Replicating a paper's Stacked Horizontal Bar graph

[Meshal Alkhowaiter]

I have the following data:

```
dataex nationality economic_activity change
input str15 nationality str33 economic_activity long change
"foreign" "Agriculture, forestry and fishing" 4945
"foreign" "Mining and quarrying" -13730
"foreign" "Manufacturing" -169127
"foreign" "Construction" -1802976
"foreign" "Retail/trade" -600744
"foreign" "Total (all sectors)" -1857160
"national_male" "Agriculture, forestry and fishing" -961
"national_male" "Mining and quarrying" 11407
"national_male" "Manufacturing" 528
"national_male" "Construction" -123294
"national_male" "Retail/trade" 5815
"national_male" "Total (all sectors)" 23689
"national_female" "Agriculture, forestry and fishing" 407
"national_female" "Mining and quarrying" 1920
```

And I am trying to a graph similar to the one below:




My current code is below:

```
graph hbar (mean) change, over( nationality ) over( economic_activity )
```
which produces the graph in the file attached, which is close to the graph above, but not quite the same.

[Nick Cox]

There are some errors in the data. 4945 should be -4945. -123934 should be 123934. There is no total for Saudi females.


This may help:

clear
input str15 nationality str33 economic_activity long change
"foreign" "Agriculture, forestry and fishing" -4945
"foreign" "Mining and quarrying" -13730
"foreign" "Manufacturing" -169127
"foreign" "Construction" -1802976
"foreign" "Retail/trade" -600744
"foreign" "Total (all sectors)" -1857160
"national_male" "Agriculture, forestry and fishing" -961
"national_male" "Mining and quarrying" 11407
"national_male" "Manufacturing" 528
"national_male" "Construction" 123294
"national_male" "Retail/trade" 5815
"national_male" "Total (all sectors)" 23689
"national_female" "Agriculture, forestry and fishing" 407
"national_female" "Mining and quarrying" 1920
"national_female" "Total (all sectors)" 2327
end

separate change, by(nationality) veryshortlabel
label variable change1 "Non-Saudis"
label variable change2 "Saudi women"
label variable change3 "Saudi men"
graph hbar (mean) change1 change3 change2,  over(economic_activity) stack legend(order(1 "Non-Saudis" 2 "Saudi men" 3 "Saudi women"))  ///
bar(1, color(ltblue)) bar(2, color(navy)) bar(3, color(orange_red)) yla(-2e6 "2 m" -15e5 "1.5 m" -1e6 "1 m" -5e5 "500000" 0)  ytitle(Change)

The unstacked version looks clearer to me.

Please note our request here for full real names, typically a family name and a given name. This was flagged previously to you in https://www.statalist.org/forums/for...n-monthly-data