Quote:
Originally Posted by Gregatron
vhawk: I'd like to think that the idea of being "clutch" can go beyond post hoc-ness and our natural tendency to create narrative. Before I elaborate on this, I want to clarify: I think both of these (post hoc theorizing and creating narratives) are the main culprits for the idea of "clutchness" as it is popularly presented. Most people that believe in "teh clutch" have no idea what a standard deviation is.
I'm actually coming from this from the prospective of an academic social scientist. People, hell even members of the same species of mammal, can handle the same situation differently. I propose there are theoretical reasons (yes, involving real theory) that there will be individual variation among individual players. (And I don't think I am superimposing a narrative when I say this!)
The problem, as you have pointed out, is identifying such people based on objective and theoretical criteria. This is problematic, and may just be impossible. There are also a host of other methodological issues exacerbating this.
I agree with you wholeheartedly. As others have mentioned, there seems to be evidence supporting SOME role for "clutch" in baseball, and if I had to guess I would say that the evidence is probably underreporting. And the reason for that is methodological. While I dont think its theoretically impossible to measure just about ANYTHING, I do think its currently way beyond our ability to accurately measure and isolate everything that goes with "clutch." As such, we are stuck with a situation where "pretending clutch doesnt exist" does a MUCH better job of approximating reality than "any model or theory or prediction based on clutch that vhawk has ever heard." Its a matter of practicality. Do I really think clutch absolutely doesnt exist and (gasp) that all players are robots? No, I do not. I just dont think that there is any current method for taking clutch into consideration that isnt WORSE than just pretending they are all robots.
But I am not an expert on everything sabermetric, far from it, and I am certainly not abreast of every new study or analysis that comes out. I imagine that at some point there will be a nice, handy statistic measuring clutch and choke-factor that will join the playbook. I expect it to be entirely marginalized by more important stats, though.