[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: tokenize and missing

From   "Nick Cox" <[email protected]>
To   <[email protected]>
Subject   st: RE: tokenize and missing
Date   Mon, 28 Jan 2008 16:48:54 -0000

I am pleased to see -nmissing- being useful to anyone. But the problem
here, as I understand it, can be tackled more directly: 

*! NJC 1.0.0 28 January 2008 
program listmissing	
	version 8 
	syntax [varlist] [if] [in] [, *] 
	marksample touse, novarlist 
	qui count if `touse' 
	if r(N) == 0 error 2000 

	qui foreach v of local varlist { 
		count if `touse' & missing(`v') 
		if r(N) local vlist `vlist' `v' 

	local vlist : subinstr local vlist " " ",", all 
	list `vlist' if missing(`vlist'), `options'  

[email protected] 	 

Ronnie Babigumira

I have a quarterly data set and I am trying to identify missing cases

houscode       qtr   wage_ld  biz_lead
       72         1         0         0
       81         1         0         1
       81         2         .         1
       11         3         0         1
       10         4         .         .
        9         1         1         0

I started by using Nick Cox's very useful -nmissing- (ssc install

Which gave me

wage_ld             2
biz_lead            1

Knowing which variables have missing cases, I

list if mi(wage_ld, biz_lead)

However, to generalize it (I will be doing it by quarter) I used

tokenize `r(varlist)'
list if mi(`1' `2')

This works for the data as it is, however, you will notice that the
missing cases are in qtr2. So, if I did this 
analysis by quarter and I

keep if qtr == 1
tokenize `r(varlist)'
list if mi(`1', `2')

I get ........invalid syntax...............

I reckon this is coming from the fact that indeed for qtr == 1, I do not
have two variables with missing data. So, is 
there a way I can tweak this to allow it to proceed?

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index