Graph bar for multiple categories

[Keira Gb]

Hello everyone,
I am brand new to Stata, so apologies if my question seems easy to some of you.
I have the following variables:

School: an encoded variable ranging from 1-13 for different school names
Jobs: a dummy variable with 0= no job, 1=job
Sex: a second variable with 1= M, 2= F.

I want to classify how the school impacts the job search, for males and females. To do this, I want to make two horizontal bar graphs with blue bars for males, red for women, and for every categories of schools: one graph for Job Yes and one graph for Job No.
I manage to get what I want with this command:

graph hbar (percent), over(Sex) over(School) over(Jobs)

and I get this "nice" graph:



Now, since the orders of magnitude are very different, I don't see anything in the second part of the graph. So I want just to split the two graphs, to make the bars visibles for Job too.
If I simply do:

graph hbar (percent) if Jobs==1, over(Sex) over(School)

I get this, which is not exactly what I want cause I don't need all the M F, I just would like a legend as in the previous graph.



And if I do

graph hbar (percent) School if Jobs==1, over(Sex)

, then is completely not what I want:



Where am I wrong?
Thank you to anyone who can help!!

[Carole J. Wilson]

See a recent post here for an explanation of the options below: http://www.statalist.org/forums/forum/general-stata-discussion/general/1336245-loosing-precision-in-y-axis-large-number?p=1336679#post1336679


Separating sex into two variables gets you the different colors with only one over() variable

separate Sex, by(Sex)
graph hbar (percent) Sex0 Sex1  , over(School) by(Jobs, xrescale)

When working with graph bar or hbar, be sure that you are using the correct code, otherwise your percentages will be off.

****I just committed this error!
For example, the above will give the the percent of those with Jobsthat are men/women and have gone to school #. (the row/col percentage), not the percent of those in the whole sample who are m/w & school # & job/no job (cell).

[Carole J. Wilson]

To get the percent of those in the whole sample:

separate Sex, by(Sex)
preserve
collapse (percent) Sex0 Sex1, by(Jobs School)
graph hbar (percent) Sex0 Sex1  , over(School) by(Jobs, col(1) xrescale) legend(lab(1 "new label 1") lab(2 "new label 2")) ysize(10)
restore

[Nick Cox]

catplot (SSC) offers a variety of percent controls.

Elementary tutorial:

sysuse auto, clear
ssc inst catplot 
catplot rep78 foreign, percent
catplot rep78 foreign, percent(foreign)
catplot rep78 foreign, percent(rep78)

[Keira Gb]

Thank you both for your help, but unfortunately I'm still confused.

catplot is very temptating to use, as it doesn't need to create new variables. But if I do

catplot Sex School Jobs

, I get the same result (with frequency instead of %) as with
[CODE]

 
 graph hbar (percent), over(Sex) over(School) over(Jobs)

Still not what I am looking for.

On the other hand, separating Sex gives something similar, but you'r right Carole, that's not the correct %, as I need the percentage over the WHOLE number of individuals. And if I do your second suggestion with

col(1) rescale

I get this, which is still very weird


I am puzzled 'cause it should not be so complicated to separate my initials graphs...

[Nick Cox]

I am not clear exactly what you want. Why not post your data? You could do this

 
contract Sex Jobs School  
ssc inst dataex 
dataex

to get something to post. See also

http://www.statalist.org/forums/help#stata

[Keira Gb]

Thank you Nick, I'll have a look on how to post my data.
What I want is just the first graph I posted

but either with a different scalebar on the lower panel (to be able to visualize the data), or two separate graphs for Job & No Job, where percentages are calculated on the entire sample (the sum of ALL the bars Jobs+NoJobs should give 100).

Not sure if it is more clear.
If not, I'll try to post my data.
Thank you very much!!!

[Nick Cox]

It will be much easier to answer if we can see your data and I've already explained how to do it. Should take two minutes to learn how.

[Keira Gb]

Thank you Nick. Here you are the code. I am actually not sure anymore that the percentage on the graph I just posted are correct, again.
What I really need is as I said, an overview of the WHOLE sample (sum of all the bars=100), classified by Schools and Sex and Jobs Y/N.

