PFF isn't really "'statistical analysis,'" i don't think. IIRC they grade players' performance on each snap all season and simply rank players based on the accumulation of positive and negative grades for each snap over the whole season. in other words, they evaluate with their eyeballs. i often disagree with them too, but from what i know it ain't like they're using football sabermetrics or something.
as for martin's performance vs long's, anyone who thinks martin was better than long at LT is crazy IMO. martin was OK vs SF, fine against jax and buf, and hideously godawful against NE (at least as bad as long was against indy). martin doesn't run block as well as even a diminished long and probably never will. he has the tools to handle speed rushers better than long can at this point in long's career, but those tools aren't as developed as they need to be yet. long was unbelievably bad against freeney but the rest of the season he was more of an average performer IMO (honestly probably a hair above average). problem with that is that mere average-ish performance makes him massively overpaid.
martin is a better choice than long cuz he's a better value for his performance level and a better fit for the system. but martin is not generally a better tackle than jake long, not yet anyway.
