Catplot over option

[Nettie vd Merwe]

Hi Statalist

I am using Stata/SE 17.0. I want to make a graph using the user-written demand catplot.

Let's say I have a variable called sample_groups that splits my sample into three groups and then I have a categorical variable that we can call category_var numbered from 1 to 5 (it is numbered numerically but each number shows a different category, e.g. 1 = like it very much, 2 = like it a bit, 3 = neutral,, etc)

you can use this to generate some data

clear
set obs 120
egen sample_groups = seq(), from(1) to(3)
generate category_var = runiformint(1,5)



I want to create a graph that looks something like this:

Meaning, the count in each category over the sub-samples shown in different colours

I tried:
catplot category_var , over(sample)
but that just makes three graphs (almost the same as using the by option)


Any advice would be super helpful!

[Nick Cox]

Thanks for the clear data example.

catplotis from SSC. (It's where commands come from that should be mentioned.)


For what you want, you need to offer sample_groups as if it were the outcome, regardless of how you think of it.

Although catplot supports over() by virtue of being a wrapper for (in this case) graph hbar, the syntax was intended to follow a pattern

catplot outcome [predictors] , [options]

clear
set obs 120
egen sample_groups = seq(), from(1) to(3)
generate category_var = runiformint(1,5)

catplot sample_groups category_var,  asyvars

[Nettie vd Merwe]

Works like magic!! Thanks Nick - exactly what I needed