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte Sex float Jobs long School byte _freq
1 0  1 1
1 0  2 1
1 0  3 1
1 0  4 1
1 0  5 1
1 0  6 1
1 0  7 1
1 0  8 1
1 0  9 1
1 0 10 1
1 0 11 1
1 0 12 1
1 0 13 1
1 0  . 1
1 1  1 1
1 1  2 1
1 1  4 1
1 1  5 1
1 1  8 1
1 1  9 1
1 1 10 1
1 1 11 1
1 1 12 1
1 1 13 1
1 1  . 1
1 .  1 1
1 .  2 1
1 .  3 1
1 .  4 1
1 .  5 1
1 .  6 1
1 .  7 1
1 .  8 1
1 .  9 1
1 . 10 1
1 . 11 1
1 . 12 1
1 . 13 1
1 .  . 1
2 0  1 1
2 0  2 1
2 0  3 1
2 0  4 1
2 0  5 1
2 0  6 1
2 0  7 1
2 0  8 1
2 0  9 1
2 0 10 1
2 0 11 1
2 0 12 1
2 0 13 1
2 0  . 1
2 1  1 1
2 1  2 1
2 1  3 1
2 1  4 1
2 1  5 1
2 1  6 1
2 1  8 1
2 1  9 1
2 1 10 1
2 1 11 1
2 1 12 1
2 1 13 1
2 1  . 1
2 .  1 1
2 .  2 1
2 .  3 1
2 .  4 1
2 .  5 1
2 .  6 1
2 .  7 1
2 .  8 1
2 .  9 1
2 . 10 1
2 . 11 1
2 . 12 1
2 . 13 1
2 .  . 1
. 0  1 1
. 0  2 1
. 0  8 1
. 0  . 1
. 1  . 1
. .  1 1
. .  2 1
. .  3 1
. .  4 1
. .  5 1
. .  6 1
. .  7 1
. .  8 1
. . 12 1
. . 13 1
. .  . 1
end
label values Sex S
label def S 1 "M", modify
label def S 2 "F", modify
label values Jobs j
label def j 0 "no job", modify
label def j 1 "job", modify
label values School sch
label def sch 1 "1", modify
label def sch 2 "2", modify

[Nick Cox]

Sorry, but that can't be your data. No cross-combination of categories occurs more than once. I don't know what you did wrong, but there's no useful graph from that.

[Carole J. Wilson]

Keira, does the solution in #3 not work?

[Keira Gb]

Sorry, Nick, I might have launched twice the command contract. Is it now better?

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte sex float jobs long(school _freq)
1 0  1  2579
1 0  2 31890
1 0  3    73
1 0  4  2109
1 0  5  6408
1 0  6    16
1 0  7  1163
1 0  8 34427
1 0  9  1392
1 0 10   847
1 0 11   669
1 0 12    85
1 0 13  7946
1 0  .  4480
1 1  1    33
1 1  2   104
1 1  4    10
1 1  5     4
1 1  8    27
1 1  9    56
1 1 10     8
1 1 11     4
1 1 12     3
1 1 13     2
1 1  .    31
2 0  1  1401
2 0  2 26951
2 0  3 10445
2 0  4  1830
2 0  5  3850
2 0  6  7459
2 0  7  1535
2 0  8 35771
2 0  9  1521
2 0 10   625
2 0 11   500
2 0 12    47
2 0 13  1563
2 0  .  4199
2 1  1    20
2 1  2    52
2 1  3     4
2 1  4     4
2 1  5     1
2 1  6     5
2 1  8    33
2 1  9    27
2 1 10     1
2 1 11     1
2 1 12     3
2 1 13     2
2 1  .    16
. 0  1     1
. 0  2     1
. 0  8     1
. 0  .     5
. 1  .     1
end
label values sex sex
label def sex 1 "M", modify
label def sex 2 "F", modify
label values jobs j
label def j 0 "no job", modify
label def j 1 "job", modify
label values school sch
label def sch 1 "1", modify
label def sch 2 "2", modify
label def sch 3 "3", modify
label def sch 4 "4", modify
label def sch 5 "5", modify
label def sch 6 "6", modify
label def sch 7 "7", modify
label def sch 8 "8", modify
label def sch 9 "9", modify
label def sch 10 "10", modify
label def sch 11 "11", modify
label def sch 12 "12", modify
label def sch 13 "13", modify

[Keira Gb]

Carole, solution #3 gives the graph I showed in answer #5.... something still not working but I can't figure out what.

[Nick Cox]

Good. Is this closer?

catplot sex school [fw=_freq], by(jobs, note("")) percent(jobs school) asyvars bar(1, bfcolor(blue*0.5))  bar(2, bfcolor(pink*0.5))

[Keira Gb]

Something like that, but not exactly. I need percentages on the total size of the sample, not on each category. Here it looks like that if I take M+F let's say, with job from school 8, then I have 100% (as from i.e. F at school 3, they already fill the 100%). So it looks like this graph allows me to see on a specific category, what is the percentage of M and of F.
I would like to classify the categories on the ENTIRE set of data, so that I could see on the entire population what is the percentage of let's say M, that have a job and comes from school 3.