Variance is massively underestimated at all stakes in poker.
This myth mainly stems from being unaware of how big variance actually is. At some point in your career you are going to run worse than you thought was humanly possible. It is possible to simulate how your earnings might look given an expected winrate and standard deviation (which can be obtained using HEM or PT3) (You can run your own simulation on evplusplus.com)
In the above example we have selected a standard deviation of 80bb/100 (around normal for micro-stakes), and a winrate of 6bb/100 which is by no means a bad winrate for any stakes. We simulated 100 possible outcomes and have plotted the best and worst results. The dotted black line represents the expected winnings assuming 0 variance.
Bear in mind that the green/blue lines represent identical players. Yet one is up nearly 120 buyins (100bb), and the other is down perhaps 10. Hopefully this should give you a brief picture that variance is a lot bigger than people imagine. If someone loses over 100k hands at a very low micro-stake, you might immediately assume they are a bad player, when in actual fact they could be a 6bb/100 winner!
Conversely, marginal losers have gone on 100k upswings, turned a nice profit and promptly quit their day-job - before realising they are perhaps not as good as they initially thought.