01-18-2018 08:58 AM
Hello, so I am working with the Health and Retirement study dataset and I am trying to compare the proportion of males from the 2006 wave to the 2008 wave. These variables are gender06 and gender08 (male or female). I have not been able to figure out how to test whether or not the proportion of males are the same in both waves. I tried a chisq but since most observations have either gender06 or gender 08, this is not feasible as shown here: (I am also running into the same problem with race06 and race08, which has 3 different options)
proc freq data=ya; tables gender06 gender08 gender06*gender08 /chisq; tables race06 race08 race06*race08 /chisq; run;
01-18-2018 10:27 AM
I suggest that you go back into your processing and instead of naming variables by year to have an additional variable in the data to indicate year.
Then the chi-square with year*gender would make a lot more sense.