I'm not sure that would be empirical, because when there are no ramification to an action many people will choose to do what is convenient.
We would have to agree beforehand that those who did not take the money were loyal. We would predict that if there were loyal people that some of the people tested would not take the money. We then test the hypothesis, and the results of the experiment would be independent of either of our beliefs.
There was also a famous test that bore this out. I can't remember the proper name of it, but it had people who would apply electrical shock to people if they were told to do so. However, it wasn't real electricity and the person hooked up to the machine receiving the "shock" was acting in a lot of pain. The people were told that there wouldn't repercussions if they used this device to discipline somebody for giving a wrong answer.
This experiment demonstrated, empirically, that people will ignore their own judgement if someone in authority tells them to.