A curious question that I don't have time to research right now: How does the substantial percentage of people voting early in some states affect the conclusions drawn by exit polling on election day?