Hypotheses: conformism although we control for a lot
testing mechanism:
-
consider ALL decisions after voting (2x7 per person)
-
dependent variable: do people change after voting
-
independent variable: was person in minority
-
allow for inter person fixed effects (maybe also for session effects) (these help us to take out any selection effects)
prediction
coefficient for minority is positive and significant
treatment of jar
We have two options
- exlude jar data
- include jar data but give them special treatment
- include domain dummies
- include interaction term for jar and minority (variable of interest) I am tending toward the second option.
Edited by Wendelin Schnedler