Plotting multiple concentration curves in one graph

[Ashani Abayasekara]

Hi everyone,

I'm conducting some inequality analysis and using concentration indices and curves to depict inequalities in outcomes. I'm using the user contributed conindex command for this purpose. While I am able to produce these curves for a specific population, I have not been able to use it to graph the curves for multiple groups (e.g., by sex or age) in one graph. I tried using another user contributed command glcurve for this purpose, but was unable to get this to work properly.

Below is the code I'm currently using to produce two curves for males and females in two separate graphs. How can I adjust this to produce both curves in the same graph? I've also included a sample of my data.

conindex pgsi [pweight=scq_wgt] if sex==1, rankvar(hheqinc) robust cluster(xwaveid) truezero bounded limits(0 27)erreygers graph
conindex pgsi [pweight=scq_wgt] if sex==2, rankvar(hheqinc) robust cluster(xwaveid) truezero bounded limits(0 27)erreygers graph

* Example generated by -dataex-. For more info, type help dataex
clear
input long xwaveid float(pgsi scq_wgt sex hheqinc)
1500001  .         . 2 36191.332
1500002  .         . 2 36191.332
1500003  .         . 1 36191.332
1500004  . 1336.3737 1  43259.33
1500006  .         0 2   44685.5
1500007  .         . 2   44685.5
1500010  .         . 1 126668.66
1500012  0 1209.0953 2  45549.33
1500018  0  897.1745 2     34340
1500023  0 1879.3213 1 67506.664
1500024  .         0 1  35680.77
1500025  0 1954.5593 2  100449.6
1500028  0 1095.2823 2  49080.67
1500031  .         . 2     44462
1500034  0 1225.8156 2  80826.66
1500035  0  694.3892 1  46307.78
1500038  5 1395.4056 1 15129.333
1500039  0 1581.5642 1  57854.67
1500042  0 1588.5952 1   33541.6
1500045  0  1817.037 1  48392.14
1500046  .         . 1  48392.14
1500047  .         . 1 28703.334
1500051  0   737.827 2     63013
1500053  0  2861.205 1 27301.785
1500054  0  4960.978 2 27301.785
1500055  0  4686.794 2  48662.82
1500056  0 4799.8564 1  48662.82
1500060  0   8072.21 2  48662.82
1500062  0  810.4601 2     35148
1500064  .         0 2  30672.22
1500066  0  161.7171 1 12819.333
1500068  .         . 1   94625.5
1500069  .         . 2   94625.5
1500070  .         0 2   53920.5
1500071  0 1117.5151 2  65297.33
1500072  0  2202.835 1 29333.334
1500073  0  762.3912 2 34087.332
1500076  0 1028.1517 2  40757.08
1500078  0   1418.98 1 24608.666
1500082  0  495.1196 2  60330.67
1500083  .         . 1  56118.23
1500084  .         . 2  56118.23
1500086  4 1142.2218 1 37424.668
1500087  .         . 1     20885
1500088  .         . 2     20885
1500089  0  368.1432 1     42028
1500090 11 1230.8286 1  40152.67
1500091  .         . 1     96545
1500092  0 10203.215 2 35772.855
1500094  .         . 2  93092.66
1500095  0  661.3651 2 26825.334
1500096  0 1060.0026 1  60813.33
1500097  .         . 2  51146.09
1500098  .         . 1  51146.09
1500100  0   1615.66 2  43363.48
1500101  0 1418.0198 1  43363.48
1500103  3 1435.0386 1   73576.8
1500105  .         0 1     21431
1500106  0 105.06734 2     21431
1500107  0 140.27469 1     21431
1500108  0 260.66693 1 15757.143
1500111  .         0 2 27531.305
1500114  0  616.7527 1 71536.664
1500115  0  1364.354 1  9333.333
1500116  0   845.457 2  65547.86
1500120  0  849.9174 1     42798
1500121  1  694.4107 1  79206.66
1500123  1  1598.022 2  87631.34
1500126  0  919.6907 2     55508
1500127  0 2722.1455 1  41018.33
1500129  .         . 1     30267
1500130  .         . 2     30267
1500131  .         . 2 67339.336
1500132  0 1638.4905 1   91041.2
1500133  0    1577.7 2   91041.2
1500134  0  661.7795 2     47628
1500136  0  912.3063 2  50053.33
1500137  1  691.1695 2     51034
1500140  0 1219.2117 1  30307.31
1500142  .         0 2   31309.5
1500143  .         . 2   57014.5
1500145  0 2075.8484 2     41572
1500146  .         0 2   59285.5
