SPSS is treating marital status as an interval variable. You should tell SPSS that it is categorical.

Daniel Wright

Thanks a lot for your answer, but how can I tell SPSS that is a categorical variable?

Syed Azamussan

Thanks a lot, dr.

What is the type of regression give me the beta coefficient for each category?

I don't use SPSS, but I assume there is something to tell it that the variable is categorical. Alternatively, just create dummy variables (with the value 1 for 1 category and 0 for the others) for all but on of the values, then enter these into the regression.

The REGRESSION command in SPSS treats all variables as quantitative variables, so if you wish to include a categorical variable with k categories, you have to compute k-1 indicator variables. (Of course, you can use some other coding scheme if you wish). The table of coefficients from REGRESSION does include standardized regression coefficients (with no way to suppress display of them, AFAIK).

You can use either UNIANOVA or GLM to estimate the same model. These commands do allow you to treat categorical variables as "fixed factors", meaning there is no need to generate your own indicator variables. The table of coefficients is not displayed by default, but there is an option to ask for it. (I think it is /PRINT=PARAMETER, or something like that). And IIRC, UNIANOVA/GLM does not have an option to display standardized regression coefficients.

HTH.

• https://www.ibm.com/support/knowledgecenter/SSLVMB_26.0.0/statistics_reference_project_ddita/spss/base/syn_unianova.html
• https://www.ibm.com/support/knowledgecenter/SSLVMB_26.0.0/statistics_reference_project_ddita/spss/base/syn_regression.html

Bruce Weaver

Many thanks, dr.