Skip to main content

Table 4 Covariates balancing test of PSM: Mean differences before and after matching

From: The impact of medical insurance on family financial support: evidence from CHARLS in China

Variable

Mean

% reduction

 

Treated

Control

% bias

|bias|

Age

U

60.664

61.509

-8.4

46.0

M

60.664

61.12

-4.5

Male

U

0.480

0.428

10.5

89.6

M

0.480

0.475

1.1

Urban resident

U

0.203

0.312

-25

95.5

M

0.203

0.208

-1.1

Marriage status

U

0.883

0.806

21.3

82.5

M

0.883

0.869

3.7

No formal education

U

0.245

0.302

-12.9

95.1

M

0.244

0.242

0.6

Can read and write

U

0.197

0.221

-6

96.9

M

0.197

0.197

0.2

Primary school

U

0.226

0.200

6.5

72

M

0.226

0.219

1.8

Secondary school or above

U

0.332

0.277

12

80

M

0.332

0.343

-2.4

Public pensions

U

0.602

0.294

65.1

98.5

M

0.602

0.606

-1

Self-rated health status

U

3.042

3.083

-4.2

46.8

M

3.042

3.020

2.3

Number of children

U

3.029

3.162

-7.7

74

M

3.029

3.063

-2

Living arrangement

U

0.280

0.349

-15

80.8

M

0.279

0.293

-2.9

Family financial assets

U

0.047

0.031

6.6

89.2

M

0.047

0.045

0.7

  1. % bias denotes mean standardized difference in percentage. “U” represent “the unmatched subsample”, and “M” represent “the matched subsample”. The balancing test results here are from nearest neighbor matching (k = 4) of the whole sample. The matching test results of radius matching, kernel matching and other sub-groups are also effective, which are available upon request