I recall there was a story from Predictably Irrational where the experimenters were trying to figure out how to get participants to avoid double-dipping tortilla chips.
Along with a control condition, they tried setting up a sign that said "NO DOUBLE DIPPING", and I think they also tried paying people or getting them to promise not to double dip, stuff like that.
The thing they found most successful was to set up two bowls of dip: One labeled "For double-dipping", and one "Not for double-dipping".
They supposed that once they had to do a physical action where they sorted themselves according to "what kind of person they are", they wanted all of their visible actions to be consistent with that.
Side note, I believe the mythbusters did double dipping, and as I recall, it's really not a problem unless you stir or leave the chip in the dip/salsa for an extended period.
It's been a long time but I recall a study featured on Freakonomics where a national park tried different signs to get people to not steal rocks. Signs like, "Taking rocks hurts the ecosystem" and "Taking rocks is a crime."
The only effective one was something along the lines of, "A million people visit this park every year and leave things alone." Suggesting that telling people to do the right thing is less effective than peer pressure.