I've seen some interesting things just getting 2003-2006 done and, to be honest, there doesn't appear to be a great correlation between negative VORP and Wins.  At least not for the bulk of the teams.  The teams that have the worst negative VORP each season are pretty much going to be the worst so there is some correlation, but not as much as I'd hoped to find anyway.  

There shouldn't be much difference between the leagues though.  Replacement players are defined as the 20% of the players who played the least at each specific position.  Over a long period of time, all teams are going to have roughly 20% of their innings and at-bats given to below replacement level players and, I'd assume, have a similar amount of negative VORP.  That would certainly be true if we looked at 50 years worth of data, but over 5 or 7 years we should see some interesting information.

For example, during the 2003-06 years I have completed right now, there are three teams in baseball that clearly do pay attention to negative VORP.  Two of the teams you could probably guess--Oakland and Boston.  The third team suprised me--Milwaukee.  These 3 teams in this span (granted, not much data, but the advanced statistics era has taken hold during this span also) have shown the ability to reduce negative production beyond what should be expected.  Interestingly, all 3 of these teams are quite different.

Boston spends a lot of money while Oakland spends about league average and Milwaukee is generally in the bottom third.  

As for what stats are my favorite...I don't know.  I don't really have a favorite.  I can say that I prefer Baseball Prospectus' stats (WARP, VORP, etc) over Win Shares and other similar stats, but that's only because I have more knowledge about BP's stats.  I think if you're going to do a comparison between two players who play the same position, Batting Runs Above Replacement (BRAR) and Fielding Runs Above Replacement (FRAR) is a pretty good way to do it.  It's rather simple and it gives us a pretty good idea about each player's value offensively and defensively (VORP doesn't consider defense).  We always have to remember what are rate stats and what are counting stats (batting average is a rate stat and RBIs are a counting stat for example).  Rate stats don't consider playing time so they may be the better stat to use from time to time.  If we're comparing two players to league average (or replacement level), VORP is probably the way we'd want to do that, or VORPr (which is just VORP Rate).  

And we also have to remember that all stats are fallible and the more we have the less flaws there will be in predicting future performance (that is the ultimate goal of stats, after all...along with the historical record they create).  Now, not all stats are equally flawed though.  Ideally, we want stats to represent production, which is either runs created or runs prevented.  The less a stat has to do with one of those two things, the more flawed it is.  For example, an error does indeed have an impact on runs scored or runs prevented (both actually), but not all players get an error for the same play because of a player's range.  It's distinctly possible that a SS with little range could make as few as 8 or 10 errors and be considered by some to be the superior defensive SS although a SS with superior range who makes 12-15 errors is almost certainly the better defender because he has a greater impact on runs prevented (he's turning more balls hit near him into outs) than the other player.  An error is a terribly flawed stat.  It tells us very little about the impact that player has had on what is important--runs.