I have a Stata dataset that was exported from SurveySolutions. The survey contains a Yes/No multiple selection question like in question 4.11 in the below image. For each item on that list I have a separate variable that contains the value 0 or 1.
The PDF of the instrument only states “multi-select: yes/no” but does not explain the mapping to the numeric codes used in the dataset. I need to know whether 0=Yes or 0=No.
Note that for questions like 4.13 in the below image, the PDF clearly states the mapping, i.e., 1=Yes and 2=No.
More generally could you add the mapping for the multi-select Yes/No questions to the PDF output?
Since the codes for ‘yes’ and ‘no’ are not user-changeable - they are not part of this output.
You can assign different numeric codes, like 88 to items, that’s why they are shown in the PDF preview, and you can also assign different labels to the categories, that’s why they are also shown in the PDF preview, but you can’t change how the program exports the data hence 0 and 1 are not shown there for ‘no’ and ‘yes’ options.
Note also that if the question is ordered, the 0/1 notation is no longer true, so take caution.
Could you let me know what the mapping is for these kinds of questions? Is it that 0=Yes or 0=No?
The only values I see are 0 and 1. I just don’t know which of them means Yes or No.
See the Export section on multi-select questions here
The quick answer: Yes = 1, No = 2.
Thanks for your response, much appreciated.
I initially also thought it would be Yes=1 and 2=No because that appears to to be the standard for single yes/no questions, as in question 4.13 in my initial screenshot. But that does not appear to be the correct mapping for multi-select yes/no questions.
Instead, the screenshot for the multi-select yes/no in the link you shared suggests that it is Yes=1 and No=0. Unanswered questions have a missing value. I repost those screenshots below.
I see 1’s and 0’s in my data (there are no values=2) so I’ll go with that.
(Sorry, can only post 1 image at a time)
provided that you have a simple multiselect question with y/n mode and no further tweaks to it, the data will be exported to dummy variables named mcq__optioncode1 mcq__optioncode2 etc, where mcq is the variable for the multiselect question entered in the designer, and optioncode1, optioncode2 are numeric integer codes of each option of the multiselect.
the values in these dummy variables are:
0 - negative response selected on that item
1 - positive response selected on that item
ext missing - no response was selected on that item (.a in Stata, -999999999 in other formats).
system missing - question didn’t apply to this case (suppressed by the logic, should occupy all mcq__* variables of the same row of data).
NB: Other types/options of multiselect may change how it saves the data.