Design data for p under RDHet

[aber_r16]

HI,

I'm confused as to the structure of the design data for p created under the RDHet model. It does not include variables for time, but only for mixture and session. The results don't show well in the post, but both the time and Time variables are all zeros. I would like to include models that explore session and time additive and interactive effects with the 2 mixtures (as I have done in the robust model without heterogeneity, p=c).

Code: Select all

par.index model.index group time mixture session Time
1          1          47     1    0       1    2009    0
2          2          48     1    0       2    2009    0
3          3          49     1    0       1    2010    0
4          4          50     1    0       2    2010    0
5          5          51     1    0       1    2011    0
6          6          52     1    0       2    2011    0
7          7          53     1    0       1    2012    0
8          8          54     1    0       2    2012    0
9          9          55     1    0       1    2013    0
10        10          56     1    0       2    2013    0
11        11          57     1    0       1    2014    0
12        12          58     1    0       2    2014    0


What am I missing?

Thanks, R

[egc]

The results don't show well in the post...

They do, if you add the appropriate tags -- see viewtopic.php?f=21&t=1851

I went ahead and edited your post by adding the appropriate BBCode code tag.

[jlaake]

RDHet doesn't have a separate p for each time. Notice in your design data that there are just 2 parameters per session. It is only a mixture model. You want RDFullHet if that is what you want to do.

--jeff

[aber_r16]

Right thanks!

Just to make sure I get it...

If I don't expect a behaviour effect within the capture probabilities then I set p=c (share=TRUE) and model p based on the various conditions (constant, ~time, ~session, ~time+session, ~time*session) with RDFullHet. This is what I did with the Robust model, I see now that it would be the same with the RDFullHet. RDHet is just a reduced parameter version of RD thus taking away the flexibility of modelling effects on p, only letting you model the mixture effects. I misunderstood - I initially interpreted the lack of the c parameter for RDHet as saying no behaviour effect, but you could still model p over time.

Thanks Jeff and egc!