Questionnaire/vocabulary test page
If possible, modular questionnaire construction:
- Table "Test"/"ItemTest": defines a questionnaire/test as a combination of items (e.g., items ID 1, 11-40)
- Table "Item": defines a specific item, which can be:
- word (written) + 4-5 options (pictures)
- word (audio with , autoplay if possible) + 4-5 options (pictures)
- word (audio or written) + true/false (for now) for each, define which option == correct
behaviour:
- Auto move to next item after response selected (if not too complicated)
- No possibility to go back or forward (no correcting a question, no skipping)
- ResponseTime recorded for each item
participant response for each item is recorded as ( user_id, item_id, response_id, correct=T|F, responseTime )
=> This way, possible to build an adaptive vocabulary test with ELO ranking system (same ELO ranking system as for chess, but with each participant competing against each item -> increasing/decreasing each other "skill" ranking).
Activité
added To do label
Important note: after discussing this with Hubert, and realizing you were particularly efficient developing the typing test without requiring any other framework, I think it might be better NOT to implement SurveyJS at all after all, and rather develop each questionnaire internally.
If possible, modular questionnaire construction:
- Table "Test"/"ItemTest": defines a questionnaire/test as a combination of items (e.g., items ID 1, 11-40)
- Table "Item": defines a specific item, which can be:
- word (written) + 4-5 options (pictures)
- word (audio with , autoplay if possible) + 4-5 options (pictures)
- word (audio or written) + true/false (for now)
- Auto move to next item after response selected (if easy)
- ResponseTime recorded for each item
=> This way, possible to build an adaptive vocabulary test with ELO ranking system (same ELO ranking system as for chess, but with each participant competing against each item -> increasing/decreasing each other "skill" ranking).
added 1 deleted label and removed To review label
added To do label and removed 1 deleted label
assigned to @sbibauw
assigned to @bridubois and unassigned @sbibauw
One more thing to implement, if possible:
-
Compute item score for every question with a correct response given (
0|1) -
Compute mean score (
0.49) + display mean score (e.g., as percentage:49%) after finishing the test (this can be refined in the future, but at least we'll have the basic behaviour)
Modifié par Serge Bibauw-
Compute item score for every question with a correct response given (
added 1 deleted label and removed To do label
added To do label and removed 1 deleted label
mentioned in commit eddfb20f
Except the scoring part, everything should be working now, mainly thanks to eddfb20f and 686987b5.
The tables and inheritance has been rewritten, in a way that now surveys, groups and items are independent, and can be linked to each others, rather than trough a top-down dependency. That way, it's now possible to use multiple times an item in groups, and multiple times a group in surveys.
Also, a script (in /scripts/surveys) can convert a bunch of csv-like files into surveys, groups and items. To take advantage of the feature explained before, the data are split in three files:
- items.csv to create items.
- groups.csv to create groups, where items are just appended at the end of each line.
- surveys.csv to create surveys, where groups are appended at the end of each line.
Even if the script do some checks, it's important to double check that there is no duplicates in ID for each field, or it will fail (or not provide the expected output). It's also important to know that running again a script on the same survey/group/item, will first delete it, before recreating it. So, be cautious!
In
items.csv, blanks fields are replaced by files from static/surveys/items, with the structure<item_id>/<option_id|q>.<jpeg|mp3>. For now, only those extensions are supported, but others will be added in a future.To simplify development, questions have for now a static amount of options of 8, but the structure is done in a way that it would be possible to evolve to any number of options if it become necessary. Also, for now it's not possible to have multiple items sharing a same static item.
Finally, to configure the script, three variables have been added at the top of the script. The most important one is probably
API_URL, which is set by default for local use. In a future version, it would be possible to pass them trough CLI arguments. For security reasons, the email and password the user used to create/delete items/groups/surveys is asked during execution.
