Table 2.

Growth model for life satisfaction over time-to-death including individual-level and county-level factors

Variable	Estimate	SE
Fixed effects
Intercept, π 000	44.51*	1.47
Linear slope, π 100	−1.44*	0.18
Quadratic slope, π 200	−.06*	0.003
Individual level: intercept
Age at death, γ 01	0.03*	0.01
Gender, γ 02	1.50	2.32
Education, γ 03	0.47*	0.06
Disability, γ 04	−5.52*	0.32
Individual level: linear slope
Age at death, γ 11	−0.01*	0.00
Gender, γ 12	0.98*	0.31
Disability, γ 13	−0.25*	0.04
County level: intercept
Affluence, π 001	0.01	0.01
Proportion of individuals aged 75+, π 002	0.12	0.20
Care ratio inpatient, π 003	5.76*	2.53
Care ratio outpatient, π 004	1.27	3.00
Number of inpatient institutions, π 005	1702.71*	863.93
Number of outpatient institutions, π 006	−314.89	281.17
Main task “physical care” inpatient, π 007	−0.03	0.05
Main task “physical care” outpatient, π 008	0.08	0.07
Main task “administration” inpatient, π 009	0.16*	0.06
Main task “administration” outpatient, π 0010	−0.13	0.08
Number private beds, π 0011	−3.03	1.77
Number non-private beds, π 0012	−27.07	17.57
Flexibility in bed provision, π 0013	−0.01	0.08
Capacity, π 0014	0.01	0.03
County level: linear slope
Proportion of individuals aged 75+, π 101	0.04	0.02
Main task “physical care” outpatient, π 104	−0.00	0.01
Cross-level interactions: level
Gender × proportion of individuals aged 75+, π 021	0.00	0.31
Gender × main task “physical care” outpatient, π 022	0.06	0.08
Cross-level interactions: slope
Gender × proportion of individuals aged 75+, π 111	−0.12*	0.04
Gender × main task “physical care” outpatient, π 112	0.02*	0.01
Random effects: individual level
Variance intercept, ${\mathit{\sigma }}_{\text{u0ic}}^2$	85.18*	2.45
Variance linear slope, ${\mathit{\sigma }}_{\text{u1ic}}^2$	0.66*	0.04
Covariance intercept, linear slope	4.59*	0.25
Random effects: county level
Variance intercept, ${\mathit{\sigma }}_{\mathit{\nu }00\text{c}}^2$	3.83*	1.01
Variance linear slope, ${\mathit{\sigma }}_{\mathit{\nu }10\text{c}}^2$	0.04*	0.02
Covariance intercept, linear slope	0.07	0.09
Residual variance	63.20*	0.58
− 2LL	230,545
AIC	230,618

Unstandardized estimates and standard errors are presented. Scores are standardized to a T metric (M = 50; SD = 10) based on 2002 SOEP sample (M = 6.90, SD = 1.81 on 0–10 scale). AIC Akaike’s information criterion; −2LL = −2 log-likelihood. Intercept is centered at 2 years prior to death. Slopes of change are scaled in T units per year

* p < 0.05