Are you sure those numbers are correct, as you seem to have a contradiction... You note that 6282 people bought nothing, while 1496 people bought a purse, even if we assume that no one bought only sunglasses (i.e. if you bought sun glasses, you also bought a purse), then the sum of people who bought something and people who didn't is larger than your stated sample size. Note also, that you calculate (logically) "total purchasers of at least one" to be 1316, but you already know that 1496 people bought at least a purse.