Testing Equivalence in Communication Research: Theory and Application

Abstract

Although equivalence testing is preferred when a researcher's goal is to support the null hypothesis (i.e., no substantial effect), equivalence tests are virtually unknown and unused in the communication field. This article provides the rationale for and theoretical background of equivalence testing and offers examples of equivalence tests for the independent and dependent groups t-test and tests of association using Pearson's coefficient or correlation. From a review of meta-analyses, we provide tables of commonly observed effect-sizes across subdisciplines and topic areas in communication and offer these as a guideline for choosing minimum substantial effects (Δ) in equivalence testing when no other information source is available. To facilitate the adoption of equivalence tests in future research, we provide easy-to-use custom dialogs for SPSS which greatly simplify their computation and application.

Notes

¹A list of all included meta-analyses is available from the authors upon request.

²The SPSS database with all effect sizes is available from the authors upon request.

³It should be noted that recent publications in the field of Psychology suggest using a Bayesian approach, and particularly a Bayes Factor Analysis, for the test of alternative theoretical models that may include a null-model, i.e., a model that predicts effects that are inconsequential on practical grounds (Dienes, 2011). Bayesian effect- and equivalence testing is currently becoming more accessible with the development of free software (e.g., WinBUGS; Lunn, Spiegelhalter, Thomas, & Best, 2009) and the availability of online calculators (e.g., http://pcl.missouri.edu/bayesfactor). While we see the clear benefits of such an approach (e.g., the fact that Bayes Factors avoid irrational test-decisions; Bayes factors are insensitive to multiple testing and the timing of statistical testing), a Bayesian test for statistical equivalence in communication research is beyond the scope of this paper. For more information on a Bayesian approach for the test of statistical equivalence, see Rouder, Speckman, Sun, Morey, and Iverson (2009).