From the article:

I will also point out that "clutch hitter" doesn't mean a player who hits well in the clutch – it means a hitter who performs *better* in clutch situations than normal.

This definition almost guarantees that you will never find "clutch hitters" even if they exist, because the distribution of performance in "clutch situations" is centered at a different point than the distribution of performance across "normal" situations. A hitter can be on the positive side of the "clutch" curve and still not hit better than he does in "normal" situations.

-- MWE

Right. You have to adjust for the fact that clutch performance is different overall than regular performance, which I think all of the quoted studies did. I'll update the original post.

Thanks, Mike ...

It should go without saying that it's after adjusting for the bias of the mean point. It's always performance minus mean in situation 1 compared to performance minus mean in situation 2.

The mean is different because of the relief pitcher, or the men on base situation, or really, whatever. It doesn't matter. You simply account for this bias.

It should go without saying that it's after adjusting for the bias of the mean point.

It should - but it doesn't always, which is why I like to say it :)

EDIT: You (maybe not Tango, but the rest of you) would be surprised at how many of the so-called "no clutch effect" studies miss this point.

-- MWE

studies, I think, have already convincingly shown that clutch talent doesn’t exist

I haven't read most of those studies, but I feel confident in saying that no study has shown that clutch talent doesn't exist, because its impossible to show. You can say that no study has shown that it does exist, but that's not the same thing. You can also say that the evidence shows that if it does exist, its not common, or that most of what appears to be clutch performance can be explained in other ways. But once again, you can't prove that it doesn't exist.

5/Dizzypaco: you can show, with statistical significance, that if it does exist, it must be very, very small. And the practical difference between "very, very small" and "doesn't exist" is zero.

However, yes, it's still possible that (for instance) one player in baseball history was capable of raising his OPS by 30 points in clutch situations. And, yes, if that's true, it would be impossible to prove it exists. But it's like the invisible pink unicorn in my garage -- it's impossible to prove it exists, and there's good reason to believe it doesn't.

It should also go without saying that anything that involves human beings is never random. Clutch hitting exists because people are involved. Pitcher skill on balls in play exists because pitchers are people too.

The question in play is always the degree to which this exists.

If a well-constructed study with granular data is having a hard time finding it, it likely means that there's very little of it to find.

As Phil noted, Andy was able to detect a clutch skill. However, this amount of skill that is detected in the performance data will barely alter any decision-making process.

More however, your gut feel as to who is clutch (Crede, Scutaro, etc) may have an even larger effect in terms of decision-making process, that a guy with a .340 OBP may be a better choice than a .355 OBP (and corresponding SLG for each), if you "feel" someone is a better clutch player, as my Clutch Project may be suggesting.

you can show, with statistical significance, that if it does exist, it must be very, very small. And the practical difference between "very, very small" and "doesn't exist" is zero.

Except that it doesn't take much of an effect, in terms of raw numbers, to have a potentially big effect on the outcome of games, because of *when* these situations are occurring. It doesn't have to exist at the level of 30 points of OPS to have the potential for a measurable swing in game results.

If a well-constructed study with granular data is having a hard time finding it, it likely means that there's very little of it to find.

Normally, when you can't find something, it means one of two things:

1. it's not there
2. you're not looking for it in the right way - either you're not looking for it in the right place, or your conception of how it might be made visible is false, or you haven't looked deeply enough or in the proper direction to see it yet, etc. etc.

(Have you ever had a kid say, "Dad, I can't find my favorite shirt" only to discover that he hadn't looked in the most obvious place, his pile of cleaned and folded laundry?)

We ought to at least rule out (as best we can) the possibility of the second before we accept the first. If "clutch ability" manifests itself primarily along one dimension of performance - say, for example, in-play slugging percentage - and we are trying to measure it by measuring aggregate performance, we might not be able to find it not because it isn't there, but because we're not looking in the right place for it.

-- MWE

Well, okay, but at what point are you willing to say to rule out the second possibility? After we check in-play slugging percentage? After we check fly ball distance with a runner on second? After we check DIPS on days when there's a full moon?

I think it's incumbent on anyone who thinks clutch hitting exists to

(a) accept the studies that have already been done;
(b) say where he thinks we should look; and
(c) say in advance what evidence is enough for us to finally be able to conclude that clutch hitting doesn't exist.

If you will NEVER be happy concluding that clutch hitting doesn't exist -- because, no matter what, you still can't prove it and you've never checked absolutely everywhere it could be -- then you're not really doing science. You're just arguing a point in logic, that it's POSSIBLE that there's always some obscure place that clutch hitting COULD be hiding. And we can all agree on that logical point -- but it's not really the question at hand.

In spite of the plethora of studies and discussion on clutch hitting, I seldom see much theoretical discussion of possible origins of clutch hitting. The assumption that all participants begin the debate with, as near as I can tell, is that it's something like a "nerves of steel" versus "choker" issue.

If the effect exists, my first guess is that would be measuring "lollygagger" versus "ass in gear". I read some Brooks Robinson quote some years ago (sorry, can't find it with google) where he admitted that in the course of a full season, he couldn't take every at bat with full intensity, which I find hardly surprising. But maybe some people can keep the intensity better than others. Then the "lollygaggers" show up as the clutch hitters.