In my previous post I explained how I had gone about customizing the Assignment Module. In this post I’ll show you how I visualized the data and explain briefly what I did.
I started out by setting a few properties. I did this just as a baseline, knowing that my client may change them.
The first properties I set were the above and below values. I knew that there would be three groups of Users. The first group would be all the Users below 49%. The second group would be between 50% and 80%, and the last group would be 80% and above.
The second property I set was for the number of random Users to select in each of the three groups. I set this value to 2 because I only had 22 Users.
The first image shows three things:
- A count of the number of Users in each of these three groups.
- The lowest, average and highest grades achieve.
- Six Users, two from each group that have been randomly selected for moderation.
So far that’s pretty easy, and a few simple queries will get the job done.
Next I plotted these counted values (10, 9, 3) on a line chart. It wasn’t really necessary, but it’s nice to show the data visually.
The blue line showed an actual count of the Users in each of the three groups. The red line showed the desired count – more Users in the middle group. I obtained the red line by grading the students on a curve, and recounting the number of Users in each group.
Adjusting the grades:
The first thing I did was set the desired minimum and maximum values.
Once those were set, I used the formulate from the wikihow site, and regraded each student. (http://www.wikihow.com/Curve-Grades)
f(x) = y1 + ((y2-y1)/(x2-x1)) (x-x1)
Actual vs Adjusted
The graph below shows the initial grades in the Column Chart and the adjusted grades in the Line Chart. The red indicating all Users below 49%, while the green showing all Users above 79%.
The adjust graphs in the second and third images as really a ‘what if’, allowing the Moderator to see what the effect of a grade adjustment would look like.
My property values are fixed at this stage, but a slider may be ideal. This would allow for a real-time update of the data as the slider is moved along.