How to model time-dependent covariate in random effect model?

[nahh]

I am using a random effect model to examine the effect of several covariates on the outcome variable over time. Some of the covariates are time-dependent. I think the regular random effect model is not sufficient in this case. Is there any special technique to address the time-dependent covariates? Are there any book or articles on this topic? Many thanks!

[danielkeeton]

Are you suggesting that by having time dependent covariates you have a bimodal distribution (or tri, etc.)? I'm curious as to why you would need to consider time when couldn't you take the mean of (T1, T2, etc.) and use that as your covariate...Or covary out those variables at each change then collapse. Maybe i'm unsure of the exact nature of the dilemma.

[TheAnalysisFactor]

Lesa Hoffman has lecture audios and slides for her courses. Week 11 in her Longitudinal Course is about time varying covariates: http://psych.unl.edu/psycrs/944/index.html

Karen

[nahh]

Are you suggesting that by having time dependent covariates you have a bimodal distribution (or tri, etc.)? I'm curious as to why you would need to consider time when couldn't you take the mean of (T1, T2, etc.) and use that as your covariate...Or covary out those variables at each change then collapse. Maybe i'm unsure of the exact nature of the dilemma.

Sorry I should've provided more information. The dependent variable was measured at 3 different time points, so were the time-dependent covariates. I would like to estimate the cross-sectional effect of covariates on the dependent variable at each time point, as well as longitudinal within-subject effect of covariates on the dependent variable.

If I simply take the mean of the time-dependent covariates over the 3 time points, I would lose too much information and not be able to estimate their within-subject effect.

I think Karen's link is very helpful for me. Thank you both for answering my questions!