We experimentally study how receiving information about tax compliance of others affects individuals' occupational choices and subsequent evading decisions. In one treatment individuals receive information about the highest tax evasion rates of others in past experimental sessions with no such social information; in another treatment they receive information about the lowest tax evasion rates observed in the past sessions with no such social information. We observe an asymmetric effect of social information on tax compliance. Whereas examples of high compliance do not have any disciplining effect, we find evidence that examples of low compliance significantly increase tax evasion for certain audit probabilities. No major differences are found across countries. data source: lab experiment data, available from authors upon request