1500147  .         0 1   59285.5
1500148  0  3977.003 2  121845.6
1500149  0 1122.8436 2     33800
1500150  0  957.3792 2     34002
1500152  0 2998.9736 2     48864
1500155  0  1948.939 2  63473.43
1500159  .         . 2   58896.8
1500160  .         . 2   58896.8
1500161  .         . 1   58896.8
1500163  0 1041.3129 2 36480.668
1500166  .         0 2 73296.664
1500168  0  530.2118 2 31999.445
1500169  .         0 1  44980.29
1500170  .         0 2  44980.29
1500175  0 291.78607 1 34042.668
1500177  0  736.3896 1     41518
1500178  0  1946.247 1     84972
1500181  0 566.26697 2     38094
1500182  0  709.2154 2  56356.67
1500184  .         0 1   39208.5
1500185  0 2192.5264 1    138347
1500186  0  1841.427 2    138347
1500188  0  17918.41 1 68476.664
1500189  .         0 1   54096.5
1500190  .         0 1   54096.5
1500191  0 152.17635 1 24402.666
1500192  0 1345.4823 2     34191
1500193  0  3569.598 2  113069.6
1500195  .         . 2  52657.33
1500196  .         . 2     99932
1500198  0 1659.8325 1 13285.714
1500199  .         0 2 13285.714
1500201  0 2004.9705 2  48497.33
1500202  .         . 2     50868
1500203  0  530.0276 2  46667.33
1500204  0 148.26433 1     30676
1500208  .         0 2  7638.889
1500210  0 1194.7814 1  96360.66
1500211  0  495.9803 2     53865
1500214  0  461.6954 2  51433.33
1500217  0 2341.9646 1   38964.8
1500218  0  704.5697 2     44603
1500219  0  635.3337 2     44603
1500221  0  4327.951 2 12666.667
1500223  0 2030.2954 1     62950
1500224  0 2087.5125 2     62950
1500225  0 1558.2502 1     62950
1500226  0  583.8546 1 65964.664
1500227  .         . 2     78375
1500230  0  1906.377 1     46094
1500231  .         . 1     97661
1500232  .         . 1 30560.666
1500234  0 2144.5833 1 36821.332
1500236  0 2281.3767 1  61991.33
1500241  0   501.775 2   60688.5
1500242  0  909.5589 1  49498.57
1500243  0   856.088 2  49498.57
1500244  0  877.5878 2  49498.57
1500245  .         . 2     13276
1500246  .         . 1     13276
1500247  .         . 1     90156
1500248  0 3387.8596 1  126306.4
1500250  .         0 2  43162.38
1500253  .         . 2    279449
1500254  .         . 1    279449
1500255  .         . 2    279449
1500257  0  572.0884 2  49018.09
1500261  0  795.1612 2     68888
1500262  .         . 1     75276
1500265  .         . 1   94914.2
1500268  .         . 1 108807.78
1500272  .         0 2  50244.78
1500275  0 1511.9425 1  41306.19
1500276  0 141.28271 2 18979.334
1500278  0 1641.2476 2     35480
1500280  .         . 1  22742.22
1500282  0   903.976 2  97278.66
1500284  0   1411.44 1     56247
1500286  0  1225.103 2   60911.5
1500287  0 1802.3956 2    118198
1500289  0 3129.8506 1  43111.33
1500290  1  495.4764 2 29400.666
1500293  0  842.8392 1     58900
1500294  0  964.5455 2    108602
1500296  0  870.3943 2 70995.336
1500298  3  724.2965 2   44676.4
1500299  0  748.8852 1  48543.33
1500300  0 538.77875 1 33445.715
1500302  0 1172.0776 2  95058.66
1500303  .         0 1  61375.77
1500305  0    3396.4 2     27766
1500306  .         . 1     27766
1500307  .         . 2 28960.666
1500308  .         0 1     74698
1500311  .         . 1  52048.67
1500313  0 1378.3098 2 108124.66
1500315  0  714.8577 1 26568.666
1500318  0 1334.6116 2    301610
1500319  1  8893.913 2     27800
1500320  .         . 1 72841.336
1500321  2  1309.949 2   32530.5
1500323  9 1864.9603 1     37426
1500325  0 1696.5016 2     61830
1500326  0 1260.3423 1 71605.336
1500328  .         0 2 26741.334
1500329  0  961.4438 2 77503.336
1500330  0  2063.097 1  53430.67
1500333  0         0 2     30888
1500334  .         0 1  23659.13
1500335  2  1348.435 2  23659.13
1500337  0  1688.963 1     62170
1500338  .         0 1  34405.24
1500339  0   1663.13 2     46430
1500340  0  1611.695 2     46430
1500344  0  875.5526 1     41934
1500346  0 525.39276 2 67377.336
1500347  0 304.66434 1  55663.67
end

