TABLE 3.

Change in session-by-session relationship happiness for veterans and partners in the full sample and treatment completer subsample

Variable	Full sample			Treatment completers
	B	SE	p	B	SE	p
Fixed effects
InterceptVeteran	3.34	0.10	< .001	3.38	0.13	< .001
InterceptPartner	3.20	0.10	< .001	3.26	0.14	< .001
TimeVeteran	0.02	0.01	.007	0.03	0.01	< .001
TimePartner	0.02	0.01	.005	0.03	0.01	.002
Random effects
Var (InterceptVeteran)	0.95	0.14	< .001	0.94	0.19	< .001
Var (InterceptPartner)	1.07	0.16	< .001	1.05	0.22	< .001
Var (TimeVeteran)	0.00	0.00	< .001	0.00	0.00	< .001
Var (TimePartner)	0.00	0.00	< .001	0.00	0.00	< .001
Cov (InterceptVeteran, InterceptPartner)	0.81	0.14	< .001	0.84	0.19	< .001
Cov (InterceptVeteran, TimeVeteran)	−0.01	0.01	.155	−0.02	0.01	.115
Cov (InterceptVeteran, TimePartner)	−0.03	0.01	.002	−0.03	0.01	.006
Cov (InterceptPartner, TimeVeteran)	−0.01	0.01	.508	−0.01	0.01	.213
Cov (InterceptPartner, TimePartner)	−0.03	0.01	.004	−0.03	0.01	.009
Cov (TimeVeteran, TimePartner)	0.00	0.00	< .001	0.00	0.00	< .001
Residuals
Var (Veteran)	0.29	0.02	< .001	0.27	0.02	< .001
Var (Partner)	0.31	0.02	< .001	0.30	0.02	< .001
Cov (Veteran, Partner)	0.09	0.01	< .001	0.09	0.01	< .001
ρa	0.19	0.03	< .001	0.23	0.03	< .001

Note. Var = variance; Cov = covariance.

^aThe correlation between adjacent elements in the first-order autoregressive covariance structure (AR(1)) specified in the model.