Graphing just the interaction variable using coefplot

[Jim Callegari]

Hi All,

I have a regression and I would like to plot the coefficients so that I can see how they change over time. My code is:

reg rel_px soft_drink_new##ym if country=="Ireland" & test_include_prod==1 & year>=2015 , robust

and I only want to plot where the output is "soft_drink_new#ym." The regression output also includes "ym" constant, and others. Because of the ease of plotting this vertically, I was trying to use

coefplot, keep(soft_drink_new##ym) xline(0) vertical

But STATA outputs:

(.: no coefficients found, all dropped, or none kept)
(nothing to plot)

I have attached a snippet of my data below, and am happy to provide any additional information as well.

Many thanks,
Jim

* Example generated by -dataex-. To install: ssc install dataex
clear
input str21 prod float rel_px str7 country float ym
"Allitems"  .997 "Ireland" 648
"Allitems" 1.002 "Ireland" 649
"Allitems" 1.009 "Ireland" 650
"Allitems"  1.01 "Ireland" 651
"Allitems" 1.009 "Ireland" 652
"Allitems" 1.011 "Ireland" 653
"Allitems" 1.009 "Ireland" 654
"Allitems" 1.011 "Ireland" 655
"Allitems" 1.008 "Ireland" 656
"Allitems" 1.006 "Ireland" 657
"Allitems" 1.003 "Ireland" 658
"Allitems"  .999 "Ireland" 659
"Allitems"  .991 "Ireland" 660
"Allitems"  .997 "Ireland" 661
"Allitems" 1.003 "Ireland" 662
"Allitems" 1.003 "Ireland" 663
"Allitems" 1.007 "Ireland" 664
"Allitems" 1.009 "Ireland" 665
"Allitems" 1.007 "Ireland" 666
"Allitems" 1.011 "Ireland" 667
"Allitems" 1.006 "Ireland" 668
"Allitems" 1.004 "Ireland" 669
"Allitems" 1.001 "Ireland" 670
"Allitems"     1 "Ireland" 671
"Allitems"  .992 "Ireland" 672
"Allitems"  .996 "Ireland" 673
"Allitems"     1 "Ireland" 674
"Allitems" 1.002 "Ireland" 675
"Allitems" 1.007 "Ireland" 676
"Allitems" 1.014 "Ireland" 677
"Allitems" 1.012 "Ireland" 678
"Allitems"  1.01 "Ireland" 679
"Allitems" 1.006 "Ireland" 680
"Allitems" 1.001 "Ireland" 681
"Allitems"     1 "Ireland" 682
"Allitems"     1 "Ireland" 683
"Allitems"  .995 "Ireland" 684
"Allitems" 1.001 "Ireland" 685
"Allitems" 1.007 "Ireland" 686
"Allitems" 1.011 "Ireland" 687
"Allitems" 1.009 "Ireland" 688
"Allitems"  1.01 "Ireland" 689
"Allitems"  1.01 "Ireland" 690
"Allitems" 1.014 "Ireland" 691
"Allitems" 1.008 "Ireland" 692
"Allitems" 1.007 "Ireland" 693
"Allitems" 1.005 "Ireland" 694
"Allitems" 1.004 "Ireland" 695
"Allitems"  .997 "Ireland" 696
"Allitems" 1.006 "Ireland" 697
"Allitems" 1.009 "Ireland" 698
"Allitems" 1.007 "Ireland" 699
"Allitems" 1.013 "Ireland" 700
"Allitems" 1.014 "Ireland" 701
"Allitems" 1.018 "Ireland" 702
"Allitems" 1.021 "Ireland" 703
"Allitems" 1.017 "Ireland" 704
"Allitems" 1.016 "Ireland" 705
"Allitems" 1.011 "Ireland" 706
"Allitems" 1.011 "Ireland" 707
"Allitems" 1.004 "Ireland" 708
"Allitems" 1.012 "Ireland" 709
"Allitems"  1.02 "Ireland" 710
"Allitems" 1.024 "Ireland" 711
"Allitems" 1.023 "Ireland" 712
"Allitems" 1.025 "Ireland" 713
"Allitems" 1.023 "Ireland" 714
"Allitems" 1.028 "Ireland" 715
"Allitems" 1.026 "Ireland" 716
"Allitems" 1.023 "Ireland" 717
"Allitems" 1.022 "Ireland" 718
"Allitems" 1.024 "Ireland" 719
"Allitems" 1.017 "Ireland" 720
"Allitems" 1.023 "Ireland" 721
"Allitems" 1.027 "Ireland" 722
"Allitems" 1.023 "Ireland" 723
"Allitems" 1.018 "Ireland" 724
"Allitems" 1.021 "Ireland" 725
"Allitems" 1.019 "Ireland" 726
"Allitems" 1.018 "Ireland" 727
"Allitems" 1.014 "Ireland" 728
"Bread"    1.056 "Ireland" 648
"Bread"    1.058 "Ireland" 649
"Bread"    1.051 "Ireland" 650
"Bread"    1.045 "Ireland" 651
"Bread"    1.049 "Ireland" 652
"Bread"    1.047 "Ireland" 653
"Bread"    1.047 "Ireland" 654
"Bread"    1.045 "Ireland" 655
"Bread"    1.043 "Ireland" 656
"Bread"    1.044 "Ireland" 657
"Bread"    1.032 "Ireland" 658
"Bread"    1.017 "Ireland" 659
"Bread"    1.019 "Ireland" 660
"Bread"    1.022 "Ireland" 661
"Bread"    1.016 "Ireland" 662
"Bread"    1.015 "Ireland" 663
"Bread"    1.013 "Ireland" 664
"Bread"    1.006 "Ireland" 665
"Bread"    1.006 "Ireland" 666
end
format %tm ym

