Sampling Sites and Interview Times Simultaneously: Application to Fishery Intercept Surveys with Limited Staff

NOAA Fisheries intercept surveys provide a special challenge for probability sampling of sites and observation periods to be covered by small teams of data collectors.  The problem is illustrated by application to a North Carolina pilot study.  The paper presents and compares sample designs that can be implemented in a coastal region assigned to a team of data collectors for a specified month. Within this region, a probability sample of sites (or site clusters) must be distributed across fishing modes and selected with probabilities proportional advance estimates of the fishing effort (based on fishing pressure codes). The sample must also be distributed across observation periods by kind of day and time of day. The sampling frame for observation periods is based on four six-hour periods per day.  The implementation of the design is particularly challenging when coordinating the selection of site clusters with the selection of time periods and ensuring that the assignment can reasonably be conducted by the interviewing team. Consideration is also given to having data collectors work in two-person teams during some time periods (e.g., night time) and individually during other periods.