Table 4.

Multilevel Analysis of Change Over Time–Dependent Variable (DV)=Log of Total Symptom

	Unconditional growth model		Growth model with GSE 2 groups		Growth model with all level 2 predictors and covariates
	Model A		Model B		Model C		Interpretation
	Estimate	S E	Estimate	S E	Estimate	S E	Model C
Intercept	−0.079	0.009	−0.090a	0.011	1.206a	0.039	There is significant variation about the intercept in symptom intensity
Time	1.377	0.020	1.399a	0.027	−0.067a	0.012	There is a significant decrease in symptom intensity in both groups over the 3-month period
Black					0.095b	0.042	Black individuals had a significantly higher symptom intensity rating at baseline compared with other ethnicities
Female					0.021	0.042	Females had a significantly higher symptom intensity rating at baseline compared with male and transgender individuals
[gse2grp=0.00]			0.137a	0.034	0.087b	0.042	Individuals in the low self efficacy group had a significantly higher symptom intensity at baseline compared with the high GSE group
Activities_eff					0.008b	0.004	Individuals who rated activities effectiveness higher had increased symptom intensity scores at baseline
Activities_used					0.008	0.004	Individuals who used more symptom management activities had increased symptom intensity scores at baseline
[gse2grp=1.00]a activities_useda black					−0.010b	0.004	Black individuals in the high selfefficacy group used significantly fewer activities compared with non-black individuals at baseline
[gse2grp=1.00]a timea activities_useda black					0.001	0.003	Black individuals in the high self- efficacy group remained constant over 3 months in using fewer activities compared with non-black individuals, but not significantly fewer activities
[gse2grp=0.00]a activities_used					0.010	0.006	The low self efficacy group did not use more strategies at baseline compared with the high self efficacy group
[gse2grp=0.00]a activities_eff					−0.008	0.006	The low self efficacy group did not rate strategy effectiveness differently at baseline compared with the high self efficacy group
Model fit statistics
−2 log likelihood	1996.0		1409.9		1219.2		The model fit improved with the addition of explanatory level 2 variables
Akaike information criterion (AIC)	1986.0		1397.9		1255.2
Schwarz's Bayesian criterion (BIC)	2023.2		1441.0		1347.8
Residuals analysis
Residual in model	0.003	0.013	0.090	0.005	0.088	0.005
Residual variance - intercept	0.005	0.0009	0.132	0.012	0.104	0.010	21% of the variance has been explained from model B with the addition of the additional level 2 variables
Residual variance - time	0.195	0.087	0.004	0.003	0.002	0.003	50% of the variance in time is explained by the addition of level 2 variables

^ap≤0.001, ^bp≤0.05.