June 13, 2020

Qualitative Research


  • Process tracing
  • Grounded truth
  • Criticisms
  • Lacks reproduceability
  • Lacks generalizability
  • Researcher bias


Natasha Rostova https://statmodeling.stat.columbia.edu/2017/01/11/29593/:

I always think of qualitative data’ as humans attempt to generate prior distributions based on our neural/linguistic model of the world, which is frequently more accurate in aggregate, but with insane human-to-human variance and sensitivity to initial conditions. I think that’s why psych research sucks so badly. Because all the researchers first create a qualitative prior’ by using their brain to simulate another brain experiencing a cognitive bias. Often they (rightly or wrongly) place massive prior power on this, and then do the formal stats’ as a formality. Cannot decide sample size in advance must only stop when you are getting no new info saturation’