. Author manuscript; available in PMC: 2012 Sep 1.

Published in final edited form as: Biometrics. 2010 Oct 29;67(3):876–885. doi: 10.1111/j.1541-0420.2010.01500.x

Table 4.

Analysis results for CPP data

Partial linear model

Linear Model

Methods

AMSE

ABIAS

AVAR

MEAN

\hat{SE}

{\hat{β}}_{P}

SE ({\hat{β}}_{P})

95% CI

{\hat{β}}_{V}

SE ({\hat{β}}_{V})

95% CI

{\hat{β}}_{L}

SE ({\hat{β}}_{L})

95% CI

PCB

See Figure 2

EDU1

2.79

1.04

(0.75, 4.83)

2.73

1.24

(0.30, 5.16)

2.72

1.04

(0.68, 4.76)

EDU2

10.44

1.66

(7.19, 13.69)

9.39

2.02

(5.43, 13.35)

10.47

1.66

(7.22, 13.72)

SES

1.39

0.20

(1.00, 1.78)

1.31

0.24

(0.84, 1.78)

1.40

0.20

(1.01, 1.79)

RACE

−7.97

0.75

(−9.44, −6.50)

−7.73

0.89

(−9.47, −5.99)

−7.82

0.75

(−9.29, −6.35)

SEX

−0.81

0.69

(−2.16, 0.54)

−0.75

0.84

(−2.40, 0.90)

−0.79

0.69

(−2.14, 0.56)

Note 4: The result of linear model is obtained using the Zhou et al. (2002) method, and in the case of linear model, g(PCB) = α₁ + α₂PCB; ${\hat{β}}_{P}$ and ${\hat{β}}_{V}$ correspond to the estimates obtained by the P and V methods respectively; $SE ({\hat{β}}_{P})$ and $SE ({\hat{β}}_{V})$ are the estimated standard errors of corresponding estimators.