Why this is mistaken
An observed difference is not a causal effect. Establish timing, assignment or comparison design, confounders, mechanisms, and rival explanations before attribution. Historical association also does not become a forecast or policy choice automatically.
Where to watch
When this mistake may appear
- A period with a policy change also has a different sample mean.
- Two variables move together.
Check your work
Your work may contain this mistake if:
- Says Juniper's policy caused faster settlement from two year samples alone.
- Converts historical association into a forecast or recommendation.