Lots of things could go wrong with this study. I haven't looked at their data, so I'm not saying any of these infirmities are present (maybe this is absolutely true and should've brought about a paradigm shift in the lawl school world), but the following spring to mind:
1. I just don't buy, off the bat, the assumption that a student who gets a 3.5 at UF would get a 3.75 at Rutgers and a 2.75 at Boalt (or the equivalent thereof; they don't have numerical GPAs, do they?).
2. The discussion of that (in the blurb, at least) betrays a real confounding effect between GPA and School Rank. Generally, better schools curve to higher GPA medians. I know it's not true all the time, but there's definitely a trend as you move up. So, maybe in a bizarre way, it makes sense to say that a student with a 3.75 at Rutgers Newark is going to do as well, earning-wise, as someone with a 3.75 at Cornell, because the first student's top, what... 2%?
3. It's much less useful to talk about lawyer salaries in terms of averages than it is in terms of percentiles. It's not like jumping from a 3.4 to a 3.6 increases your earnings by 13%; it's that at a 3.4 at a given school you have 80% chance of making $50k off the bat and 20% chance of making ~$160k, and a 3.6 changes it to 50/50 or something. When taking the average of the salaries and trying to draw that across the board, it just becomes more likely the real thrust of a conclusion like "If student had gone to Rutgers, she'd make 13% more than if she'd've gone to UF" loses its meaning, especially with different schools attracting student bodies that focus on different things.
4. As was said, this article is 2 years old, and probably based on data from the boomtimes. I still wouldn't trust most of it, but its conclusions probably make more sense if we're in a time when there's not a shortage of jobs.