It really depends on the sampling characteristics and the observed population distribution. If iPhone buyers are very consistent and follow a normal distribution, it could be ok.