Hi, I have a question in my survey asking about race/ethnicity. The respondents can choose either from one of the given options or select 'other' and type in their answer.
The data for this question in spss is presented as two variables:
- one variable has all the given answer options and the 'other' option as missing data
- the second variable is a string variable with only the written answers for the 'other' option and the rest is shown as missing data.
Apart from doing it manually is there a way to merge these two variables to make a single variable for race/ethnicity?
Hello Astha,
I would inspect whether the string variables have a substantial number of assignments that you could code as a further category of the categorical variable. The rest will simply be represented as mixed, miscellaneous, or rest with--of course--the unfortunate implication to not be very informative.
Perhaps you can present a count / frequency table of the categorical variable and the string variable. The goal should be to have no unnecessary missings.
Best
--Holger
In line with what Holger Steinmetz said, you need a strategy that balances between finding any categories that are frequent enough that they should be included, versus creating too many small categories. In the end, this is likely to require doing things by hand.
One option might be to create a variable that contains only the string values for "other" (everything else is treated as missing), and then sort that variable alphabetically to see where large groups of responses occur.
Because you have mentioned 'Other - please specify', you need to include all respondents specified categories in your classification and analysis. Good luck.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey