Imply the late penalty more fairly

[hxue]

Currently in Canvas, late submission penalties are applied based on full days, and the system rounds partial days up. For example, a submission that is 1.01 days or 1.99 days late is both rounded to 2 days late.

This approach can feel unfair to students when they are only slightly over the 1-day mark.

Is there a way to implement fairer rounding, such as:

• Round 1.49 days or less to 1 day

• Round 1.5 days or more to 2 days

Or alternatively, allow institutions to customise the rounding threshold for late penalties?

Keen to hear if others have encountered this, or if there are any existing workarounds or planned improvements in this area.

[Gabriel33 · Accepted solution]

The best I can suggest is to set the penalty by hour instead, but that may limit you to multiples of 24% of penalty per day.

[Gabriel33 · Accepted solution]

The best I can suggest is to set the penalty by hour instead, but that may limit you to multiples of 24% of penalty per day.

[chriscas · Accepted solution]

I agree with this.  As much as more flexibility with the late penalties could be a good thing, I also know there may be drawbacks as well.

Since I'm a coder at heart, I often think about the extremes/edge cases with something like this.  While having some kind of rounding sounds good in my head when thinking about your example of 1-2 days, if that login were applied universally, something that was 0.49 days late would be rounded to 0, and I'm pretty sure most instructors would disagree with that.  That gets into special exceptions being made for that case, and thinking about that further, it would mean there would essentially be a 36 hour period where the same late penalty would apply when the setting says would say "1 day" so that would cause confusion too...

I know people often don't like to dive too far into the details like this, but it's something Instructure definitely has to do with everything they code.  I don't always agree with their final decisions, but I know there is usually some kind of method to the apparent "madness" we sometimes observe as end users.

-Chris