For stratified performance, such as how well behavioural data explains transmission across different pandemic periods and age groups, there are limitations with fitting a full population-level model to the entire 2020-2022 period and post-hoc stratifying
Determine how to meaningfully identify failure nodes and interpret why performance drops or improves
May need to use hierarchical model and pandemic period specific terms to capture these heterogeneities within the model fitting process, rather than post-hoc stratification.
Consider data availability as different subsets will have different counts (sparse data is an issue)
Consider interactions, such as how we have assortative mixing which predicts transmission (as per Munday et al), and therefore may not be appropriate to model every age group independently
For stratified performance, such as how well behavioural data explains transmission across different pandemic periods and age groups, there are limitations with fitting a full population-level model to the entire 2020-2022 period and post-hoc stratifying
Determine how to meaningfully identify failure nodes and interpret why performance drops or improves
May need to use hierarchical model and pandemic period specific terms to capture these heterogeneities within the model fitting process, rather than post-hoc stratification.
Consider data availability as different subsets will have different counts (sparse data is an issue)
Consider interactions, such as how we have assortative mixing which predicts transmission (as per Munday et al), and therefore may not be appropriate to model every age group independently