A company proposes to destroy a rocky shoal, which will be dangerous to container ships coming into a harbour where a new port is being constructed. The company collects samples of the fauna living on and around the shoal and compares them to similar samples taken at three other rocky shoals in the area.

1. The company decides (without consulting anyone else) on a minimum effect size of 200% for the above test - that is, the shoal targeted for potential destruction has to have three times as many species as the average of the three other shoals in the area for it to be considered worth saving. Does this seem reasonable or unreasonable? Why? (4 marks)
2. The company carries out the statistical test to see if the shoal to be destroyed is significantly different from other shoals in species diversity. The null hypothesis is disproven. However, the actual effect size is smaller than the company's decision (above) - i.e. the targeted shoal has less than three times the number of species as the average of the other three shoals - so the company says this means it can go ahead and destroy the shoal anyway even though the null hypothesis was rejected. What is the drawback with the company's process of decision-making? (4 marks)
3. Suppose that a detailed examination of the sampling process shows that each rocky shoal was sampled by divers counting animals in places they chose because, from visual inspection, those places seemed to represent the abundances of animals on the shoal overall. What is the effect on the reliability of the statistical test on the data, and is there a simple correction that could be applied? If so, what? (2 marks)