06-14-2018 04:13 AM - edited 06-14-2018 10:53 AM
I am trying to pre-specify / get a better understanding of how to perform a multivariate (3 or 4 outcomes) longitudinal repeated measures analysis where outcomes are measured daily over many months.
If I had just a few time points then my research has guided me towards a random coefficient models in PROC MIXED such as that specified in “Analyzing Multivariate Longitudinal Data Using SAS® “.
But I am struggling to find details of models to account for "continuous" time. I have seen mention of modelling time as a continuous variable, or the use of a splined covariance structure. Would someone mind pointing me in the direction of some appropriate literature in relation to these, or other suggested methodologies please.
06-14-2018 12:40 PM
time is continuous in the random coefficients model that you mention. You might want to look at brown and prescott's book: https://onlinelibrary.wiley.com/doi/book/10.1002/9781118778210
i'm pretty sure they cover SAS (ie give code examples) and discuss random coefficient models and building a polynomial model for nonlinear relationship with time. Although, im not sure if they discuss the multivariate case
06-14-2018 01:08 PM - edited 06-14-2018 01:14 PM
Great thank you so much for your reply and guidance ... I will be looking at the book right away .