Normal values on y-axis when using logged data in marginsplot

[Annabelle Coleman]

Hello Stata List users,

I am trying to plot my predicted margins using the command marginsplot for a variable NfL that has been logged (logNfL) on the y-axis against Age on the x-axis.

I would like to run the regression and predict the margins based on the logged NfL values, however I would like to graphically represent the y-axis on a non-logged scale. Would anyone know how to do this?

Here is my code below:

regress logNfL Age CAG
levelsof Age
margins, at(Age=(`r(levels)')) saving(model_cag, replace)
est store model_cag
marginsplot, recast(line) noci plotopts(lcolor("cranberry'")) title("CAG") ytitle("NfL pg/mL")  xtitle("Age (years)") legend(off) aspectratio(1)

[Andrew Musau]

Perhaps consider poisson instead of linear regression to avoid such issues. See https://blog.stata.com/2011/08/22/us...tell-a-friend/