Survey research and design in psychology/Tutorials/Psychometrics/Composite scores

This tutorial explains how to create composite scores in SPSS, using the compute function.


 * 1) Before creating composite scores, consider conducting exploratory factor analysis and/or analyses of internal consistency to help identify the extent to which the items you are wishing to combine are correlated with one another.
 * 2) Use the Compute function in SPSS (via syntax or pull-down menus) to create unit-weighted composite scores. Alternatively you can create (regression-weighted) composite scores via exploratory factor analysis (not shown in this tutorial).
 * 3) Obtain descriptive statistics (M, SD, Skewness, and Kurtosis) e.g., via Analyze - Descriptive Statistics - Descriptives - Options - Distribution - Kurtosis and Skewness

Exercise 1: Openness
Data file: data_15_1.sav Allen and Bennett 15.2, pp. 209-221

COMPUTE Openness=MEAN(Open1,Open2R,Open3,Open4R,Open6R,Open7,Open8R,Open9,OPen10R). EXECUTE.
 * 1) Create a composite satisfaction score as the average of responses to the nine items: Open1, Open3, Open7, Open9, Open2R, Open4R, Open6R, Open 8R, Open10R (where the R indicates a recoded item).
 * 2) Open5 is dropped because analyses of internal consistency indicated that it didn't belong (it has negative and low correlations with other items and the Cronbach's alpha increases when it is removed). Why do you think it doesn't belong? (examine the wording of the item - "liberal" has different meanings in different contexts, e.g., in Australian politics, the Liberal Party is right-wing, whereas in North American politics, liberal means left-wing)
 * 3) Open7 (carry conversation to a higher level) could be dropped or included - it has small positive correlations with the other items and the Cronbach's alpha would basically stay the same whether it was included or dropped

See extra exercises