Package "memisc": Sandwich estimators and 'mtable'

Posted on Fri, 13 May 2011 21:46:19 +0100

Recently I was asked whether it was possible to use with 'mtable' sandwich estimators of variance for lm and glm models, as they are provided by the R packages 'sandwich' and 'lmtest'.

An R script that provides this extension can now be downloaded from here.

One trick to make use of this extension is to add a class attribute "lm_sandwich" or "glm_sandwich", respectively, to your lm or glm object like in this example:

library(memisc)
library(sandwich)
library(lmtest)
source("../extras/getSummary-glm-sandwich.R")
data(Mandible)
fm1 <- lm(length ~ age, data=Mandible, subset=(age <= 28))
fm1.sw <- fm1
class(fm1.sw) <- c("lm_sandwich",class(fm1))
mtable(
  "Model-based"=fm1,
  "Sandwiched-based"=fm1.sw,
   summary.stats=c("R-squared","N"))

The output of this example looks like this:

==============================================
               Model-based    Sandwiched-based
----------------------------------------------
(Intercept)     -11.953***       -11.953***   
                 (0.976)          (1.010)     
age               1.773***         1.773***   
                 (0.048)          (0.054)      
----------------------------------------------
R-squared          0.898            0.898     
N                158              158         
==============================================

Alternatively, one can use an appropriate 'getSummary' argument to 'mtable' like in this example:

library(memisc)
library(sandwich)
library(lmtest)
source("../extras/getSummary-glm-sandwich.R")
data(Mandible)
fm1 <- lm(length ~ age, data=Mandible, subset=(age <= 28))
fm2 <- lm(length ~ age+I(scale(age)^2), data=Mandible, subset=(age <= 28))
# Using 'conventional' i.e. model-based standard errors:
mtable(fm1,fm2,summary.stats=c("R-squared","N"))

======================================
                    fm1        fm2
--------------------------------------
(Intercept)      -11.953*** -11.303***
                  (0.976)    (1.060)
age                1.773***   1.754***
                  (0.048)    (0.049)
I(scale(age)^2)              -0.362
                            (0.236)
--------------------------------------
R-squared           0.898      0.900
N                 158        158
======================================

# Using sandwich-based standard errors:
mtable(fm1,fm2,getSummary=getSummary.lm_sandwich,
      summary.stats=c("R-squared","N"))

======================================
                    fm1        fm2
--------------------------------------
(Intercept)      -11.953*** -11.303***
                  (1.010)    (1.348)
age                1.773***   1.754***
                  (0.054)    (0.063)
I(scale(age)^2)              -0.362
                            (0.258)
--------------------------------------
R-squared           0.898      0.900
N                 158        158
======================================

Tags: memisc

Package "memisc": Sandwich estimators and 'mtable'

Research

Software