R help - Feb 2010 - "Types" of missingness

Dear R-List,

My questions concerns missing values. Specifically, is is possible to 
use different "types" of missingness in a dataset and not a 
one-size-fits-all NA?
For example, data may be missing because of an outright refusal by a 
respondent to answer a question, or because she didn't know an answer, 
or because the item simply did not apply. In later analysis it is 
sometimes useful to be able to distinguish between the cases, but 
nonetheless have them all treated as missing when using, say, lm( ).
In Stata this is possible by using different missing value indicators. 
The standard one is a period '.' whereas '.a' and '.b'
etc are treated
as missing too, but can all be distinguished from another (they are even 
ordinal such that . < .a < .b).
To give a simplistic example in R, let

 > dat <- data.frame(
+ hours = c(36, 40, 40, 0, 37.5, 0, 36, 20, 40),
+ wage = c( 15.5, 7.5, 8, -1, 17.5, -1, -2, 13, -2))
 > dat
   hours wage
1  36.0 15.5
2  40.0  7.5
3  40.0  8.0
4   0.0 -1.0
5  37.5 17.5
6   0.0 -1.0
7  36.0 -2.0
8  20.0 13.0
9  40.0 -2.0


where for wages -1 indicates "didn't work" and -2 indicates
"refused to
respond". How could I replace the negative values for wages with 
missingness indicators to use the data frame in for instance lm( ), but 
later operate only on those observations who "refused to respond"?
Of course I can always work around this somehow, especially in this easy 
example, but as data frames get larger and cases more complex the 
workarounds seem more and more klutzy to me.
So, if there is an easy way to do this that I have overlooked, I would 
be grateful for any advice or references.

Best,
Christian

-- 
Christian Raschke
Department of Economics
and
ISDS Research Lab (HSRG)
Louisiana State University
Patrick Taylor Hall, Rm 2128
Baton Rouge, LA 70803
crasch2@lsu.edu


	[[alternative HTML version deleted]]

Check out:
https://stat.ethz.ch/pipermail/r-help/2010-February/228228.html

On Sun, Feb 28, 2010 at 2:39 AM, Christian Raschke
<crasch2 at tigers.lsu.edu> wrote:
> Dear R-List,
>
> My questions concerns missing values. Specifically, is is possible to
> use different "types" of missingness in a dataset and not a
> one-size-fits-all NA?
> For example, data may be missing because of an outright refusal by a
> respondent to answer a question, or because she didn't know an answer,
> or because the item simply did not apply. In later analysis it is
> sometimes useful to be able to distinguish between the cases, but
> nonetheless have them all treated as missing when using, say, lm( ).
> In Stata this is possible by using different missing value indicators.
> The standard one is a period '.' whereas '.a' and
'.b' etc are treated
> as missing too, but can all be distinguished from another (they are even
> ordinal such that . < .a < .b).
> To give a simplistic example in R, let
>
> ?> dat <- data.frame(
> + hours = c(36, 40, 40, 0, 37.5, 0, 36, 20, 40),
> + wage = c( 15.5, 7.5, 8, -1, 17.5, -1, -2, 13, -2))
> ?> dat
> ? hours wage
> 1 ?36.0 15.5
> 2 ?40.0 ?7.5
> 3 ?40.0 ?8.0
> 4 ? 0.0 -1.0
> 5 ?37.5 17.5
> 6 ? 0.0 -1.0
> 7 ?36.0 -2.0
> 8 ?20.0 13.0
> 9 ?40.0 -2.0
>
>
> where for wages -1 indicates "didn't work" and -2 indicates
"refused to
> respond". How could I replace the negative values for wages with
> missingness indicators to use the data frame in for instance lm( ), but
> later operate only on those observations who "refused to
respond"?
> Of course I can always work around this somehow, especially in this easy
> example, but as data frames get larger and cases more complex the
> workarounds seem more and more klutzy to me.
> So, if there is an easy way to do this that I have overlooked, I would
> be grateful for any advice or references.
>
> Best,
> Christian
>
> --
> Christian Raschke
> Department of Economics
> and
> ISDS Research Lab (HSRG)
> Louisiana State University
> Patrick Taylor Hall, Rm 2128
> Baton Rouge, LA 70803
> crasch2 at lsu.edu
>
>
> ? ? ? ?[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>