Meta-analysis can't solve the problems with statistical studies. Disparate small studies don't correlate well with a single larger study, given different standards for measurements, participants, different and possibly overlapping pools, etc.
Meta-analysis also suffers from publication bias, agenda based bias, etc.
How does running a bunch of replications of a study ever falsify it? Based on statistical probability it's always possible the original study was accurate, and it's the replications that cheated. The realization that the results of these studies are essentially random goes a long ways towards establishing how unfalsifiable any of them are.
And beyond statistical studies into analytic theories, how do you falsify the theory of the "Id, ego and super-ego"? Transfer neuroses? How do you falsify "Dialectical behavior therapy"? It worked for someone somewhere, so it must be true.
The only thing that's come out of the whole of psychiatry and psychology that has any scientific merit is Skinners behavioral experiments.
Meta-analysis also suffers from publication bias, agenda based bias, etc.
How does running a bunch of replications of a study ever falsify it? Based on statistical probability it's always possible the original study was accurate, and it's the replications that cheated. The realization that the results of these studies are essentially random goes a long ways towards establishing how unfalsifiable any of them are.
And beyond statistical studies into analytic theories, how do you falsify the theory of the "Id, ego and super-ego"? Transfer neuroses? How do you falsify "Dialectical behavior therapy"? It worked for someone somewhere, so it must be true.
The only thing that's come out of the whole of psychiatry and psychology that has any scientific merit is Skinners behavioral experiments.