With all due respect you need a lesson in statistics. 15,000 people can be a great sample size, and saying that 56% of republicans are over 50 implies nothing about the percentage of people over 50 who are republicans.