Bar graph with second y

[Giorgio Di Stefano]

Dear all,

I have a macro country panel, where a country is classified into groups and two types. type1 and type2. The two types are indicated by a dummy variable. My six variables of interest are: fd ka ext dom pr la.
I provide a sample of my data below

I would like to approach the following bar chart below, with a double right-hand column y, indicator1, this time , for each of the variables, by decade, for each group and each type.

indicator1 is a kind of special measure in percentage with both negative and positive values. Changing it by logarithmic scale or other way is not possible as that will altrer the entire project.

An alternative way to go, it could show a bar graph of the pair of each variable in each decade, by group and type, with indicator1.

Consider the pair of each variable by adding the indicator1 to the corresponding group and typing by decade, as in this example.

https://www.statalist.org/forums/fil...0007&type=full


or this one without the line

https://journals.sagepub.com/doi/pdf...867X1101000411


I am not interested in tabplot or catplot

Thanks in advance
Giorgio

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(id year) str97 country str1 group float(fd ka ext dom pr la type1 type2 indicator1 decade) byte(decade1 decade2 decade3)
1 1992 "Country1" "A"   .524175   2.321955  .75 .7777778 .071428575        1 0 1       12.2 1990 1 0 0
1 1993 "Country1" "A"  .5846328   2.321955  .75 .7777778 .071428575        1 0 1  17.553352 1990 1 0 0
1 1994 "Country1" "A"  .6478685   2.321955  .75 .9444444 .071428575 .7912164 0 1      19.11 1990 1 0 0
1 2000 "Country1" "A"  .8414965   1.048859  .75        1  .53571427 .7782844 0 1     28.634 2000 0 1 0
1 2001 "Country1" "A"   .873997   1.048859  .75        1   .6190476 .7782844 0 1   26.41307 2000 0 1 0
1 2002 "Country1" "A"  .8985876   1.048859  .75        1   .6190476 .7661784 0 1   5.536341 2000 0 1 0
1 2003 "Country1" "A"  .9015374   1.048859  .75        1   .6190476 .7661784 0 1   5.536341 2000 0 1 0
1 2004 "Country1" "A"  .9041975   1.048859  .75        1   .6190476 .7661784 0 1   7.350179 2000 0 1 0
1 2010 "Country1" "A"   .941813   1.048859 .875        1   .8452381 .7704133 0 1   6.241802 2010 0 0 1
1 2011 "Country1" "A"   .934702   1.048859 .875        1   .8452381 .7704133 0 1      5.068 2010 0 0 1
1 2012 "Country1" "A"  .9079577   1.303478 .875        1   .8452381 .7704133 0 1      5.068 2010 0 0 1
1 2013 "Country1" "A"  .8901507  1.5580974 .875        1   .8452381 .7694158 0 1   7.929047 2010 0 0 1
2 1992 "Country2" "A"   .399736   .6099069 .875 .7777778          0 .5697657 0 1        1.6 1990 1 0 0
2 1993 "Country2" "A"   .461959  2.0673358 .875 .9444444          0 .5697657 0 1   6.996447 1990 1 0 0
2 1994 "Country2" "A"  .4953545   2.321955 .875 .9444444          0 .5697657 0 1    8.69136 1990 1 0 0
2 2000 "Country2" "A"  .7018992   2.321955    1        1   .4404762 .5697657 0 1   2.364474 2000 0 1 0
2 2001 "Country2" "A"  .7035186   2.321955    1        1   .5952381 .5697657 0 1   2.364474 2000 0 1 0
2 2002 "Country2" "A"  .6993845   2.321955    1        1   .5952381 .5457992 0 1  -.7668521 2000 0 1 0
2 2003 "Country2" "A"  .7077955   2.321955    1        1   .5952381 .5457992 0 1 -2.9592714 2000 0 1 0
2 2004 "Country2" "A"   .728983   2.321955    1        1   .5952381 .5457992 0 1  -2.950783 2000 0 1 0
2 2010 "Country2" "A"  .7894757   2.321955    1 .9444444   .9166667 .5718316 0 1      3.891 2010 0 0 1
2 2011 "Country2" "A"  .8007692   2.321955    1 .9444444   .9166667 .5718316 0 1      3.891 2010 0 0 1
2 2012 "Country2" "A"  .7593077   2.321955    1 .9444444   .9166667 .5718316 0 1 -2.5410414 2010 0 0 1
