forming summated scales from Likert-type items in SPSS

#1
Howdy, I have a data set of mostly Likert-type items. In particular, I have 18 4-point items that measure 3 constructs (7, 6, and 5 items per construct). The database has 278 respondents, although not every one answered every question. My question is how to get SPSS to provide mean and std dev on the summated scales, not just the individual items.
Then, what is the best way to analyze the data? I don't think anova is a good method (not sure it will meet the requirements), but what is?
Also, is anyone familiar with Borich's Mean Weighted Discrepancy Scores? I am calculating and analyzing these, but am not sure what stats to use. I think the MWDS is at least interval, if not ratio, so I could use regression. Maybe I need to use Ordinal Logistic Regression? I've never done that before.
Any help is greatly appreciated.
 

Lazar

Phineas Packard
#2
Depending on what version of SPSS you have you can impute missing values. Note sure if you know much about multiple imputations (MI) but it is a thorny issue and I would read into it before submitting it to your data.

In terms of forming scales:
COMPUTE scaleA = mean (item1, item2, item3.....item7).
EXECUTE.

DESCRIPTIVES scaleA.

and so on with all the scales of interest.

In terms of analysis cannot tell you unless we know what the research questions is.