Wait, UConn was getting points from Duke? Who set that line?
We're not talking about a bettor's point spread, which I'm sure had UConn by 12-15 points. This is the Sagarin ratings, which tries to evaluate teams by analyzing a team's W-L, W-L point margins, and home\away factors to sort out the rankings for 355 D1 teams. By taking the difference between two teams' rating scores and adjusting with the home\away factor, you get a decently accurate prediction for an upcoming contest, and when you average in all the games over the course of a season, Sagarin is very good. On Tuesday morning the main score that Sagarin ranks by had UConn up 3 over Duke, but with the -4 court effect thrown in, UConn was the underdog. One of the rating scales Sagarin uses called the Golden Mean or Dimin Curve had UConn favored even with the court factor, but that's just one of the scales that is used.
As I mentioned elsewhere, Sagarin is not really able to handle a tremendous outlier team like UConn because it tries to adjust all those weirdly huge results that teams occasionally have during the season down to something more normal. But for those once-every-few-years spectacular freak-out teams like 2013-14 UConn, every game has those hugely weird results, and the team is basically off the charts there. UConn's current +8 advantage straight-up over Duke just does not begin to describe the true distance between the teams.