stacked graph bar over different years and age groups

[Paula de Souza Leao Spinola]

I have the dataset below: proportion of the population (variable called prop_pop) in different age groups (variable called grupo) over different years (variable called ano). They sum up to 1 (100%) for a given year.

I would like to create a graph that shows the proportion of the population in each age group for the years available. Hence, each of the years in the horizontal axis should be a stacked bar. I am lost in how I can mix a stacked graph bar with the twoway command and the over option. Could anyone help? Many thanks!

grupo    ano    prop_pop
0-4    2000    0.1008
0-4    2010    0.0762
0-4    2020    0.0696
0-4    2030    0.0606
0-4    2040    0.0535
0-4    2050    0.0499
0-4    2060    0.0472
40-64    2000    0.2127
40-64    2010    0.2588
40-64    2020    0.2952
40-64    2030    0.3253
40-64    2040    0.3421
40-64    2050    0.3293
40-64    2060    0.3155
5-39    2000    0.6284
5-39    2010    0.5917
5-39    2020    0.5369
5-39    2030    0.4786
5-39    2040    0.4303
5-39    2050    0.4022
5-39    2060    0.3825
65-84    2000    0.0547
65-84    2010    0.0666
65-84    2020    0.0885
65-84    2030    0.1209
65-84    2040    0.1508
65-84    2050    0.1835
65-84    2060    0.2088
85+    2000    0.0033
85+    2010    0.0066
85+    2020    0.0098
85+    2030    0.0145
85+    2040    0.0232
85+    2050    0.0351
85+    2060    0.0461

[Nick Cox]

Before we can do anything serious, we need to sort out your age variable, which looks all too much like a string variable that can only sort in a dopey order, as your data example implies. (But please in future do use dataex as requested: FAQ Advice #12.)

Then I wouldn't use twoway at all -- but graph bar -- and in fact I wouldn't use graph bar really, but rather tabplot (Stata Journal). Naturally stuff in your own language (Portuguese?) is up to you.

Here's all my code:

* data 
clear 
input str5 grupo    ano    prop_pop
0-4    2000    0.1008
0-4    2010    0.0762
0-4    2020    0.0696
0-4    2030    0.0606
0-4    2040    0.0535
0-4    2050    0.0499
0-4    2060    0.0472
40-64    2000    0.2127
40-64    2010    0.2588
40-64    2020    0.2952
40-64    2030    0.3253
40-64    2040    0.3421
40-64    2050    0.3293
40-64    2060    0.3155
5-39    2000    0.6284
5-39    2010    0.5917
5-39    2020    0.5369
5-39    2030    0.4786
5-39    2040    0.4303
5-39    2050    0.4022
5-39    2060    0.3825
65-84    2000    0.0547
65-84    2010    0.0666
65-84    2020    0.0885
65-84    2030    0.1209
65-84    2040    0.1508
65-84    2050    0.1835
65-84    2060    0.2088
85+    2000    0.0033
85+    2010    0.0066
85+    2020    0.0098
85+    2030    0.0145
85+    2040    0.0232
85+    2050    0.0351
85+    2060    0.0461
end 

set scheme s1color 

* prepare stacked bar 
label def grupo 1 "0-4" 2 "5-39" 3 "40-64" 4 "65-84" 5 "85+"
encode grupo, gen(Grupo) label(grupo)

* do stacked bar 
graph bar (asis) prop_pop, over(Grupo) over(ano) asyvars stack ///
legend(row(1)) ytitle(Cumulative proportion) yla(0 1 0.25 0.5 0.75, ang(h))

* prepare tabplot 
gen pc_pop = prop_pop * 100 
label def grupo2 5 "0-4" 4 "5-39" 3 "40-64" 2 "65-84" 1 "85+"
encode grupo, gen(Grupo2) label(grupo2)

* do tabplot 
tabplot Grupo2 ano [iw=pc_pop] , subtitle(%) showval  bfcolor(ltblue) blcolor(blue) ytitle(age) xtitle("")

Notes:

1. I haven't worked at the colours for the stacked bar but using different colours is essential.

2. It is hard to read off changes in the smallest categories in the stacked bar chart and indeed it isn't really good for quantitative interpretation.

3. Depending on your readers 2 decimal places in the second graph from tabplot might be too many.

4. You can have different colours in the tabplot, but they aren't the default.

5. I love to lose the legend (kill the key) which you can do with use of the axis labels.

6. Strictly tabplot is an application of twoway, but perhaps not what you were imagining.

Many more examples on this site of which first is https://www.statalist.org/forums/for...updated-on-ssc See also the Stata Journal write-up.

[Paula de Souza Leao Spinola]

Many thanks!!