In other words, some amigo might run a study showing that NL-AL crossover players didn't show an "adjustment" year followed by a "bounce" year.  But they'll forget that they're measuring a group ... and that an individual player (say, Adrian Beltre) or type of player might indeed show the effect.

Of course, when only 1 or 5 or 12 players show the effect, they'll cross it against the group, declare "sample size" issues and blow off the exceptions to the rule.

In other words, if Mark Prior drops his arm slot 10 pitches before getting lit up, you still might not see the tendency across the group of all MLB pitchers.  Does that mean that the info is useless to the Cubs?

=======

As Jeff S is well aware, as an ex-pitcher, the state of pitching is so complex that it may be literally impossible to get our arms around it.  

You simply can never isolate the variables.  Do 120 pitches (or a dropped arm slot, or a wobbly release point) matter for young pitchers?  Left-handed pitchers?  Curveball pitchers?  Pitchers who have thrown 150 innings that year?  112 pitches that day?  Pitchers who have been in a close game that day?  Tall pitchers?  94 mph+ pitchers?

The biggest problem with Carroll's general reaction (as to uselessness of release-point info) is that he shows little respect for (1) human intuition (Sullivan zero'ing on Prior's fatigue) and (2) the complexity of the problem in general.

But then, this debate about dogmatism in baseball didn't start this week.  :-)