- As datasets grow in size, it's going to become trivial to find "significant" effects (i.e. non-zero).
- That isn't a problem that can be fixed by just shrinking α down.
- We need to ask ourselves:
- Are the effects we're observing large enough to be interesting?
- How big did we expect them to be?
- To answer (2), we need an articulated theory that can make quantitative predictions.
- I walk through two examples where I try to predict effect sizes given background theory.
- link: https://jofrhwld.github.io/papers/plc39_2015/
Last active
August 29, 2015 14:17
-
-
Save JoFrhwld/f3783b9375364bc0e504 to your computer and use it in GitHub Desktop.
Big Data and Sociolinguistics
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment