This research is interesting and certainly points to a potential issue for MOOCs.
Social Influence Bias: A Randomized Experiment
“Prior ratings created significant bias in individual rating behavior, and positive and negative social influences created asymmetric herding effects. Whereas negative social influence inspired users to correct manipulated ratings, positive social influence increased the likelihood of positive ratings by 32% and created accumulating positive herding that increased final ratings by 25% on average.
http://www.sciencemag.org/content/341/6146/647.abstract
There is a risk of ‘set and forget’ in some MOOCs, which relies on the notion of the ‘self correcting’ crowd. While this research isn’t specifically about online education I think the basic findings still apply.
What we have known for a long time is that online discussions/bulletin boards/chats need to have a teacher actively involved, not necessarily commenting all the time, but monitoring and stepping when needed or requested.