In the case of left-censored data, several statistical analysis and visualization methods depend on how do we treat those observations:
Threshold levels may not be computed for left-censored data, since when doing so we are assuming that the exposure levels by the time they adopted were exactly those that we observed in the first time we observe them.
Visualizing the diffusion process. Currently, we are considering individuals with toa = 1 as adopters in that time point, so individuals that actually adopted earlier should be represented with other colors in plot_diffnet and plot_diffnet2.
In regression analysis, we cannot use observations of individuals that adopted in time 1 if the data is censored.
Currently, netdiffuseR has no way of specifying this. To build diffnet objects the user must provide a time of adoption, and what traditionally is done is just set left-censored as adopters in the firts wave of data.
The problem
In the case of left-censored data, several statistical analysis and visualization methods depend on how do we treat those observations:
Threshold levels may not be computed for left-censored data, since when doing so we are assuming that the exposure levels by the time they adopted were exactly those that we observed in the first time we observe them.
Visualizing the diffusion process. Currently, we are considering individuals with toa = 1 as adopters in that time point, so individuals that actually adopted earlier should be represented with other colors in
plot_diffnetandplot_diffnet2.In regression analysis, we cannot use observations of individuals that adopted in time 1 if the data is censored.
Currently, netdiffuseR has no way of specifying this. To build diffnet objects the user must provide a time of adoption, and what traditionally is done is just set left-censored as adopters in the firts wave of data.