Skip to main content

Table 6 Estimates of parameters, standard errors (s.e.) and P values in the bivariate selection model using two-step and one-step methods

From: Sample selection bias with multiple dependent selection rules: an application to survey data analysis with multilevel nonresponse

Reason

Parameter

Two-step

One-step

Estimate

s.e.

P value

Estimate

s.e.

P value

Noncontact

Intercept

1.605

0.127

0.000

1.604

0.127

0.000

 

Org. 1

− 0.971

0.297

0.000

− 0.970

0.297

0.001

 

Org. 7

− 1.019

0.199

0.000

− 1.012

0.199

0.000

 

Org. 11

− 1.892

0.504

0.000

− 1.943

0.509

0.000

 

Previous status in survey

− 0.378

0.186

0.042

− 0.377

0.185

0.041

 

Not having ancillary unit

0.379

0.176

0.031

0.378

0.179

0.035

 

Position in industrial zone

0.356

0.140

0.011

0.358

0.141

0.011

Refusal

Intercept

1.155

0.082

0.000

1.184

0.080

0.000

 

Org. 3

0.343

0.207

0.096

0.471

0.208

0.023

 

Org. 10

0.285

0.162

0.079

0.153

0.166

0.358

 

Org. 12

0.694

0.289

0.016

0.675

0.284

0.018

 

Org. 20

− 1.083

0.186

0.000

− 1.076

0.181

0.000

 

Org. 23

− 0.615

0.112

0.000

− 0.681

0.110

0.000

 

Org. 25

− 1.072

0.201

0.000

− 1.059

0.195

0.000

 

Previous status in survey

− 1.076

0.152

0.000

− 1.024

0.162

0.000

 

Previous status in sample

− 0.595

0.093

0.000

− 0.567

0.093

0.000

 

Not having ancillary unit

0.240

0.099

0.015

0.146

0.102

0.151

Correlation between noncontact and refusal

− 0.440

0.535

0.411

− 0.657

0.785

0.402

  1. The sample size is 1236 establishments with 865 respondents, 55 noncontacts and 316 refusals