How can I increase the number of characters on axis labels on a graph?

[Aidan Horn]

I have slightly long value labels for a variable I have created a horizontal bar chart for, and they have been truncated. I would like to increase the length of the axis labels, if possible, as they are more important information in this case than the bars (i.e., I would like the bars to take up less space in the graph).

graph hbar (asis) prop in 1/20, over(occup2merge, sort(prop) descending label(labsize(small))) title("Modal two-digit occupations in PALMS", span) note("Total proportion of categories represented here: `represent'")

Dataex (I collapsed the dataset with a sampling weight, and generated the percentages. I have just looked up -contract- and I see that I should have used contract.):

* Example generated by -dataex-. To install: ssc install dataex
clear
input float occup2merge double count float prop
91 159901414.10819277     .16708773
51  94526189.24348901      .0987744
41  65583660.85793375     .06853113
93  59233561.85169494     .06189565
71  58507001.71862424     .06113643
92  58345606.10650118     .06096778
83 51823326.948460504     .05415238
12  47010386.76683183     .04912313
52  40218623.46498255     .04202614
72 39016734.760210834     .04077023
42   33545790.1701924     .03505341
34  31543574.66525055    .032961212
33 28412273.521815125     .02968918
13 26106903.896305665    .027280204
82 22430265.660911735     .02343833
24 21995550.273464024     .02298408
31 19696664.311145265    .020581875
32  17166525.78292244    .017938027
74  16765460.62872613    .017518936
23 12755740.775627147     .01332901
21  9954382.979502445    .010401755
81  8386879.132879108    .008763803
61  8167152.061751989    .008534201
62  8015457.204215655    .008375689
22  7372757.707365178    .007704105
73  5115432.935445052    .005345332
11 3441729.1718450896    .003596408
 8 1426840.1302517399   .0014909654
53  282556.7655838961   .0002952555
35  161876.3596279645  .00016915143
15  27176.76826377258  .00002839815
98   25693.6422316163  .00002684837
99 13908.765224794282 .000014533855
57  4631.417922723242  4.839564e-06
77 1393.7234084775369 1.4563603e-06
84 1157.9445286330504  1.209985e-06
40 1059.1637348730678 1.1067648e-06
44  651.9258829926297  6.812248e-07
56  643.7014519852817  6.726308e-07
14  629.4893419596493    6.5778e-07
55  628.4369225556962  6.566802e-07
47  623.5352522169171  6.515583e-07
68    502.07629356924  5.246407e-07
43 474.04928602656105  4.953541e-07
67  403.4298282043139  4.215608e-07
30 354.26303438678246  3.701844e-07
29    272.17477357745  2.844069e-07
58 249.17024150356258  2.603685e-07
end
label values occup2merge occup2merge
label def occup2merge 12 "12 Corporate managers", modify
label def occup2merge 13 "13 General managers", modify
label def occup2merge 23 "23 Teaching professionals", modify
label def occup2merge 24 "24 Other professionals", modify
label def occup2merge 31 "31 Natural science and engineering associate professionals", modify
label def occup2merge 32 "32 Life science and health associate professionals", modify
label def occup2merge 33 "33 Teaching associate professionals", modify
label def occup2merge 34 "34 Other associate professionals", modify
label def occup2merge 41 "41 Office clerks", modify
label def occup2merge 42 "42 Customer service clerks", modify
label def occup2merge 51 "51 Personal and protective services", modify
label def occup2merge 52 "52 Models, salespersons and demonstrators", modify
label def occup2merge 71 "71 Extraction and building trades workers", modify
label def occup2merge 72 "72 Metal and machinery workers", modify
label def occup2merge 74 "74 Other craft workers", modify
label def occup2merge 82 "82 Machine operators and assemblers", modify
label def occup2merge 83 "83 Drivers and mobile-plant operators", modify
label def occup2merge 91 "91 Elementary sales and services", modify
label def occup2merge 92 "92 Agricultural and fishery labourers", modify
label def occup2merge 93 "93 Labourers in mining, construction, manufacturing and transport", modify

[Nick Cox]

If you are willing to sacrifice the first three characters, then you can make progress like this:

decode occup2merge, gen(toshow)
replace toshow = substr(toshow, 4, .)

graph hbar (asis) prop in 1/20, over(toshow, sort(prop) descending label(labsize(small))) title("Modal two-digit occupations in PALMS", span) note("Total proportion of categories represented here: `represent'") yla(0(0.05)0.15) xsize(7)

[Sergiy Radyakin]

On a slightly different issue, I find ordering by value here makes it hard to interpret the multiple 'other' categories here, such as 'Other associate professionals'. It is straightforward when you see the 30-s group:

31 Natural science and engineering associate professionals
32 Life science and health associate professionals
33 Teaching associate professionals
34 Other associate professionals

But on the chart it is all chaotically interleaved with other groups, so interpretation of 'other' is confusing. Perhaps, present it as separate charts by first digit code?? Maintaining the last position for other, regardless of its value.

Best, Sergiy Radyakin

PS: in other words, interpretation of "Other X" is possible only in the context of already presented X1, X2,... and (imho) is not possible independently.

