While the picture seems bad for Ego-Depletion, I find the counter arguments put forward in the article not wholly satisfying, either. So they tried to replicat

While the picture seems bad for Ego-Depletion, I find the counter arguments put forward in the article not wholly satisfying, either. So they tried to replicate it in a large effort and failed, but they didn't use the same test as in the original paper. How is that a replication study? I think to properly fail a replication study, you should use the same test as in the original study. In the replication the used many participants, but developed a cheaper test to administer (wholly computer based). As it is, they have another indicator that Ego Depletion might not be real, but just as well there could be some other effect that explains why the replication failed. Likewise for meta analysis: it looks good on paper, how could an average of n studies be wrong? Surely the mass of papers studied will cancel out all the flaws of the individual studies? But on the other hand it seems to me meta analysis would overestimate the impact of bad studies, as they are easier to make so there a probably more of them. A meta study is certainly an interesting data point, but I feel it can not always be the end of the story. It can maybe point out the need for deeper questions or further research, but not be proof on it's own.