Hello, I have a question concerning the procedure compute in SPSS. How can I create an overall scale “perceived injustice”, combining the subscales perceived personal and perceived group injustice?
You may use Data / Transform / Compute... from the main menu in spss.
At the top you'll see two boxes. Enter the desired target variable name for the overall/summated score in the left-hand box, and the arithmetic statement combining the relevant subscores in the right-hand side.
For example:
Overall (=) Personal + Group
(assuming that the two constituent variable scores are "Personal" and "Group", and the desired target score is "Overall").
In ordinary syntax (don't forget the period at the end):
if you do not want to sum the scores up to form an index (as David Morse has explained excellently how to do that in SPSS) but rather want to have an average score (either for each dimension or combining both dimensions) you can do the same procedure, using the following steps:
Compute the average scores for "Personal" and "Group" separately by first summing all items for each variable and dividing the sum by the number of items.
Sum up the two average scores and divide the overall score by 2 (the number of variables).
Alternatively, and to account for differences in explained variance within the two variables, you could also run an exploratory factor analysis to see how the items from the two scales relate to each other.
One concern that I have when you just combine both average scores is that you assume they share the same properties, even though they cover two different facets of a similar concept. There is good literature out there on how to use multi-factorial constructs in research that may be worth having a look at (e.g., Hair et al., "Multivariate Data Analysis", 7th edition, Pearson Publishers).
Let me know if this helps or if you have more questions.
Thanks for answering the question David Morse and Alexander Mafael
The problem is that I do not have the different items, only the two metric subscales. I am conducting a secondary data analysis. Perceived personal injustice and perceived group injustice are two high related variables.The two subscales have a high correlation, hence I want to merge them into one overall variable. In addition I standardized the overall scale. Now the thing is that when I'm conducting a GLM analysis, the two subscales do not have a significant effect, but the overall scale does so. This is the reason why I doubt the correctness of the merge.