Using catplot to graph

[Anne Wanyonyi]

Hello,
I have three variables that I would like to graph. They are school attendance(hv121), access to electricity(hv206), gender(hv104).
P/S - all the variables are dummy variables

I used the following commands:

ssc inst catplot

catplot hv104, hv206, percent(hv121) asyvars

This is my output:

Kindly, how do I label the y-axis to be no electricity instead of no and electricity instead of yes?



P/S. I did most of the manipulations from the graph itself

[Anne Wanyonyi]

The graph is in posting #4

[Nick Cox]

The graph is (so far) unreadable. See http://www.statalist.org/forums/help#stata for how to post graphs (save them first as .png).

The syntax is illegal (too many commas). Do check carefully what you post.

[Anne Wanyonyi]

Apologies. I have posted it again.

[Nick Cox]

Just change the value labels, i.e whatever is giving you "yes" or "no".

NB the vertical axis here is not to Stata the y axis. See the help for graph hbar for an explanation.

[Anne Wanyonyi]

Hello Nick,

Here is a snapshot of the graph again (I tried to do some corrections and re-upload but it ended up getting lost from the post #4)



I finally managed to change the value labels but I realized the plotted values do not correspond to the values that I should expect:

bysort hv206: tab hv104 hv121

-> hv206 = no

           |    member attended
    sex of | school during current
 household |      school year
    member |        no  attended  |     Total
-----------+----------------------+----------
      male |       326      4,199 |     4,525 
    female |       298      4,083 |     4,381 
-----------+----------------------+----------
     Total |       624      8,282 |     8,906 


----------------------------------------------------------------------------------------------------------
-> hv206 = yes

           |    member attended
    sex of | school during current
 household |      school year
    member |        no  attended  |     Total
-----------+----------------------+----------
      male |        41        824 |       865 
    female |        46        846 |       892 
-----------+----------------------+----------
     Total |        87      1,670 |     1,757

hv121 is school attendance
hv206 is access to electricity
hv104 is gender

If I use the above results from tab command, the fractions should be:
Electricity: school attendance of males should be 0.51
school attendance of females should be 0.4

No electricity: school attendance of males should be 0.49
school attendance of females should be 0.51

The corrected formula I used for the catplot is :

catplot hv104 hv206, fraction(hv121) asyvars

What could be bringing about the incorrect values in the graph? and how can I go about addressing it?

Thank you.

[Nick Cox]

This is hard to follow, for several reasons:

1. Contrary to advice in #3 and in the Statalist FAQ Advice, you are not posting graphs as .png attachments. Images are more difficult to read without clicking on them separately, which slows down understanding.

2. You don't post example data, contrary to the Statalist FAQ Advice. It's possible to reconstitute your data from your tables, however.

3. You don't make clear (to me) precisely what you want to plot. Which variable is response? Which are explanatory? If it's attendance as response, then it's clear from your tables that most children attend, which is far from your results implying about half.

4. You don't seem troubled by the fact that you have a 3 way table with 2 x 2 x 2 = 8 cross-combinations, but your graph shows only 4 bars,

5. Dopey variable names such as hv121 hv206 hv104 are completely unmemorable, at least to readers here, as shown by the fact that you have to keep explaining them.

Here is a stab at what you may want. Fond though I am of catplot, it seems much simpler here, just to plot the mean non-attendance percents. As attendance averages about 95%, it is a much better use of space to plot the complement.

clear

input attendance electricity female freq 
0 0 0   326 
0 0 1   298 
0 1 0    41 
0 1 1    46 
1 0 0  4199  
1 0 1  4083  
1 1 0   824  
1 1 1   846 
end 

label def female 0 male 1 female 
label val female female 
label def attendance 0 no 1 attended 
label val attendance attendance 
label def electricity 0 no 1 yes 
label val electricity electricity 

tab female attendance if !electricity [fw=freq] 

bysort electricity : tab female attendance [fw=freq] 

gen nonattendance = 100 * (1 - attendance) 

graph bar (mean) nonattendance [w=freq] , over(female) over(electricity) ///
ytitle(did not attend (%))

[Nadeeka Weera]

Can I use markers option with catplot?

[Nick Cox]

#8 If you use recast(dot) then the bars become point symbols or markers and you can change them. I am not clear whether that is what you are asking.