trvrb
commented
6 years ago I believe that replacing current figure 3 (mers_epi) with a 3 panel version that accumulates "successful" simulations across bias levels would significantly improve narrative flow. This is proper model averaging and can be thought of having a uniform prior on bias of 1, 2 or 3. Our prior is 1/3, 1/3, 1/3; our posterior upweights bias of 2 and 3.
In this case, narrative becomes all about model and results. After this, you can separate out different bias levels in an extended data figure and discuss what this means (we find evidence of sequencing bias). Otherwise it's just too much information all at once.
p6-7, I found the description of the simulation study and simulation parameters difficult to follow. For example, greater clarity would be appreciated for terms like 'bias level' in Figure 3. These terms are clarified in the supporting methods, but I found the main text incomprehensible without seeing that section first. I would make an effort to transfer more detailed methods related to the branching process model here.