CF Smit Department of Statistics, University of Pretoria, Pretoria, 0002, South Africa and V Peddemors Department of Zoology, University of Durban-Westville, Durban, 4000 South Africa.
In two series of tests of a new electric shark repellent (the SharkPOD), data was collected on the time needed to attack the bait, under power-off and power-on (active) conditions. Various approaches were followed to estimate the relevant probabilities and their standard errors, e.g. through Bayesian updating and resampling methods (especially jackknifing).
Conclusions were separately drawn after completion of the first experiment (in which there were 8 successful attacks in 98 five minute active periods), and after completion of the second experiment (in which no successful attacks were reconied in 24 ten minute active periods).
In general it was concluded that the probability of an attack in at most 5 minutes was reduced from about 0.70 in power-off mode to about 0.08 in power-on mode and in a period of at most 10 minutes from 0.90 to 0.16.