Ok, thanks for the example. I think I understand now.
Does the following code accomplish what you want (It assumes you
already know that the maximum number of categories is 26):
clear
input str2 tests noof pos str4 timetest
bb 3 4.96 sysa
hh 4 4.98 sysa
ss 6 4.56 sysa
uu 7 21.33 sysa
zz 5 145.19 sysa
a4 5 37.69 sysa
a5 4 9.98 sysa
a6 7 6.85 sysa
a7 5 15.71 sysa
a8 4 7.02 sysa
a9 5 4.81 sysa
b1 7 4.61 sysa
d1 9 12.51 sysa
e1 3 4.75 sysa
f1 12 6.96 sysa
g1 4 6.95 sysa
m3 11 7.32 sysa
n1 7 5.69 sysa
r1 7 3.93 sysa
aa 5 6.70 sysb
dd 4 4.25 sysb
ii 5 3.95 sysb
jj 5 4.26 sysb
kk 14 3.14 sysb
oo 5 5.68 sysb
pp 5 2.90 sysb
ss 6 3.71 sysb
uu 7 5.48 sysb
b6 7 3.52 sysb
d1 17 28.71 sysb
e1 5 7.23 sysb
f1 12 4.59 sysb
g1 4 5.63 sysb
h1 28 4.96 sysb
i1 15 4.60 sysb
j1 3 6.50 sysb
l1 4 7.79 sysb
l2 4 4.88 sysb
m1 11 3.96 sysb
m2 11 4.96 sysb
n2 7 3.76 sysb
o1 3 4.39 sysb
o4 4 5.10 sysb
o5 12 3.25 sysb
q1 7 3.54 sysb
cc 12 9.77 sysx
nn 9 7.26 sysx
uu 7 6.98 sysx
yy 3 7.03 sysx
b2 6 4.21 sysx
d1 10 6.31 sysx
j1 5 6.49 sysx
o3 3 9.08 sysx
o6 7 3.77 sysx
q2 7 3.81 sysx
ee 12 30.95 sysz
kk 10 3.33 sysz
ll 5 4.53 sysz
qq 9 4.17 sysz
rr 6 3.73 sysz
uu 7 4.83 sysz
yy 3 4.86 sysz
a3 3 5.69 sysz
b3 5 4.62 sysz
d1 33 10.30 sysz
j1 11 6.20 sysz
k1 3 4.62 sysz
k1 3 4.69 sysz
l1 4 4.64 sysz
n3 3 6.39 sysz
o2 6 4.65 sysz
o8 7 3.46 sysz
p1 4 9.66 sysz
q2 7 3.13 sysz
s1 12 4.71 sysz
ff 12 6.09 sysk
gg 6 38.33 sysk
nn 9 4.24 sysk
a1 4 8.76 sysk
c1 5 18.41 sysk
j1 5 4.06 sysk
o7 7 4.38 sysk
ff 12 9.35 sysL
gg 6 30.83 sysL
mm 5 8.40 sysL
nn 9 6.34 sysL
tt 6 4.67 sysL
uu 7 3.81 sysL
a2 8 10.94 sysL
c1 8 6.33 sysL
d1 20 5.85 sysL
j1 5 5.16 sysL
n3 3 5.45 sysL
o2 6 36.70 sysL
o9 7 5.67 sysL
end
levelsof time,local(levels)
foreach l of local levels {
count if time == "`l'"
if r(N) !=26 {
local size = ((26-2)/r(N))*2.5
}
else {
local size = 1
}
disp "`size'"
gr hbar noof if timetest=="`l'", over(test, sort(pos) gap(*`size')
label(labsize(*.65))) ///
ytitle("No. of studies") ysc(r(0 35)) ylab(0 (5) 35) ///
name(gr_`l',replace)
local name "`name' gr_`l'"
}
graph combine `name'
On 8/21/07, b. water <[email protected]> wrote:
<snip>
> i want to produce gr hbar ala [G] page 119 for each varlists in timetest,
> which i can accomplish easy enough, ending up with 6 hbar graphs. naturally,
> the size of the hbar would be different for each graph. what i want to
> achieve is these six graphs but with consistent hbar size accross the six
> graphs. so overall, i would end up with graphs that have the x-axis
> (horizontal perimeter) of the graph consistent because i am fixing it with
> ysc(r(0 35)) ylab(0 (5) 35) but that the y-axis (vertical perimeter) of
> differing lengths depending on obviously on the no. of tests.
>
> scott's suggested approach, while feasible, is not ideal because some of the
> tests are not applicable to some of the varlists, resulting, if scott's
> suggested approach was used, redundant gaps in the various graphs.
> furthermore, there will be more tests and different testtime varlists to be
> added in the future.
>
> not only i wanted to do the hbar in the way i have described with consistent
> hbar size, i was hoping that the answer from statalisters would enable me to
> apply it to creation of consistent gr hbox size, ala [G] page 152, but again
> with differing y-axis (vertical perimeter) lengths depending on the no. of
> tests.
>
> down the road, i am want to combine the six graphs (for now) with gr combine
> so that i have two panels ala [G] page 142 and page 158 respectively but
> obviously then each column in the respective panels would be of different
> vertical size.
>
> after going through [G] i am not certain that i wanted is achievable, at
> least not on stata 8.2 hence the question to statalisters. there might be a
> 'lateral thinking' solution to achieve what i envisage but at the moment i
> am drawing blanks.
>
> hope i have made it clearer. all [G] references are of course stata 8.2.
> again, thank you in advance.
>
> regards,
> bw
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
