Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: Re: ST: npresent/nmissing

From   "Nick Cox" <>
To   <>
Subject   st: RE: Re: ST: npresent/nmissing
Date   Wed, 9 Feb 2005 20:50:16 -0000

Thanks for the positive comments. 

For those in the dark, -nmissing- (and -npresent-, 
really the same program, if you look closely) 
were published in STB-49, then STB-60, then 
SJ 3-4. 

I can see why Ric wants this. I have thoughts
under various heads:

1. Yes, any one can clone the code and modify the 
program to do this. It's just that -nmissing- is 
my name (unless StataCorp in turn grab it off me, 
or grab it off of me, as they might say, being Texan). 

2. I don't want to do that myself. It could be done 
by an option that a user can easily ignore, but at 
the same time my thought is that for what it does 
-nmissing- is already a bit more complicated than 
I would really like. There is often a small design 
issue with commands, especially relatively minor 
utilities: ideally each should do one thing, and 
that one thing well, and designers should resist 
urges to add just one more option. "Creeping featuritis"
is I believe Eric Raymond's term. I wanted -nmissing- 
to be minimal in its output: adding scope to 
complicate that output is moving in the wrong direction
as far as I am concerned. 

3. Nevertheless I have changed -nmissing- so 
that the variable list is returned in -r(varlist)-. 
This allows users to fire up -describe- afterwards, 

4. If you learn -foreach- the caprice of programmers
and what they will or won't do is not an issue. 

Just a minor tweak to some code posted yesterday 
in answer to Ric's original question gives
part of an answer: 

foreach v of var * { 
	qui count if mi(`v') 
	if r(N) >= 80 local badlist "`badlist'`v' " 
d `badlist' 

and something like that can go in a do file or 


Eric Uslaner
> Nick Cox kindly (as always) responded to my query about listing
> variables with a minimum number of cases through npresent and nmissing
> (which he wrote).  nmissing and esp. npresent are wonderful.  
> But I was
> wondering if there is anyway to hack them to have them place variable
> labels on the RHS in addition to the variable names and 
> number of cases
> (as fsum does).  Thanks much for any help.

*   For searches and help try:

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