A novices’ information to causal inference strategies: randomized managed trials, difference-in-differences, artificial management, and A/B testing
This text is meant for novices who need a complete introduction to causality and causal inference strategies, defined with minimal math.
On the subject of causality, we merely can’t keep away from this basic assertion: “Correlation doesn’t indicate causation.” And a basic instance is that simply because ice cream gross sales and drowning incidents are correlated, one doesn’t trigger the opposite. You’ve in all probability heard many such examples illustrating the distinction between the 2. Whereas these examples are sometimes simple, the excellence can grow to be blurred in precise analyses.
With no clear understanding of how causality is measured, it’s simple to make incorrect causal inferences. On this regard, one query I usually encounter is, “Sure, we all know that correlation doesn’t imply causation, however what a few regression evaluation?”. The brief reply is that linear regression, by default, doesn’t present any causal statements until we undertake acceptable steps — that is the place causal inference strategies come into play.