Download study survey data

Export result of a study in CSV format. One line per response, one CSV per study.

study, test, group, item, user, type item (gap|mcq|...), response, correct, response time

assigned to @dvanrossum

@bridubois For this issue you advise me to look at the way this was done to download session data. I tried it, but the url doesn't work. If I add a v1 in the url it's fixed. Is this a fix that makes sens to you, or this is a bad work around the new way we deal with call to the backend api?

The API format is /tmp-api/v1/..., so it makes much sense. However, depending on where you do your queries, you would probably use /api/... instead on the frontend side (in lib/api), because the mapping to /tmp-api/v1/... is done by the api proxy (defined in hooks.server.ts). This logic will change in the consolidation work (the "tmp" was there to be ... temporary ... which is not really the case ), to increase the consistency, but the actual move of the API isn't done yet.

Thank you for the explanation !

added Studies & tests To do labels

@sbibauw Is this the expected format for the csv for the studies ? I wasn't sure if you wanted the test id and group id rather than the title, but this can be change really easily.

@bridubois I have a little issue, I can't seem to figure out how to get which entry is related to which study. I know which test is part of a study and which user. But if there are two studies with the same user and a test is also present in both, I don't know to which study the entry correspond. Could I maybe add a study_id column in the tests_entry table ? Or is there another relation between the tables I didn't find ?

You can definitly add this column, I completly forgot it. There is also the "code" given by the user, and a "rid" (random id) that change at each trial. But I don't know if they need to be part of the export.

(user) code is needed, yes. rid is not needed to export (but indeed necessary to match certain tables)

Thanks @dvanrossum! Looking great!

I realized I forgot to share my notes with you. I just added them to Teams and here is the section about the CSV export:

CSV long format ← prioritaire

• for study Y
• 1 line per response-item
• header (check if I didn't forget sth):
  • (study_id,) test_id, group_id, item_id, user_id, item_type (gap|mcq|...), response, correct (0/1), response time

Concretely:

• test_id rather than title (only used for display, not unique enough)
• great idea to add the group_name, so keep that one, but also add group_id
• user_id: OK but also include user_code
• correct, use 0|1 (rather than false|true)
• try avoiding spaces in headers (user_id, item_type, response_time...)

This is the modified version with your feedback :

created branch 170-download-study to address this issue

mentioned in merge request !40 (merged)

closed with merge request !40 (merged)

mentioned in commit 549fe1d9