Using group_by with multiple factors in a calculation of z-scores

[FunnyBikeRider]

I am trying to calculate z scores (in order to detect outliers) for a number of groups in my data set split by two different factors (e.g., a study in which 2 distinct groups of subjects participated in a listening task in two different conditions). I see that splitting by one factor is straightforward, with a command like Z(uLevs, group_by=CondCode) but is it not possible to split by more than one factor? This is a very common situation and I cannot find any relevant documentation about it. Thanks!

[jonathon]

so one way you could do it is create a column with a level for each group combo. so with the Tooth Growth data set for example, we could create a new computed column with the formula `supp + dose` ... this will result in a column with 6 levels with values like `VC 500`, `OJ 2000`, etc. and then you could use that column with a group_by. (note that the supp + dose, only works because one of the columns is text ... if both supp and dose were numeric, you could have to convert one or both of them to text, i.e. TEXT(supp) + dose.)

or, you could do it all in one formula like this:

Screenshot 2025-08-06 at 10.05.46.png (168.29 KiB) Viewed 3853 times