Skip to contents

Get started with pyramid_chart

Source: vignettes/pyramid_chart.Rmd
pyramid_chart.Rmd

pyramid_chart turns three variables into a population pyramid: a numeric value, the categories stacked on the vertical axis, and the two groups drawn left and right. This page walks through preparing the data and then builds up the chart one option at a time. All code is Stata.

Preparing the data

The command needs long data: one row per category × group. Stata’s bundled pop2000 dataset is wide (separate maletotal and femtotal columns), so reshape it first.

sysuse pop2000, clear
keep agegrp maletotal femtotal
rename maletotal population1
rename femtotal  population2
reshape long population, i(agegrp) j(sex)
label define sexlbl 1 "Male" 2 "Female"
label values sex sexlbl
label variable population "Population"
label variable sex "Sex"
describe
Variable Storage Display format Value label Variable label
agegrp float %12.0gc agelbl Age group
sex byte %10.0g sexlbl Sex
population float %12.0gc Population

Two requirements matter:

  • The by() variable must be coded 1 and 2, each with a value label. Level 1 is drawn on the left, level 2 on the right, and the labels become the legend.
  • The over() variable’s value labels become the vertical-axis labels, so label it before plotting.

Percentages are computed within each by() group: each bar is that category’s share of its own group’s total.

1. The minimal call

Only the three variables and the number of decimals are required.

pyramid_chart population, over(agegrp) by(sex) dec(0)

2. Titles and decimals

Any twoway title option passes straight through. dec(1) adds one decimal to the bar labels; the x-axis stays in whole numbers.

pyramid_chart population, over(agegrp) by(sex) dec(1) ///
    title("Population Pyramid for year 2000") subtitle("in percentages") ///
    xtitle("Percentage") ytitle("Age groups")

3. Styling the bar labels

The percentage labels are a scatter layer. Pass scatter options to it with sctopt().

pyramid_chart population, over(agegrp) by(sex) dec(1) ///
    title("Population Pyramid for year 2000") subtitle("in percentages") ///
    xtitle("Percentage") ytitle("Age groups") ///
    sctopt(mlabcolor(blue))

4. Styling each side

bar1() and bar2() take bar options for the left and right bars respectively, so the two groups can be coloured independently of the scheme.

pyramid_chart population, over(agegrp) by(sex) dec(1) ///
    bar1(color(navy%80)) bar2(color(orange%80)) ///
    sctopt(mlabcolor(gs4) mlabsize(small))

5. Schemes

Because everything is a regular twoway graph, any installed scheme applies.

pyramid_chart population, over(agegrp) by(sex) dec(1) ///
    title("Population Pyramid for year 2000") subtitle("in percentages") ///
    xtitle("Percentage") ytitle("Age groups") ///
    sctopt(mlabcolor(blue)) scheme(white_tableau)

pyramid_chart population, over(agegrp) by(sex) dec(1) ///
    title("Population Pyramid for year 2000") ///
    xtitle("Percentage") ytitle("Age groups") scheme(economist)

What you can’t override

The legend, x-axis labels and y-axis labels are generated from the data (the x-axis is mirrored so both sides read as positive percentages), so legend(), xlabel() and ylabel() are not intended to be changed. Everything else, including name(), saving(), note(), graphregion() and xsize()/ysize(), works as usual.

Getting help

In Stata, help pyramid_chart. Bugs and feature requests: GitHub issues.