Wednesday, August 3, 2011

Incentives for cheating and for cheating-detection

A chilling and surprisingly complicated story from Panos Ipeirotis at NYU Stern (who blogs under the title "A Computer Scientist in a Business School): Why I will never pursue cheating again (since deleted, but that URL now has related links). Apparently he got a cease and desist letter, and the story has since attracted some press.

Here's one story about it: NYU Professor Catches 20% Of His Students Cheating, And He's The One Who Pays For It,

and here's another: NYU Prof Vows Never to Probe Cheating Again—and Faces a Backlash

Ultimately he draws a market design conclusion:
"In Mr. Ipeirotis’ view, if there’s one big lesson from his semester in the cheating trenches, it’s this: Rather than police plagiarism, professors should design assignments that cannot be plagiarized.

"How? He suggested several options. You could require that projects be made public, which would risk embarrassment for someone who wanted to copy from a past semester. You could assign homework where students give class presentations and then are graded by their peers, ratcheting up the social pressure to perform well. And you could create an incentive to do good work by turning homework into a competition, like asking students to build Web sites and rewarding those that get the most clicks."

1 comment:

Zach Mayer said...

For statistics/econometrics classes, Kaggle in Class is a pretty cool tool. Basically, it lets you setup data prediction competitions where the students are competing against each other for the top spot (and thus top grade).

http://inclass.kaggle.com/