Module 07 Activity

Scenario

Turn a test result into something a decision-maker can act on.

What you build

An effect size in real units, with its interval, framed for a non-technical reader.

Steps

1. Report the difference in the units the reader already uses, with its interval, before any standardised measure.
2. If you report Cohen's d, give the pooled SD alongside so the raw difference can be recovered.
3. For a binary outcome, give the absolute difference in percentage points, the relative change, and one practical translation such as events per 1000 or number needed to treat.
4. If you report an odds ratio, say that it is one and give the baseline rate so a reader can judge how far it sits from a risk ratio.
5. Name the smallest effect worth acting on before analysing, and compare the whole interval to it.
6. State the result in at least two framings, one of them absolute, and carry the interval into the units of the decision.

Evidence to hand in

• The raw difference with its interval.
• Cohen's d with the pooled SD, if used.
• The absolute, relative and practical framings for any proportion.
• The odds ratio with its baseline rate, if used.
• The pre-set threshold and the verdict against it.
• The two framings with uncertainty attached.

Review checklist

• The raw difference in real units is reported, not only a standardised one.
• The practical threshold was set before the analysis.
• Any odds ratio is labelled as one and accompanied by the baseline rate.
• The confidence interval is carried into the decision's units.
• No statistically significant but practically irrelevant result is reported as a success.

Mark complete