Many thanks,
Ashani

[Andrew Musau]

conindex is from SSC (FAQ Advice #12). The command includes an option to save the data used to create the plots, but this applies only to the line data and does not include the confidence interval data.

keepgraphdata(string) creates a new dataset storing the data used to graph the Lorenz or concentration curve, where string is the name of the dataset to be created.

Here is a way to combine the graphs by directly extracting data from the plots. However, with overlapping confidence intervals, I’m not convinced it’s a good idea to display multiple graphs in a single plot. If you include more than two graphs, the result could be a confusing mess that your readers may find difficult to interpret.

* Example generated by -dataex-. For more info, type help dataex
clear
input long xwaveid float(pgsi scq_wgt sex hheqinc)
1500001  .         . 2 36191.332
1500002  .         . 2 36191.332
1500003  .         . 1 36191.332
1500004  . 1336.3737 1  43259.33
1500006  .         0 2   44685.5
1500007  .         . 2   44685.5
1500010  .         . 1 126668.66
1500012  0 1209.0953 2  45549.33
1500018  0  897.1745 2     34340
1500023  0 1879.3213 1 67506.664
1500024  .         0 1  35680.77
1500025  0 1954.5593 2  100449.6
1500028  0 1095.2823 2  49080.67
1500031  .         . 2     44462
1500034  0 1225.8156 2  80826.66
1500035  0  694.3892 1  46307.78
1500038  5 1395.4056 1 15129.333
1500039  0 1581.5642 1  57854.67
1500042  0 1588.5952 1   33541.6
1500045  0  1817.037 1  48392.14
1500046  .         . 1  48392.14
1500047  .         . 1 28703.334
1500051  0   737.827 2     63013
1500053  0  2861.205 1 27301.785
1500054  0  4960.978 2 27301.785
1500055  0  4686.794 2  48662.82
1500056  0 4799.8564 1  48662.82
1500060  0   8072.21 2  48662.82
1500062  0  810.4601 2     35148
1500064  .         0 2  30672.22
1500066  0  161.7171 1 12819.333
1500068  .         . 1   94625.5
1500069  .         . 2   94625.5
1500070  .         0 2   53920.5
1500071  0 1117.5151 2  65297.33
1500072  0  2202.835 1 29333.334
1500073  0  762.3912 2 34087.332
1500076  0 1028.1517 2  40757.08
1500078  0   1418.98 1 24608.666
1500082  0  495.1196 2  60330.67
1500083  .         . 1  56118.23
1500084  .         . 2  56118.23
1500086  4 1142.2218 1 37424.668
1500087  .         . 1     20885
1500088  .         . 2     20885
1500089  0  368.1432 1     42028
1500090 11 1230.8286 1  40152.67
1500091  .         . 1     96545
1500092  0 10203.215 2 35772.855
1500094  .         . 2  93092.66
1500095  0  661.3651 2 26825.334
1500096  0 1060.0026 1  60813.33
1500097  .         . 2  51146.09
1500098  .         . 1  51146.09
1500100  0   1615.66 2  43363.48
1500101  0 1418.0198 1  43363.48
1500103  3 1435.0386 1   73576.8
1500105  .         0 1     21431
1500106  0 105.06734 2     21431
1500107  0 140.27469 1     21431
1500108  0 260.66693 1 15757.143
1500111  .         0 2 27531.305
1500114  0  616.7527 1 71536.664
1500115  0  1364.354 1  9333.333
1500116  0   845.457 2  65547.86
1500120  0  849.9174 1     42798
1500121  1  694.4107 1  79206.66
1500123  1  1598.022 2  87631.34
1500126  0  919.6907 2     55508
1500127  0 2722.1455 1  41018.33
1500129  .         . 1     30267
1500130  .         . 2     30267
1500131  .         . 2 67339.336
1500132  0 1638.4905 1   91041.2
1500133  0    1577.7 2   91041.2
1500134  0  661.7795 2     47628
1500136  0  912.3063 2  50053.33
1500137  1  691.1695 2     51034
1500140  0 1219.2117 1  30307.31
1500142  .         0 2   31309.5
1500143  .         . 2   57014.5
1500145  0 2075.8484 2     41572
1500146  .         0 2   59285.5
1500147  .         0 1   59285.5
1500148  0  3977.003 2  121845.6
1500149  0 1122.8436 2     33800
1500150  0  957.3792 2     34002
