I got the impression Edwards's calculations were mostly dictated by--surprise!--self-interest. Early on, he wasn't sure Obama was tough enough to beat Hillary. Or to reassure voters and superdelegates that he'd be able to win the general. And what good does it do you to endorse a guy who's going to lose?
Since then, Obama's obviously become the favorite to win the Democratic nomination, which has changed Edwards's calculus. The risk is no longer endorsing a guy who may lose. (At least in the primaries.) It's that you won't get credit for helping Obama win. Endorsing Obama at this point would basically mean jumping on a bandwagon, and there's no percentage in that. So I'm guessing Edwards is biding his time until there's a moment when his endorsement would matter--for example, when it could help bump Hillary from the race. (Say, after a loss in the North Carolina primary.)