Separate an anomaly from a structural change

Why this lab Separate a one-off anomaly from a structural change in the signal.

One unusual point may be a transient shock, measurement problem, or genuine event. A level shift persists. Inspect the before-and-after evidence and classify each pattern before deciding how to annotate it.

Timing is not cause. When a level shift lines up with a date, such as a new policy, the chart shows the timing: the change happened after that date. Alignment is not cause. Other events may share the date, and a single series has no comparison group showing what would have happened without the policy. Caption the shift as a change aligned with the date, and leave the causal claim to a design that can support it.