------> Item 1  (item 1 and item 1 on scale 1 tend to correlate on scale 2)
> Item 2 <-----------> item 2 (item 2 and item 2 on scale 1 tend to correlate on scale 2)
> Item 3 }____________> item 3 (item 3 and item 4 on scale 1 tend to correlate with item 3 on scale 2)
> Item 4} item 4 (item 4 on scale 2 doesn't correlate with anything)
>
> A correlation matrix would a bit cumbersome to determine if groups of items also correlated with a single or another group of items.
> I tried using a principle components analysis but the rotation just gives me the two scales as factors, so I don't rotate them, but didn't get much in the way of results.
> I was thinking of multi-dimensional scaling, but SPSS didn't like the setup indicating that there too few cases.
> I was thinking about canonical correlation but I'm not familiar with that analysis.
>
> Ideas? Thoughts?
> Much appreciation in advance.
>
>
> Peter Ji, Ph.D
> Associate Professor, Clinical Psychology Department
> Adler University
> 312-662-4354
> [hidden email]
>

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD