what about share of learners in 3rd hypothesis
idea run regression with
- dependent variable whether learner (Min, Maj, Maj)-Signature
- on sample of all in the minority
- control for domain, session etc
variable of interest
- coefficient should be not significant because we do not expect the share to change