30. A study finds no statistically significant difference. Can it conclude that two methods are equivalent?
HardStatistics & inference
Technical
Show model answer
Not from that result alone. The study may be too imprecise to detect meaningful differences. Equivalence requires an appropriate design and a justified margin defining differences small enough for the intended purpose.
Examine the estimated difference and uncertainty relative to that margin, using the planned analysis. Do not choose the margin after seeing the results merely to obtain equivalence or describe lack of evidence as proof of sameness.
What interviewers listen for
- Non-significance is not equivalence
- Justify a meaningful margin
- Assess precision against the margin