Hi all, Hans and Kristine,

I have a slightly different way of handling multiple response questions that seems to work well for data entry, but is possibly a little more tricky to set up initially, and also you may want to convert it to a Boolean/dichotomous form afterward.

You will need as many variables as there possibly will be answers given to the question (so at most as many variables as possible responses). Using Hans' example of symptoms, you may call the variables symptom1, symptom2, symptom3, etc.

These variables do not correspond directly to responses, but instead each of them has a value list equal to all possible responses (i.e. if the five possible symptoms are coughing, sneezing, aching, bleeding and vomiting, then you will need all of these to be the value lists of each of the variables symptom1, symptom2, etc.)

The thing that makes this tricky to set up but then easier for data entry, is that you will want to set up the .chk file logic such that once the data entry operator has entered as many options as were ticked, they should skip over the rest of the variables so that they only need to enter as many values as were ticked on the form (plus one for the skip). Also, you may want to add some extra logic to check for duplicate entries which can get fairly complicated depending on how many variables you have.

I will put together an example of this as Epidata files - can I attach this to the list or somewhere else?

If anything I've said here is not clear and you would like more info I'll be happy to provide it.

Cheers,

Charity Jenkins Data Processing Officer Centre for Behavioural Research in Cancer The Cancer Council Victoria http://www.cancervic.org.au/cbrc

On 2/16/07, epidata-list@lists.umanitoba.ca epidata-list@lists.umanitoba.ca wrote:

Dear Kristine:

As one variable can take only one value, you must use as many variables as there are possible values. If you have 5 possible responses ("Do you have any of the following 5 symptoms?"), you need 5 variables, each with 3 possible values, "present", "absent", "unanswered". I don't see an easier way.

Regards,

Hans

Dear Epi-data users:

What is the best way to deal with multiple response questions?

Thanks, Kristine Hopkins Population Research Center, University of Texas at Austin