While I don't disagree with the results in the article there is something that is ignored in articles like these.
When I have sold a covered call that could be called at expiration I tend to cover it and move to a future month. I will do this if the move meets my profitability standards and if I believe that the new option has a good chance of expiring worthless. Doing this takes the call out of the catagory of expiring in the money. Thus the statistics could be out of whack by a unknown amount as I presume that others do this also.
How large a percentage is this? I don't have the faintest idea but could it be substantial?