[New Quizzes] Improved Statistics View for Numerical Questions
When looking at the quiz statistics, an instructor should easily be able to tell the spread of answers for a quiz question to know how to tailor learning. In multiple choice, you see how many learners (and who) chose each option, informing teaching strategy. For numeric questions, it shows the number of people with the exact correct answer, and the number of 'other answers', which is not helpful in quickly assessing classroom mastery. If the numeric question has a margin of error, some of those 'other answers' may also be correct, skewing the results unfavorably. A possible workaround is the item analysis report, but this report does not populate for numerical questions, so there is no meaningful way to analyze the effectiveness of these questions or see class understanding at a glance (or even at more than a glance, since the report doesn't help either).
Problem: there can be all sorts of answers, obviously we can't list every possible answer a student typed in the box and list how many students chose it, the list could become uselessly long and bog down loading of the page
Five categories for each numeric question with margin of error, three for numeric questions without.
For with margin of error:
1. Lower than range
2. Lower, but in range
3. Correct
4. Higher, but in range
5. Higher, but out of range
For straight numeric -
Lower than range
Correct
Higher than range
This gives a limited number of easy to predict options, while still giving teachers significant insight into how their students are performing at a glance. In many cases, the direction that a student is incorrect is just as important as that it was incorrect. This would bring numeric quiz statistics into line with the insights able to be gleaned from other question types from the quick view of the quiz statistics page.
admin
