The problem is that we are failing to set the conditions because each individual has 10 calls and these 10 calls need to be sampled (as a block) when the individual is chosen in the random sampling. What we have so far in R can select single rows. Therefore the procedure we have devised is a lengthy one.

My Data is as follows:

I have data in excel comprising of female and male sound parameters of echolocation calls of bats. This data is organized into Indviduals (identified by rows), Individuals calls (10 calls for each individual), and the acoustic parameters belonging to each call (identified by columns). I have 88 individuals (43males, 45females) (880 calls-10 calls per individual). I want to randomly sample three sets (replicated three times) of 6, 12, 18,24,30,36,and 42 individuals This means for 6 individuals I would have 60 calls.  These sets need to be made up of equal numbers of males and females. To do this I first need to sample from each of the sexes and then combine this sample.

So for example if I want to randomly sample the acoustic parameters of 6 individuals I need to first randomly sample 3 females (from the 45 female individuals in my data set) and then randomly sample 3 males (from the 43 males in my data set). The same would be done with a sample of 12 (first sample 6 males and 6 females and add these together).

I then want to run a DFA on each of these random samples (i.e a dfa on each of  the three sets or replicates of 6 randomly sampled individuals, three sets of 12 randomly sampled individuals.. and so forth).

The goal is determine if classifcation success decreases with increasing the numbers of indviduals included in the analysis as well as if classifcation success depends on which individuals you randomly sample.

Any help with this would be greatly appreciated. This is for my friend Nikita Finger.