. Author manuscript; available in PMC: 2013 Dec 20.

Published in final edited form as: Stat Med. 2012 Jul 24;31(29):4074–4086. doi: 10.1002/sim.5510

Table II.

Mean squared error with varied sample sizes (r = 0.25, p = 0.30).

Parameter

Poisson regression model

ZIP model

Hurdle model

N = 100

N = 200

N = 400

N = 100

N = 200

N = 400

N = 100

N = 200

N = 400

The Poisson component

β₁

0.0720

0.0675

0.0554

0.0138

0.0042

0.0020

0.0089

0.0039

0.0019

β₂

0.3266

0.2848

0.2813

0.0172

0.0057

0.0022

0.0134

0.0044

0.0023

β₃

0.0355

0.0252

0.0200

0.0150

0.0043

0.0016

0.0090

0.0033

0.0016

β₄

0.0141

0.0118

0.0081

0.0110

0.0050

0.0018

0.0081

0.0038

0.0016

β₅

0.1577

0.1529

0.1384

0.0158

0.0082

0.0020

0.0100

0.0050

0.0019

g(t)*

197.06

147.67

61.58

24.07

22.56

19.86

23.79

21.53

20.34

σ_{a}^{2}

0.2247

0.1709

0.1780

0.7660

0.7663

0.7282

0.8870

0.7708

0.6959

The zero component

γ₁

N/A

0.9796

0.1446

0.0305

0.0486

0.0145

0.0068

γ₂

N/A

3.1837

0.3632

0.0820

0.0674

0.0245

0.0114

γ₃

N/A

0.6031

0.1361

0.0447

0.0294

0.0143

0.0067

γ₄

N/A

0.3763

0.0721

0.0295

0.0228

0.0136

0.0064

γ₅

N/A

1.5438

0.2396

0.0429

0.0512

0.0187

0.0094

h(t)*

N/A

20.08

18.42

13.98

13.83

13.12

11.14

σ_{b}^{2}

N/A

3.1340

1.7310

1.4238

1.0860

1.0362

1.0361

σ_ab

N/A

2.1606

0.3449

0.1214

0.0979

0.0534

0.0179

The average of the integrated mean squared error was calculated for the spline function.