2 2013 "Country2" "A"  .7435991   2.321955    1 .9444444   .9166667 .5718316 0 1  -6.333579 2010 0 0 1
3 1992 "Country3" "B"  .5840359   2.321955    1 .7777778  .08333334 .6281998 0 1   .3835026 1990 1 0 0
3 1993 "Country3" "B"  .6147648   2.321955    1 .8888889  .08333334 .6281998 0 1   .3835026 1990 1 0 0
3 1994 "Country3" "B"   .632942   2.321955    1 .8888889  .08333334 .6260909 0 1   .8169867 1990 1 0 0
3 2000 "Country3" "B"  .7692201   2.321955    1 .8888889        .75 .6297867 0 1  1.9049623 2000 0 1 0
3 2001 "Country3" "B"  .7502941   2.321955    1 .8888889        .75 .6297867 0 1  1.9049622 2000 0 1 0
3 2002 "Country3" "B"  .7244223   2.321955    1 .8888889   .8333333 .6297867 0 1   1.283737 2000 0 1 0
3 2003 "Country3" "B"  .7401305   2.321955    1 .8888889   .8333333 .6297867 0 1 -1.3254085 2000 0 1 0
3 2004 "Country3" "B"  .7518873   2.321955    1 .8888889   .8333333 .6547069 0 1 -1.3254085 2000 0 1 0
3 2010 "Country3" "B"  .7356597   2.321955    1 .9444444          1 .6547069 0 1  1.1606808 2010 0 0 1
3 2011 "Country3" "B"  .7201834   2.321955    1 .9444444          1 .6547069 0 1  1.1606808 2010 0 0 1
3 2012 "Country3" "B"  .7283527   2.321955    1 .9444444          1 .6547069 0 1  1.1606808 2010 0 0 1
3 2013 "Country3" "B"  .7077386   2.321955    1 .9444444          1 .6547069 0 1  1.1165322 2010 0 0 1
4 1992 "Country4" "B"  .3452282   .6099069 .875 .5555556 .071428575 .6121462 0 1   2.615249 1990 1 0 0
4 1993 "Country4" "B"  .3790323  2.0673358    1 .7222222 .071428575 .6121462 0 1  1.4437795 1990 1 0 0
4 1994 "Country5" "C"  .4175321   2.321955    1 .7222222 .071428575 .6121462 0 1  4.0116615 1990 1 0 0
4 2000 "Country5" "C"   .759881   2.321955    1 .8888889   .7619048 .6121462 0 1    5.97681 2000 0 1 0
4 2001 "Country5" "C"  .7410694   2.321955    1 .8888889   .7619048 .6121462 0 1   7.402936 2000 0 1 0
4 2002 "Country5" "C"  .7281089   2.321955    1 .8888889   .8333333 .6121462 0 1      8.534 2000 0 1 0
4 2003 "Country5" "C"   .735869   2.321955    1 .8888889   .8333333 .6263371 0 1      8.534 2000 0 1 0
4 2004 "Country5" "C"  .7334849   2.321955    1 .8888889   .8333333 .6263371 0 1      8.534 2000 0 1 0
4 2010 "Country5" "C"  .7798166   2.321955    1 .9444444   .9166667 .6263371 0 1     -16.25 2010 0 0 1
4 2011 "Country5" "C"  .7637033   2.321955    1 .9444444   .9166667 .6263371 0 1 -14.330358 2010 0 0 1
4 2012 "Country5" "C"  .7608096   2.321955    1 .9444444   .9166667 .6263371 0 1          0 2010 0 0 1
4 2013 "Country5" "C"  .7455279   2.321955    1        1   .9166667  .607849 0 1   .6806767 2010 0 0 1
5 1992 "Country6" "C"  .1757301          .    .        .          .        . 0 1          . 1990 1 0 0
5 1993 "Country6" "C"  .1761862          .    .        .          .        . 1 0          . 1990 1 0 0
5 1994 "Country6" "C"  .1562944          .    .        .          .        . 1 0          . 1990 1 0 0
5 2000 "Country6" "C"  .2008834 -1.2261546    .        .          .        . 1 0          . 2000 0 1 0
5 2001 "Country6" "C" .17261477 -1.2261546    .        .          .        . 1 0          . 2000 0 1 0
5 2002 "Country6" "C" .21249928 -1.2261546    .        .          .        . 1 0          . 2000 0 1 0
5 2003 "Country6" "C"  .2206576 -.02334493    .        .          .        . 1 0          . 2000 0 1 0
5 2004 "Country6" "C" .23541167 -.02334493    .        .          .        . 1 0          . 2000 0 1 0
5 2010 "Country6" "C" .27683625   1.249751    .        .          .        . 0 1          . 2010 0 0 1
5 2011 "Country6" "C" .27487376   1.249751    .        .          .        . 0 1          . 2010 0 0 1
5 2012 "Country6" "C" .27767062   1.249751    .        .          .        . 0 1          . 2010 0 0 1
5 2013 "Country6" "C"  .2884341   1.249751    .        .          .        . 0 1          . 2010 0 0 1
