Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: eiv
See also: reca

Quantlet: simex
Description: SIMEX (SIMulation EXtrapolation) is a simulation-based method of estimating and reducing bias due to measurement error. simex is applicable to general estimation methods, for example, least-squares, maximum likelihood, quasi-likelihood, etc.

Reference(s):

Link:
Usage: {simexl,simexq} = simex(y, w, z, su2, lam, b)
Input:
y n x 1 matrix, the design variables
w n x 1 matrix, surrogate variable of unobservable variable
z n x 1 matrix, covariate measured without error
su2 the variance of the measurement error
lam pseudo-parameter for generating pseudo-errors
b the number of replication in each simulation
Output:
simexl the estimate based on linear extrapolant function
simexq the estimate based on quadratic extrapolant function

Example:

library("xplore")

library("eiv")

n=100

randomize(n)

y=floor(uniform(n)+0.3)

w=uniform(n)^2

z=floor(uniform(n)+0.45)

su2=var(w)/4

lam=aseq(0.01,6,0.5)

b=20

gest=simex(y,w,z,su2,lam,b)

gest.simexl

gest.simexq

Result:

Contents of simexl

[1,]  -2.2569  0.92894  0.90323

Contents of simexq

[1,]  -2.265   0.95407  0.90822


Library: eiv
See also: reca

Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Author: Hua Liang, 970818
(C) MD*TECH Method and Data Technologies, 21.9.2000