Neighborhood differences in social capital in Ghent (Belgium): a multilevel approach

Table 3 Fixed and random parameters of the social cohesion multilevel models

	Model 0	Model 1a	Model 1b	Model 2a	Model 2b
	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)
Fixed effects
Constant	14.241 (0.215)***	14.903 (0.246)***	14.920 (0.199)***	15.038 (0.219)***	15.121 (0.170)***
Level 1
Female		−0.031 (0.199)		−0.034 (0.198)
65 years or older		0.086 (0.257)		0.050 (0.256)
Low education level		−0.001 (0.288)		0.058 (0.283)
Single		−0.572 (0.231)*	−0.527 (0.228)*	−0.493 (0.230)*	−0.491 (0.226)*
Not Belgian		0.076 (0.329)		0.276 (0.329)
Weak neighborhood attachment		−1.941 (0.236)***	−1.962 (0.233)***	−1.891 (0.233)***	−1.901 (0.230)***
<5 years of residence		−0.009 (0.236)		0.076 (0.235)
Level 2
Deprived neighborhood				−0.960 (0.337)*	−0.977 (0.351)**
Percentage of elderly				−0.156 (0.142)
Turnover				−0.552 (0.173)***	−0.557 (0.143)***
Physical disorder				−0.188 (0.177)
Random effects
Level 1
Constant	7.852 (0.414)***	7.244 (0.384)***	7.246 (0.383)***	7.232 (0.383)***	7.253 (0.383)***
Level 2
Constant	1.510 (0.425)***	0.964 (0.300)***	0.944 (0.295)***	0.247 (0.143)	0.298 (0.154)
Intraclass correlation	0.161	0.117	0.115	0.033	0.039
Log likelihood	3795,593	3688,729	3714,429	3656,402	3685,974
∆ Log likelihood (∆ df)		106,864	81,164	139,191	109,619
P		< 0.001	< 0.001	< 0.001	< 0.001

Note: *p < 0.05, **p < 0.01, ***p < 0.001.
Model 0: null model without level 1 and level 2 variables; Model 1a: model with only level 1 variables; Model 1b: parsimonious model with only level 1 variables; Model 2a: model with level 1 and level 2 variables; Model 2b: parsimonious model with level 1 and level 2 variables.

ISSN: 1476-072X