6 1992 "Country7" "C" .25534174  -.1539507 .625       .5          0 .8271327 1 0          . 1990 1 0 0
6 1993 "Country7" "C" .28078705  -.1539507 .625       .5  .16666667 .8271327 1 0          . 1990 1 0 0
6 1994 "Country7" "C"  .3567602  -.1539507 .625 .5555556  .16666667 .8271327 1 0          . 1990 1 0 0
6 2000 "Country7" "C"  .4190661 -1.2261546 .625 .6111111  .23809524 .8271327 0 1          . 2000 0 1 0
6 2001 "Country7" "C"  .3738855 -1.2261546 .625 .6666667   .3928571 .8271327 0 1          . 2000 0 1 0
6 2002 "Country7" "C"   .404098 -1.2261546 .625 .6666667   .3928571 .8271327 0 1          . 2000 0 1 0
6 2003 "Country7" "C"  .4157713 -1.2261546 .875 .6666667   .3928571 .7653575 0 1          . 2000 0 1 0
6 2004 "Country7" "C"  .4085605 -1.2261546 .875 .6666667  .53571427 .7653575 0 1          . 2000 0 1 0
6 2010 "Country7" "C"  .4888775 -.02334493 .875 .6666667   .7619048 .7653575 0 1          . 2010 0 0 1
6 2011 "Country7" "C"  .4789392 -.02334493 .875 .6666667   .7619048 .7653575 0 1          . 2010 0 0 1
6 2012 "Country7" "C"  .4922497 -.02334493 .875 .6666667   .7619048 .7653575 0 1          . 2010 0 0 1
6 2013 "Country7" "C"   .508882 -.02334493  .75 .6666667   .7619048 .7653575 0 1          . 2010 0 0 1
7 1992 "Country8" "D" .56565696   2.321955    1 .9444444   .7738096 .8009957 0 1  13.403038 1990 1 0 0
7 1993 "Country8" "D"  .6403931   2.321955    1 .9444444   .7738096 .8070084 0 1       12.1 1990 1 0 0
7 1994 "Country8" "D"  .6921942   2.321955    1 .9444444   .8452381 .8070084 0 1       12.1 1990 1 0 0
7 2000 "Country8" "D"  .8552948   2.321955    1        1   .9285714 .8314556 0 1      1.806 2000 0 1 0
7 2001 "Country8" "D"  .8805222   2.321955    1        1   .9285714 .8314556 0 1   1.958802 2000 0 1 0
7 2002 "Country8" "D"  .8265579   2.321955    1        1   .9285714 .8314556 0 1      2.076 2000 0 1 0
7 2003 "Country8" "D"  .8580942   2.321955    1        1          1 .8314556 0 1      2.076 2000 0 1 0
7 2004 "Country8" "D"  .8708485   2.321955    1        1          1 .8314556 0 1      2.076 2000 0 1 0
7 2010 "Country8" "D"  .8683383   2.321955    1        1          1 .8314556 0 1   3.564247 2010 0 0 1
7 2011 "Country8" "D"  .8672714   2.321955    1 .9444444          1 .8314556 0 1   5.454876 2010 0 0 1
7 2012 "Country8" "D"  .8684435   2.321955    1 .8888889          1 .8429364 0 1   5.454876 2010 0 0 1
7 2013 "Country8" "D"  .8866873   2.321955    1 .9444444          1 .8429364 0 1   5.454876 2010 0 0 1
8 1992 "Country8" "D"  .5748482   2.321955    1 .8888889  .52380955 .9702252 0 1       14.9 1990 1 0 0
8 1993 "Country8" "D"  .6702072   2.321955    1 .8888889  .52380955 .9702252 0 1   .0375137 1990 1 0 0
8 1994 "Country8" "D"  .6812267   2.321955    1 .8888889  .52380955 .9702252 0 1      -.781 1990 1 0 0
8 2000 "Country8" "D"  .8790284   2.321955    1        1        .75 .9702252 0 1      1.463 2000 0 1 0
8 2001 "Country8" "D"   .889005   2.321955    1        1        .75 .9702252 0 1   5.542341 2000 0 1 0
8 2002 "Country8" "D"  .8849987   2.321955    1        1        .75 .9702252 0 1      5.767 2000 0 1 0
8 2003 "Country8" "D"  .8916724   2.321955    1        1        .75 .9702252 0 1      5.767 2000 0 1 0
8 2004 "Country8" "D"  .8910962   2.321955    1        1        .75 .9702252 0 1      5.767 2000 0 1 0
8 2010 "Country8" "D"  .8951228   2.321955    1        1   .8333333 .9702252 0 1       -.36 2010 0 0 1
8 2011 "Country8" "D"  .8785759   2.321955    1        1   .8333333 .9702252 0 1       -.36 2010 0 0 1
8 2012 "Country8" "D"  .8894278   2.321955    1        1   .8333333 .9702252 0 1       -.36 2010 0 0 1
8 2013 "Country8" "D"  .8901961   2.321955    1        1   .8333333 .9702252 0 1   1.298654 2010 0 0 1
end

