Premium
Management: Briefs Selection of a Minimum Sample Size for Application of the Regression‐Line–Percentile Technique
Author(s) -
Brown Michael L.,
Murphy Brian R.
Publication year - 1996
Publication title -
north american journal of fisheries management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 72
eISSN - 1548-8675
pISSN - 0275-5947
DOI - 10.1577/1548-8675(1996)016<0427:mbsoam>2.3.co;2
Subject(s) - percentile , statistics , mathematics , variance (accounting) , regression , sample size determination , sample (material) , regression analysis , population , standard error , linear regression , exponential function , sample variance , demography , mathematical analysis , chemistry , chromatography , accounting , sociology , business
Proper application of the relative weight ( W r ) index depends on development of a standard weight ( W s ) equation for the target species. Development of W s equations has been facilitated by the regression‐line–percentile (RLP) technique, but the minimum data needed to develop a W s equation with the RLP technique has not been investigated. We applied the bootstrap technique to data sets containing slope values from population log 10 weight–log 10 total length regressions for nine fish species (50–292 populations each). We randomly selected regression slopes at sequential intervals of five, and each iteration was repeated 100 times to determine the variance about the regression slope means. The sample variance for slopes was observed to decline in an exponential fashion. A minimum level of precision based on the sample variance (<0.002) provided a conservative estimate of necessary sample size ( N ≥ 50). This variance benchmark is sufficient for the current level of precision required in the development and application of RLP‐generated equations.