Discrimination factor query

[MattLewisCT]

Hi Brian

I see in Stock et al. 2018 (PeerJ), in reference to discrimination factors, you wrote: "In other words, these parameters are difficult to estimate simultaneously, and one or the other is generally fixed (in food web studies, the discrimination factor is typically specified as fixed a priori). At present, MixSIAR does not provide the option to estimate discrimination from user-provided data, although such functionality could easily be added; we anticipate adding this functionality into a future software release."

My question related to this is: does the model treat all discrimination factor values within the range defined a priori (using mean and SD) as having equal probability of being "true"? (Apologies if my use of terminology here is inappropriate.)

Best, Matt

[ericward-noaa]

The prior isn’t really a fixed range - it’s a probability, so that values close to the mean are more probable than those in the tails. Increasing the SD both widens the tails and makes values close to the mean less likely relative to smaller SD

[MattLewisCT]

Right. Thanks very much for clarifying, Eric.