Handles few samples

Option	Description
Yes CT design	1 Control vs 1 Treatment design or self‑controlled designs that work with very small n.
Yes ≤ 10	Can handle 10 samples or less reliably.
10 to 100	Requires between ten and one hundred samples.
No	Fails or produces unstable estimates when n is small.

Definition

Whether the method remains effective when the total number of units / observation sites (n) is small.

Explanation

In contexts with limited data (like rare events or expensive field observations), you need methods that do not break down under a few samples. Some designs can still produce valid inference, whereas many machine‑learning approaches require large samples to avoid over‑fitting.

Tools/rationale for helping assessment

Compare directly against threshold rough estimates.

Example

An ecologist with just 6 camera-trap sites finds standard GAMs unstable, so they choose a Control Treatment design proven for very small n.