Labeling With Coefplot and Interactions

[justin winikoff]

Hello, I am running a panel fixed effects regression where I'd like to plot the interaction effects of a 0/1 dummy with a year fixed effect. The regression ultimately looks like:

xtreg inc i.year i.ogj#i.year, fe

.

I then store important estimates using

estimates store interaction

and then would like to plot using

coefeplot (interactions), keep(1.ogj#*year)

.

This works fine except the labels are tedious. They read

"Overlays Large Field#1yearyear=1975"

and so on. I'd like for them just to say the year, but I can't quite figure it out without manually going through them and changing them one by one. I have to run many varieties of this equation, so a quick way of doing this would be great. Any suggestions?

(As a side note, I am on a restricted data portal and cannot post images/code here. I also cannot download any online packages that may be better).

[Marcos Almeida]

Welcome to Stata Forum / Statalist.

It is a pity you cannot share data, even a toy example, neither download files.

In spite of that, I tried to share the help files of the user-written coefplot with you.

http://repec.sowi.unibe.ch/stata/coe...help-file.html

Be sure to take a look at the coeflabels option. You may wish to fiddle with it in order to reach your goal.

Hopefully that helps.

[Weiwen Ng]

Hello, I am running a panel fixed effects regression where I'd like to plot the interaction effects of a 0/1 dummy with a year fixed effect. The regression ultimately looks like:

xtreg inc i.year i.ogj#i.year, fe

.

I then store important estimates using

estimates store interaction

and then would like to plot using

coefeplot (interactions), keep(1.ogj#*year)

.

This works fine except the labels are tedious. They read and so on. I'd like for them just to say the year, but I can't quite figure it out without manually going through them and changing them one by one. I have to run many varieties of this equation, so a quick way of doing this would be great. Any suggestions?

(As a side note, I am on a restricted data portal and cannot post images/code here. I also cannot download any online packages that may be better).

As Marcos said, the coeflabels option would do it. You have to refer to the coefficients by the correct name and do them one by one, e.g.

 
 coefeplot (interactions), keep(1.ogj#*year) coeflabel(1.ogj#1975year = "1975")

However, I suspect that using the rename option plus regex would be easier if you've got a lot of year dummies. Do you know how to write a regular expression? They can be very powerful when used in renaming text. I think that something like this might do it:

 
 coefeplot (interactions), keep(1.ogj#*year) rename(^1.ogj#([0-9]+)year$ = \1, regex)

The phrase

([0-9]+)

should indicate to Stata to look for any number of digits. You could change the + to a {4} to tell it to look for 4 digits specifically. The

\1

after the = sign means to substitute the matched subexpression. That might work, or you might have to play around.

[justin winikoff]

Hello Weiwen. Thank you for your response. I've tried both of these methods and neither works. The output is identical to what it was before.

[EDIT] Weiwen, this worked. The problem had been that the estimates were actually stored as 1.ogj#1975.year and so on, so the "." was throwing it off. Including it in the rename works.

Thank you so much for your help!!

[Katherine Adams]

Hello,

I have a question related to this thread:
https://www.statalist.org/forums/for...-with-coefplot

I would be happy to get any advice.

Thank you!