U-Care Test 3.SR Chi2 vs 2-sided test discrepancy

[simone77]

Hi,

I don't understand a result I get when testing the survival heterogeneity with Test 3.SR in U-Care.
This is the m-array relative to the group (female rabbits) I am testing:


And this is the Test 3.SR result:


This seems to be just the opposite situation that occurred in this case where Chi2 and G2 were non significant but two sided tests are, and in fact it was suggested that Normally the two sided test z-statistics has more power than the chi-squared statistics.

I can't figure out what is going on, I'd really appreciate any suggestion.
Thanks in advance

Simone

[CHOQUET]

Hello,
in our case, looking for the z statistics (by occasion) gave an explanation of
this apparent contradiction. In fact, the z statistics have different signs, which sum together
are non significant because no real tendency of the transient nature can be given.
At occasions 2 3 4 and 6, there is something called transient whereas in the other case,
there is "anti-transient".
To conclude, there is some heterogeneity but this is unclear which biological cause is behind.
Perhaps, looking for the other components of the test can give some complementary informations.
Sincerely,
Rémi

[simone77]

Hello,

Thanks for answering, now I understand why there is this apparent discrepancy.
I understand what a transient effect is but it is not clear to me what the anti-transient means, I have been looking elsewhere but I can't find anything.

As shown below, the same transient-antitransient temporal pattern exists also for males (same signs but with a different intensity). As this is a population of rabbits relased inside an enclosure there should not be transients individuals; also, either adults as juveniles are captured and marked at each occasion and some epidemic spread of mixomatosis and RHD have been registered (and population prevalences measured for each occasion). Here below are reported the other tests' results.









As you can see, there is trap-happiness effect for females and heterogenity for survival rate of males.
Even though, I don't catch if there is further information that is escaping to my understanding and the pattern you pointed out seems to suggest something more. If you could give me some hints/suggestion on what could be going on, I really would appreciate it.

Thank you in any case.

[CHOQUET]

Hello

In your case, there is certainly at least two kind of individuals (heterogeneity of classes).
See for example pledger (2003). In the case of heterogeneity of classes, the two test2.CT
and 3.SR are significant. If there is no clue of the states of the rabbit, you can try a model
with two class of heterogeneity (can be easily done with E-SURGE) with exchanges between
the two classes if the two classes correspond the two states "safe" and "infected",
see for example Péron, G., Crochet, P.A.C., Choquet, R., Pradel, R., Lebreton, J.-D. and O. Gimenez. (2010). "Capture-recapture models with heterogeneity to study survival senescence in the wild." Oikos 119(3): 524-532. DOI:10.1111/j.1600-0706.2009.017882.x. If there is some clue about the states, see
Conn and Cooch (2009) for an application on epidemiology.

Sincerely,

Rémi