As N grows sufficiently large the sample mean will begin to reflect the true population parameter. I know how statistics works. I also know that if I have seen something twice I stop operating on best practices for statistical analysis and start operating as a human being. The number one killers in the US are...
Cancer: 576,691
Chronic lower respiratory diseases: 142,943
Stroke (cerebrovascular diseases): 128,932
Accidents (unintentional injuries): 126,438
In my whole life I know 1 cancer victim, 1 lower respiratory disease victim, 1 stroke victim (who was 90 and smoked 2 packs a day) and 1 accident victim.
So even though it is not statistically valid my count of 2 seems extraordinarily high, yes I know it can be attributed to chance but I also know that that chance is fairly small.