I have conducted pairwise relatedness estimations between over 100 samples from one termite population, which is divided into five colonies. The estimator basically looks at the genotypical information from any sample, compares it to all other samples, and calculates how high the probability is for the genotypic information of both samples to be identical by descent. So in the end I get an estimation of this probability for every possible pairwise combination (dyad) of the samples. I'm mostly interested in the relatedness within and between those colonies as a covariate in later investigations. So in the end I simply subset all the pairwise estimations into groups of interest and took the average of those. This gives me a rough number for every group to work with.
Now, technically I would want to report on the performance of these estimations. It's not the focus of my work, but I'd still like to mention it somewhere. During the estimation process, 95% confidence intervals were calculated for each dyad, by bootstrapping with 1000 iterations. I am not sure how to incorporate this in my report, since the CI's can differ greatly between the pairwise estimates. Besides, the of number of possible pairs to report on would be too large, obviously. I have thought about using the coefficient of variation instead of confidence intervals for each group, to give a rough approximation of reliability of the estimations. Or maybe the relative standard error?
If anyone has any ideas on how to approach this, please let me know. I'm generally curious on how one would investigate a data set which is produced by so many individual estimates.