Dear all,

I would like to view the interaction effect between education level and age while trying to predict their health status (continuous variable) with the Oaxaca-Blinder decomposition approach (linear model).
Education level is measured by 3 dummy variables for lower than high school degree, high school degree and university degree. Age is measured by 8 age groups: 25-29, 30-34, 35-39, 40-44, 45-49, 50-54, 55-59 and 60-64.

I am unsure how to measure this interaction. Options I see are:
1) 3x8 interaction, but this seems over-complicated.
2) Interacting the two categorical variables directly as follows: educationlevel##age, but this is not allowed in the oaxaca command
3) Treating the age variable as a continuous variable and writing the code as follows: normalize(lessthanhighschool highschool university) normalize(lessthanhighschoolxage highschoolxage universityxage # age), but this does not seem technically correct.

What do you think?