I am interested in understanding if two predictors variables are associated with treatment outcome (DV). I included two covariates that 1) differed across my levels of IV (age) and 2) another that is strongly associated with the DV (baseline functioning). I ran a simultaneous multiple regression with the covariates and both IVs entered into the same time to estimate the unique relationships between the IVs with the DV.

A reviewer implied that hierarchical regression was needed (enter covariates into step 1 and IVs into step 2. My understanding is this would be asking a different question and that it is totally appropriate to include covariates in a single step in simultaneous regression.

Another reviewer suggested that only ANCOVA was appropriate to address this question. I'm pretty confident that this is a consequence of the reviewer being trained in ANOVA but not regression models, and is a misunderstanding.

Welcoming feedback and any academic citations to support the statistical approach.

  • $\begingroup$ Welcome to Cross Validated? Is an IV an instrumental variable in this case? $\endgroup$
    – Dave
    Feb 29 at 15:33

1 Answer 1


Why not do the first reviewer the favor and run the regression with just the covariates in Step 1, then add the substantive predictors in Step 2? From my perspective, I don't see how it could hurt anything. Or, put differently, it makes the reviewer happy and (probably) does not make you unhappy ;-)

ANCOVA would be equivalent to linear regression (both fall under the general linear model) and you would get the exact same results, so that other reviewer's point is unclear to me.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.