2017
DOI: 10.31234/osf.io/v3zkt
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

Equivalence Testing for Psychological Research: A Tutorial

Abstract: Psychologists must be able to test both for the presence of an effect and for the absence of an effect. In addition to testing against zero, researchers can use the Two One-Sided Tests (TOST) procedure to test for equivalence and reject the presence of a smallest effect size of interest (SESOI). TOST can be used to determine if an observed effect is surprisingly small, given that a true effect at least as large as the SESOI exists. We explain a range of approaches to determine the SESOI in psychological scienc… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

3
285
0
1

Year Published

2019
2019
2022
2022

Publication Types

Select...
8

Relationship

0
8

Authors

Journals

citations
Cited by 216 publications
(289 citation statements)
references
References 21 publications
3
285
0
1
Order By: Relevance
“…Third, the present study employed two different devices to record resting‐state heart rate. The TOST results showed differences in the two sets of HRV data collected [all P ‐values >0.05; Lakens et al, ]. We cannot determine whether this is related to either the HRV devices or differences in the individuals.…”
Section: Discussionmentioning
confidence: 90%
See 1 more Smart Citation
“…Third, the present study employed two different devices to record resting‐state heart rate. The TOST results showed differences in the two sets of HRV data collected [all P ‐values >0.05; Lakens et al, ]. We cannot determine whether this is related to either the HRV devices or differences in the individuals.…”
Section: Discussionmentioning
confidence: 90%
“…There were no interaction effects with group for any indices of HRV by the device and there was no main effect of device variable (see supplementary Tables S1a,b and Table ). The equivalence test used a two‐one sided (TOST) analysis [Lakens, Scheel, & Isager, ] and the equivalence bounds suggested differences between the data acquired by the two different devices or populations. Log 10 SDNN( t (61) = 0.09, P = 0.59,equivalence bound −0.069 and 0.069), log 10 RMSSD, ( t (44) = 0.01, P = 0.49, equivalence bound −0.12 and 0.12), log 10 HF‐HRV ( t (60) = −0.82, P = 0.79, equivalence bound −0.27 and 0.27), and log 10 LF ( t (41) = −0.53, P = 0.70, equivalence bound −0.03 and 0.33) given P ‐value of 0.05.…”
Section: Methodsmentioning
confidence: 99%
“…Hypothesis 2 assumed that color‐evasion would be unrelated to cognitive, metacognitive, and behavioral CQ. Since conventional significance tests technically cannot test for the absence of an effect, we used equivalence testing with one‐sided t ‐tests to reject the presence of a smallest effect size of interest (SESOI) (Lakens, Scheel, & Isager, ). As these procedures are not yet available for multilevel regression, we only tested equivalence for the bivariate correlations between color‐evasion with self‐reported cognitive, metacognitive, and behavioral CQ as well as with the SJT score.…”
Section: Resultsmentioning
confidence: 99%
“…On the other hand, there are concerns about the limitations of NHST. Importantly, in NHST, nonstatistically significant results are typically inconclusive, so researchers cannot accept a null hypothesis (but see Machery, , and Lakens, Scheel, & Isager, ). And if we cannot accept a null hypothesis, then it is harder to evaluate and publish failed replication attempts.…”
Section: Causes Of the Replicability Crisismentioning
confidence: 99%