Description Usage Arguments Details Value Author(s) References See Also Examples
msvymean
computes robust HorvitzThompson estimates of the mean or robust weighted mean estimates for complex samples based on using Mestimation.
1 2 
y 
a formula object (only one variable) 
design 
a 
k 
robustness tuning constant 
type 
either 
na.rm 
should cases with missing values be dropped? (default 
control 
control object; see 
... 
(additional specifications which are delivered to 
msvymean
performs (inverse probability) weighted Mestimation (Huber psifunction or asymmetric Huber psifunction; asymmetric=TRUE
). The msvymean
methods supports the following two methods (depending on the underlying sampling design)
robust HorvitzThompson estimator (type="rht"
),
robust weighted mean estimator (type="rwm"
).
If y
is positively correlated with the inclusion probabilities, a "rht" type estimator should be used, and "rwm" otherwise. The initial value is a weighted median or a ratio of weighted medians. You may set steps
equal to one in order to get a onestep estimator. Variance estimates are computed as firstorder linearization using the designbasedestimation facilities in the survey package.
msvymean
allows also the estimation for domains. Use the command subset
and a design subset expression instead of the original survey.design
object in msvymean
(see examples for more details).
Users may set exact=TRUE
to compute an "exact" linearizationvariance estimate, which takes into account that the MAD has been used as preliminary scale estimate. However, the estimates may become very unstable.
Object of class "svystat.rob"
.
The following (S3) methods are defined for objects of the class "svystat.rob"
:
print
method,
summary
method,
coef
method,
vcov
method,
residuals
method,
robweights
method.
Beat Hulliger and Tobias Schoch
Hulliger, B. (1995): Outlier robust HorvitzThompson estimators, Survey Methodology 21 (1), pp. 7987.
Hulliger, B. (1999): Simple and robust estimators for sampling, Proceedings of the Survey Research Methods Section, American Statistical Association, 1999, pp. 5463.
Hulliger, B. and T. Schoch (2011): Elementary robust estimators. In: Robust methodology for Laeken indicators: AMELI Deliverable D4.2, ed. by B. Hulliger, A. Alfons, P. Filzmoser, A. Meraner, T. Schoch and M. Templ. AMELI Project.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20  ## load "api" data set from "survey" package (a description of the data
## set can be found there)
data(api)
## define "survey.design" for stratified sampling
dstrat < svydesign(id=~1,strata=~stype, weights=~pw, data=apistrat,
fpc=~fpc)
## compute a robust HorvitzThompson estimate for the mean of the
## variable "api00" (Academic Performance Index in 2000)
rht1 < msvymean(~api00, dstrat, type="rht", k=1.2)
# get a summary of the estimation
summary(rht1)
## robust HorvitzThompson estimates for a domain of the variable. Here
## we are interessted in the robust mean for api00 for
## (sch.wide == "Yes"). That is the average of the academic performance
## in 2000 only for the schools that met the schoolwide growth target.
msvymean(~api00, subset(dstrat, sch.wide == "Yes"), type="rht", k=1.2)
## to extract the estimate from the object
coef(rht1)
## to extract the variance from the object
vcov(rht1)

Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.