Ideas and plans

[mixmixmix]

What do you think?

1. Adding some correlated noise?
2. More complex models? What?
3. More datasets?

[fergusjchadwick]

I think these are great ideas. A couple of easy things to add to the existing version could be:
random effects (e.g. site level effects)
cross-correlation between predictors
differential effects based on observation method (e.g. does hunting remove more large dragons and thus have a different effect on number and mass)
Overdispersion of count data

I think once we have a core flow through the modelling bits it would be reasonably straightforward to move to other datasets by adding an additional page at the starts of the joined apps and then having that select the text and data used in subsequent pages...