Dropping subgroups with no observations (graphbar)

[Paolo Maldini]

I have a data structured as follows, and I am using the Stata graph bar guide here.

```
dataex job_search_duration educ_institution gender

----------------------- copy starting from the next line -----------------------
[CODE]
* Example generated by -dataex-. For more info, type help dataex
clear
input float job_search_duration str103 educ_institution str8 gender
1 "جامعة جازان" "ذكر"
1 "جامعة جازان" "ذكر"
1 "جامعة جازان" "ذكر"
1 "جامعة جازان" "ذكر"
1 "الكليات التقنية" "ذكر"
0 "الكليات التقنية" "ذكر"
1 "جامعة جازان" "ذكر"
```

I am trying to exclude subgroups without any observations from appearing in the bar graph, so I ran the code below.
I thought that by adding (drop if educ_institution==""), I would be able to drop such subgroups but the graph as you can

```
preserve
drop if job_search_duration<1
drop if educ_institution==""
graph hbar (mean) job_search_duration, over(educ_attainment, sort(1) descending) over(gender)
restore
```

[Nick Cox]

Sorry, but I can't read this language (Arabic?), but I guess you need the nofill option.

Note that if I am right the problem is not missing values -- observations in the dataset with missing values on one or more variables -- but absent values -- observations not in the dataset for category combinations that graph bar can't find.

[Paolo Maldini]

Thanks Nick, indeed this is Arabic. Your solution worked well!
```
preserve
drop if job_search_duration<1
drop if educ_institution==""
graph hbar (mean) job_search_duration, over(educ_attainment, sort(1) descending) over(gender) nofill
restore
```