[Aidan Horn]

I didn't really want to sacrifice the first three characters, as they are useful information for the reader to check the occupations codebook. The following example shows that there is enough room in the graph for more characters. Is it impossible in Stata to force more characters in? Where is the character restriction documented?
I am using Stata version 15.1.

graph hbar (asis) prop in 1/20, over(occup2merge, sort(prop) descending label(labsize(small))) title("Modal two-digit occupations in PALMS", span) note("Total proportion of categories represented here: `represent'") yla(0(0.05)0.15) aspectratio(1.5) xsize(7) ysize(4.5)

[Scott Merryman]

Caveat emptor.

Somewhere along the way -graph- calls _ggroup.ado which truncates the label lengths to 32 character. Lines 78-80:

        if ("`truncate'" == "") {
            local truncate 32
        }

If the change this to 65, -discard-, and then re-run your example you get:

[Nick Cox]

That is a good detail from Scott Merryman. The best fix would be if graph bar (etc.) allowed a truncate() option to be passed along. Otherwise hacking at Stata's own code is at your peril.

[Aidan Horn]

Thank you Scott. I perhaps have a different version to you, because I had `c(namelenchar)' as the value on line 41 of "C:\Stata15\ado\base\_\_ggroup.ado":

local truncate 65 // `c(namelenchar)'

Nick, I understand that changing this is risky (I'd also caution other readers of this post to be careful when changing the Stata base code, at least without remembering it). I'll keep this change I made in mind as I continue using Stata in future. I've encountered this problem of having too-short axis labels before, so I like it how I'm able to change this.

[Scott Merryman]

Yes, I am using version 16.1:

. which _ggroup
C:\Program Files\Stata16\ado\base\_\_ggroup.ado
*! version 1.0.0  14apr2020

[Aidan Horn]

Something to note is that the SSC package splitvallabels can transform the value labels onto multiple lines.
https://www.ssc.wisc.edu/sscc/pubs/stata_bar_graphs.htm

[Andrew Musau]

Hacking Stata code seems overly excessive, unless you often produce graphs with very long labels. You can simply specify labels greater than 32 characters using the -relabel- option. See #11 of the following link.

https://www.statalist.org/forums/for...rid-in-a-graph

[Aidan Horn]

Hacking Stata code seems overly excessive, unless you often produce graphs with very long labels. You can simply specify labels greater than 32 characters using the -relabel- option. See #11 of the following link.

https://www.statalist.org/forums/for...rid-in-a-graph

Andrew, could you please show that -relabel- can be applied to my example?
Thanks

[Andrew Musau]

Andrew, this question is about how to increase the length of the axis labels, and I believe that -relabel- does not achieve that.

Erm, actually it does as illustrated in the linked example.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float occup2merge double count float prop
91 159901414.10819277     .16708773
51  94526189.24348901      .0987744
41  65583660.85793375     .06853113
93  59233561.85169494     .06189565
71  58507001.71862424     .06113643
92  58345606.10650118     .06096778
83 51823326.948460504     .05415238
12  47010386.76683183     .04912313
52  40218623.46498255     .04202614
72 39016734.760210834     .04077023
42   33545790.1701924     .03505341
34  31543574.66525055    .032961212
33 28412273.521815125     .02968918
13 26106903.896305665    .027280204
82 22430265.660911735     .02343833
24 21995550.273464024     .02298408
31 19696664.311145265    .020581875
32  17166525.78292244    .017938027
74  16765460.62872613    .017518936
23 12755740.775627147     .01332901
21  9954382.979502445    .010401755
81  8386879.132879108    .008763803
61  8167152.061751989    .008534201
62  8015457.204215655    .008375689
22  7372757.707365178    .007704105
73  5115432.935445052    .005345332
11 3441729.1718450896    .003596408
 8 1426840.1302517399   .0014909654
53  282556.7655838961   .0002952555
35  161876.3596279645  .00016915143
15  27176.76826377258  .00002839815
98   25693.6422316163  .00002684837
99 13908.765224794282 .000014533855
57  4631.417922723242  4.839564e-06
77 1393.7234084775369 1.4563603e-06
84 1157.9445286330504  1.209985e-06
40 1059.1637348730678 1.1067648e-06
44  651.9258829926297  6.812248e-07
56  643.7014519852817  6.726308e-07
14  629.4893419596493    6.5778e-07
55  628.4369225556962  6.566802e-07
47  623.5352522169171  6.515583e-07
68    502.07629356924  5.246407e-07
43 474.04928602656105  4.953541e-07
67  403.4298282043139  4.215608e-07
30 354.26303438678246  3.701844e-07
29    272.17477357745  2.844069e-07
58 249.17024150356258  2.603685e-07
end
label values occup2merge occup2merge
label def occup2merge 12 "12 Corporate managers", modify
label def occup2merge 13 "13 General managers", modify
label def occup2merge 23 "23 Teaching professionals", modify
label def occup2merge 24 "24 Other professionals", modify
label def occup2merge 31 "31 Natural science and engineering associate professionals", modify

graph hbar (asis) prop in 1/20, over(occup2merge, descending label(labsize(small)))
gr save gr1, replace
graph hbar (asis) prop in 1/20, over(occup2merge, sort(prop) descending ///
relabel(5 "31. Natural science and engineering associate professionals" )label(labsize(small)))
gr save gr2, replace
gr combine gr1.gph gr2.gph

Res.:

This thread can be closed from here.

Good that you solved your problem, but bear in mind that if you post on a public forum like Statalist, the thread is not only for your own benefit but for others who will face the same problem in the future. Therefore, others can continue this thread if they have a similar issue and there is no ownership of threads as such.