graph with range in x axis

[River Huang]

Dear All, Suppose that I have this code

sysuse auto, clear
twoway (scatter mpg weight if weight < 2500) (scatter mpg weight if weight > 2500) ///
(lfit mpg weight if weight < 2500) (lfit mpg weight if weight > 2500), ///
xline(2500) legend(off)

which generates

My question is that, how can I obtain a graph with weight < 3,500 only (x axis)? Any suggestions are appreciated.

[Raymond Zhang]

Is this want you want?

sysuse auto, clear
twoway (scatter mpg weight if weight < 2500) (scatter mpg weight if inrange(weight,2500,3500)) ///
(lfit mpg weight if weight < 2500) (lfit mpg weight,range(2500 3500)), ///
xline(2500) xlabel(2000(500)3500) legend(off)

[Nick Cox]

This leaves the question open of whether you want:

1. The linear fit for values above 2500, but only drawn up to 3500.

2. The linear fit for value above 2500 and below 3500.

which in general are not the same.

[River Huang]

Dear Raymond, Thanks for your suggestion. However, as indicated by Nick, this is not what I wanted.

[River Huang]

Dear Nick, My purpose is, as indicated by you, 1. The linear fit for values above 2500, but only drawn up to 3500. That is, I have to use the "whole" data set to draw the graph, but only want to show the graph up to 3,500. Any suggestions?

[Ali Atia]

I think this is what you want:

clear *
sysuse auto
reg mpg weight if weight>2500
predict yhat
twoway (scatter mpg weight if weight < 2500) (scatter mpg weight if weight > 2500 & weight<3500) ///
(lfit mpg weight if weight < 2500) (line yhat weight if weight > 2500 & weight<3500), ///
xline(2500) legend(off) yla(10(10)40)

[River Huang]

Dear Ali, Thanks for the suggestion. Yes, the figure is what I had in mind. However, the problem is that the graph above is generated by some other (net install rdrobust, from(https://raw.githubusercontent.com/rd...t/master/stata) replace
) commands (According to your suggestion, I guess I have to modify the rdplot.ado file) to mofigy). I wonder if there is a straightforward way to do what I wanted. Since l am leaving my office, if needed, I can post an example with code/data/example/graph later.

[Raymond Zhang]

Yes

[Raymond Zhang]

@River Huang
I want to give another solution.Maybe it is better-looking than the graph in #6(the orange line seems odd) .

sysuse auto, clear
twoway (scatter mpg weight if weight < 2500) (scatter mpg weight if inrange(weight,2500,3500)) ///
(lfit mpg weight if weight < 2500) , ///
xline(2500) legend(off)
keep if weight > 2500
addplot:lfit mpg weight,range(2520,3500)

[Raymond Zhang]

Here is another solution.

sysuse auto, clear
twoway (scatter mpg weight if weight < 2500) (scatter mpg weight if inrange(weight,2500,3500)) ///
(lfit mpg weight if weight < 2500) , ///
xline(2500) legend(off)
keep if inrange(weight,2500,3500)
addplot:lfit mpg weight

[River Huang]

Dear Raymond, Thanks for this helpful suggestion. Could you have a look at this related link (https://www.statalist.org/forums/for...robust-problem).