Brendan Nyhan writes about how the need to construct a campaign narrative can lead to people substantially overestimating the importance of this or that campaign occurrence. For example, current polling makes it look likely that Hillary Clinton will beat Barack Obama by a bit more than ten points.
Now if you’d said on March 5 “looks like Clinton will win Pennsylvania by about 12 points” most people would have said “sounds about right, she has a huge advantage in the polls right now but Obama always gains ground through actual campaigning; still, demographically speaking it’s very favorable terrain for Clinton.” But today it’s essentially inevitable that any failure on Obama’s part to close the gap will be substantially attributed to “bittergate” even though failure to fully close the gap was not only predictable but widely predicted weeks ago based on Pennsylvania’s age structure, educational attainment, and African-American population.