Special values for questions other than “Numeric” have been discussed here before.
Here is a (new) case for special values in categorical questions:
Eurostat is now requiring additional options (-1 = Don’t know, -2 = Not applicable) in ALL categorical questions. Adding these as special values instead of additional options would improve readability of the questions, simplify coding the questions in Designer and facilitate verification that these options are present in all categorical questions.
Observation: Users implementing Eurostat surveys in Survey Solutions are on the increase.
As a matter of fact I didn’t find any “Don’t know” category (as I don’t equate “Not stated” with “Don’t know”). Quite a few are not including “Not applicable” category as well.
Answer categories: Wording and codes for possible answers for the question. Besides these answer categories, other codes are to be used but are not mentioned in the further descriptions of questions/variables:
• “-1” for missing values (don’t know, refusal);
• “-2” for not applicable questions (filter); and
• “-3” for the use of proxy answers (only for certain variables)
• “-4” for the optional variables when not included in the questionnaire
These codes are not specified explicitely in the questions’ answer categories, but in many cases they are mentioned in filter expressions (enabling conditions), e.g.:
FILTER
Interviewer instruction: Next question (PL2) is to be asked only for respondents who are not blind
(codes 1, 2 or -1 in PL1).
Hello @klaus , I see the document you’ve mentioned specifically says (page 9):
The manual should serve as a handbook for planning and implementing EHIS wave 4 in EU and EEA member states.
This recommendation for implementation of a particular survey is not the same as:
which is a much more generalized statement.
Survey Solutions currently allows accommodating the categories listed by @klaus (with negative codes) and quality control can be implemented on the questionnaire document to make sure that it satisfies certain requirements, (like presence of specific categories), but I see it as rather unlikely that a single set of options will satisfy all users and apply to all surveys.