Be More Than a Little Wary of Exit Polls

The media has smartened up and  is far more cautious about using polls (with very good reason) to predict the primaries. They've been wrong too often, including both

But that doesn't explain why the exit polls, taken after voters have sealed their ballots, aren't more accurate -- and more importantly, why the media treats them as gospel, often reporting them as facts when their conclusions are rough and unreliable.

As the news cycle progresses, look at how many stories state as a given, for example, that African-Americans went 8-1 for Obama and that Hispanics went 2-1 for Clinton. Unlike other poll results, these become accepted as facts because there is no way to prove them right or wrong.

But consider this: The same exit polls last night showed Clinton struggling in Massachusetts and New Jersey when she was actually winning easily. They also made California too close to call for both parties, and yet both Clinton and McCain won by a comfortable 10 and 8 percentage points, respectively. If the exit polls were off on the totals, why do we accept the breakdowns as accurate, especially when the sampling of Hispanics or blacks is far smaller than the total and has a higher margin of error? It just doesn't make much sense, does it?