How to overlay distributions using a pie chart using StataMP 17? (plot distributions on the same graph)

[Rabab Al hasni]

Dear Statalist,

I hope you are well.

Please, I need help on how to overlay distributions using a pie chart using StataMP 17?

In my paper, I made a pie chart - using the below syntax - I found the Distribution of ‘i’ more extreme for micro-businesses. Therefore, I have been advised to overlay the distributions (said: just plot both distributions on the same graph).

graph pie, over( Int_RATE_01 ) plabel(_all percent) by(, legend(on)) by( Ent_size )

Another syntax I used to make a pie chart that shows interest rate distribution by bank finance application status (whether received finance or not)

graph pie, over( Int_RATE_01 ) plabel(_all percent) by(, legend(on)) by( Finance_App_Status )

The questionnaire question of the related information is this: What was the average interest rate you paid for the loan? The nature of the data is categorical variables. The response categories are as follows:

1. 0-3% (number of the responses 51)
2. 4-5% (number of the responses 7)
3. 6-7% (number of the responses 13)
4. 8-9% (number of the responses 12)
5. Over 10% (number of the responses 0)

Here is a partial dataset of about 83 firms (Small and Medium Enterprises):

* Example generated by -dataex-. For more info, type help dataex
clear
input int firm float Ent_size long(Int_RATE_01 Finance_App_Status)
 1 4 2 0
 2 4 1 1
 3 3 1 0
 4 3 4 1
 5 4 4 1
 6 3 1 1
 7 3 4 1
 8 1 4 1
 9 1 3 1
10 1 1 1
11 1 1 1
12 1 2 1
13 3 1 0
14 1 1 0
15 1 2 1
16 3 1 1
17 4 3 1
18 4 1 1
19 4 1 1
20 1 1 1
21 3 1 0
22 1 1 1
23 4 1 1
24 1 1 1
25 3 4 1
26 1 1 1
27 3 3 1
28 1 4 1
29 4 2 1
30 3 3 1
31 4 1 1
32 1 1 0
33 3 2 1
34 4 3 1
35 1 3 1
36 3 3 1
37 1 1 1
38 3 4 1
39 3 1 1
40 3 1 1
41 3 1 1
42 1 1 1
43 3 1 1
44 3 1 1
45 1 1 1
46 1 4 1
47 1 4 1
48 4 4 1
49 4 2 1
50 1 1 1
51 4 1 1
52 3 3 1
53 1 1 1
54 4 1 1
55 3 3 1
56 4 1 1
57 4 1 1
58 4 1 1
59 3 3 1
60 1 4 1
61 3 1 1
62 1 1 1
63 3 1 1
64 4 3 1
65 4 1 1
66 3 1 1
67 3 1 1
68 1 1 1
69 1 1 1
70 3 1 1
71 4 2 1
72 3 1 1
73 4 3 1
74 4 1 1
75 1 4 1
76 3 1 1
77 1 1 1
78 1 1 1
79 1 1 1
80 3 3 1
81 4 1 1
82 4 1 1
83 3 1 1
end
label values Ent_size Ent_size
label def Ent_size 1 "Micro", modify
label def Ent_size 3 "small", modify
label def Ent_size 4 "Medium", modify
label values Int_RATE_01 Int_RATE
label def Int_RATE 1 "0 -3%", modify
label def Int_RATE 2 "4 -5%", modify
label def Int_RATE 3 "6 -7%", modify
label def Int_RATE 4 "8 -9%", modify
label values Finance_App_Status Needing_Funding
label def Needing_Funding 0 "Not Recieved", modify
label def Needing_Funding 1 "Recieved", modify

I am struggling with how to answer the above question

Greatly appreciate your help

Best regards,
Rabab

[Nick Cox]

I don't get a clear picture of how you want to modify your pie charts. I rather start from the position that even simple pie charts are inherently problematic and are not made better by making them more complicated.

Backing up, it seems that you have a 3 way table with interest rate as outcome to explain and the other variables as predictors. That suggests to me a three-way bar chart using tabplot from the Stata Journal.

