Unestimated parameters in Variance Components Output

questions concerning analysis/theory using program MARK

Unestimated parameters in Variance Components Output

Postby constant survivor » Sat Dec 05, 2020 3:36 pm

Hello,
I want to estimate mean survival rates for different species via the Variance Components approach.

Here is an example Output for one species (TSM model phi(t/t)p(effort) was used):

Beta-hat SE(Beta-hat) Label
----------- ------------ ----------------
0.251760 0.049822 Intercept

Par. Num S-hat SE(S-hat) S-tilde SE(S-tilde) RMSE(S-tilde)
-------- ---------- ----------- ----------- ----------- -------------
31 0.497489 0.228818 0.347983 0.124665 0.194662
32 0.274248 0.225705 0.244752 0.123992 0.127452
33 0.413970 0.201544 0.297568 0.120152 0.167290
34 0.580545 0.163909 0.415644 0.110898 0.198723
35 0.381847 0.165571 0.301842 0.111489 0.137224
36 0.604509 0.137999 0.449158 0.101826 0.185749
37 0.632643 0.130168 0.485555 0.098311 0.176918
38 0.444967 0.149616 0.359857 0.106128 0.136039
39 0.704313 0.259160 0.423015 0.126155 0.308292
40 0.174540 0.175824 0.222226 0.111096 0.120898
41 0.000000 0.000000 0.000000 0.000000 0.000000
42 0.810400 0.169751 0.562663 0.111807 0.271798
43 0.327060 0.273181 0.273912 0.129654 0.140125
44 0.000000 0.000093 0.000000 0.000093 0.000093
45 0.412466 0.274483 0.279174 0.130035 0.186215
46 0.545033 0.347157 0.338920 0.135140 0.246465
47 0.000000 0.000079 0.000000 0.000079 0.000079
48 0.263506 0.213345 0.201256 0.122068 0.137024
49 0.167125 0.149957 0.157959 0.106431 0.106825
50 0.626819 0.181744 0.430612 0.115714 0.227787
51 0.444153 0.145379 0.319551 0.104581 0.162675
52 0.421552 0.145117 0.311701 0.104500 0.151616
53 0.380834 0.149502 0.298808 0.106174 0.134168
54 0.402819 0.151032 0.307050 0.106797 0.143447
55 0.402859 0.128861 0.295397 0.097729 0.145255
56 0.508864 0.197327 0.343295 0.117687 0.203133
57 0.308819 0.188594 0.272576 0.116276 0.121793
58 0.268911 0.162718 0.206530 0.110075 0.126523
59 0.417501 0.217274 0.297342 0.122710 0.171744

Naive estimate of sigma^2 = 0.0079402 with 95% CI (-0.0071203 to 0.0416710)

Estimate of sigma^2 = 0.0194132 with 95% CI (0.0071456 to 0.0498797)

Estimate of sigma = 0.1393312 with 95% CI (0.0845318 to 0.2233376)
--------------------------------------------

As you see, some of the parameters are not estimated. Surely because of sparse data.
I wonder if it makes sense at all to "use" (i.e report) the estimated mean of survival if some of the parameters are not estimated. Because obviously they are anyway contributing to estimation of the mean - leading to very small value of the mean.
What should be the general approach in this case? Should I try to find a model for that less (or none at all?) parameters are non-estimable? Or simply accept it?

Thanks
Hannes
constant survivor
 
Posts: 36
Joined: Wed Dec 11, 2019 12:20 pm

Return to analysis help

Who is online

Users browsing this forum: No registered users and 13 guests