Skip to main content

Table 1 Results for parametric and smooth terms in the full and reduced model when using the second-order ordinal smoothing penalty

From: Statistical inference for ordinal predictors in generalized additive models with application to Bronchopulmonary Dysplasia

 

Full model

   

Parametric terms

    

Covariate

Estimate

Std. error

z-value

p-value

(Intercept)

6.214

2.630

2.363

0.018

Weight (g)

− 0.013

0.004

− 3.381

<0.001

SGA sym.

1.909

1.359

1.405

0.160

Sex (male)

3.022

1.114

2.712

0.007

Multiples

1.524

0.744

2.048

0.041

Steroids

− 0.241

0.090

− 2.684

0.007

Antibiotics

0.079

0.090

0.874

0.382

Smooth terms

    

Predictor

edf

Ref.df

Chi.sq

p-value

Gram negative

1.000

1.000

1.307

0.253

Gram positive

3.708

4.393

5.227

0.264

Pathogenic

5.258

5.831

13.711

0.030

 

Reduced model

   

Parametric terms

    

Covariate

Estimate

Std. error

z-value

p-value

(Intercept)

6.426

2.170

2.961

0.003

Weight (g)

− 0.012

0.003

− 3.859

< 0.001

SGA sym.

1.991

1.290

1.544

0.123

Sex (male)

2.107

0.834

2.527

0.012

Multiples

1.054

0.528

1.995

0.046

Steroids

− 0.174

0.072

− 2.432

0.015

Antibiotics

0.079

0.078

1.013

0.311

Smooth terms

    

Predictor

edf

Ref.df

Chi.sq

p-value

Gram negative

–

–

–

–

Gram positive

–

–

–

–

Pathogenic

4.973

5.696

13.573

0.027