This Beyond the Box Score story has been vaguely bugging me for a few days now, but I couldn't quite put my finger on why.  I've been really busy at work, and I've only had about 5 hours a day to play around on the intertubes.  And what with all the March Madness and my passion for pornography, I haven't been able to post a response over there.  The quotation from your diary above finally helped me nail down what was bothering me.

Turnbow is not "significantly worse when there are runners on base."  He puts significantly more runners on base when he is at his worst.  He frequently comes in and shuts down the side, or maybe gives up a single with two outs before striking out the next guy.  Or he comes in, throws the ball all over the park, walks two batters, strikes one out, gives up a single, walks another, grooves a fast ball for a home run, then gets yanked. His problem on these days is not that he is unable to pitch with runnters on base or that he's unlucky.  His problem is he can't get the ball ove the plate, and when he does he has a harder time missing bats.  If you cleared the bases before each hitter, he'd still have a higher probability of putting each hitter on base than he would on one of his effective days.  We've all seen this happen week in week out for the last two years.

Beyond the reasons you list, which are very good ones, that's why we can't measure clutch in (or assign bad luck to) a pitcher's performance using Anderson's method, especially for a short reliever who has a tendency to blow up like Turnbow.

As has been pointed out a number of times on this site (and I think by Turnbow himself), the problem may not be mental but physical.  He seems to have many of his worst days when pitching for the second consecutive day.