glcurve

[Zuhumnan Dapel]

Dear All,
I'm exploring 5-year stacked dataset. Trying to compare distribution over this period using absolute Lorenz curves. I used

glcurve income , split by(year)

. I got a graph with this but without the line of perfect equality (similar to 45-degree line). Any help on how to add this?

Thanks,
Dapel

[Nick Cox]

Please explain where user-written programs you refer to come from, as requested in the FAQ Advice.

glcurve is from SJ/SSC.

This is a documented example in the help. Use the option

plot(function equality = x)

[Stephen Jenkins]

As the help file for glcurve states:

The graphs drawn by glcurve are relatively basic. For graphs with full user control over formatting and
labelling, users are recommended to use glcurve to generate the ordinates of the graph required using the
pvar(newvarname) and glvar(newvarname) options, and then to draw the graph using graph twoway.

This is how you can add things such as the line of "perfect equality".
You can find illustrative code in, for example, the appendix to https://www.iser.essex.ac.uk/files/i...rs/2005-05.pdf
NB As per Forum FAQ, you should say where you found user-written programs. (glcurve is on SSC.)

[Zuhumnan Dapel]

Thanks Dr Cox.

Here is are some

glcurve pce [aw = popwt2] , by(year) split pvar(prl) glvar(rl) lorenz nograph 
lab var rl_1980 "1980"
lab var rl_1985 "1985"
lab var rl_1992 "1992"
lab var rl_1996 "1996"
lab var rl_2004 "2004"
lab var rl_2010 "2010"

lab var prl"Cumulative population share"
sort prl
graph twoway(line rl_1980 prl, yaxis(1 2)) ///
(line rl_1985 prl, yaxis(1 2)) ///
(line rl_1992 prl, yaxis(1 2)) ///
(line rl_1996 prl, yaxis(1 2)) ///
(line rl_2004 prl, yaxis(1 2)) ///
(line rl_2010 prl, yaxis(1 2)) ///
(function y = pce, range(0 1) yaxis(1 2))///
,aspect(1) xtitle("Cumulativepopulation share, p") ///
ytitle("Incomeshare of poorest 100p%", axis(1)) ytitle(" ",axis(2)) ///
legend(label(1 "1980") label(2 "1985") label(3 "1992") label(4 "1996") label(5 "2004") label(6 "2010")label(7 "Equality") ///
region(lstyle(none))) saving(rl80_10, replace)

.

I got

. graph twoway(line rl_1980 prl, yaxis(1 2)) ///
> (line rl_1985 prl, yaxis(1 2)) ///
> (line rl_1992 prl, yaxis(1 2)) ///
> (line rl_1996 prl, yaxis(1 2)) ///
> (line rl_2004 prl, yaxis(1 2)) ///
> (line rl_2010 prl, yaxis(1 2)) ///
> (function y = pce, range(0 1) yaxis(1 2))///
) required
r(100);

end of do-file

r(100);

. Still needing some help

[Zuhumnan Dapel]

Thank Prof Jenkins. I have followed this in post #4. Still needing some help

[Nick Cox]

I guess Stata is getting confused. The command twoway function should make reference to a generic x.

It's incidental, but you can have several y variables in a line plot.

[Zuhumnan Dapel]

Dear Dr Cox, I'm yet to get that

[Stephen Jenkins]

You are asking readers to look at diagnose detailed errors in code when we don't have access to your data -- which is quite a Big Ask. In this sort of situation (relatively lengthy graph code), it is usually best to add each line one at time (or remove selected lines) in order to isolate where the problem(s) lie. In your particular case, look at the line

(function y = pce, range(0 1) yaxis(1 2))///

I suspect that you require a space after the last ")" and before the "///".

[Zuhumnan Dapel]

Thank you Prof Jenkins. I have confirmed that

glcurve pce [aw = popwt2] , by(year) split pvar(prl) glvar(rl) lorenz nograph 
lab var rl_1980 "1980"
lab var rl_1985 "1985"
lab var rl_1992 "1992"
lab var rl_1996 "1996"
lab var rl_2004 "2004"
lab var rl_2010 "2010"

lab var prl"Cumulative population share"
sort prl
graph twoway(line rl_1980 prl, yaxis(1 2)) ///
(line rl_1985 prl, yaxis(1 2)) ///
(line rl_1992 prl, yaxis(1 2)) ///
(line rl_1996 prl, yaxis(1 2)) ///
(line rl_2004 prl, yaxis(1 2)) ///
(line rl_2010 prl, yaxis(1 2)) ///
(function y = pce, range(0 1) yaxis(1 2) ) ///
,aspect(1) xtitle("Cumulativepopulation share, p") ///
ytitle("Incomeshare of poorest 100p%", axis(1)) ytitle(" ",axis(2)) ///
legend(label(1 "1980") label(2 "1985") label(3 "1992") label(4 "1996") label(5 "2004") label(6 "2010")label(7 "Equality") ///
region(lstyle(none))) saving(rl80_10, replace)

is OK.

The problem is with the last command line

region(lstyle(none))) saving(rl80_10, replace)

I still find it difficult to resolve this.

[Nick Cox]

The specific problem here is that your parentheses in

  
region(lstyle(none)))

don't match as you have two (( and three ))).

We've already made suggestions to simplify your line call and correct the call to function which you haven't yet implemented. With some further simplifications and corrections, consider something like this:

twoway function y = x, range(0 1)  ///
|| line rl_1980 rl_1985 r1_1992 r1_1996 r1_2004 r1_2010 prl, yaxis(1 2) ///
aspect(1) xtitle("Cumulative population share, p") ///
ytitle("Income share of poorest 100p%", axis(1)) ytitle(" ",axis(2)) ///
legend(order(2 "1980" 3 "1985" 4 "1992" 5 "1996" 6 "2004" 7 "2010" 1 "Equality") ///
region(lstyle(none)) saving(rl80_10, replace)

[Zuhumnan Dapel]

Thanks Dr Cox. Your suggest worked but I added ) to

 region(lstyle(none))) saving(rl80_103, replace)

so that we now have

glcurve pce [aw = popwt2] , by(year) split pvar(prl) glvar(rl) lorenz nograph
lab var rl_1980 "1980"
lab var rl_1985 "1985"
lab var rl_1992 "1992"
lab var rl_1996 "1996"
lab var rl_2004 "2004"
lab var rl_2010 "2010"

twoway function y = x, range(0 1)  ///
|| line rl_1980 rl_1985 rl_1992 rl_1996 rl_2004 rl_2010 prl, yaxis(1 2) ///
aspect(1) xtitle("Cumulative population share, p") ///
ytitle("Income share of poorest 100p%", axis(1)) ytitle(" ",axis(2)) ///
legend(order(2 "1980" 3 "1985" 4 "1992" 5 "1996" 6 "2004" 7 "2010" 1 "Equality") ///
region(lstyle(none))) saving(rl80_103, replace)

Dapel

[Nick Cox]

I see now that region() qualifies the legend(). I guessed wrongly that you wanted otherwise. In that case, there must have been some other mistake that has been fixed now.

[Zuhumnan Dapel]

Dear All,
How do I increase the with between the curves in such away that the trend being depicted will be very clear?

Thanks,
Dapel