[Bruce Weaver]

Hi Jim. You've made it a bit difficult for people to help you figure things out because your dataset is missing variables that are needed by the -regress- command you posted.

• Variables in your dataset: prod, rel_px, country, ym
• Variables needed for your -regress- command: rel_px, soft_drink_new, ym, country, test_include_prod year
• Variables missing from your dataset: soft_drink_new, test_include_prod, year

HTH.

[Jim Callegari]

So sorry about that--I've included them below. Please let me know if this additional information is helpful! Or if it is easier, I can remove the "if ..." part of my regression--the main issue I'm concerned with is graphing the interaction variable coefficients themselves.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(rel_px soft_drink_new ym) str7 country float test_include_prod int year
 .997 0 648 "Ireland" 0 2014
1.002 0 649 "Ireland" 0 2014
1.009 0 650 "Ireland" 0 2014
 1.01 0 651 "Ireland" 0 2014
1.009 0 652 "Ireland" 0 2014
1.011 0 653 "Ireland" 0 2014
1.009 0 654 "Ireland" 0 2014
1.011 0 655 "Ireland" 0 2014
1.008 0 656 "Ireland" 0 2014
1.006 0 657 "Ireland" 0 2014
1.003 0 658 "Ireland" 0 2014
 .999 0 659 "Ireland" 0 2014
 .991 0 660 "Ireland" 0 2015
 .997 0 661 "Ireland" 0 2015
1.003 0 662 "Ireland" 0 2015
1.003 0 663 "Ireland" 0 2015
1.007 0 664 "Ireland" 0 2015
1.009 0 665 "Ireland" 0 2015
1.007 0 666 "Ireland" 0 2015
1.011 0 667 "Ireland" 0 2015
1.006 0 668 "Ireland" 0 2015
1.004 0 669 "Ireland" 0 2015
1.001 0 670 "Ireland" 0 2015
    1 0 671 "Ireland" 0 2015
 .992 0 672 "Ireland" 0 2016
 .996 0 673 "Ireland" 0 2016
    1 0 674 "Ireland" 0 2016
1.002 0 675 "Ireland" 0 2016
1.007 0 676 "Ireland" 0 2016
1.014 0 677 "Ireland" 0 2016
1.012 0 678 "Ireland" 0 2016
 1.01 0 679 "Ireland" 0 2016
1.006 0 680 "Ireland" 0 2016
1.001 0 681 "Ireland" 0 2016
    1 0 682 "Ireland" 0 2016
    1 0 683 "Ireland" 0 2016
 .995 0 684 "Ireland" 0 2017
1.001 0 685 "Ireland" 0 2017
1.007 0 686 "Ireland" 0 2017
1.011 0 687 "Ireland" 0 2017
1.009 0 688 "Ireland" 0 2017
 1.01 0 689 "Ireland" 0 2017
 1.01 0 690 "Ireland" 0 2017
1.014 0 691 "Ireland" 0 2017
1.008 0 692 "Ireland" 0 2017
1.007 0 693 "Ireland" 0 2017
