The thing is, if you randomly created sample sizes in clutch situations- many which are comparatively small, you would not end up with every hitter batting exactly like his career norm any more than you would, if you looked at his Monday numbers vs Wednesday. Some would do better- some would do worse.
Just the fact that some players have better numbers is this highly selective sample sizes, does not prove clutch is a skill.