In 1980, some states were early called to Reagan, and his margin was very small.
In 1992, some states that Clinon won narrowly were called early to Clinton, and some states that Bush won were considered too close to call.
In 2004, some states that Kerry won narrowly were called early to Kerry, and some states that Bush Jr. won narrowly were considered too close to call.
Hypothesis:
Exit polls do not reach well the rural areas of the states. So, they overestimated Clinton and Kerry, and overestimated Reagan too. In the southern states, Carter did better in the rural areas.
Why would exit polls be less reliable in rural areas? From my understanding of how exit polls work, people ask random voters as they are leaving the polling station, who they voted for. Would rural voters be more inclined to lie than say urban or suburban voters? Or is it simply a numbers issue whereby there are fewer volunteers in rural areas and hence the sample size is too small to make a reliable conclusion?
Exit pollsters are less likely to sample rural precincts.
