Changed health behavior improves subjective well-being and vice versa in a follow-up of 9 years

Table 5 Linear regression models in which baseline subjective well-being (SWBscore₂₀₀₃) or its change (SWBchange) predicts follow-up health behavior (HBSS₂₀₁₂)

Model	SWBscore₂₀₀₃ β p-value (standard error)	SWBchange p-value	HBSS₂₀₀₃ β p-value	Age p-value	Gender p-value	Education p-value	Diseases p-value	SWB₂₀₀₃* Education p-value	Negative life events p-value	AIC
Model 2.0: Linear model, no covariates	− 0.038	–	–	–	–	–	–	–	–	28,000
	< .001
	(0.0027)
Model 2.1: Model 2.0 + Age, gender, education, diseases, SWB₂₀₀₃*education	− 0.040	–	–	< 0.001	< 0.001	0.041	0.002	0.003	–	27,300
	< .001									27,300
	(0.0048)
Model 2.2: Model 2.1 + SWBchange	− 0.049	< .001	–	< 0.001	< 0.001	0.024	0.007	0.009	–	27,200
	< .001
	(0.0049)
Model 2.3: Model 2.1 + HBSS₂₀₀₃	− 0.019	–	0.49	0.040	< 0.001	0.55	0.020	0.003	–	23,100
	< .001		p < .001
	(0.0031)		(0.0088)
Final model 2.4: Model 2.1 + SWBchange + HBSS₂₀₀₃	− 0.025	< .001	0.48	0.093	< 0.001	0.49	0.040	0.005	–	23,000
	< .001		p < .001
	(0.0045)		(0.0088)
Model 2.5: Model 2.4 + negative life events	− 0.026	< .001	0.48	0.079	< 0.001	0.50	0.036	0.005	0.52	23,000
	< .001		p < .001
	(0.0045)		(0.0088)
Model 2.6: SWBchange + Age, gender, education, diseases, SWBchange*education	–	.070	–	< 0.001	< 0.001	< 0.001	< 0.001	0.72^b	–	27,400
Model 2.7: Model 2.6 + HBSS₂₀₀₃	–	< .001	0.49	0.022	< 0.001	< 0.001	0.001	< 0.55^b	–	23,000
			p < .001
			(0.0088)

HBSS, Health behavior sum score i.e. no. of protective health behaviors; the SWBscore, Subjective well-being score (lower scores indicating better SWB); SWBchange, Change in health behavior sum score during follow-up; SWBchange*education, Interaction term in the model 2.6 used because SWB₂₀₀₃ was excluded

ISSN: 1477-7525