svyCooksD: Modified Cook's D for models fitted with complex survey data

In svydiags: Linear Regression Model Diagnostics for Survey Data

Modified Cook's D for models fitted with complex survey data

Description

Compute a modified Cook's D for fixed effects, linear regression models fitted with data collected from one- and two-stage complex survey designs.

Usage

svyCooksD(mobj, stvar=NULL, clvar=NULL, doplot=FALSE)

Arguments

mobj	model object produced by svyglm in the survey package
stvar	name of the stratification variable in the svydesign object used to fit the model
clvar	name of the cluster variable in the svydesign object used to fit the model
doplot	if TRUE, plot the modified Cook's D values vs. their sequence number in data set. Reference lines are drawn at 2 and 3

Details

svyCooksD computes the modified Cook's D (m-cook; see Atkinson (1982) and Li & Valliant (2011, 2015)) which measures the effect on the vector of parameter estimates of deleting single observations when fitting a fixed effects regression model to complex survey data. The function svystdres is called for some of the calculations. Values of m-cook are considered large if they are greater than 2 or 3. The R package MASS must also be loaded before calling svyCooksD. The output is a vector of the m-cook values and a scatterplot of them versus the sequence number of the sample element used in fitting the model. By default, svyglm uses only complete cases (i.e., ones for which the dependent variable and all independent variables are non-missing) to fit the model. The rows of the data frame used in fitting the model can be retrieved from the svyglm object via as.numeric(names(mobj$y)). The data for those rows is in mobj$data.

Value

Numeric vector whose names are the rows of the data frame in the svydesign object that were used in fitting the model

Author(s)

Richard Valliant

References

Atkinson, A.C. (1982). Regression diagnostics, transformations and constructed variables (with discussion). Journal of the Royal Statistical Society, Series B, Methodological, 44, 1-36.

Cook, R.D. (1977). Detection of Influential Observation in Linear Regression. Technometrics, 19, 15-18.

Cook, R.D. and Weisberg, S. (1982). Residuals and Influence in Regression. London:Chapman & Hall Ltd.

Li, J., and Valliant, R. (2011). Linear regression diagnostics for unclustered survey data. Journal of Official Statistics, 27, 99-119.

Li, J., and Valliant, R. (2015). Linear regression diagnostics in cluster samples. Journal of Official Statistics, 31, 61-75.

Lumley, T. (2010). Complex Surveys. New York: John Wiley & Sons.

Lumley, T. (2021). survey: analysis of complex survey samples. R package version 4.1-1.

See Also

svydfbetas, svydffits, svystdres

Examples

require(MASS)   # to get ginv
require(survey)
data(api)
    # unstratified design single stage design
d0 <- svydesign(id=~1,strata=NULL, weights=~pw, data=apistrat)
m0 <- svyglm(api00 ~ ell + meals + mobility, design=d0)
mcook <- svyCooksD(m0, doplot=TRUE)

    # stratified clustered design
require(NHANES)
data(NHANESraw)
dnhanes <- svydesign(id=~SDMVPSU, strata=~SDMVSTRA, weights=~WTINT2YR, nest=TRUE, data=NHANESraw)
m2 <- svyglm(BPDiaAve ~ as.factor(Race1) + BMI + AlcoholYear, design = dnhanes)
mcook <- svyCooksD(mobj=m2, stvar="SDMVSTRA", clvar="SDMVPSU", doplot=TRUE)

svydiags documentation built on April 28, 2022, 1:07 a.m.