This is still only for 1 number.  It's not for a RANGE of numbers missing.  Imagine this and add a power of 100 to it.  Thats your probability for all the missing numbers I believe.  I'm certainly no math major but even a 0.0000827% chance of it happening.... to me and multiple people?  Within the same century?  Really?  heh  Start adding in other user's roll sample size and that number gets REALLY small.  I dont have Matlab nor know how to use it,  but I assume thats what you're using here.

EDIT:  Oh nevermind,  you're just using a spreadsheet.  Thats probably not the most accurate way to calculate the range variance,  even so it does still illustrate just how improbable the situation is yet it's obviously showing up more than it should in real world practice.