Different y-axis scaling with the twoway by() option

[Klaus Steitzel]

Dear Statalist

I have a figure that kind of looks like this (for code see below):

A reviewer is now saying -- and I see their point -- that I should plot no. 3 with a different, say log axis scaling. Is there any way to do this within the -twoway, by()- option framework?

I can imagine doing what the reviewer wants with -graph combine-, but unlike the example I have given, my actual figure is very large and complex and I can already see that going the -graph combine- route will be a total nightmare leading to results that no one will be happy with.

Thanks
Klaus

clear
input id x y
1 1 .1
1 2 .2
1 3 .3
2 1 .3
2 2 .2
2 3 .3
3 1 .008
3 2 .08
3 3 .8
end

twoway line y x, by(id, col(3))

[Nick Cox]

Log scale applies to all panels or none. Perhaps there is a case for a compromise scale that makes equal sense to all panels, say square root or cube root.

Otherwise if I were in your situation, I would fight back with a reasoned response. I have had all the roles (reader, author, reviewer, editor) and respect authors who explain why the reviewer is wrong or being impractical.

[Klaus Steitzel]

Thank you very much, Nick, I had not considered finding a compromise scale, I'll see if I can find one. I actually agree with the reviewer's point, I'm just a bit sad that solving the problem seems to be much easier with ggplot than with -twoway, by()-.

Cheers
KS