Having Y-axis in exponent scales

[Ofir Gefen]

Dear Statalist,

I've been looking at the graphs manual and couldn't find an answer to this.

I am reporting the mean of a variable (on the Y-axis) by year (on the X-axis).
The mean varies hugely, from ~0 to ~500. If I do nothing, STATA's output isn't very informative, and you cannot make anything from it. However, I do not want to take the Natural log or scale that variable.

In R, one can scale the Y-axis itself so instead of normal increments of say 0, 10, 20, 30 it will grow exponentially: 2, 4, 8, 16, 32 etc, with the space space between each tick.

Can you please tell me if/how I can do that in STATA? (I am using the latest version).STATA is giving me this.docx

I'm attaching a drawing of my issue. I hope it will make it clearer.

Thank you in advance.

[Andrew Musau]

Please upload images in .png format, as advised in the FAQs.

webuse grunfeld, clear
line invest year if company==1, ysc(log) scheme(s1mono) ylab(200 (200) 800 1200 1500)

[Nick Cox]

If your data really include 0, then a logarithmic scale won't work any way, although there are twists you can apply. I don't see the connection between 2 4 8 16 32 and your description of your data. Negative values (alone) could be plotted in terms of -log(-data).