Figure 1. Boxplots of the results of the analysis of one of the simulation sets in which the fixed effect slope (a₁) is 0.02, the error variance (ɛ) is 16 and the random effects covariance matrix takes the form: Var(b₀) = 32, Var(b₁) = 0.002 and Cov(b₀, b₁) = 0.06.

(A) λ₁ estimates. (B) s.e. (λ₁ estimates). (C) τ estimates. (D) s.e. (τ estimates). λ₁, Parameter for the effect of the time-dependent predictor on survival (true value = −0.03); τ, Parameter for the group effect on survival (true value = 0); Cox.b, Cox regression using only the baseline values of the time-dependent predictor; Cox.tp, Cox regression using all of the longitudinal values of the time-dependent predictor; JM.pc, JM package with a piecewise-constant baseline hazard; JM.sp, JM package with the baseline hazard specified by regression splines; JM.Cox, JM package with an unspecified baseline hazard.