RT @drmollycarlyle
Question to the academic twittersphere 🛸: when publishing statistical models, do you choose to report descriptive means or estimated marginal means? And why?
@sirileknes usually EMMs because the model is constructed to more accurately target the effects of interest.
@sarahmcintyre that’s making the rather trusting assumption that people’s models are well constructed 🤷🏼‍♀️
@sirileknes well I also plot individual data points, and share data files, so that risk is somewhat mitigated.
@sirileknes mean and SD is just a simple model, and may not be a particularly good one.