Small Treatment Groups Limited the THC–CBD Trial’s Null Findings
The trial’s planning calculation targeted a large standardized difference. With fifteen people per treatment group, a nonsignificant comparison should not be treated as proof that a smaller effect was absent.
What the calculation described
The paper stated that fifteen participants per group could detect a standardized difference of approximately 1.1 with 80% power using a two-sided, two-sample Wilcoxon comparison at a 5% error level, without multiplicity correction in that planning calculation. The completed experiment had four groups and several cognitive and emotional outcomes.
Planning and final analysis were different steps
The reported analyses used a closed-testing procedure to control familywise error. That is distinct from assuming the original planning calculation guaranteed equal sensitivity for every final contrast. Power depends on the effect being tested, variability, analysis and sample size. An experiment can find some effects while leaving smaller differences poorly resolved.
What a null comparison supports
The results did not establish that the tested CBD pretreatment prevented THC effects. A particular nonsignificant contrast also does not demonstrate equivalence between treatments. An equivalence claim would require its own margin and supporting analysis. These limits matter when a group result is used to assert that an individual experienced no effect, or that all alternative dose combinations must behave the same way.
Related analysis: the oral CBD pretreatment trial.
Research cited
Woelfl and colleagues (2020). Effects of Cannabidiol and Delta-9-Tetrahydrocannabinol on Emotion, Cognition, and Attention: A Double-Blind, Placebo-Controlled, Randomized Experimental Trial in Healthy Volunteers. DOI: 10.3389/fpsyt.2020.576877.