1500152  0 2998.9736 2     48864
1500155  0  1948.939 2  63473.43
1500159  .         . 2   58896.8
1500160  .         . 2   58896.8
1500161  .         . 1   58896.8
1500163  0 1041.3129 2 36480.668
1500166  .         0 2 73296.664
1500168  0  530.2118 2 31999.445
1500169  .         0 1  44980.29
1500170  .         0 2  44980.29
1500175  0 291.78607 1 34042.668
1500177  0  736.3896 1     41518
1500178  0  1946.247 1     84972
1500181  0 566.26697 2     38094
1500182  0  709.2154 2  56356.67
1500184  .         0 1   39208.5
1500185  0 2192.5264 1    138347
1500186  0  1841.427 2    138347
1500188  0  17918.41 1 68476.664
1500189  .         0 1   54096.5
1500190  .         0 1   54096.5
1500191  0 152.17635 1 24402.666
1500192  0 1345.4823 2     34191
1500193  0  3569.598 2  113069.6
1500195  .         . 2  52657.33
1500196  .         . 2     99932
1500198  0 1659.8325 1 13285.714
1500199  .         0 2 13285.714
1500201  0 2004.9705 2  48497.33
1500202  .         . 2     50868
1500203  0  530.0276 2  46667.33
1500204  0 148.26433 1     30676
1500208  .         0 2  7638.889
1500210  0 1194.7814 1  96360.66
1500211  0  495.9803 2     53865
1500214  0  461.6954 2  51433.33
1500217  0 2341.9646 1   38964.8
1500218  0  704.5697 2     44603
1500219  0  635.3337 2     44603
1500221  0  4327.951 2 12666.667
1500223  0 2030.2954 1     62950
1500224  0 2087.5125 2     62950
1500225  0 1558.2502 1     62950
1500226  0  583.8546 1 65964.664
1500227  .         . 2     78375
1500230  0  1906.377 1     46094
1500231  .         . 1     97661
1500232  .         . 1 30560.666
1500234  0 2144.5833 1 36821.332
1500236  0 2281.3767 1  61991.33
1500241  0   501.775 2   60688.5
1500242  0  909.5589 1  49498.57
1500243  0   856.088 2  49498.57
1500244  0  877.5878 2  49498.57
1500245  .         . 2     13276
1500246  .         . 1     13276
1500247  .         . 1     90156
1500248  0 3387.8596 1  126306.4
1500250  .         0 2  43162.38
1500253  .         . 2    279449
1500254  .         . 1    279449
1500255  .         . 2    279449
1500257  0  572.0884 2  49018.09
1500261  0  795.1612 2     68888
1500262  .         . 1     75276
1500265  .         . 1   94914.2
1500268  .         . 1 108807.78
1500272  .         0 2  50244.78
1500275  0 1511.9425 1  41306.19
1500276  0 141.28271 2 18979.334
1500278  0 1641.2476 2     35480
1500280  .         . 1  22742.22
1500282  0   903.976 2  97278.66
1500284  0   1411.44 1     56247
1500286  0  1225.103 2   60911.5
1500287  0 1802.3956 2    118198
1500289  0 3129.8506 1  43111.33
1500290  1  495.4764 2 29400.666
1500293  0  842.8392 1     58900
1500294  0  964.5455 2    108602
1500296  0  870.3943 2 70995.336
1500298  3  724.2965 2   44676.4
1500299  0  748.8852 1  48543.33
1500300  0 538.77875 1 33445.715
1500302  0 1172.0776 2  95058.66
1500303  .         0 1  61375.77
1500305  0    3396.4 2     27766
1500306  .         . 1     27766
1500307  .         . 2 28960.666
1500308  .         0 1     74698
1500311  .         . 1  52048.67
1500313  0 1378.3098 2 108124.66
1500315  0  714.8577 1 26568.666
1500318  0 1334.6116 2    301610
1500319  1  8893.913 2     27800
1500320  .         . 1 72841.336
1500321  2  1309.949 2   32530.5
1500323  9 1864.9603 1     37426
1500325  0 1696.5016 2     61830
1500326  0 1260.3423 1 71605.336
1500328  .         0 2 26741.334
1500329  0  961.4438 2 77503.336
1500330  0  2063.097 1  53430.67
1500333  0         0 2     30888
1500334  .         0 1  23659.13
1500335  2  1348.435 2  23659.13
1500337  0  1688.963 1     62170
1500338  .         0 1  34405.24
1500339  0   1663.13 2     46430
1500340  0  1611.695 2     46430
1500344  0  875.5526 1     41934
1500346  0 525.39276 2 67377.336
1500347  0 304.66434 1  55663.67
end


