comment on the paragraph on the simulated dataframe

[DelphineDeMoor]

'This dataframe is an example of the format that bisonR uses' -> here I'd refer to the term edgelist, which is what most people who use SNA would call this, and I'd specify that the necessary columns are 1. event (which is 1/0 OR the weight) 2. ID1 3. ID2.

[DelphineDeMoor]

'See the explanation of the bisonR framework here for more details on this.' -> what does the here refer to?

[DelphineDeMoor]

in the little R block for manually updating the prior for the fixed effect it should be 2 instead of 2.5

[DelphineDeMoor]

Might be worth adding a line or 2 at the beginning of the edge model paragraph to explain that the edge model is estimating the edges, and not an inferential model using the edges

[DelphineDeMoor]

'In bisonR, the left hand side of the ~ describes the sampling data, and uses the (event | duration) notation. In this notation, event corresponds to the name of the column in the dataframe that represents the measure of social events, such as 1 and 0 in the binary model, or frequencies 0, 1, 2, … in the count model. If using an unaggregated dataframe, the duration corresponds to the durations of each observation. In our case this is fixed, but this will depend on the data at hand.'

-> this paragraph is a bit confusing, I would stick to your example of the binary model so not talk about the frequencies in the count model here. Also, 'the duration of each observation' is confusing, because it could be interpreted as 'how long did they groom?', maybe call it instead the duration of observation effort, i.e. how long the dyad was observed for in that event, from which then rates can be calculated?

[DelphineDeMoor]

In the standard BISoN model this is primarily the edge weight that we use to build the network, but this can also include additional effects such as age, sex, or even observation-level factors such as location, time of day, or weather, to name a few. -> I think this might be confusing to some people, that' you're estimating the edge weights (kind of what is before the ~), based on the edge weights.. Might be worth a bit more explicit explanation of what this model is really doing, how it is estimated the edges by calculating rates with uncertainty.

[DelphineDeMoor]

To see a summary of edge weights and their credible intervals, we can use the summary() function:To see a summary of edge weights and their credible intervals, we can use the summary() function

-> might be worth adding a line here saying what the summary is showing you (for each dyad a directed edge weight distribution, represented by the median, the 5% CI and the 95% CI)

[DelphineDeMoor]

The predictions from the model are shown in blue and the real data are shown in black. -> Can you change the real data to red or so and have it be plotted after the blue lines, just so it's clearer?

[DelphineDeMoor]

Uncertainty is visualised by showing the lower and upper bounds as overlapping edges in the network. -> would be worth explaining more here, saying that the the inner grey lines are the 5% CI for the edge (an estimation of how small the edge could be), and the dark lines are the 95% CI for the edge (an estimation of how large the edge could be)