suspected of being problematic, one is a common but questionable model, and an adjustment may be required to address all the nonsignificant differencesβ how they nonetheless constitute important arbitrary criterions for equivalence
the significance test based on observational data is susceptible to (errors of) interpretation over the question at issue namely, do case differences arise because of exposure to a comparatively small sample or because of another variable?
Exposure can be only mediated by crude estimates and so may be misleading during the forming of the hypothesized model of one that describes the association between exposure, bias, and the variables, and reconciles difference with equivalence significantly.
The model provides little information that is incontrovertible but the results suggest if adjustment for the variable makes no substantive difference ignore it
but if your knowledge indicates the adjusted variable to be preferable