Report of the second Workshop on Practical Implementation of Statistical Sound Catch Sampling Programmes (WKPICS2)

This workshop, chaired by Jon Helge Vølstad (Norway) and Mike Armstrong (UK) was held at ICES headquarters, Copenhagen, from 6–9 November 2012.

WKPICS2 has developed guidelines for “best practice” that covers the design, implementa-tion and analysis stages of catch sampling schemes, assuming that regional objectives and data needs are clearly defined. Ideally, all national surveys should clearly document the sampling frame, sample selection procedures, response rates (e.g. refusals to take observers), imputation methods for missing data and weighting procedures employed to derive national estimates. Best practice can be defined as sampling designs, implementation and data analysis that minimize bias and maximize precision, and which makes the most efficient use of sampling resources. For example, probability-based sampling with accurate control of the inclusion probabilities would be considered an example of best practice. However, if logistical, legal, and economic constraints dictate the use of a non-probability based scheme to select primary sampling units (for example legal requirements in the selec-tion of a reference fleet), it is good practice if the selection is done in a way that ensures representative coverage of the target population and minimises bias, and if this can be demonstrated with suitable diagnostics. Bad practice would be an ad-hoc, non-probability based sampling scheme, particularly where there are no census data to show how repre-sentative the samples are of the population or to re-weight the samples during analysis.

WKPICS2 also proposes revised data quality indicators, including a simple one-page form that can be used to evaluate quality of data used for stock assessments. It is recommended that the quality indicators be further refined through practical testing by Regional Coordi-nation Groups and stock assessment working groups, based on several case studies.

WKPICS2 advises on future development of regional databases (RDB) and analysis soft-ware (FishFrame, Intercatch, and COST) to accommodate analysis of data that are collected according to best-practices survey sampling methods advocated by ICES WKPICS and SGPIDS. In particular, it is recommended that the RDB and analysis framework be further developed so that catch sampling schemes that result in clustered samples of trips can be accommodated.