1.005 0 694 "Ireland" 0 2017
1.004 0 695 "Ireland" 0 2017
 .997 0 696 "Ireland" 0 2018
1.006 0 697 "Ireland" 0 2018
1.009 0 698 "Ireland" 0 2018
1.007 0 699 "Ireland" 0 2018
1.013 0 700 "Ireland" 0 2018
1.014 0 701 "Ireland" 0 2018
1.018 0 702 "Ireland" 0 2018
1.021 0 703 "Ireland" 0 2018
1.017 0 704 "Ireland" 0 2018
1.016 0 705 "Ireland" 0 2018
1.011 0 706 "Ireland" 0 2018
1.011 0 707 "Ireland" 0 2018
1.004 0 708 "Ireland" 0 2019
1.012 0 709 "Ireland" 0 2019
 1.02 0 710 "Ireland" 0 2019
1.024 0 711 "Ireland" 0 2019
1.023 0 712 "Ireland" 0 2019
1.025 0 713 "Ireland" 0 2019
1.023 0 714 "Ireland" 0 2019
1.028 0 715 "Ireland" 0 2019
1.026 0 716 "Ireland" 0 2019
1.023 0 717 "Ireland" 0 2019
1.022 0 718 "Ireland" 0 2019
1.024 0 719 "Ireland" 0 2019
1.017 0 720 "Ireland" 0 2020
1.023 0 721 "Ireland" 0 2020
1.027 0 722 "Ireland" 0 2020
1.023 0 723 "Ireland" 0 2020
1.018 0 724 "Ireland" 0 2020
1.021 0 725 "Ireland" 0 2020
1.019 0 726 "Ireland" 0 2020
1.018 0 727 "Ireland" 0 2020
1.014 0 728 "Ireland" 0 2020
1.056 0 648 "Ireland" 0 2014
1.058 0 649 "Ireland" 0 2014
1.051 0 650 "Ireland" 0 2014
1.045 0 651 "Ireland" 0 2014
1.049 0 652 "Ireland" 0 2014
1.047 0 653 "Ireland" 0 2014
1.047 0 654 "Ireland" 0 2014
1.045 0 655 "Ireland" 0 2014
1.043 0 656 "Ireland" 0 2014
1.044 0 657 "Ireland" 0 2014
1.032 0 658 "Ireland" 0 2014
1.017 0 659 "Ireland" 0 2014
1.019 0 660 "Ireland" 0 2015
1.022 0 661 "Ireland" 0 2015
1.016 0 662 "Ireland" 0 2015
1.015 0 663 "Ireland" 0 2015
1.013 0 664 "Ireland" 0 2015
1.006 0 665 "Ireland" 0 2015
1.006 0 666 "Ireland" 0 2015
end
format %tm ym

[Justin Niakamal]

I can't test this on your example because soft_drink_new takes only one value, but try

reg rel_px soft_drink_new##ym if country=="Ireland" & test_include_prod==1 & year>=2015 , robust

matrix A = e(b)
local namecol "`: colnames A'"

tokenize `namecol'

forvalues i = 1 / `=colsof(matrix(A))' {
    local mv = A[1,`i']
    if `mv' != 0 & strmatch("``i''" , "*#*") {
        local coefinter `coefinter' ``i''
    }
}

coefplot, keep(`coefinter')

Source:
https://stackoverflow.com/questions/...ms-in-coefplot

[Jim Callegari]

Thank you so much! That works perfectly.