[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: RE: AW: creating an ordered month variable over several years

From   Michael McCulloch <>
To   Statalist Statalist <>
Subject   Re: st: RE: AW: creating an ordered month variable over several years
Date   Sat, 28 Nov 2009 13:46:57 -0800

Thank you Austin,
Your suggestion of the -scatter- command with a -recast- option provides exactly what I want to display:
	sc sampled screened noreply m, recast(bar)
This command is the last line of the following complete simulation example.

However, I have been unable to find how to control color, having explored both the option commands for -scatter- and -bar-. Can you suggest how to do that? I'm seeking a green/yellow/red color scheme for sampled/screened/noreply.

*** begin example
clear all
set seed 1
set obs 1000
gen id=_n
gen month=1 + int(12*uniform())
gen year=2008 + int(2*uniform())
gen subject_type=1 + int(3*uniform())
la def lab_subjects 1 "no reply" 2 "screened" 3 "sampled", modify
la val subject_type lab_subjects
g m=(year-1960)*12+month-1
format m %tmm_CY
la var m "Calendar month"
levelsof m, loc(vs)
foreach v of loc vs {
la def m `v' "`:di `:format m' `v''", modify
la val m m
* insert Nick Cox's cum freq table
by m, sort: gen freq = _N // sort by the variable of interest and get the frequencies by m: gen cumfreq = _N if _n == 1 // get the cumulative frequencies replace cumfreq = sum(cumfreq) // The trick here is to put the frequencies once more into a variable, /// but only once within each group. We chose to do this for the first observation in each group; that /// is what is meant by _n == 1. In many other problems, we could do it also for the last observation /// in each group, but that is not good in this problem. The cumulative sum produced by the sum() function /// treats all the missing values produced by the previous command as 0, which is precisely what we want. /// That is, the cumulative frequency is, as its definition requires, the cumulative sum of just one group ///
         frequency from each group.
tabdisp m, cell(freq cumfreq)

byhist m, by(subject_type) disc tw(xla(576/599,angle(90) val))
contract m subject_type, z f(one)
reshape wide one, j(subject_type) i(m)
g noreply=max(0,one1)
g screened=max(0,one1)+max(0,one2)
g sampled=max(0,one1)+max(0,one2)+max(0,one3)
sc sampled screened noreply m, recast(bar) xla(576/599,angle(90)) name(s)
*** end example

Best wishes,

Michael McCulloch
Pine Street Foundation
124 Pine Street
San Anselmo, CA 94960-2674
tel:	415-407-1357
fax: 	206-338-2391

*   For searches and help try:

© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index