Number of repeat visits per samling units non-random

[mondulkiri]

Dear Phidot,

I am hoping to use information collected during law enforcement patrols in South-East Asian protected areas to estimate occupancy of ungulates. Protected areas will be divided into spatial sampling units (of a size relevant to the biology of the species of interest) with independent visits to each sampling unit by law enforcement teams acting as sampling sessions during which a novel data collection method will be used to obtain 1 or 0 (species presence/absence) data. However the number of independent visits to each sampling unit will be neither equal or random – certain units (where illegal activity is highest/suspected; closer to ranger stations) are likely to be visited more often than others (remoter; less illegal activity). It is possible that there may be a negative relationship between number of visits per sampling unit and probability of species presence in that unit.

Assuming that the main assumptions of occupancy studies are met (i.e. no false positives; closure within sampling units; detection at one unit is independent of detection at all other units) would such a sampling design, where the number of repeat visits per sampling unit is non-random, be statistically valid?

Regards

Tom Gray

[darryl]

Number of surveys aren't likely to be an issue, but sounds like there may be a general selection issue where certain areas are going to be targeted for patrols rather than areas being selected at random (or at least according to a good probabilistic sampling scheme).

Cheers
Darryl