So you think college-football recruiting is an inexact science? Several economists would beg to disagree.

Two years ago, a trio of economists borrowed a software program designed for the business community and turned it on the recruiting world.

They sorted through hundreds of factors that highschool kids use to make their college choice, plugged in factors that proved significant, and presto ? a model that correctly predicts the recruit?s decision about 70 percent of the time.

"I use similar sorts of models in the stuff I do every day," said Mike DuMond, who works for the employment-litigation company ERS Inc. in Tallahassee, Fla. "I guess I got tired of doing real work and decided to use this for sports."

His partners in the project are two professors: Allen Lynch of Mercer University in Macon, Ga., and Jennifer Platania of Elon (N.C.) University.

They use the nation?s top 100 recruits as ranked by rivals.com. For the initial study, they used data from 3,395 recruits in the 2002 to 2004 classes and found they correctly predicted the college choices for 68 percent.

In 2005, their success rate for the top 100 was 71 percent. After taking 2006 off, they are 70 for 92 in this recruiting class, or 76 percent.

There were some surprises when they analyzed which factors were important to kids and which weren?t, DuMond said.

For instance, factors such as school graduation rate, number of bowl appearances, depth at the player?s position, number of players drafted by the NFL, and number of national championships were not significant.

What does matter? Winning programs that are close to home, have good physical facilities, and are in good graces with the NCAA.

"On the surface (recruiting) doesn?t seem to lend itself to scientific analysis," DuMond said. "You hear all the anecdotal stories about players picking schools because they liked the colors, or a jersey number. But from a statistician?s point of view, it has proved interesting. Recruits tend to follow a logic that makes sense."

The current year?s model can be viewed at http://ssbea.mercer.edu/lynch/CFRPM.htm

There are misses, of course. This year, for example, the model predicted five of the rivals.com list would choose Ohio State, but only two did: Brandon Saine and Eugene Clifford. It was wrong in the cases of Ben Martin (Tennessee), Dionte Allen (Florida State) and Anthony Davis (Rutgers).

Of the eight remaining uncommitted players in the top 100, only one is considering OSU: Belle Glade (Fla.) receiver Deonte Thompson. The model predicts he will commit to Florida, with OSU coming in fourth.

But fourth is where the model had Southern California on the list of Muskegon (Mich.) receiver Ronald Johnson (Michigan was first). On Sunday, Johnson chose USC.

So there is hope for Buckeyes fans. Statistically speaking, of course.

kgordon@dispatch.com?