My question might be too simple but I couldn´t figure out the answer online or by only reading the help function of metafor package. While conducting a meta-analysis trough escalc function what is the correct order for both test and control variables?

The code I´m trying to use is this:

es_mean<-escalc(measure ="SMD", m1i=Clear_mean, sd1i=Clear_dev, n1i=Clear_n, m2i=turbid_mean, sd2i=turbid_dev, n2i=turbid_n, data = effect_chl, append = T, vtype="LS")

Clear is my control environment and turbid corresponds to the effect I´m testing for. Is this the correct way to use this function, I mean, are m1, sd1 and n1 used for control and m2, sd2 and n2 used for what I´m testing?

Thanks for any help.


1 Answer 1


For measure="SMD", the escalc() function puts m1i-m2i in the numerator (which is divided by the pooled standard deviation). It is up to you to decide what you want m1i and m2i to represent. For example, if large values of the mean are a "good" thing, then using m1i for the treatment and m2i for the control group means would lead to values above 0 if the treatment is working (assuming that increasing the mean is the goal of the treatment).

  • $\begingroup$ And what about the interpretation of estimate and p-value given by the test? My model returned: Estimate: -0,32 and p-value = 0,003. Does it allow me to say that my treatment has a negative and significant effect on the variable? $\endgroup$ Dec 17, 2021 at 19:08
  • 1
    $\begingroup$ I know nothing about your data, analysis, model, etc., so I cannot help with this. This aside, this is a different question. $\endgroup$
    – Wolfgang
    Dec 17, 2021 at 20:05
  • $\begingroup$ @PatriciaNunes can you not just look at your data to see which condition has the larger mean on average? $\endgroup$
    – mdewey
    Dec 18, 2021 at 14:34

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.