That's either dangerously close to the line of the unprovable, or simply on the wrong side of it. Or "undisprovable", if you prefer. At some point a sufficiently "subtle" impact on behavior simply drops below the noise floor of normal human interactions, which aren't exactly a continuous party of peace and love no matter how you slice it.
Another study that showed that whites with criminal records have a better change of getting a job callbacks compared with blacks without criminal records when qualifications were the same: http://ftp.iza.org/dp4469.pdf (the study was also done in Chicago with the same results)
