hypothesis: caring conformism exists
Hypothesis:
some people are not social learners but willing to respect the majority decision.
Statistical procedure
- core question: how likely is an unmotivated switch?
- take all consecuitive decisions without new information
- initial individual decision before vote
- later individual decision before next vote
- dependent variable: person switches between these two
- independent variable: person has been conforming in first decision (was in minority and adjusted)
- control variable:
- domain dummy
- dummy whether first or second change
- allow for errors to be clustered on individual (and perhaps also session) level
Hypothesis
coefficient should be positive and significant
Interpretation
baseline is error rate coefficient: how many percentage points difference to error rate
Explanatory Text
How can we identify whether there are conformists who do this just because they care for others? The overall percentage of people who return to their original choice when voting the second time on an issue only gives a rough indication. These people may just have switched back erroneously by clicking on the wrong button, etc. We can estimate how often such errors occur by looking at all situations in which no new information arrives and measuring how often people switch behavior in these situations. There is no arrival of new information between the individual decision and the first voting decision and the dictatorial decision and the second voting decision. The intercept in our regression will tell us the base rate of switching behavior and measure how often such errors occur. We also include a dummy if the situation is of the second type. The coefficient of this dummy then captures how much more often switches occur the second time around. Note that this may be different, for example, if people are getting tired, etc. Most importantly, we include a dummy for all those observation in which someone beforehand was in the minority and then confirmed in the ensuing dictatorial decision. The coefficient of this dummy then tells us how many more often we see switches when people gave in to a majority previously. It measures the excess rate of switching after conforming. If this excess rate is positive and significant, this suggests that switching back after such a history occurs more often than expected. Accordingly, the choices that identify a caring conformist do not appear to be the result of error or randomness but something more systematic.