View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|05668||Development||Import/Export||public||2012-01-16 17:50||2012-12-16 19:44|
|Target Version||Fixed in Version|
|Summary||05668: Easier support for repeating groups|
|Description||Now that EM supports group-level relevance, we can make it easier to support repeating groups where you want to ask the same general set of questions for each of several people, products, etc.|
Say you're trying to conduct a survey like the US Census, you need to do the following:
(1) Identify how many people in the household
(2) For each member of the household, ask a set of 10+ questions, which will have embedded conditional logic - like ask the person's age and gender and customize the follow-up questions based upon age/gender. Possibly even tailor the questions based upon those variables, name, etc.
Here's how we could make this easier in LS:
(1) Have person customize their group, with all desired conditional logic as relevance verified for the group
(2) Add option to SaveGroup to add a prefix/suffix to all question codes used in the Group. This would also add the same prefix/suffix to the same question codes used in tailor, relevance, or validation.
(3) Add option in Group Import to set replacement value for that prefix/suffix. That way, when you import the group, you automatically get newly named question codes with substituted prefix/suffix. This ensures that all of the relevance/validation/tailoring works within the newly imported group, but you don't have to modify the question codes, relevance equations etc.
(4) Then, all the author has to do is to update the group-level relevance to the desired value (like numCohabitants >= 5)
|Tags||No tags attached.|
This will be very hard to fit in current survey behavior.
David Olivier found a nice solution in my opinion: use sub surveys.
Adding a question type for this would be nice, for example
Question type: Multiple subsurvey entries.
Where a user would then be able add multiple records to another survey which will contain an identifier to the parent survey.
To implement this only a few changes are needed:
- We must add a column to the survey responses that keeps the parent response id and survey.
- We probably will have to improve the export functionality, on the other hand exporting vertical rows for this kinds of multi-dimensional data doesnt really work anyway.
||Passing all context data to a subsurvey would be a pain, as would normalizibf the data dor ecport to an analytic package.|