Here is a sample effort. The major paper was in 2016, the latest software update was in 2020 and there's an executive summary at https://www.statalist.org/forums/for...updated-on-ssc

. search tabplot, sj

Search of official help files, FAQs, Examples, and Stata Journals

SJ-20-3 gr0066_2 . . . . . . . . . . . . . . . . Software update for tabplot
(help tabplot if installed) . . . . . . . . . . . . . . . . N. J. Cox
Q3/20 SJ 20(3):757--758
added new options frame() and frameopts() allowing framing
of bars and so-called thermometer plots or charts

SJ-16-2 gr0066 . . . . . . Speaking Stata: Multiple bar charts in table form
(help tabplot if installed) . . . . . . . . . . . . . . . . N. J. Cox
Q2/16 SJ 16(2):491--510
provides multiple bar charts in table form representing
contingency tables for one, two, or three categorical variables

tabplot Int_ Ent_, by(Fin, t1title(Needing funding) note("")) yreverse bfcolor(green*0.2) percent(Ent_ Fin) showval  xtitle(Entity size) ytitle(Interest rate) subtitle(, fcolor(blue*0.1)) scheme(s1color)

I made some tweaks to your data example (thanks!) removing spaces and fixing typos. More importantly, there is plenty of scope to modify this plot, e.g. by showing frequencies rather than percents or basing the percents differently.

[Maarten Buis]

I agree with Nick (again). However, visually the variable needing funding is given a special role (there is a collored box around its values), what you realy want is two predictors to have equal "visual weight" (if one gets a box than the other gets a box as well). This is what twby prefix can do. To install type in Stata ssc install twby . You have to do more work with twby, so it is not as convenient as tabplot.

clear
input int firm float Ent_size long(Int_RATE_01 Finance_App_Status)
 1 4 2 0
 2 4 1 1
 3 3 1 0
 4 3 4 1
 5 4 4 1
 6 3 1 1
 7 3 4 1
 8 1 4 1
 9 1 3 1
10 1 1 1
11 1 1 1
12 1 2 1
13 3 1 0
14 1 1 0
15 1 2 1
16 3 1 1
17 4 3 1
18 4 1 1
19 4 1 1
20 1 1 1
21 3 1 0
22 1 1 1
23 4 1 1
24 1 1 1
25 3 4 1
26 1 1 1
27 3 3 1
28 1 4 1
29 4 2 1
30 3 3 1
31 4 1 1
32 1 1 0
33 3 2 1
34 4 3 1
35 1 3 1
36 3 3 1
37 1 1 1
38 3 4 1
39 3 1 1
40 3 1 1
41 3 1 1
42 1 1 1
43 3 1 1
44 3 1 1
45 1 1 1
46 1 4 1
47 1 4 1
48 4 4 1
49 4 2 1
50 1 1 1
51 4 1 1
52 3 3 1
53 1 1 1
54 4 1 1
55 3 3 1
56 4 1 1
57 4 1 1
58 4 1 1
59 3 3 1
60 1 4 1
61 3 1 1
62 1 1 1
63 3 1 1
64 4 3 1
65 4 1 1
66 3 1 1
67 3 1 1
68 1 1 1
69 1 1 1
70 3 1 1
71 4 2 1
72 3 1 1
73 4 3 1
74 4 1 1
75 1 4 1
76 3 1 1
77 1 1 1
78 1 1 1
79 1 1 1
80 3 3 1
81 4 1 1
82 4 1 1
83 3 1 1
end
label values Ent_size Ent_size
label def Ent_size 1 "Micro", modify
label def Ent_size 3 "small", modify
label def Ent_size 4 "Medium", modify
label values Int_RATE_01 Int_RATE
label def Int_RATE 1 "0 -3%", modify
label def Int_RATE 2 "4 -5%", modify
label def Int_RATE 3 "6 -7%", modify
label def Int_RATE 4 "8 -9%", modify
label values Finance_App_Status Needing_Funding
label def Needing_Funding 0 "Not Recieved", modify
label def Needing_Funding 1 "Recieved", modify

