There is a compelling case to be made that manipulating the reporting of results (rather than the results themselves) would be more in line with your goals of public perception and deniability.