[RecFIN News] Weight Estimates in 1993-2000

[Wade VanBuskirk <wade@psmfc.org>]

*** RecFIN News Service ***

The mean weights were recalculated for the 1993-2000 MRFSS catch and applied to the 
total weight estimates for catch type A and A+B1 for the same time period. 

During this time period only actual weights of fish were used for the mean weight
calculation. The new method also includes calculated weights taken from fish that
were measured but not weighted.

The aggrate difference in the weights is available on this page:
http://www.psmfc.org/recfin/examples/recalc_wgts_1993-2000.html

A breakdown example by stratum for bocaccio is available on this page:
http://www.psmfc.org/recfin/examples/recalc_wgts_1993-2000_bocaccio.html

The purpose of the recalculation was to fill in as many "missing" 
weight estimates as possible. Missing weight estimates are those strata for which 
there is an estimated number of fish caught, but there were no fish weighed and 
so, there is no mean weight to calculate a weight estimate (catch # estimate X 
mean weight = catch weight estimate). 

In the traditional MRFSS other mean weights are applied to "missing" 
weight estimators within a wave and mode from means calculated in the same state 
or subregion. This method is incomplete as it still allowed an estimate for a 
number of fish without an estimate for total weight to occur on taxa that were 
relatively uncommon in a particular wave and mode. The search for a usable mean 
weight did not extend far from the stratum for which a catch estimate was 
calculated. This had caused summaries of total weight to be below what they 
should have been because of lack of data on weights of the catch on some taxa 
in certain strata.

A revised method for gathering mean weights has been applied to the 1993-2000 data 
to fill in mean weights from other waves, years and modes. This insures that a 
mean weight is nearly always made available to estimate total weight of the catch 
when a number of fish has been estimated. The mean weight from other more 
distant strata may not accurately represent the weights of the fish counted, 
but it is much better then a zero weight estimate.

The most specific mean weight available has been applied to each catch estimator.
If there was no mean weight for the stratum for which a number estimate exists, 
then other waves (two-month periods) are examined in the year. Then other 
water areas (inland and near or off-shore), then other states and subregions, 
then other fishing modes in the same state and subregion, and finally the the 
entire data series is examined for weights on that taxa. The 'type' variable in 
the estimate data indicates which summary level was used to calculate the mean 
weight and total weight estimate. 

The table below shows the values for the type variable and the strata keys used 
(or not used) in the summary of mean weight for a taxa.

 TYPE*                   EXCLUDED KEYS - RETAINED KEYS
  32  mode_fx year sub_reg area_x wave - 
  44               mode_fx area_x wave - year sub_reg
  56               sub_reg area_x wave - mode_fx year
  60                       area_x wave - mode_fx year sub_reg
  62                              wave - mode_fx year sub_reg area_x
  63                                   - mode_fx year sub_reg area_x wave

*Higher 'type' numbers are more specific.


*** RecFIN News Home -> http://www.psmfc.org/recfin/service/news/ ***