Power, precaution, type ii error and sampling design in assessment of environmental impacts

Solo disponible en BuenasTareas
  • Páginas : 41 (10012 palabras )
  • Descarga(s) : 0
  • Publicado : 12 de marzo de 2012
Leer documento completo
Vista previa del texto
Journal of Experimental Marine Biology and Ecology 296 (2003) 49 – 70 www.elsevier.com/locate/jembe

Power, precaution, Type II error and sampling design in assessment of environmental impacts
A.J. Underwood *, M.G. Chapman
Centre for Research on Ecological Impacts of Coastal Cities, Marine Ecology Laboratories A11, University of Sydney, NSW 2006, Australia Received 27 February 2003; receivedin revised form 5 June 2003; accepted 13 June 2003

Abstract Increasingly, environmental managers attempt to incorporate precautionary principles into decision making. In any quantitative analysis of impacts, precaution is closely related to the power of the analysis to detect an impact. Designs of sampling to detect impacts are, however, complex because of natural spatial and temporalvariability and the intrinsic nature of the statistical interactions which define impacts. Here, pulse and press responses and impacts that affect time courses (temporal variance) were modelled to determine the influences of increasing temporal replication—sampling more times in each of several longer periods before and again after an impact. Increasing the number of control or reference locations andnumber of replicate sample units at each time and place of sampling investigated the influence of spatial replication on power. From numerous scenarios of impacts, with or without natural spatial and temporal interactions (i.e. not caused by an impact), general recommendations are possible. Detecting press impacts requires maximal numbers of control locations. Shorter-term pulse impacts are bestdetected when the number of periods sampled is maximized. Impacts causing changes in temporal variance are most likely to be detected by sampling with the greatest possible number of periods or times within periods. To allow precautionary decision making, the type of predicted impact should be specified with its magnitude and duration. Only then can sampling be designed to be powerful, therebyallowing precautionary concepts to be invoked. D 2003 Elsevier B.V. All rights reserved.
Keywords: Type II error; Power; Precaution

* Corresponding author. Tel.: +61-2-9351-2590; fax: +61-2-9351-6713. E-mail address: aju@bio.usyd.edu.au (A.J. Underwood). 0022-0981/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0022-0981(03)00304-6

50

A.J. Underwood, M.G. Chapman /J. Exp. Mar. Biol. Ecol. 296 (2003) 49–70

1. Introduction 1.1. Precaution and decision-making Increasing global effort is being made in attempts to improve decision-making in environmental management. There is, however, often a lack of clear and logically consistent definitions of precaution leading to contradictions in practical attempts to implement precautionary decision-making.Particularly, there is confusion about the roles (or lack of roles) of science in legal decisions about precaution (Bazelon, 1981; Bodansky, 1991). The relationship between precautionary ideas and decision-making has been discussed in detail by several authors (e.g. Cameron and Abouchar, 1991; Myers, 1993; Gray, 1996; Gray and Bewers, 1996). A consensus from these discussions is that precautionary principlesrequire or demand one operational consequence. If decision-making is potentially erroneous—and it is difficult to imagine many environmental scenarios where environmental decision-making can be guaranteed to be free of errors—it is appropriate (by principle) to err ‘‘in favour’’ of environmental well-being. How to maximize precaution is directly related to how likely it is that there would be animpact due to any activity or development. This is intimately related to the wellunderstood probabilities of errors in decisions about ecological (and many other scientific) theories based on statistical tests. The general outcome of decision-making under inherent uncertainty has been illustrated as a 2 Â 2 matrix of decisions and outcomes. When testing any null hypothesis (e.g. Cohen, 1977;...
tracking img