I wouldn't poll the players, at least not nakedly, since that is vulnerable to the stated preference bias. Tracking actual play would be more likely to demonstr

I wouldn't poll the players, at least not nakedly, since that is vulnerable to the stated preference bias. Tracking actual play would be more likely to demonstrate revealed preference.