conindex pgsi [pweight=scq_wgt] if sex==1, rankvar(hheqinc) robust cluster(xwaveid) truezero bounded limits(0 27)erreygers graph
gr save gr1, replace

conindex pgsi [pweight=scq_wgt] if sex==2, rankvar(hheqinc) robust cluster(xwaveid) truezero bounded limits(0 27)erreygers graph
gr save gr2, replace

tempfile graphdata
forval s= 1/2{
    gr use gr`s'
    serset use, clear
    rename _* (cil ciu rank cshare)
    gen series=`s'
    cap append using `graphdata'
    save `graphdata', replace
}

use `graphdata', clear
tw (rarea cil ciu rank if series==1, fcolor(blue%40) lc(blue%0)) ///
   (rarea cil ciu rank if series==2, fcolor(grey%40) lc(grey%0)) ///
   (line cshare rank if series==1, color(blue)) ///
   (line cshare rank if series==2, color(grey)), ///
   ytitle(Cumulative share of pgsi) xtitle(Rank of hheqinc) ///
   leg(order(1 "95% CI Female"  3 "L(p) Female"  2 "95% CI Male" 4 "L(p) Male"))

Res.:

[Ashani Abayasekara]

Thank you very much for this, much appreciated. Yes, I also realised it'll look too messy with the confidence intervals, so I might omit those and just show the curves.

[Ashani Abayasekara]

Hi Andrew, thanks again for this code which works for me. I have one more question - I need to add the 45 degree diagonal line of equality in the graph, how do I do this?

[Clyde Schechter]

I have one more question - I need to add the 45 degree diagonal line of equality in the graph, how do I do this?

I don't think this makes any sense for this data. The cumulative share variable ranges from about 0 to 1, whereas the rank ranges from 0 to 100. How would the cumulative share be even close to the rank except perhaps at the first data point of the extreme left end of the graph? Am I missing something?

In any case, the line of equality would not be anything like a 45 degree diagonal with this data. It would be a line at the left end of the graph whose slope is so large that it would appear to be completely vertical. It would probably be visually indistinguishable from the y-axis of your graph, or very nearly so.

[Ashani Abayasekara]

Hi Clyde,

Thank you for pointing that out and you are correct in that the axis values don't match for a line of equality. I think this has to do with this specific command and the way it graphs the data. For example, the individual graphs produced by the first two commands are two separate figures like the one below (smoother here than the sample data one since I'm using all the data), which despite the different scales produces a dashed diagonal line. I just want to retain this line when combining two or more curves in one graph.

 
 conindex pgsi [pweight=scq_wgt] if sex==1, rankvar(hheqinc) robust cluster(xwaveid) truezero bounded limits(0 27)erreygers graph gr save gr1, replace  conindex pgsi [pweight=scq_wgt] if sex==2, rankvar(hheqinc) robust cluster(xwaveid) truezero bounded limits(0 27)erreygers graph gr save gr2, replace

[Clyde Schechter]

Well, I see the diagonal line, but it is clearly not a line of inequality. The same mismatch in the range of values I pointed out in #5 obtains here as well. It seems to be a line like y = x/100. Assuming you want something like that, you can do this:

use `graphdata', clear
gen rank100 = rank/100
tw (rarea cil ciu rank if series==1, fcolor(blue%40) lc(blue%0)) ///
   (rarea cil ciu rank if series==2, fcolor(grey%40) lc(grey%0)) ///
   (line cshare rank if series==1, color(blue)) ///
   (line cshare rank if series==2, color(grey)) ///
   (line rank100 rank, color(black) lpattern(dash)), ///
   ytitle(Cumulative share of pgsi) xtitle(Rank of hheqinc) ///
   leg(order(1 "95% CI Female"  3 "L(p) Female"  2 "95% CI Male" 4 "L(p) Male"))

[Nick Cox]

See also https://journals.sagepub.com/doi/pdf...6867X241276116 for a miniature review of adding lines to graphs.

twoway function and twoway scatteri are in my experience often neglected here.

[Ashani Abayasekara]

Thank you very much Clyde for the code, and thank you Nick, I'll have a look at these options.