Hi all,
I am doing an interrupted time series with Poisson regression in a similar way to the paper "Interrupted time series regression for the evaluation of public health interventions: a tutorial; IJE 2016; J. Lopez Bernal, S. Cummins, A. Gasparrini"
I also want to include two variables in the model that are time-varying covariates. I have found in the literature that you can just add confounding variables to the log-linear regression of the Poisson time series and have found an example elsewhere where a covariate has been added to the time series as a categorical variable, so in Stata using i.var to approximate a seasonal effect for month.
My time series is in days and the time-varying covariates are in days and change every day. Is it okay to add the time-varying covariate as a continuous variable c.var or do I need to add it to the model as a categorical variable i.var?
Any help much appreciated,
many thanks,
Anna
I am doing an interrupted time series with Poisson regression in a similar way to the paper "Interrupted time series regression for the evaluation of public health interventions: a tutorial; IJE 2016; J. Lopez Bernal, S. Cummins, A. Gasparrini"
I also want to include two variables in the model that are time-varying covariates. I have found in the literature that you can just add confounding variables to the log-linear regression of the Poisson time series and have found an example elsewhere where a covariate has been added to the time series as a categorical variable, so in Stata using i.var to approximate a seasonal effect for month.
My time series is in days and the time-varying covariates are in days and change every day. Is it okay to add the time-varying covariate as a continuous variable c.var or do I need to add it to the model as a categorical variable i.var?
Any help much appreciated,
many thanks,
Anna
Comment