statistical equivalence
What is statistical equivalence and how do researchers check for it?
Statistical equivalence is a concept in statistics that suggests that two or more groups or samples being compared have no meaningful difference. In other words, the groups are considered ‘equivalent’ from a statistical standpoint.
Determining statistical equivalence often involves constructing confidence intervals around the difference between group means and then assessing whether this interval falls within a pre-specified range of practical equivalence (ROPE). If it does, the groups are considered statistically equivalent.
Researchers check for statistical equivalence usually in two ways:
-
TOST (Two One-Sided Tests) Procedure: This method involves conducting two one-sided hypothesis tests. The null hypothesis is that the true mean difference is either greater than the positive limit or less than the negative limit of the ROPE. If we reject both null hypotheses, we conclude that the groups are statistically equivalent.
-
Confidence Interval Approach: Here, researchers calculate a confidence interval for the mean difference between groups. If this interval falls completely within the ROPE, we conclude that the groups are equivalent.
These methods require researchers to define beforehand what they consider to be a ‘practically significant’ difference, which can sometimes be subjective and depends on the context of research.
#researchlandscape #inferentialstatistics