The sample size is fine but a survey was done with a group of a thousand Americans. Do you really think those people represent the correct population? There are millions of iPhone users in America. Don't you think a sample from those users would be more accurate than just 1000 random people?