I decided to ponder some new methods of grading in school based on only rewarding students based on performance and the N-Effect.
Linear Rewards:
This system entails rewarding those who do better with more
extra credit (A minor amount).
For example, a simple design could be implemented would be to give one extra credit point for
every 10 points you earn on an assignment (Every assignment gets a free 10%
point boost based on the amount of points earned).
Someone with a 70/100 on an assignment would instead earn a
77.
This percentage can range from any percentage really, as
long as it provides the incentive to keep doing better.
This system entices people to do better on all assignments
because there are better rewards if they do better. It’s better than a flat reward system
(Or otherwise an inflation of the grade) because some students with low grades
benefit more from flat bonuses and therefore may not try as hard and bank on
the inflation rather than their own skills.
Rivalry System:
 |
| N-effect of Rivals put into place on Students |
This system pits two anonymous students of a similar caliber
to compete for a few extra credit points.
This system should require a preliminary system that
somewhat ranks the students based on their performance on the first major
assessment (This is not entirely necessary, however it will make it more fair).
These rankings will allow for setting up the Rivalry System:
Consecutively ranked students will be paired up anonymously (only
the professor or teacher will know who is paired with whom)
On every assessment, each student will be shown their own
grade (of course) as well as JUST the grade of their anonymous rival. If the student scores higher than their rival
does, that student earns an extra X percentage points on their Linear Reward.
For example, Student A earns a 70/100 and student B earns
75/100, and the Linear Reward is set at 5%. Since Student B earned more than Student A, Student B earns an extra 2 percentage points, setting his or her
Linear Reward to 7% instead. Student A would not earn this extra bonus.
Student A: 70*1.05 = 73.5
Student B: 75*1.07 = 80.25
Student B: 75*1.07 = 80.25
This system encourages students to score better with both the Linear Reward design as well as puts them in a small competitive setting
that is most effective (N-effect with rivals by Steven Garcia).
What do you guys think of these designs?
I think these are two very interesting propositions to our existing grading schemes which i'm sure no one is ever truly happy with in terms of fairness or accuracy. Two most popular ones now as you touched upon are grade inflation and the quintessential bell curve. The linear rewards system to me represents a perfect remedy for grade inflation because in this case you see your grade being inflated directly to your performance thereby giving you an incentive to work hard instead of hoping for an arbitrary boost. The rivalry system which correlates with the bell curve system however seems to be a little to difficult to implement. For one, its very hard to judge a student's overall abilities based on one assignment, especially the very first assignment of a class. And to thereby to rank people based on groups of similar abilities is even harder to measure. I think in this instance, the bell curve is better because it pits everyone against each other and everyone is aware of this fact, incentivizing them to work harder than their fellow peers.
ReplyDeleteOne potential problem I could see arising with the rivalry system is the fact that hostilities between students could develop because built intrinsically into the system is the concept of dominance and superiority. Even though there is no concrete reason, people don't like to be told "Alex is smarter than you." You address this in part by suggesting that performance is anonymous, but in my opinion, anonymity in this setting detracts from competition because there is no personal element to it, which is essential to the very concept of "rivalry." Another problem is pairing up students. It is very difficult to objectively determine the caliber of students. For example, I might score higher on the SAT or IQ test than my peer, but if he's a hard working guy, he's going to beat me every single time because I'm also lazy. There are simply too many factors to take into account. Also, such a system would make it incredibly difficult to create a fair class-wide curve, which is the standard in a very large number of college-level courses.
ReplyDelete