Differences from Distance for Windows vignette

[dill]

Create new documentation object which describes differences between mrds/Distance and Distance for Windows.

Inaugural topics:

• ER variance differences #40 and #52
• calculation of p DistanceDevelopment/Distance#135

[dill]

Following from #63, setting varflag=1 leads to Buckland et al (2001) estimation of variance. BUT in mrds (here), we add in group size variance. This doesn't seem right if we had size as a covariate in the detection function. Need to tell people that this is the case.

[erex]

@dill help me understand your comment above; if varflag=1 is used when size is in the detection function, variance estimate will be incorrect; is that right?

[erex]

the er_method argument to ds() is equivalent to the varflag argument to mrds() with the same meanings for values 1 and 2, correct?

[dill]

I think this was based on a conversation between me and @lenthomas, but I can't remember how we came to this conclusion. Looking through both Marques and Buckland (2003) and Innes et al (2002) doesn't make this any clearer for me.

Looking at this now, I think my comment above is wrong? If we have group size in the detection function that will tell us about model-based uncertainty in the parameter related to group size in the detection function, but it will not give us any uncertainty due to sampling variance in the group sizes we saw, so I'd think we need both.

@lenthomas do you remember this discussion?

[dill]

(Note that the differences vignette is here.)