6. Likelihood of Introducing Nonindigenous Organisms
89
Figure 6.3. Uniform distribution from APHIS’s Mexican avocado risk assessment
(USDA 1995c, Table 8) used to estimate the probability that fruit would be transported to a
habitat suitable for establishment of the avocado seed moth (Stenoma catenifer) under
program option A. Distribution minimum = 0.25, maximum = 0.75.
indicated by the shape of the curve, which is determined by the relative position of
the three parameters. Values at or near the most likely value are selected for
calculations more often than values at or near the minimum and maximum.
Triangular distributions can be used when it is fair to assume that some values are
more likely than others but have the disadvantage of not selecting values below
the minimum and maximum. Choice of the minimum and maximum values can be
considered more important than choice of the most likely value because the
minimum and maximum imply a high degree of assurance about the limits of the
underlying distribution. We use triangular distributions occasionally but avoid
them because biological systems are seldom well represented by distributions
with strict limits. The arbitrary shape of the triangular distribution and its strict
limits are indications that the precise nature of the distribution is not known.
Figure 6.4 shows an example of a triangular distribution used when assessing the
risk of importing Japanese Unshu orange fruit into citrus-producing states (USDA
1995b).
Normal Distribution
Normal distributions have the advantage of being representative of many types of
biological phenomena. They also have the advantages of not being limited by
minimum or maximum values. However, when the parameter being estimated is a
probability—which has a domain from 0 to 1—care must be taken to avoid
including “probability” values below 0 and above 1. The risk is greatest with
probability values greater than 0.1. Figure 6.5 shows an example of a normal
probability distribution used during assessment of risks associated with importation of the Japanese Unshu orange (USDA 1995b, Table 11). This distribution was
89
Figure 6.3. Uniform distribution from APHIS’s Mexican avocado risk assessment
(USDA 1995c, Table 8) used to estimate the probability that fruit would be transported to a
habitat suitable for establishment of the avocado seed moth (Stenoma catenifer) under
program option A. Distribution minimum = 0.25, maximum = 0.75.
indicated by the shape of the curve, which is determined by the relative position of
the three parameters. Values at or near the most likely value are selected for
calculations more often than values at or near the minimum and maximum.
Triangular distributions can be used when it is fair to assume that some values are
more likely than others but have the disadvantage of not selecting values below
the minimum and maximum. Choice of the minimum and maximum values can be
considered more important than choice of the most likely value because the
minimum and maximum imply a high degree of assurance about the limits of the
underlying distribution. We use triangular distributions occasionally but avoid
them because biological systems are seldom well represented by distributions
with strict limits. The arbitrary shape of the triangular distribution and its strict
limits are indications that the precise nature of the distribution is not known.
Figure 6.4 shows an example of a triangular distribution used when assessing the
risk of importing Japanese Unshu orange fruit into citrus-producing states (USDA
1995b).
Normal Distribution
Normal distributions have the advantage of being representative of many types of
biological phenomena. They also have the advantages of not being limited by
minimum or maximum values. However, when the parameter being estimated is a
probability—which has a domain from 0 to 1—care must be taken to avoid
including “probability” values below 0 and above 1. The risk is greatest with
probability values greater than 0.1. Figure 6.5 shows an example of a normal
probability distribution used during assessment of risks associated with importation of the Japanese Unshu orange (USDA 1995b, Table 11). This distribution was