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

RE: st:easy way to deal with "paired" variables?

From   "Nick Cox" <>
To   <>
Subject   RE: st:easy way to deal with "paired" variables?
Date   Mon, 2 Feb 2009 16:48:02 -0000

Maarten gave very good advice. It's perhaps worth pointing out as well
that it is possible to work "row-wise", which is the title of my next
Speaking Stata column in Stata Journal 9(1) (2009). The topic was
sparked by several postings last month on this list. 

You could loop over the variables, say like this: 

gen doingtime = 0 

qui foreach r of var reason* { 
	local g : subinstr local r "reason" "gap" 
	replace doingtime = doingtime + 
			cond(`r' == 4 & !missing(`g'), `g', 0) 

Or suppose you know in advance that there are reason1-reason42,

gen doingtime = 0 

qui forval i = 1/42 { 
	replace doingtime = doingtime + 
		cond(reason`i' == 4 & !missing(gap`i'), gap`i', 0) 

The main trickiness that can be guessed from Mandy's example is the
possibility of missing values. That's tackled above. 

Maarten buis

--- Mandy fu <> wrote:
> The data set I'm working on includes 100 variables measuring the time
> spell of gaps between jobs. And for each gap there's a variable
> explaining why the person does not work during that gap. I want to
> calculate the total length of  any gaps caused by being in
> prison(reason=4) .

Problems like these can sometimes become a lot easier when the data is
transformed from a wide format to a long format. As you seem to want to
avoid creating a prison dummy, I used in the example below a logical
statement (reason==4) inside the -total- command. Logical statements
evaluate to 1 if true and 0 if false, so this statement plays the same
role as the dummies in your method.

*-------------- begin example --------------------
drop _all
input id gap1 reason1 gap2 reason2              
      1  30   1       20   2       
      2  25   2       180  4 
reshape long gap reason, i(id) j(spell)
bys id : egen prisontime = total((reason==4)*gap)
*----------------- end example -------------------
(For more on how to use examples I sent to the Statalist, see )

*   For searches and help try:

© Copyright 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index