I have a 37-question scale, where each item belongs to one of four scales.
The items for each scale are interspersed.
How can I compute a scale score for each of the four scales?
I need to be able to analyse each subscale independently and then compare scale responses on and independent variable.
Here are two posts where I describe the process of computing scale scores for multiple-item, multiple-scale tests:
There are many things to consider when creating scales (e.g., should the items be equally weighted? do you have missing data? do you want a mean or sum? etc.), and there are several tricks for doing it efficiently and reliably (e.g., using loops, using syntax, automatically generating syntax from metadata), but the posts above describe this in detail.
This code divides the sum of scores (or of weighted scores) by the number of valid values a given case has for that set of variables.
In this example, if a case has values only for verbal and quant, then the score will be simply the sum of those divided by 2.
Compute SCORE = sum (.05* jobs, .15* exper, verbal, quant, intvw, college, gpa)/
nvalid (jobs, exper, verbal, quant, intvw, college, gpa).
Execute.
The benefit of using this over simpler syntax is that cases will not be left out of the SCORE even if they are missing one or more values on the component variables.
Your question seems quite easy, I hope I have not misunderstood. If you work in SPSS it is suggestble to calculate the mean of every subscale.
However, it would be better not to have items belonging to more than one subscale, as this could affect correlations and regressions you are going to run.
GP
