Drawing a 2 Y-axis graph with an X variable in string format

[Aziz Essouaied]

Hello Stata community;

I dispose of this data:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str7 date float(unemployment employment)
"Q1 2008"   5  71.50778
"Q2 2008" 5.3  71.25295
"Q3 2008"   6  70.73177
"Q4 2008" 6.9  69.96442
"Q1 2009" 8.3  68.65476
"Q2 2009" 9.3   67.8991
"Q3 2009" 9.6 67.302895
"Q4 2009" 9.9   66.5618
"Q1 2010" 9.8  66.73094
"Q2 2010" 9.6  66.79226
"Q3 2010" 9.5 66.724525
"Q4 2010" 9.5  66.44049
"Q1 2011"   9 66.655556
"Q2 2011" 9.1  66.49724
"Q3 2011"   9  66.58442
"Q4 2011" 8.6 66.779106
"Q1 2012" 8.3  66.96114
end

My goal is to draw a 2 Y axish graph (my two Y variables are unemployment & employment) using the X variable date. The problem is that this date variable is in string format.
I've tried this code: graph twoway (scatter unemployment date, connect(direct) msymbol(circle) msize(large) mcolor(green) yaxis(1) ytitle("Unemployment", axis(1))) (scatter employment date, connect(direct) msymbol(diamond) msize(small) mcolor(black) yaxis(2) ytitle("Employment", axis(2))), but I guess it didn't work out because the date variable is in string format.

Should I change the date variable format? If so, how? Or, is there another way to get to my goal?

Thanks for your help.

[Nick Cox]

Your date is a string variable; but that is not a format in Stata terms. Either way, dates are documented in detail at

help datetime

which should be consulted here. Once you have a numeric date variable, you have a choice of graphical representation. I recommend multiline from SSC over any abomination with two y axes, hard to get right and all too easy to make cryptic or confusing.


More at https://www.statalist.org/forums/for...ailable-on-ssc

* Example generated by -dataex-. To install: ssc install dataex
clear
input str7 date float(unemployment employment)
"Q1 2008"   5  71.50778
"Q2 2008" 5.3  71.25295
"Q3 2008"   6  70.73177
"Q4 2008" 6.9  69.96442
"Q1 2009" 8.3  68.65476
"Q2 2009" 9.3   67.8991
"Q3 2009" 9.6 67.302895
"Q4 2009" 9.9   66.5618
"Q1 2010" 9.8  66.73094
"Q2 2010" 9.6  66.79226
"Q3 2010" 9.5 66.724525
"Q4 2010" 9.5  66.44049
"Q1 2011"   9 66.655556
"Q2 2011" 9.1  66.49724
"Q3 2011"   9  66.58442
"Q4 2011" 8.6 66.779106
"Q1 2012" 8.3  66.96114
end

gen qdate = quarterly(substr(date, 2, .), "QY")
* ssc install multiline

set scheme s1color
multiline u e qdate, xla(192(4)208, grid glc(gs12) format(%tq)) yla(#6) recast(connected) xtitle("")

[Nick Cox]

The graph doesn't show up well in what I can see. Here it is again.

[Aziz Essouaied]

Thanks very much for the help, I appreciate it.
Yet, my goal is to have the two lines on the same graph actually, because I think it's better for comparative and interpretation reasons.

[Nick Cox]

Well, in that case I solved about half of your problem. Good luck with the other half.

Wainer, H. 1991. Double Y-axis graphs. Chance 4(1): 50-51 is pertinent.

[Aziz Essouaied]

Thanks very much