Bar graph?

[River Huang]

Dear All, I have the following data

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(lly btot privo) int rgdpsh
 18.3381  74.1596 15.68565 4462
 51.7266  92.6655 54.81786 7782
 67.4966  98.4359 65.29842 5143
 24.6914  86.3752 13.54584  952
 51.5927  91.8502 40.64384 2666
 49.0197  92.0092 25.64779 5495
 16.3877  31.5974 13.14186 1148
 19.1631  61.6758 21.44617 1784
56.50429   88.997 60.86423 7258
 22.9597  52.8085  27.8104 2885
 22.4096  80.1646 22.08222 1684
 29.3846  72.8176 21.77378 2096
 74.4912    92.72 62.01511 2037
 49.4787  88.0999 42.44699 6760
 20.5826  73.3352 19.11292 1195
 20.0486  62.1616 17.98507 1461
 26.9358  72.0441 22.84953 1427
 37.4766  96.9463  23.6507 2108
  45.349  97.2227 51.78182 5291
   63.37  96.5411 75.46823 5823
57.46272   97.571 76.45579 6570
  17.577  39.7841  5.07367  894
 53.3381  74.3563 36.72324 2093
 20.2198  75.1588 13.31726 1660
 52.9648   58.018  20.5166 1596
 22.6035  23.7174   7.7135  924
 23.0376  76.5371  23.8559 1039
 31.7611  88.9397 34.79329 4964
 32.9494  63.6846 19.53084  766
 54.7421  94.7271 49.14149 3311
 51.9463  84.2815 37.43197 3477
77.47569  87.7728 59.08989 4564
 36.8505  78.0906 24.54591 1773
125.9392  96.7181 128.3783 2954
 35.7355  81.2692 22.78104  659
 41.0236  83.9456  65.4788  904
  9.7347  37.9002 10.16035  717
 63.7392  96.4534 46.89995 1420
143.4259   92.565 43.97008 1374
 46.8707  82.2124 24.36066 2862
 25.5744  69.3824 22.89418 2836
 20.2658  57.8637  7.72334  628
71.41078  98.0962 86.69409 6077
  49.626    82.43 37.59236 7960
 14.4336  83.8875 13.05017  532
 54.0413  90.0224 81.62302 5610
 38.6757  67.8879 20.76761  638
 33.3723   71.971 40.22082 1575
 31.0542   89.117  20.8444 1235
 17.6203  65.2917 14.51989 1177
 18.5189  86.0365 13.32469 2019
 27.5044  81.4047 26.97613 1133
  78.022  90.3459 55.01267 1869
 22.7968   84.538 27.50557 1047
 16.8288  49.5504  5.06637  878
 51.4366  94.7736 71.93586 2191
 70.3053  92.7402 65.04914 3123
 30.3415  57.7968 16.20979 1259
 53.4937  88.9401 89.11234 7592
123.4055  98.9879 141.2944 9409
 43.3227  46.9346   8.8252 1575
 66.9684  95.7224 57.85528 1256
 47.7859  84.6563 47.27738  943
    32.9  82.0514 21.88465  367
 37.4615  91.7183 31.39167 5627
 48.6267  83.5528 46.31212 6823
 62.1195  93.1142 113.0658 9895
 29.4668  59.2383 21.21116 3968
 36.8425  90.8998 33.12155 6338
 16.0277  28.5825  4.08484  489
46.92189 75.80044 23.03637  989
end

where the financial development indicators are

1. lly: LIQUID LIABILITIES
2. btot: COMMERCIAL-CENTRAL BANK
3. privo: PRIVATE CREDIT

I use the following code to generate four income groups (where `rgdpsh' is the income per capita):

xtile rgdpsh4= rgdpsh, nq(4)

and I'd like to have a figure like the following:

Any suggestion is highly appreciated!

[ram singh]

Dear River,
The following code or similar may give you the graph you want ( I have not labelled the variables)
I have used the mean which gives very similar graph to the one you have posted

xtile rgdpsh4= rgdpsh, nq(4)
*then
rangestat (mean) lly btot privo, interval(rgdpsh4, 0, 0)
collapse (max) btot_mean privo_mean lly_mean, by( rgdpsh4)
xpose, clear varname
drop in 1
gr bar v*, over(_varname)

Apologies, I am sure there may be more eloquent ways of doing it than my clunky suggestion
Kind regards
RS

[Nick Cox]

You don't say what you want the graph to show: means? totals? something else?

Regardless, statplot (SSC) can help. See e.g.

https://www.stata.com/statalist/arch.../msg00447.html

http://researchnotes.eric-a-booth.co...-examples.html

http://researchnotes.eric-a-booth.co...es-part-2.html

Notes:

1. VARIABLE LABELS LIKE THIS are surely dysfunctional. They are harder to read, take up more space and for many create a synesthetic sense of SHOUTING.

