Setting up model for method of treatment

[Happyme]

Good afternoon all,

I am wanting to do an ANOVA on the estimates for my trapping methods. I have four trapping methods: Control, Glow, Bait, Glow+Bait. I have the data setup in excel in the same format you would your presence/ absence data. I am not sure though where this data should go? Do I run a separate model for each treatment and treat it like presence/absence data and run a constant model?

[darryl]

I'm not quite sure what you mean about your data format, but if you want to conduct and ANOVA using occupancy models it's essentially the same approach as for any other maximum likelihood based approach. Fit a range of models representing the different combinations of the factors of interest (and interactions), then compare the differences is the -2log-likelihood values that are reported. For more details, google 'analysis of deviance' and you should get something. Probably not in a occupancy modelling context, but the steps are the same.

Any particular reason why you want to go down the ANOVA path compared to other options?

[Happyme]

Hi Darryll,

Sorry I was not very clear. My treatment is set up in excel as
S.1 S.2 S.3 S.4
0 0 1 0
0 0 0 0

0 for no catch and 1 for catch with say bait for this one. I thought I could put this into PRESENCE in the presence/absence section to get site estimates for each of my treatments. So I would have to create four separate models for each of my treatments. Bailey et. al ran an anova testing effects of year and sampling method on species detectability estimates p(.) using model psi(.), p(.). I thought I could do the same for sampling period and treatment. Maybe?

[cooch]

... For more details, google 'analysis of deviance' and you should get something.

Brett Sandercock provided a very nice summary of ANODEV (which is a hard-coded option in MARK) -- viewtopic.php?f=1&t=3736&p=12263&hilit=ANODEV#p12263

[Happyme]

Ok thank you!