---------

[Nick Cox]

Sorry, but I can't follow what you want here. The examples given are quite different and none seems to be about data with the structure you have. You have multiple variables for multiple countries and multiple years (with some big gaps). That would usually call for some variation on twoway bar or twoway line. Interest in a second y axis usually obliges use of twoway.

[Giorgio Di Stefano]

Sorry, but I can't follow what you want here. The examples given are quite different and none seems to be about data with the structure you have. You have multiple variables for multiple countries and multiple years (with some big gaps). That would usually call for some variation on twoway bar or twoway line. Interest in a second y axis usually obliges use of twoway.

For the variables in consideration, I would like to create bar graphs by groups over decates, having a second y indicator 1.


In the image provided in #1 as an example, that would be adding indicator1 as a second y.

Hope now it is clear to you. Thank you for your time and any help you can provide.

[Nick Cox]

I understood most of that, thanks, but the devil is the details, so I don’t have anything useful to add. Other people here may be able to help.

[Giorgio Di Stefano]

I understood most of that, thanks, but the devil is the details, so I don’t have anything useful to add. Other people here may be able to help.

Thanks for your time, Nick. I have contacted the stata tech support and was told that the graph I want could be created using -twoway bar-, but I would need to manually create the -over- groups to be plotted as separate x-axis values. I do not know how doing that.

