The New Republic's Walter Shapiro gives several explanations. Here's his take on how Iowa has distorted the media's view of the race:
Why have the media hordes been so consistently off-base in handicapping the GOP horserace? Virtually everyone from blonde midday cable TV anchors to polling gurus like Nate Silver (his January 16 blog post was headlined, “National Polls Suggest Romney Is Overwhelming Favorite for G.O.P. Nomination”) bought into the Mitt-placed confidence in Romney’s inevitability. . . . Romney remains the favorite for the nomination, but it is not likely to be the quickee January coronation that was forecast just a few days ago.
Part of the explanation for the bum predictions has been a false sense of historical determinism by political reporters who should know better based on Romney’s ersatz Iowa victory. (As recently as Wednesday night in Irmo, the candidate was still chortling over his now-vanished 8-vote validation in the caucuses). The coverage coming out of New Hampshire was so tilted towards a Romney cakewalk that the other candidates were consigned to Ron Paul spoiler territory. . . .
Of course, we now know that Rick “We Wuz Robbed” Santorum ended up with 34 more votes in Iowa than Romney. But the idea that Romney or Santorum ever “won” Iowa was always ludicrous. Given the amateur-night vote-counting methods in Iowa combined with the statistical improbability of sorting out an election that close under optimum conditions, it should have been apparent for weeks that Iowa was a tie. But the oddball conventions of political journalism demanded that Iowa crown a winner because even false certainty is required when deadlines loom. (The counting of ballots in the 1988 Democratic caucuses was also a mess—and it is still ambiguous whether the anointed Dick Gephardt actually beat Paul Simon.) The Iowa caucuses should not be equated with the 2000 Florida deadlock since, in that tragic case, someone had to win the state’s electoral votes. But for all their symbolic importance, the formal purpose of the caucuses is to allocate Iowa’s 28 delegates to the GOP Convention. And guess what—Romney and Santorum were always going to be awarded the exact same number of delegates. Only in the phantasmagorical world of media perceptions does it matter which candidate had a tiny edge when the counting stopped.
At Bain Capital, Romney would never have made an investment decision based on the small sample of data from Iowa and New Hampshire. But the press corps made him the prohibitive favorite in both South Carolina and nationally in part because he is running the kind of on-message campaign that political consultants fantasize about. Romney is the perfect paint-by-numbers candidate: He is smart, disciplined, malleable and equipped (counting his personal fortunate and his SuperPAC) with a formidable bankroll. Since most political reporters uncritically accept the governing premises of campaign professionals, how could Mitt go wrong?