# Use a contemporaneous comparison carefully

Illustrative planning brief; no automatic product import.

Synthetic planning case: In a toy pilot, a changed-protocol group goes from ten to eight mean minutes. A contemporaneous unchanged-protocol group goes from twelve to eleven. Their comparability is not established.

## Decision

The changed group decreases two minutes and comparison group one; the difference of changes is minus one minute using after-minus-before. Keep the comparability assumptions unresolved rather than declaring a causal effect.

## Owned work

- Confirm contemporaneous windows
  - Owner role: Pilot owner
  - Acceptance evidence: Both groups’ observations use the same declared before/after periods.
- Calculate both changes
  - Owner role: Reviewer
  - Acceptance evidence: The unchanged group’s one-minute decrease is not ignored.
- Review inference assumptions
  - Owner role: Method owner
  - Acceptance evidence: Group comparability and other changes remain explicit before any causal interpretation.

## Workflow

1. Confirm contemporaneous windows. Check: Both groups’ observations use the same declared before/after periods.
2. Calculate both changes. Check: The unchanged group’s one-minute decrease is not ignored.
3. Review inference assumptions. Check: Group comparability and other changes remain explicit before any causal interpretation.

## Judgment

This is an arithmetic example, not an identified causal estimate. No population, trend or assignment evidence is supplied.


## Filled manual planning note

The changed group decreases two minutes and comparison group one; the difference of changes is minus one minute using after-minus-before. Keep the comparability assumptions unresolved rather than declaring a causal effect. Difference of changes=(8−10)−(11−12)=−2−(−1)=−1 minute. A shared time period alone does not prove the groups would otherwise have moved comparably. This is an arithmetic example, not an identified causal estimate. No population, trend or assignment evidence is supplied.


## Before / after

Before: Changed group drops two minutes; that alone is called the protocol effect.

After: The comparison group drops one too; the minus-one change difference is labelled descriptive with unverified assumptions.


## Workflow questions

### Does a comparison group automatically establish causation?

No. Selection, different trends and other interventions can invalidate the intended comparison.

### Why retain the unchanged group’s movement?

It shows that time-related changes also occurred outside the changed protocol; interpretation still requires a justified design.

## Product connection

Use the owned checks and downloaded brief to discuss this planning decision alongside your TeamBoostAI tasks. Confirm available fields, roles and account features separately. The example is manual; it does not calculate live analytics, create work or run an experiment in the product.

Confirm account availability before adopting this manual outline.

## Original worked case

Synthetic records, manual planning only. No account import or live analytics.

### Inspect the invented case records

Group | Before mean minutes | After mean minutes | After minus before
--- | --- | --- | ---
Changed protocol | 10 | 8 | −2
Unchanged protocol | 12 | 11 | −1

### Reasoning

Difference of changes=(8−10)−(11−12)=−2−(−1)=−1 minute. A shared time period alone does not prove the groups would otherwise have moved comparably.

### Bounded result

The changed group decreases two minutes and comparison group one; the difference of changes is minus one minute using after-minus-before. Keep the comparability assumptions unresolved rather than declaring a causal effect.

### Distinct decision

The artifact inspects two group changes and their inference boundary, beyond a single before/after result.

### Limits

This is an arithmetic example, not an identified causal estimate. No population, trend or assignment evidence is supplied.

### Definitions and method context

- NIST combinatorial testing — https://www.nist.gov/publications/combinatorial-testing — Combination-coverage context. Invented review/state examples do not prove full fault coverage, certification or product behavior.