Probit Model: Liklihood Ratio Significant, Very Low Pseudo R-Square

I'm running a probit model in Stata, and I'm a bit confused by the results. The LR test is highly significant, .0000, but the Pseudo R-Square is very low at .05. In addition, some of the coefficients have the wrong sign, albeit of small magnitude.

I have two questions here. 1) What to make of the low Pseudo R-Square. I know people disagree about the value of this statistic. Does the low value (in the face of the significant LR test) tell me I have problems with my model? Or can I just ignore this low R-Square?

2) If there is a problem with low R-Square, could this account for the wrong signs?

My data set is 1000 cases, and I'm examining the effects of various college institutional variables on graduation rates.