decile by decile graph

[Julian Schaper]

Hey guys,

I am looking for a way to plot income deciles on emission deciles. I have about 10.000 observations and created deciles as follows:

*make income deciles
xtile net_income_month_dec = net_income_month, nq(10)

*make deciles for emissions
xtile e_md_dec = e_md, nq(10)

If I now plot them using

scatter e_md_dec net_income_month_dec

I get something that looks like a chess board, such that x,y = 1,1; 1,2; ...; 1,10. x,y = 2,1; 2,2; ...; 2,10 and so on and that is obviously not what I want.

What I would like is a scatter in which every dot represents 1% of the population and which makes visible the different levels of dispersion in emission levels among the income deciles. I guess I took a wrong turn in thinking somewhere and I am thankful for another idea!

Best,
Julian

[Nick Cox]

Your design is to me dubious as in practice many cells will contain (much) less than 1% of the total. You perhaps need some kind of bar chart or heat map.

[Julian Schaper]

Dear Nick,

I would like something like in the image below, only that I would like to have 10 points per decile instead of a boxplot.

[Nick Cox]

If you only want ten points per decile group of X, what are they to be e.g.

1. 5(10)95 percent points of Y as the median of each decile group on Y

2. Means of each decile group on Y for each decile on X.

The second is going to be something like

egen toshow = mean(emd), by(e_md_dec net_income_month_dec)
egen tag = tag(e_md_dec net_income_month_dec)
scatter toshow net_income_dec if tag,  xla(1/10) ytitle("`: var label emd'")

and the first just needs median() to be used, not mean().