Social science epistemology, afaict:

1. Ignore all the interesting and replicable phenomena

2. Find data sets that can be interpreted in isolation as Cutting Social Commentary

3. Wonder why your interventions never work, findings never replicate, and you can't predict anything.

How it should be done:

1. Separate experimental studies from social commentary. Build a corpus of replicated phenomena that need explanation

2. Find a-priori plausible explanatory hypotheses or laws that explain the replicated phenomena.

3. Favor hypotheses that predict better, are simpler, and integrate with other theories

4. Check your work by building working social technology, making novel predictions, or otherwise demonstrating veracity

Note that my recommended social science epistemology is physics epistemology. Soc sci needs more physics envy and fewer linear regressions.

Oops forgot step 0. ASSUME AN UNUPPORTED IMPLAUSIBLE FALSEHOOD (SSSM) AS AXIOMATIC

Nyan Sandwich