I believe that there is a fundamental error in an assumption made in Canvas that all questions of the types: missing_word, multiple_answers, and multiple_choice must have a correct answer and if they do not, the code turns the first answer into a "correct" answer - even though it is incorrect by using: "@answers.set_correct_if_none"

See app/models/quizzes/quiz_question/answer_parsers/

missing_word.rb
multiple_answers.rb
multiple_choice.rb

This is quite a disturbing finding since it means that the system overrides the setting of the answer_weight in the answers.

https://kth.test.instructure.com:443/api/v1/courses/2139/quizzes/2476/questions

Response Body

[
  {
    "id": 17983,
    "quiz_id": 2476,
    "quiz_group_id": 1393,
    "assessment_question_id": 16971,
    "position": 1,
    "question_name": "Q5",
    "question_type": "multiple_answers_question",
    "question_text": "<p>The question is xxxxx?</p>",
    "points_possible": 1,
    "correct_comments": "",
    "incorrect_comments": "",
    "neutral_comments": "",
    "correct_comments_html": "",
    "incorrect_comments_html": "",
    "neutral_comments_html": "",
    "answers": [
      {
        "id": 8617,
        "text": "first incorrect answer",
        "comments": "",
        "comments_html": "",
        "weight": 0
      },
      {
        "id": 8697,
        "text": "second incorrect answer",
        "comments": "",
        "comments_html": "",
        "weight": 0
      },
      {
        "id": 3447,
        "text": "first correct answer",
        "comments": "",
        "comments_html": "",
        "weight": 100
      }
    ],
    "variables": null,
    "formulas": null,
    "answer_tolerance": null,
    "formula_decimal_places": null,
    "matches": null,
    "matching_answer_incorrect_matches": null
  },
  {
    "id": 17984,
    "quiz_id": 2476,
    "quiz_group_id": 1393,
    "assessment_question_id": 16972,
    "position": 2,
    "question_name": "Q6",
    "question_type": "multiple_answers_question",
    "question_text": "<p>The question is xxxxx?</p>",
    "points_possible": 1,
    "correct_comments": "",
    "incorrect_comments": "",
    "neutral_comments": "",
    "correct_comments_html": "",
    "incorrect_comments_html": "",
    "neutral_comments_html": "",
    "answers": [
      {
        "id": 9963,
        "text": "first incorrect answer",
        "comments": "",
        "comments_html": "",
        "weight": 100
      },
      {
        "id": 9301,
        "text": "second incorrect answer",
        "comments": "",
        "comments_html": "",
        "weight": 0
      }
    ],
    "variables": null,
    "formulas": null,
    "answer_tolerance": null,
    "formula_decimal_places": null,
    "matches": null,
    "matching_answer_incorrect_matches": null
  }
]

The code that puts these two questions into the quiz is insert_multiple_answers.py'

The resulting quiz can be found at: sandbox-ik2554-ht17-quiz-export-turning-incorrect-to-correct.zip