SPSS Statistics

 View Only

 Grouping Likert Scale Variables

Jump to  Best Answer
Aidan Byrne's profile image
Aidan Byrne posted Fri December 06, 2024 06:16 AM

I’m looking to group Likert scale variables in SPSS and then carry out regression analysis on the grouped variables.

I came across a procedure using the SPSS menu path : Transform / Compute Variable / Name New ‘Target Variable’ / Select Function Group ‘Statistical’ / Select Functions & Special Variables ‘Mean’.

Is this the best procedure to use or might there be a better alternative ?

Any help would be great.

Regards,

Aidan Byrne

Dublin

Ireland

Robert Walker's profile image
Robert Walker posted Fri December 13, 2024 02:25 PM  Best Answer

So, OK, create your six categories (Values, Understanding, Enhancement, Career, Social, Protective) using the 30 VFI items (COMPUTE). Run regression using your six new variables as independents (but, what is the dependent measure, some measure of volunteer activity?) And then run the same analysis using the 30 individual items as independents to see which individual item has the most explanatory power.

Jon Peck's profile image
Jon Peck posted Fri December 06, 2024 02:09 PM

What do you mean by

"carry out regression analysis on the grouped variables."

The Compute dialog you described is the general transformation command, but how do you want to combine these?  You might use them  all as separate regressors, or you might aggregate using something like the median across the variables, or something else.

Robert Walker's profile image
Robert Walker posted Fri December 06, 2024 03:39 PM

Hi Aidan,

Sounds like you have a number of dependent measures that you want to use to create one summary dependent measure? You have a few options. Since they are Likert scales, you could simply add them together to come up with an aggregate score and use that as a dependent measure. Alternatively, you could take the mean of these measures. Alternatively, you could take the maximum across your multiple dependent measures. Of course, these all assume that all dependent measures carry an equal weight in creating your aggregated dependent measure.
Cheers,
Bob Walker
Surveys & Forecasts, LLC
Aidan Byrne's profile image
Aidan Byrne posted Sat December 07, 2024 04:17 AM

7/12/24

Hi Jon & Robert,

Thanks a million for your swift replies.

Here is a bit more detail as to what I’m trying to achieve in SPSS:

a)      Clary, Snyder & Stukas (1996) devised the Volunteer Functions Inventory (VFI) to assess motivations for volunteering. I am replicating their data analysis.

b)     The VFI uses 30 questions to gauge volunteer motivation.

c)      The VFI argues that these 30 questions can be grouped into 6 motivation categories.

d)     I have 30 Likert scale questions with 7 data points each, which I need to group into the 6 motivation categories.

e)     I need to know the best SPSS procedure to achieve (c) above, which would also allow further analysis without distorting the results that analyses of the individual questions would deliver.

f)       I applied the Transform / Compute Variable procedure to my data. This resulted in the creation of 6 new ‘Motive’ variables representing the VFI motivation categories. These categories are based upon the mean of the responses to the grouped questions.

g)      My question is: Can I rely upon the above process, using the mean, and these new ‘Motive’ variables to conduct further regression analyses or is there a more highly recommended way to group Likert scale variables to achieve (e) above ?    

For whatever tests I use, I’m thinking of running those tests on the 30 individual Likert scale questions AND running those test on the 6 grouped variables as a control to check if the grouping procedure has generated any distortion in the expected results.

Any guidance you might have would be a great help.

Regards,

Aidan

Aidan Byrne's profile image
Aidan Byrne posted Sun December 15, 2024 02:50 AM

Hi Robert,

That is very helpful.

Thanks a million.

Regards,

Aidan