That’s not how probabilities work. The market share in the US is P(has iPhone | resident of US), but we’re looking for P(resident of US | has iPhone), which according to Bayes’ law is equal to the market share in the US (see above) times P(resident of US) [aka US pop. / world pop.] divided by P(has iPhone) [aka global market share]. So essentially, while the market share in the US may be twice as high as the global average, the US has fewer than half of all people in the world - making it more likely that the person is not from the US than that they are.