Recent policies require some interactions previously conducted in close social proximity (e.g., school, workplace) to take place remotely, which motivates our investigation of how in-person versus online environments impact honesty. We modify a well-known coin-flip task and examine the influence of going from the physical laboratory environment, to online with identifiable participants (same lab subject pool), to online with anonymous participants using mTurk. Surprisingly, while a simple move from in-lab to online (using the same subject pool) appears to increase "fake effort" - those who likely never flip the coin - it does not predict more dishonest behavior when there is a monetary incentive to cheat. The most socially distant and anonymous participants (mTurk) are more likely to be deemed cheaters in our analysis - these individuals report coin flip outcomes consistent with cheating for monetary gain. Implications of our findings indicate the greatest risk of potentially costly dishonest behavior results when anonymity, not just social distance, is high.