BB - can't help but throw my $0.02 in....
From a pure statistics standpoint you are just as likely to get 3 bad sets in a row as one. AND, yes, you are also just as likely to win the lottery 3 times in a row as you are to win once. Each random draw stands on its own and is not influenced mathmatically by previous draws.
To put this in perspective - say you take a six sided die and roll it 10 times, and by chance you roll a "5" each time. On your eleventh roll you are just as likely to roll a 5 as you are any other number. The odds remain 1 in 6, regardless of the fact that 5 came up the last ten times in a row. Or in reverse - if you roll the die 100 times and never get a 5, your next roll is still just as likely to not be a 5 as it was on your first roll. Its sounds hard to believe but that's the way it is.
So - if Klipch manufactured, say, 10,000 4.1's last year, and 5% were defective, your odds of getting a bad unit are always 1 in 20, no matter how many units you buy (or return), or who you buy them from.
Now - that's assuming, I suppose, that the defects are spread uniformly across the production run so no single distribution source receives a disproportionate number of bad units. (i.e. all 500 bad units are not manufactured in a row and all sent to Best Buy). I suspect the final statistics for any specific person would be governed by production runs and shipping schedules, but from an overall perspective, your odds would always be the same.
Who was that guy in the Asimov "Foundation" series that could predict the future of mankind by applying statistics on a large enough scale?
Isn't math fun?