Last week, Sports Nut set out to determine how well sportswriters could forecast the NCAA Tournament. We combined the brackets of 15 basketball writers and columnists into one, forming a superbracket that holds their collective wisdom. How have the writers fared so far? Not particularly well. They correctly predicted only 19 of 32 first-round games and nine of the 16 second-round matchups, which is slightly worse than if they had simply picked the higher seed in each game. But the writers still have dogs in the fight, and after tonight and tomorrow's games they could nail seven members of the Elite Eight. (Click here to peruse the writers' bracket and read more about our methodology.)
What's really intriguing, though, is how often our 15 writers made the same picks. For the tournament's more lopsided first-round games—say, the No. 1 seed versus No. 16—you would expect most of the writers to vote for the heavy favorites. And they did, almost unanimously. But in the other, more evenly matched first-round games, you would expect the writers' opinions to diverge—that is, for some of them to pick one team and some to pick the other. They didn't. Look how they voted in four games in the Midwest Region:
No. 5 Florida (12 votes) vs. No. 12 Creighton (three votes)
No. 6 Texas (14 votes) vs. No. 11 BostonCollege (one vote)
No. 7 WakeForest (one vote) vs. No. 10 Pepperdine (14 votes)
No. 8 Stanford (two votes) vs. No. 9 Western Kentucky (13 votes)
The writers weren't just flocking to the favorites—Pepperdine and Western Kentucky were both underdogs. Nor, in hindsight, were they exercising some kind of divine journalistic wisdom. Three of their four picks—Florida, Pepperdine, and Western Kentucky—lost.
In fact, it happened again and again all over the bracket—games that should have divided the writers barely divided them at all:
No. 6 California (three votes) vs. No. 11 Pennsylvania (12 votes)
No. 7 OklahomaState (one vote) vs. No. 10 KentState (14 votes)
No. 7 North CarolinaState (four votes) vs. No. 10 MichiganState (11 votes)
No. 7 Xavier (five votes) vs. No. 10 Hawaii (10 votes)
No. 8 Notre Dame (11 votes) vs. No. 9 Charlotte (4 votes)
Why would 15 randomly selected writers, from different publications in different parts of country, suddenly come down with a case of groupthink? Well, the most obvious explanation is the information-sharing capabilities of the Internet, which has become an integral tool for sports journalists (and every other journalist). An underdog like, say, Pepperdine is touted in two or three big Web sites or newspapers in the week before the tournament. The information spreads like a fad, and before long every writer has penciled in the same "upset special." But without a similar pre-Web study, such a phenomenon is impossible to prove.
Another explanation is that when sportswriters predict a given game, they rely on their professional advantage: They have seen the teams in question play more than the average fan. But when the NCAAs roll around, the size of the tournament field dulls that advantage. (Assume the average writer, depending on his publication, has seen no more than half of the 64 teams play before the tournament starts.) How good is Montana or Creighton or Holy Cross? Without specific insight, the writer relies on the same kind of apocryphal "bracket wisdom" reserved for lesser mortals. For example: "The Ivy League team always pulls a first-round upset." (This has happened only twice in the last decade, but it's a still a March Madness cliché.)Or "No. 10 always beats No. 7."(It happens only slightly more than half the time.) Thus, 80 percent of our writers pick Pennsylvania, and a big majority picks all four No. 10s.