search for: varying_

...sured at event times. Chances are that the only time such a covariate is recorded at high level, the individual for which it is measured as being high is having an event. This may bias the estimated coefficient. Here is my question: How to fully use the dataset? (that is: how to have really _time-varying_ covariates (even if they change step by step, not continuously), not covariates whose changes are measured only at event time ) Ideally, the full dataset would be use to estimate the parameters, or at least to estimate the standard error of the estimated parameters. Any ideas ??? . . . A second b...