Scientific Observer Sampling Manual ... - Kimdietrich.com

3.6 Fundamental statistical concepts
Sampling fish catch may sound simple but collecting a statistically representative sample of
commercial catch is challenging. Each vessel and gear type poses unique sampling demands
and hazards.
3.6.1 What is a “statistically representative sample”?
In general, it is a selection of individual observations intended to yield some knowledge
about a population of concern. For fisheries managers/scientists, the population of concern is
all commercial catch and other biological life caught (not just retained) by fishing vessels.
When random sampling is used to sample (sometimes called a subsample), every member of
the population must have an equal probability of occurring in the sample. If every
member of the population is equally likely to occur in the sample and sampling is repeated
over time, then the collection of samples can be used to draw conclusions about the
population.
3.6.2 Why is random sampling important?
Random sampling removes subjectivity and allows managers, fishers and other end users to
feel confident that observer data is not biased. When data is collected randomly, scientists
can justify using certain statistical methods to estimate population parameters based on the
data collected. Sampling allows management agencies to estimate the total weight (or
number) of fish removed from the ocean. Sampling may also allow scientists to determine the
age of fish taken and a whole suite of other information for assessing the status of fish
populations.
Sampling can occur at multiple levels including the fishery (which fisheries will be selected
for observer coverage), vessel (which vessels take observers), trip (which trips within vessel),
haul and/or net (which hauls are selected for catch composition), and species (which species
and/or individual are selected for length, age structures and other biological sampling).
Observers will be most concerned with sampling at the haul and species level whereas the
management agency must address sampling at the level of fishery, vessel and trip.
Observers will be provided with two different random sample tables (RST) in the Observer
Logbook. RSTs prescribe which hauls to sample. Each RST is designed for a different
number of hauls per day as follows:
Hauls per day RST Target sample rate
1-2 None 100%
3-4 #1 70-75%
5+ #2 65%-70%
General guidelines for haul-level sampling:
• Collect the sample before any sorting occurs (except some shrimp trawl sampling
scenarios);
• Do not hand pick the sample. There is a subconscious tendency to select large or
otherwise obvious individuals;
• Collect sample from multiple points (on deck or from bin);
3-8