Mitigating strategic misrepresentation of values in open-ended stated preference surveys by using negative reinforcement