Output for two-species with covariates

[tajensen]

I realize this might be a very basic question, but I'm looking for some help interpreting the output below. My concern is the large standard errors for covariates G1, G2, and G3, which according to the output have essentially no effect on occupancy. In biological terms, it makes sense to me that G1 would have a more positive effect than G2, with G2 having a more positive effect than G3, but statistically it looks like they are equally zero.

Code: Select all

                                          estimate    std.error
A1   psiA                             :  -7.024987   24.273890
A2   psiBA                            :   0.387329   99.549122
A3   psiBa                            : -15.144745   13.818888
A4   psiA.G1                          :   9.022576   24.323014
A5   psiA.G2                          :   7.758922   24.264143
A6   psiA.G3                          : -23.810218  2339102.755476
A7   psiBA.G1                         :   0.521841   99.551062
A8   psiBA.G2                         :   0.312119   99.550872
A9   psiBA.G3                         :  -0.446224   10.000000
A10  psiBa.G1                         : -12.484614  898963.877790
A11  psiBa.G2                         :  14.510289   13.818778
A12  psiBa.G3                         : -17.169919   -1.#IND00
B1   pA[1]                            :   1.029209    0.710469
B2   pB[1]                            :   0.549306    1.393029
B3   rA[1]                            :   1.225671    0.471626
B4   rBA[1]                           :   2.218199    0.756472
B5   rBa[1]                           :  -0.010303    0.816919

            -,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12,
psiA          1 0 0 G1 G2 G3 0 0 0 0 0 0
psiBA         0 1 0 0 0 0 G1 G2 G3 0 0 0
psiBa         0 0 1 0 0 0 0 0 0 G1 G2 G3

            -,b1,b2,b3,b4,b5,
pA[1]         1 0 0 0 0
pA[2]         1 0 0 0 0
pA[3]         1 0 0 0 0
pB[1]         0 1 0 0 0
pB[2]         0 1 0 0 0
pB[3]         0 1 0 0 0
rA[1]         0 0 1 0 0
rA[2]         0 0 1 0 0
rA[3]         0 0 1 0 0
rBA[1]        0 0 0 1 0
rBA[2]        0 0 0 1 0
rBA[3]        0 0 0 1 0
rBa[1]        0 0 0 0 1
rBa[2]        0 0 0 0 1
rBa[3]        0 0 0 0 1


-Tim

[darryl]

Hi Tim,
Your first issue is that the SE for A12 in the output is non-numeric, suggesting to me that you probably had a warning about the variance-covariance matrix earlier in the output in which case all your SE's are junk anyway. See the PRESENE FAQ and elsewhere in the forum for ideas on trying to work around that. Also, judging by the magnitude of the coefficients I'm wondering whether your model may be overparameterised in some regard. Are the real estimates (ie the probabilities) later in the output basically coming out to be 0 or 1? If so, then you've either found the perfect predictors or you may be trying to fit too complicated a model for the data you've got.

Cheers
Darryl