I conduct a field experiment on MovieLens to examine how various factors mediate peer effects.
Experimental design
We manipulate group size, mean of group contribution, and standard deviation of group contribution to examine how these factors mediate the treatment effect of leaderboards. We also are able to examine the effect of leaderboards themselves.
Design details
Randomization method
Randomization done by a Python script
Randomization unit
Individual
Clusters
Not applicable as randomization is done at the individual level
Observations
Around 3000
Arms
The main aim is to examine how the treatment effect of a leaderboard is affected by group size, mean of group contributions, as well as standard deviation of group contributions. The treatment effect of a leaderboard will be also estimated, but through within-subject/regression discontinuity/difference-in-differences methods. As such, this section is not really applicable.
Primary outcomes
Contributions of ratings and tags, number of logins, number of movies added to the database
Cited by
No citation data available for this registration yet.