Skip to main content

Table 2 Estimation results for the conventional multilevel model and the spatially filtered multilevel model

From: A spatially filtered multilevel model to account for spatial dependency: application to self-rated health status in South Korea

Variables

Null model

Level-1 multilevel model

Level-2 multilevel model

Spatially filtered multilevel model

Individual-level variables

 

Sex (male:0; female:1)

-

– 49.88***

– 49.65***

– 49.69***

Monthly household income

–

0.10***

0.10***

0.09***

Employment status (employed:0; unemployed:1)

–

–134.10***

–134.90***

–135.30***

Perceived levels of stress (high:0; low:1)

–

154.60***

155.60***

155.70***

Neighborhood–level variables

    

Korean Deprivation Index (KDI)

–

–

–23.82*

–15.51*

The number of doctors per 1000 people

–

–

4.85*

2.60*

Degree of the Local Governments’ Financial Independence (LGFI)

–

–

0.98

0.16

Random effects

    

Variance at individual–level

66725

52226

52225

56013

Between monthly household income variance

–

0.0039

0.0036

0.0011

Variance at neighborhood–level

1591

1102

1062

555

Constant

785.31***

761.40***

747.00***

770.70***

Eigenvector selection

–

–

–

8 eigenvectors

Moran’s I of neighborhood–level residuals

–

–

0.101*

0.005

AIC

861942

830665

830650

830549

Log–likelihood

– 447328

– 415324

– 415314

– 415254

  1. ***p < =0.001 *p < =0.05.