How to fix a kink in lfit using tw graph

[Wonhee Cho]

Hello,

I am trying to produce a tw graph. I want to create a linear fit, but I see a "kink" in the middle of the line, opposed to what I would expect --- a simple straight line.
To give you an idea, my graph currently looks like this:


My code is the following:

tw (scatter arntz_auto relwage if isco1d==1, xsc(log) msymbol(circle) mfc(none)) ///
           (scatter auto rw if isco1d==2, xsc(log) msymbol(square) mfc(none)) ///
           (scatter auto rw if isco1d==3, xsc(log) msymbol(diamond) mfc(none)) ///
           (scatter auto rw if isco1d==4, xsc(log) msymbol(cross) mfc(none)) ///
           (scatter auto rw if isco1d==5, xsc(log) msymbol(triangle) mfc(none)) ///
           (scatter auto rw if isco1d==6, xsc(log) msymbol(circle) mfc(none)) ///
           (scatter auto rw if isco1d==7, xsc(log) msymbol(square) mfc(none)) ///
           (scatter auto rw if isco1d==8, xsc(log) msymbol(diamond) mfc(none)) ///
           (scatter auto rw if isco1d==9, xsc(log) msymbol(cross) mfc(none)) ///
           (lfit auto relwage if inlist(isco1d, 3,4,5,7,8,9),color(black) lwidth(vthin)), ///
            legend(order(1 2 3 4 5 7 8 9) label(1 "Manager") label(2 "Professionals") label(3 "Technicians") label(4 "Clerks") label(5 "Services and Sales") label(7 "Craft and Trades") label(8 "Plant and Machine") label(9 "Elementary")) ///
            ytitle("Automation Potential (Arntz et al)") xtitle("Relative Wage")

How do I make the lfit straight?

Thanks in advance.

[Nick Cox]

lfit is just what it is: a way to put a linear fit through a scatter of points.

If you want to fit a kinked line, use splines and regress. But I'll guess that you'll be better off with a logit model, depending on what the outcome is precisely.

[Wonhee Cho]

Thanks for your response, Nick.

So I tried to regress it, then just plot the fitted line.
When I just plot the line alone, it works fine.

But, I'm having struggles trying to make the line show in the overlays of graphs using tw...

Below is my code:

reg arntz_auto relwage if inlist(isco1d, 3, 4, 5, 7, 8, 9)
predict linearreg, xb

tw (scatter auto rw if isco1d==1, xsc(log) msymbol(circle) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==2, xsc(log) msymbol(square) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==3, xsc(log) msymbol(diamond) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==4, xsc(log) msymbol(cross) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==5, xsc(log) msymbol(triangle) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==6, xsc(log) msymbol(circle) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==7, xsc(log) msymbol(square) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==8, xsc(log) msymbol(diamond) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (scatter auto rw if isco1d==9, xsc(log) msymbol(cross) mfc(none) mlabel(country) mlabsize(vsmall)) ///
           (line linearreg rw), ///
            legend(order(4 5 7 8 9) label(1 "") label(2 "") label(3 "") label(4 "Clerks") label(5 "Services and Sales") label(7 "Craft and Trades") label(8 "Plant and Machine") label(9 "Elementary")) ///
            legend(rows(2)) ytitle("Automation Potential (Arntz et al)") xtitle("Relative Wage")

[Wonhee Cho]

Oh nevermind. I got it!
I spotted a minor coding error.
But thanks so much, Statalist community.