Overlaying the plots instead of creating side-by-side plots.

[Sean Theriault]

I figured out how to graph confidence intervals around predicted values of the regressions (dependent variable: Legislative Effectiveness Scores).

So, I can produce the following graph using the command:
.twoway lfitci les seniority if yfe>1993, by(gk104)


I would love for this to be part of one graph instead of side-by-side graphs.

Do you know if it’s possible to do this?

Thanks!

sean

[Nick Cox]

How many categories in gk104? This sounds like a prescription for a hard-to-read image.

[Sean Theriault]

Just 2!

[Sean Theriault]

This is what the side-by-side looks like.

[Nick Cox]

So, it is something like

twoway lfitci les seniority if yfe>1993 & gk104 == 0 || lfitci les seniority if yfe>1993 & gk104 == 1, legend(order(1 "0" 2 "1"))

for your own values of 0 and 1, as it were.

[Andrew Musau]

I agree with Nick, but transparency may help here.

sysuse auto, clear
set scheme s1mono
*BY GRAPH
twoway lfitci mpg weight if rep78>2, by(foreign, leg(off) note("")) saving(gr1, replace) fc(red%20)

*COMBINED
separate mpg, by(foreign) veryshortlabel
twoway lfitci mpg0 weight if rep78>2, fc(blue%20)||lfitci mpg1 weight if rep78>2, subtitle(Combined) saving(gr2, replace) fc(red%20) leg(off)

gr combine gr1.gph gr2.gph

[Sean Theriault]

Thanks, Nick and Scott! With your help, here's what I've produced... I appreciate you taking the time to help me out!

[Sean Theriault]

Again, thanks!

[Nick Cox]

I would try a quadratic model too.