Anyone willing to give confirmation I am on the right path?

[kernicholson]

I know the title to this question is odd - considering Im posting in the help forum, but Im new here and I didn't want to get "yelled" at yet. Im just starting out learning survival analysis, only I have to learn on the go and teach myself. I have no immediate resources (except internet) and no one within a 100km who has done survival analysis (or will admit they've done it). Yes, Id love to attend a workshop or conference or class - but I can't at the moment yet still need to make advances on this work. I assumed if I posted my question out right - it would be too long and overwhelming and no one would answer or Id get a partial answer and no follow up. So, I am hoping to find someone who would be willing to have a couple of short discussions through email to make sure I am going down the right track in my analysis. Just confirm that yup - that looks good.

Is this possible? Is anyone willing? Should I just post my question - risk being "yelled" at, and hope for the best?

Thanks,

K

[Eric Janney]

K,

Shoot me an email @ ejanney@usgs.gov or give me a call @ 541 273-8689 ext 202. I'd be happy to answer survival analysis questions if I can.

[kernicholson]

Thanks Eric Janney. I will give a little bit of a synopsis for anyone else willing to take on this challenge , or just in case Eric is unable to assist for whatever reason

I have marked adult female moose that have been radio collared since ~1996 till present. So, there is staggered entry and lots of censored data (a lot of unknowns as to why the moose left - i.e., mostly radio collar failure). My first goal is to estimate calf survival. No, the calves were never marked. The method was to note their birth date and then survey for them in late summer (aka fall) and then again late winter (aka spring) . In total that makes for 3 capture sessions (which is the minimum for survival analysis). Ive a feeling that the Lukcas model method will be used since I want to model young survival from marked adults and I have lots of twins or triplets in the data set. But it is likely more complicated than this.

My second goal would be to deal with the mother survival - we have weekly survival checks.

So, I hope maybe that gives a little more clue as towards what kind of advise I would be needing.

[Eric Janney]

K,

Indeed, from what you have indicated I think Lukacs's model is the most appropriate way to estimate calf survival. I haven't used that model, but after a quick read in the Gentle Intro it doesn't seem too complicated. You just need to think hard about your sampling and whether you reasobably meet the assumptions of that model. It you only have three sampling occasions you will only be able to estimate phi for the interval between occasions 1 & 2. Do the calves stay with their mother longer than one year? If so, you may be able to use the first sampling occasion the following year as the forth sampling occasion. It aslo seems that heterogeneity could be nasty with this model as the fate (in regards to survival and detection) of twins or triplets are probably not independent. I'm not sure whether Paul has evaluated this or not?

I'm not sure what the most appropriate way to estimate cow survival would be. There are a number of factors to consider such as sample size, detection probability, survey schedule, ect.