Since this is coming over and over again in the forum, I did not find a post on how to add a second y on a twoway bar. It will be a nice gesture of you if you could create a general post, ideally here, on how to add a second y in a twoway bar over groups

[Nick Cox]

I am not in general in favour of double y axes, as I have often argued on Statalist and elsewhere. Hence, sorry, but I can’t accept as an obligation to write as you wish,

[Giorgio Di Stefano]

Thanks for your time and comments. I wasn't meaning to ask you to take an obligation, just to clarify rather a gently gesture. Thank you anyway for your time and consideration. Probably I have missed your argument against the second y

[Nick Cox]

To try to summarize my views:

1 Double y axes have a bad reputation because they have been used to mislead. For one statement, see
Wainer, H. 1991. Double Y-axis graphs. Chance 4(1): 50-51. That does not mean that any researcher doing this wants to mislead. Usually they really have no intention of doing that. But it means that you should be wary of this design, and it does mean that other researchers may be wary of this design. So, you need to worry about your readers, either way.

2. Even if you think it is a good (defensible) idea, it is hard to do well. You have to explain which variable(s) refer to which axis, you are putting a burden on the reader to grapple with two kinds of information, and it is often hard to avoid a mess. Some people go for one variable as bars, one variable as a line, and so forth. That may work well, but if often just raises other problems of different kind but equal magnitude.

3. Usually there are better designs. Separate panels with separate scales are often clearer.https://www.statalist.org/forums/for...ailable-on-ssc didn't invent this idea, which is age-old.

Standardising units to the same scale (January 2020 = 100 or whatever) may be a better idea. If the real interest is in checking for correlations (simultaneous or lagged) some kind of scatter plot or trajectory in state space may be better.

If you can find past posts where I advised on how to do this, that is what I did. But as of January 2022, I don't want to write encouraging it.

[Chen Samulsion]

Agree with #8. And please allow me to refer a fact here that in the famous Stata book Statistics with STATA (sic) written by Lawrence Hamilton, the first Stata's graph (Figure 1.1) introduced to Stata beginners is just a double Y-axis one (at least the 5th edition pressed in 2006 and the 6th edition pressed in 2009). And I think anyone who read this book will have a strong impression of double Y-axis graphs, and I'm sure that some of if not most of them will play with the yaxis option repeatedly in their later Stata exercise. And another fact I must refer to is that in the last version of the book (the 8th edition pressed in 2012), the author substitute a regular (one Y-axis) line graph for the former double Y-axis one. So why? Perhaps Hamilton realized the deficit of double Y-axis graphs (?).

P.S. why so many users spell STATA instead of Stata? The textbook's title explains some variance, although some variance should be attributed to the fact that Stata actually used the spelling of STATA in the beginning. As Richard Williams have said

Incidentally, people are occasionally chastised for writing STATA instead of Stata. Actually, people who do so are actually just being old-fashioned, because it was spelled STATA in the STATA 1 Reference Manual.

c.f. https://www.statalist.org/forums/for...e-who-wants-it.
Sorry for digressing.

[Chen Samulsion]

I must add that I do not intend to criticise Hamilton, on the contrary, his book is excellent and I learned my Stata's ABC and modelling techniques from it.

[Nick Cox]

The reasons why people write STATA not Stata are

0. They are copying someone else. This may seem a circular explanation, but it holds too.

1. (Unlikely, but possible) This was the original spelling briefly used but (a) anyone who started with Stata 1.0 (not me: Stata 2.05 was my start) has had 37 years to learn better (b) it is quite hard now to find material using that spelling.

2. By a supposed analogy with SAS, SPSS and other all caps named software. (In other words, thinking Stata is an acronym.)

3. The start-up logo may seems to use some capitals.

4. To wind up people who prefer the correct spelling. (Yes, I am watching you.)

5. To reduce their chances of being hired.

(I have seen postings from employers of the form "Of course, it really doesn't matter much how you spell it, but anyone who hasn't picked up the correct spelling hasn't been around Stata long enough to be taken seriously as a competent Stata user. Worse, they are showing inattention to detail and if someone is applying for a position that involves coding, that is really serious." )