Table 1.

Parameter Estimates (Standard Error, SE) for the Full Bivariate Latent Change Score Models^a Waves 1–6 in ELSA

	Men (N = 5,110)	Women (N = 6,123)
Latent change score model	Est. (SE)	Est. (SE)
Level-to-change model pathsb
Level social isolation → Change social isolation	−0.09 (0.11)	−0.07 (0.06)
Level memory -> Change memory	0.02 (0.10)	0.15 (0.11)
Level social isolation → Change memory	−0.33 (0.15)*	−0.08 (0.16)
Level memory → Change social isolation	−0.14 (0.09)	−0.11 (0.06)
Change-to-change model pathsc
Level social isolation → Change social isolation	−0.17 (0.16)	−0.37 (0.12)**
Level memory → Change memory	−0.50 (0.49)	0.03 (0.19)
Level social isolation → Change memory	0.07 (0.33)	0.44 (0.25)
Level memory → Change social isolation	−0.16 (0.18)	0.00 (0.08)
Change social isolation → Change social isolation	−0.03 (0.36)	0.97 (0.38)*
Change memory → Change memory	1.55 (1.14)	−0.17 (0.49)
Change social isolation → Change memory	−0.78 (1.13)	−2.61 (0.75)**
Change memory → Change social isolation	0.12 (0.48)	−0.23 (0.29)

Note. −2LL = −2 log likelihood, df = degrees of freedom, AIC = Akaike’s information criterion, BIC = Bayesian information criterion.

^aAdjusted for age, education, net wealth quintile, tenure status, smoking, and physical activity from Wave 1 and time-varying variables of limiting long-term illness, depression, and working/doing voluntary work from Waves 1 to 6.

^bThe full level-to-change model fit: −2LL = 140,599.01, df = 469, AIC = 282,136, BIC = 285,202 for men; −2LL = 176,112.58, df = 469, AIC = 353,163, BIC = 356,315 for women.

^cThe full change-to-change model fit: −2LL = 140,596.49, df = 473, AIC = 282,139, BIC = 285,231 for men; −2LL = 176,095.99, df = 473, AIC = 353,138, BIC = 356,316 for women.

*p < .05. **p < .01. ***p < .001.