Feature Request - New ways to grade a Peer Instruction question

[barbarer]

Currently the student gets a point or part of a point for each vote for a Peer Instruction question. Instructors (Duke, Virginia Tech, Michigan, Berea) would like the ability to round up or down.

[bnmnetp]

Lets try to clarify what people really want. Currently we have two levers:

autograde: either peer or peer + chat
which to grade: must be All peer votes

This made total sense to me when I wrote the code, as this is the only instance where we were grading a single question on whether and when they answered the question.

The usual options for which to grade are: first answer, best answer, last answer
The usual options for autograde are: Manual, All or nothing, Percent Correct, interact

I feel like round up or down is a distractor. What are we really trying to get at?

Maybe I had it backwards in that the which to grade should be peer or peer + chat

if which to grade was peer and autograde was all or nothing then the student would get 0 points if they only voted once. If autograde was percent correct then they would get 50% for vote 1 and 100% for vote 2.
if which to grade was peer plus chat and autograde was all or nothing then the student would have to do all three. If autograde was percent correct then we would assign 33% for vote1, 66% for vote1 and chat, and 100% for all three.

I'm not inclined to get into rounding up or down as that is fraught with where we round and who agrees with how we round and for questions worth more than one point how far do we round... If someone makes a question worth 1000 points and they have a 33% score for that question what do we round up to? We would not even have to round it could be a nice even 330.

[barbarer]

How about what to grade being: single answer, two answers , two answers and message? How to grade could be interaction - all, interaction - round down, or interaction round up. Dr. Barbara Ericson Associate Professor, School of Information University of Michigan

…

On Fri, Nov 7, 2025 at 8:23 PM Bradley Miller ***@***.***> wrote: *bnmnetp* left a comment (RunestoneInteractive/rs#1058) <#1058 (comment)> Lets try to clarify what people really want. Currently we have two levers: *autograde*: either peer or peer + chat *which to grade*: must be All peer votes This made total sense to me when I wrote the code, as this is the only instance where we were grading a single question on whether and when they answered the question. The usual options for *which to grade* are: first answer, best answer, last answer The usual options for *autograde* are: Manual, All or nothing, Percent Correct, interact I feel like round up or down is a distractor. What are we really trying to get at? Maybe I had it backwards in that the *which to grade* should be peer or peer + chat if *which to grade* was peer and autograde was all or nothing then the student would get 0 points if they only voted once or did not participate in the chat. If it was percent correct then we would have to define maybe 33% for vote1, 66% for vote1 and chat, and 100% for all three? I'm not inclined to get into rounding up or down as that is fraught with where we round and who agrees with how we round and for questions worth more than one point how far do we round... — Reply to this email directly, view it on GitHub <#1058 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AKOZ7M2G5JBWBGKQLBXLBGT33VAZTAVCNFSM6AAAAACLERNMXOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTKMBVGU2TAOBYHA> . You are receiving this because you authored the thread.Message ID: ***@***.***>