Re: AI-GEOSTATS: tick sampling
I have also been struggling trying to find references on sampling, specifically sample size. The best two that I have found are "Geostatistical Error Measurement" Myers (1997) and "Geostatistics for Environmental Scientists" Webster and Oliver (2001). Webster and Oliver (2001) commit to a n=50 "rule of thumb", granted that is not a hard and fast rule. They specifically comment on experimental design in this book. The best luck I have had is to approach the problem from a simulation point of view, in that, I have simulated the distribution of distance pairs and took enough samples to make the distance distribution uniform, this is of course using stratified random sampling. The simulation is to find the density of points for a bounded area, this is a process of back-engineering sample size.
Hope this helps
Oklahoma State University
[Non-text portions of this message have been removed]