It is given while specifying the observational plan for a two state markov model, that we suppose, that we observe N i.i.d lives retrospectively. But, then how is it possible to observe the lives in any other way?
The lives might not be independent with respect to mortality (eg they may all live on the side of the same volcano), or they might not be identical with respect to mortality (eg some of them may be smokers and some not).
NO, my query was not with the "independent and identical" term, that is intuitive. But, instead my query was with the addition of retrospective term. Lives will obviously be observed by looking at their lifetimes in the past. How is it possible to observe them prospectively or any other way? Sorry for a misleading question. Thanks.
Ah right. Well, I think it's just trying to indicate to us that the number of lives in the investigation will only be known afterwards/in retrospect. This is generally the case in actuarial mortality investigations, eg if we work for an insurance company and are studying the mortality of policyholders aged 60, we don't know how many people we will observe in advance, as we can't know whether someone aged 60 will take out a policy in the future. Doing the analysis in retrospect then means that N is known, rather than random. (In certain other contexts, we might be able to specify N in advance, eg if we're conducting experiments on the survival of cells in a laboratory, we might set out specifically to run 100 such experiments.)