Date: Tuesday 11th October 2016
Time: Refreshments from 16:45, talk at 17:00 (duration of seminar 1h).
Non-members welcome (as always).
Location: Queensland University of Technology, Gardens Point Campus, S Block, Room 314.
Members and guests are welcome to join the speakers afterwards at a nearby restaurant.
Speakers: Bill Venables and Emma Lawrence, CSIRO Data61
Title: Non-standard design and analysis of a Bycatch Reduction Trial in the Gulf of Papua. R to the rescue once again.
Abstract: The design of experiments is a fascinating and mathematically mature branch of statistics, but real experiments often do not conform to the elegant template envisaged in the abstract subject. There are at least two reasons for this: logistic constraints imposed by the situation at hand, and the response itself requiring an analysis that is not the usual least squares. Nevertheless classical design principles can offer useful guidance when tackling a nonstandard design problem, as we show.
In trawling for prawns, the prawn catch is often only a very small fraction of the entire catch, with the remainder called “bycatch”. Reducing the bycatch component (whilst maintaining the prawn catch) is an important goal for the industry, primarily for environmental purposes, but also for economic reasons.
In this talk we discuss the design of a trawling study to trial various bycatch reduction devices, conducted recently in the Gulf of Papua. The study involved 4 vessels towing quad gear (4 separate, but linked nets) over 18 days. The experiment was designed to assess the effectiveness of 27 combinations of Turtle Excluder Devices (TEDs) and Bycatch Reduction Devices, (BRDs), with a control net without any devices attached. Each vessel carries its own control net, which it must use on all trawls. There is only one copy made of each of the 27 treatment nets, however, requiring the nets to be exchanged at sea. Such exchanges could only take place at specific times during the 18 day trial. These constraints made the design problem very non-standard. Nevertheless design principles, statistical intuition, and particularly R, enabled us to achieve a design that satisfied most of the natural criteria that a good design should satisfy.
The analysis also proved somewhat challenging, as is nearly always the case in fisheries. The design proved robust enough, however, to accommodate the various glitches and infelicities that always seem to happen at sea, and we found we could address all of the questions posed at the start of the project in a reasonably satisfying way.
Biographies: Bill Venables is an Honorary Research Fellow affiliated with CSIRO Data 61. He is also an Adjunct Professor at the University of Queensland. He was formerly a Senior Research Scientist with CSIRO and prior to that a Senior Lecturer at the University of Adelaide. As an Honorary Research Fellow and Adjunct Professor his main work is in research collaboration and mentoring of younger scientists, as well as limited educational contributions in the form of R workshops.
Emma Lawrence is a statistician working in the Data 61 group, with a focus on statistical issues in fisheries management and marine monitoring. She has worked previously at the (now) Australian Bureau of Agricultural and Resource Economics and Sciences on statistical projects relating to the management of Australian and International fisheries. Prior to that she had a short stint at the ABS.
|Time:||4:45 pm - 6:00 pm|