2. You don't need a legend when the different quartile bins can be labelled directly.

3. You don't need different shadings (especially those with irrational mixes of colour) for the same reason.

4. I'd add units of measurement if I knew what they were.

5. Text explaining the binning is up to you. In practice it often goes better in the Figure caption.

clear
input float(lly btot privo) int rgdpsh
 18.3381  74.1596 15.68565 4462
 51.7266  92.6655 54.81786 7782
 67.4966  98.4359 65.29842 5143
 24.6914  86.3752 13.54584  952
 51.5927  91.8502 40.64384 2666
 49.0197  92.0092 25.64779 5495
 16.3877  31.5974 13.14186 1148
 19.1631  61.6758 21.44617 1784
56.50429   88.997 60.86423 7258
 22.9597  52.8085  27.8104 2885
 22.4096  80.1646 22.08222 1684
 29.3846  72.8176 21.77378 2096
 74.4912    92.72 62.01511 2037
 49.4787  88.0999 42.44699 6760
 20.5826  73.3352 19.11292 1195
 20.0486  62.1616 17.98507 1461
 26.9358  72.0441 22.84953 1427
 37.4766  96.9463  23.6507 2108
  45.349  97.2227 51.78182 5291
   63.37  96.5411 75.46823 5823
57.46272   97.571 76.45579 6570
  17.577  39.7841  5.07367  894
 53.3381  74.3563 36.72324 2093
 20.2198  75.1588 13.31726 1660
 52.9648   58.018  20.5166 1596
 22.6035  23.7174   7.7135  924
 23.0376  76.5371  23.8559 1039
 31.7611  88.9397 34.79329 4964
 32.9494  63.6846 19.53084  766
 54.7421  94.7271 49.14149 3311
 51.9463  84.2815 37.43197 3477
77.47569  87.7728 59.08989 4564
 36.8505  78.0906 24.54591 1773
125.9392  96.7181 128.3783 2954
 35.7355  81.2692 22.78104  659
 41.0236  83.9456  65.4788  904
  9.7347  37.9002 10.16035  717
 63.7392  96.4534 46.89995 1420
143.4259   92.565 43.97008 1374
 46.8707  82.2124 24.36066 2862
 25.5744  69.3824 22.89418 2836
 20.2658  57.8637  7.72334  628
71.41078  98.0962 86.69409 6077
  49.626    82.43 37.59236 7960
 14.4336  83.8875 13.05017  532
 54.0413  90.0224 81.62302 5610
 38.6757  67.8879 20.76761  638
 33.3723   71.971 40.22082 1575
 31.0542   89.117  20.8444 1235
 17.6203  65.2917 14.51989 1177
 18.5189  86.0365 13.32469 2019
 27.5044  81.4047 26.97613 1133
  78.022  90.3459 55.01267 1869
 22.7968   84.538 27.50557 1047
 16.8288  49.5504  5.06637  878
 51.4366  94.7736 71.93586 2191
 70.3053  92.7402 65.04914 3123
 30.3415  57.7968 16.20979 1259
 53.4937  88.9401 89.11234 7592
123.4055  98.9879 141.2944 9409
 43.3227  46.9346   8.8252 1575
 66.9684  95.7224 57.85528 1256
 47.7859  84.6563 47.27738  943
    32.9  82.0514 21.88465  367
 37.4615  91.7183 31.39167 5627
 48.6267  83.5528 46.31212 6823
 62.1195  93.1142 113.0658 9895
 29.4668  59.2383 21.21116 3968
 36.8425  90.8998 33.12155 6338
 16.0277  28.5825  4.08484  489
46.92189 75.80044 23.03637  989
end
label var lly "Liquid liabilities"
label var btot "Commercial-central bank" 
label var privo "Private credit" 
xtile rgdpsh4= rgdpsh, nq(4)
set scheme s1color 
statplot lly btot privo, over(rgdpsh4) recast(bar) bar(1, bfcolor(green*0.2)) yla(, ang(h))

[Nick Cox]

ram singh

I like rangestat (SSC, as you are asked to explain) too, but

rangestat (mean) lly btot privo, interval(rgdpsh4, 0, 0)
collapse (max) btot_mean privo_mean lly_mean, by( rgdpsh4)

itself can be collapsed to

collapse (mean) lly btot privo, by(rgdpsh4)

You will have now seen in #3 a way to do it without disturbing the dataset shape or layout.

[ram singh]

Dear Nick,
Brilliant and thanks for your input. Great to learn from you as always.
Kind regards,
RS

[River Huang]

Dear Ram, Many thanks for the suggestion.

[River Huang]

Dear Nick, Sorry not indicating that the `mean' is what I had in mind, and many thanks for the wonderful solutions.

[River Huang]

Dear Nick, As I posted in #1 (Figure 1), the (4) colors in each financial indicator (Liquid liabilities, Commercial-central bank, and Private credit) denotes the mean/average of the indicator of the low, lower middle, upper middle, and high-income countries. As such, any suggestions to make them `colorful'? Thanks.

[Nick Cox]

As signalled in #3 I can't see any need for, or gain in, adding colours. If your readership doesn't understand 1 2 3 4, then how will colours help? If the 4 bars were for years 2010 2011 2012 2013 would you want different colours? Why add a legend when you don't need one?

Less (clutter and arbitrary detail) is more (intelligent), in my view.

But statplot is (here) just a wrapper for graph bar, as the help indicates, so asyvars opens up the command to fruit salad and Joseph's technicolour dreamcoat.

[River Huang]

Dear Nick, After looking at the graph, I think you were right. Thanks.