// set the scheme
set scheme s1color

// make distance between size categories equal
replace Ent_size = Ent_size -1 if Ent_size > 1
label def Ent_size 2 "small", modify
label def Ent_size 3 "Medium", modify

// add some variable labels, which will be used to label the axes
label var Int_RATE_01 "Interest rate"
label var Ent_size "Size of enterprice"
label var Finance_App_Status "Application status"

// create the table
contract Int_RATE_01 Ent_size Finance_App_Status, nomiss zero
egen tot = total(_freq), by(Ent_size Finance_App_Status)
gen perc = _freq / tot * 100

// to display the numbers
format perc %5.0f
gen y = -10

// the graph
twby Ent_size Finance_App_Status,                            ///
        compact left xoffset(0.5) legend(off)                ///
        title("Percentage in each interest category"         ///
              "given Application status and size") :         ///
    twoway bar perc Int_RATE_01,                             ///
        xlab(1/4, val) barwidth(.8) ylab(none) ytitle("") || ///
    scatter y Int_RATE_01 if perc > 0,                       ///
    msymbol(none) mlab(perc) mlabpos(0) mlabcolor(black)

[Rabab Al hasni]

Dear Nick Cox and Maarten Buis,


Many thanks for your prompt reply. your great explanation is very helpful and fruitful.

In fact, the reviewer send me this comment "Distribution of ‘i’ more extreme for micro-businesses. Could you overlay these distributions?" regarding the below pie chart:

graph pie, over( Int_RATE_01 ) plabel(_all percent) by(, legend(on)) by( Ent_size )

so I was wondering how to overlay distributions of the pie chart using StataMP 17. I thought maybe he want me to overlay two pie charts of a relevant questionnaire question.

I would like to ask please, what does mean the 100.0 that appears under each bar of those who have not received finance of the Nick Cox post #2? is it a number or percentage? How to read/ interpret it?


If I would like to overlay distributions of the pie chart using twoway bar chart of interest rate and enterprise size, is the below syantax correct (using same dataset of mine):

tabplot Int_RATE_01 Ent_size , yreverse bfcolor(green*0.2) percent(Ent_ Fin) showval xtitle( Ent_size ) ytitle( Int_RATE_01 ) subtitle(, fcolor(blue*0.1)) scheme(s1color)

How to interpret the figure that appears under each bar of the chart?


Many thanks for being available to help us in improving our analysis through Stata.

Best regards,
Rabab

[Rabab Al hasni]

Dear Nick Cox and Maarten Buis,

Below syntax shows the number of firms under each bar, how to get a percentage for the frequency of enterprises that have received or not finance based on interest rate?

tabplot Int_RATE_01 Ent_size , by( Finance_App_Status , compact note("")) showval(mlabsize(*.5)) ysize(7)

I have tried to past the graph here but it does not work so I upload it as an attachment in this post.


Greatly appreciate your help

Best regards,
Rabab

[Nick Cox]

Various comments on #4 and #5.

The graphs in #2 and #3 show the probability of a value in each interest rate category, given entity size and finance received, as a %. All the values in the data example for micro and finance not received fall into the lowest interest rate category, so 100% is shown. And so on. What is shown follows in my case from the call to the option percent() and in Maarten's case from his calculating the percent concerned before he calls up the graph. In a paper I would explain that on the graph or through a text caption.

If you want to see frequencies, omit the percent() call in tabplot or for Maarten's approach calculate the frequencies beforehand.

If you wish to see different percents, vary the percent() call in tabplot. See the help for examples.

Sorry, but I don't understand what the reviewer is asking for any more than I did in #2. You will need to contact them. If the review was anonymous, the journal editors may be willing to forward a question to the reviewer. They may be wanting you to add numeric annotation or to add something graphical. There is no point in my trying to guess, not least because the reviewer appears to think that pie charts are a good start, and I strongly disagree with that.

