Unequal, within Treatment replication (observations)
Jul 18, 2018 12:46 PM(329 views)
I have a research design in which individual mosquitoes are released, one at a time into their own observational cage. In each observational cage is a choice of 4 places to lay eggs. Each of the 4 egg-laying spots has water with different characteristics but the choice is among 4 simultenously-presented egg dishes, not sequentially presented. The first block effect is cohort or simultaneously-reared mosquitoes of the same age, same rearing environment, though the number of mosquitoes within such a block is variable from one block to another. Each mosquito has an unknown (ahead of time) and uncontrolled (or variable from mosquito to mosquito) number of eggs she can lay in some distribution across the 4 egg-laying sites in her cage. We observe or count the number of eggs layed by each female in each of the egg-laying dishes. The data matrix can (I think) be expressed as 1 line per egg cross classified in columns by mosquito ID (categorical), cohort (categorical) and egg-laying dish or choice (categorical), in which case summing lines across similarly cross-classified eggs can yield data for calculations of means between egg-laying sites or frequency counts within a mosquito, between choices. The ultimate objective is a rather straight forward comparison of the attractiveness of the egg-laying site choices. A shorter data matrix might be represented with the same categorical variables: Treatment (Choice 1, Choice 2, Choice 3 or Choice 4); Cohort; Individual Mosquito and a response variable with the number of eggs in each of the treatment levels, thus 4 lines or rows per mosquito. Individual mosquitoes represent replication (of a sort because they don't necessarily have the same number of eggs to lay) within each cohort. I really need to figure out the rather basic way to specify the model - not a very sophisticated question but it has been years since I did this with JmpIn - so as to control for the mosquito effect (if any) within cohort effect (if any) and get at differences among mean numbers of eggs layed per choice (should be easy but JmpIn onboard documentation is not that straightforward) and frequencies, given potentially unequal numbers of eggs per female to start. Thanks.