PARAMETER-ESTIMATION IN ENCOUNTER SAMPLING SURVEYS

Encounter sampling surveys collect data from individuals who are encountered during an activity or behavior (event) by a surveyor that travels a random route through a survey area. Such surveys represent an alternative approach for estimation of parameters such as average duration of the entire event and auxiliary characteristics of individuals in the population. When the complete duration time of the event is unavailable, and/or the auxiliary variable is correlated with the duration of the event, traditional estimators taken from sampling theory and weighted distribution theory are inappropriate. Thus, we adapted earlier results of Cox (1969) for estimation of the average complete duration time of an event and present new estimators for the average value of auxiliary parameters. The performance of the new estimators was evaluated via Monte Carlo simulation, and procedures are illustrated using data from a survey of black-billed magpies (Pica pica) in a riparian area. Use of this survey design and associated parameter estimators provides an efficient estimation technique for researchers who study events or behavior that occur infrequently and/or are of short duration. Future work should address improvements in bias reduction and confidence interval procedures.