The graph attachment in #6 doesn't display properly because you need to attach .png not .gph -- this is explained in the FAQ Advice

[Rabab Al hasni]

Various comments on #4 and #5.

The graphs in #2 and #3 show the probability of a value in each interest rate category, given entity size and finance received, as a %. All the values in the data example for micro and finance not received fall into the lowest interest rate category, so 100% is shown. And so on. What is shown follows in my case from the call to the option percent() and in Maarten's case from his calculating the percent concerned before he calls up the graph. In a paper I would explain that on the graph or through a text caption.

If you want to see frequencies, omit the percent() call in tabplot or for Maarten's approach calculate the frequencies beforehand.

If you wish to see different percents, vary the percent() call in tabplot. See the help for examples.

Sorry, but I don't understand what the reviewer is asking for any more than I did in #2. You will need to contact them. If the review was anonymous, the journal editors may be willing to forward a question to the reviewer. They may be wanting you to add numeric annotation or to add something graphical. There is no point in my trying to guess, not least because the reviewer appears to think that pie charts are a good start, and I strongly disagree with that.

The graph attachment in #6 doesn't display properly because you need to attach .png not .gph -- this is explained in the FAQ Advice

Dear Nick Cox,

Many thanks for your clarification

I will try to re-attach the graph again using gph

Thank you very much

Best regards,
Rabab

[Nick Cox]

No; use .png not .gph; that is the point. You can save your graph as .png within Stata from a Graph window.

[Rabab Al hasni]

Ok, Nick
Thank you very much for giving your time to help and sorry for my silly mistakes when I post enquiries through the Stata forum.

Best regards,
Rabab

[Rabab Al hasni]

No; use .png not .gph; that is the point. You can save your graph as .png within Stata from a Graph window.

Dear Nick Cox,

For one of the attached graphs, I reached the answer regarding how to make the bars of the graph vary in percentage - thank you for that.

Regarding the other attachment, I would like to ask if the below syntax describes the twoway table of interest rate and enterprises size:

tabplot Int_RATE_01 Ent_size , by( Finance_App_Status , compact note("")) showval(mlabsize(*.5)) ysize(7)

I have tried to make the bars of this graph vary in percentage so to be all in a total of 100% but could not. Could you please help with that?


Many thanks,

Best regards,
Rabab

[Nick Cox]

We can see a graph now, which is good progress.

I can't see that the code you show bears any exact relation to the graph you show. First off, in your code the option by() would imply a graph with two panels. Second off, the subtitle shown does not appear in your code and (believe me) would not appear unless you asked for it.

Backing up, various possibilities include

percent

each bar shows a percent of the total for all data

percent(Ent_size)

percents add to 100 within groups of Ent_size

similar comments for any other variable

percent(Ent_size Fin)

percents add to 100 within cross-combinations of those two variables.

It seems to me that you are guessing wildly and trying all kinds of syntax and getting confused even on which syntax produces which graph.

That's no good for you, and for anyone trying to understand what is going on.

When you're developing graph code you can do this: add an option like

name(G1, replace)

and then that graph will appear in window G1 and you can follow what that syntax does. Similarly you can have other graphs with commands including say

name(G2, replace)

[Rabab Al hasni]

Dear Nick Cox,

Thank you for your patient

I am so sorry, I pasted in my post #10 the wrong syntax. The syntax that is referred to the attached graph in my post #10 is this

tabplot Int_RATE_01 Ent_size , yreverse bfcolor(green*0.2) percent(Ent_ Fin) showval xtitle( Ent_size ) ytitle( Int_RATE_01 ) subtitle(, fcolor(blue*0.1)) scheme(s1color)

In which I was asking if it describes a two-way or one bar chart.

Your explanations have helped me to answer my research question and solved its related issue. indeed, I value and respect your help so much


Deeply, thank you very much for giving your attention and time

Best regards,
Rabab

[Nick Cox]

Thanks for the warm closure. Glad it worked.