Hi Statalist
Apologies, inevitably, if this has been covered before - I've tried searching but suspect I'm not using the correct terms.
I'm trying to draw some simple bar graphs of the frequencies of values for a particular variable A, by another variable B, while making the sizes of the labels on the x-axis tiny. I also want to do this where values of A exceed 0, as I'm using datasets where different types of missingness are coded as negative.
However, when I specify label size using graph bar, the negative codes are included on the graph. See for example:
sysuse auto
graph bar if rep78>2, over(rep78)by(foreign)
graph bar if rep78>2, over(rep78, label(labsize(tiny)))by(foreign)
This isn't a disaster - I can recode negative values to . values, but this seems like an odd quirk.
Does anyone have any suggestions?
Apologies, inevitably, if this has been covered before - I've tried searching but suspect I'm not using the correct terms.
I'm trying to draw some simple bar graphs of the frequencies of values for a particular variable A, by another variable B, while making the sizes of the labels on the x-axis tiny. I also want to do this where values of A exceed 0, as I'm using datasets where different types of missingness are coded as negative.
However, when I specify label size using graph bar, the negative codes are included on the graph. See for example:
sysuse auto
graph bar if rep78>2, over(rep78)by(foreign)
graph bar if rep78>2, over(rep78, label(labsize(tiny)))by(foreign)
This isn't a disaster - I can recode negative values to . values, but this seems like an odd quirk.
Does anyone have any suggestions?
Comment