Table 2.

Comparison of multilevel models for the effect of current age on overall affective tone of life narratives

	Model 0		Model 1		Model 2		Model 3		Model 4		Model 5
	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI
Fixed Part
Intercept	22.22* (1.48)	[19.28, 24.83]	18.85* (2.04)	[14.50, 23.16]	25.11* (2.59)	[20.09, 30.46]	24.95* (2.61)	[20.22, 30.07]	19.65* (3.14)	[13.26, 25.74]	20.38* (3.56)	[13.45, 27.61]
Male			6.89* (2.92)	[0.57, 12.37]	6.96* (2.82)	[0.89, 12.50]	7.00* (2.82)	[1.23, 12.49]	7.17* (2.82)	[1.61, 12.87]	7.17* (2.83)	[1.84, 13.09]
Agea					−2.40* (0.65)	[−3.55, −1.19]	−2.55* (0.70)	[−3.76, −1.16]	3.18 (1.94)	[−0.51, 7.04]	1.71 (3.89)	[−5.73, 9.42]
Ageb							0.08 (0.15)	[−0.21, 0.33]
Age2									−0.85* (0.28)	[−1.41, −0.31]	−0.25 (1.40)	[−2.84, 2.50]
Age3											−0.06 (0,14)	[−0.32, 0.19]
Random Variance
Residual ${\mathrm{\sigma }}_e^2$	384.84 (19.62)	[18.34, 20.77]	384.51 (19.61)	[18.52, 20.87]	383.01 (19.57)	[18.31, 20.83]	382.88 (19.57)	[18.44, 20.70]	376.27 (19.40)	[18.24, 20.49]	375.86 (19.39)	[18.23, 20.60]
Intercept ${\tau }_1^2$	273.55 (16.54)	[14.24, 19.04]	262.72 (16.21)	[13.83, 18.36]	239.50 (15.48)	[12.87, 17.90]	239.41 (15.47)	[12.92, 17.50]	241.47 (15.54)	[12.94, 17.62]	242.35 (15.57)	[12.71, 17.69]
Model Fit
Deviance	6364.42		6358.93		6345.41		6345.15		6336.15		6335.96
Δχ2			5.49*		13.52*		0.26		9.27*		0.19
df			1		1		1		1		1

CI = confidence intervals (based on bootstrapping). N = 698 life narratives. Sequentially adding predictors was tested with Δχ² tests based on model deviance (−2Log-Likelihood), comparing it to the previous model (except Model 4 was compared with Model 2).

^aAge was divided by 10 and centred at 0.8 decades (according to the youngest age group) to aid the estimation of quadratic and cubic effects.

^bAge was within-person baseline-centred at the individually youngest age at the first measurement.

*p < .05, one-tailed