When detection probability is small, the results yielded by the two analysis methods available in GENPRES ("Analyze w/ expected values" or "Analyze w/ simulated values") can be quite different. In design of a study, which method is more reliable?

I found empirically that the results of the "Analyze w/ expected values" option were equivalent to the equations for asymptotic variance calculation from MacKenzie and Royle (2005). Is there any guidance in the literature for deciding whether asymptotic variance calculations or simulation studies are more appropriate for use in designing occupancy studies?