Equivalence Testing
Use Equivalence Tests to determine if treatments are effectively the same. The same AOV letters imply that treatments are "not significantly different" but this is not the same as "significantly the same" which is the goal of Equivalence Tests.
Step 1: Researcher defines how large of a difference is insignificant
- This Limit creates an equivalence interval about the standard treatment for comparison.
Step 2: Hypothesis Test: the difference between treatments is outside the equivalence interval
- Two one-sided tests determine if the difference is "outside the interval"
Define tests on Settings > Statistics tab
Limit can be entered as:
- Percent of Standard
- Absolute Value
- Cohen's d
(a measure of effect size)
Analysis is performed when AOV option to Include equivalent tests is selected.
P-value for "smaller than lower limit" (1) and "larger than higher limit" are listed. Need both to be significant to conclude equivalence.
See
Equivalence Tests for more examples from literature and real-life, as well as statistical calculations and theory.
