Not missing at random

Tue, Jun 7, 2011 7:05 AM

Hi Blaz,

What do you do if the number of values sampled to be set missing
(e.g., 4) is greater than the number of values for a given case that
are less than your < 3 threshold?  If no special considerations are
needed for that, I do not see why you cannot apply the same technique
you did below with MCAR to MNAR.

Best regards,

Josh

On Tue, Jun 7, 2011 at 12:17 AM, Blaz Simcic <blazsimcic at yahoo.com> wrote:

________________________________
From: Joshua Wiley <jwiley.psych at gmail.com>
To: Blaz Simcic <blazsimcic at yahoo.com>
Cc: r-help at r-project.org
Sent: Mon, June 6, 2011 10:34:38 PM
Subject: Re: [R] Not missing at random

Hi Blaz,

See below.

x <-
matrix(c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,3,3,3,4),
nrow = 7, ncol=7, byrow=TRUE) ####matrix

pMiss <- 30? ? ####percent of missing values

N <- dim(x)[1]? ####number of cases

candidate <- which(x[,1]<3 | x[,2]<3 | x[,3]<3 | x[,4]<3 | x[,5]<3 | x[,6]<3
|
x[,7]<3)? ? #### I want to sample all cases with at least 1 value
lower than 3, so I have to find candidates

## easier to use this
## find all x < 3 and return their row and column indices
## select only row indices, and then find unique
candidate <- unique(which(x < 3, arr.ind = TRUE)[, "row"])

idMiss <- sample(candidate, N * pMiss / 100)? #### I sampled cases

## from the subset of x cases that will be missing
## find all that are < 3 and set to NA
x[idMiss, ][x[idMiss, ] < 3] <- NA

## If you are going to do this a lot, consider a function
nmar <- function(x, op = "<", value = 3, p = 30) {
? op <- get(op)
? candidate <- unique(which(op(x, value), arr.ind = TRUE)[, "row"])
? idMiss <- sample(candidate, nrow(x) * p / 100)
? x[idMiss, ][op(x[idMiss, ], value)] <- NA
? return(x)
}

nmar(x)

## has the advantage that you can easily change
## p, the cut off value, the operator (e.g., "<", ">", "<=", etc.)

Cheers,

Josh

On Sun, Jun 5, 2011 at 11:17 PM, Blaz Simcic <blazsimcic at yahoo.com> wrote:


Hello!

I would like to sample 30 % of cases (with at least 1 value lower than 3 -
in
the row) and among them I want to set all values lower than 3 (within
selected
cases) as NA (NMAR- Not missing at random). I managed to sample cases, but
I
don?t know how to set values (lower than 3) as NA.

R code:

x <-

matrix(c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,3,3,3,4),
?nrow = 7, ncol=7, byrow=TRUE) ####matrix

pMiss <- 30???? ####percent of missing values

N <- dim(x)[1]?? ####number of cases

candidate<-which(x[,1]<3 | x[,2]<3 | x[,3]<3 | x[,4]<3 | x[,5]<3 | x[,6]<3
|
x[,7]<3)??? #### I want to sample all cases with at least 1 value lower
than 3,
so I have to find candidates

idMiss <- sample(candidate, N * p / 100)??? #### I sampled cases

Now I'd like to set all values among sampled cases as NA.

Any suggestion?

Thanks,
Bla?
? ? ? ?[[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.





--
Joshua Wiley
Ph.D. Student, Health Psychology
University of California, Los Angeles
http://www.joshuawiley.com/

Joshua Wiley
Ph.D. Student, Health Psychology
University of California, Los Angeles
http://www.joshuawiley.com/