1. Both Grandjean et al. (1998) and Myers et al. (2003) investigated the hypothesized relation between methylmercury exposure during pregnancy and cognitive ability. But these authors came to quite different conclusions. Grandjean et al. (1998) found an effect, while Myers et al. (2003) did not. The present question asks you to explore the possible reasons for this apparent discrepancy. (If I use this question on the final exam, I will provide you with copies of both Grandjean et al. (1998) and Myers et al. (2003).)

1a. What confounding factors did Myers not control for? Which factors did he control for? What factors did Myers control for that Grandjean did not? And vice versa? Is one study stronger than the other in this respect? (one paragraph answer)

1b. How did Myers et al. (2003) measure exposure to methylmercury? Is his measure any better or worse than the measure of exposure used by Grandjean et al. (1998)? If these authors measured methylymercury exposure differently, could this explain their differing results? If so, explain. (one paragraph answer)

1c. Why did Myers et al. (2003) come to a different conclusion than did Grandjean et al. (1998)? List the likely potential explanations for this difference, and then state the evidence for and against each. (2 paragraph answer)

1d. It is possible that the results of Grandjean et al. (1998) and Myers et al. (2003) really are consistent with one another? Explain. (one paragraph answer)

Answer. Yes, it is possible. These results would be consistent if there really was no effect of methylmercury exposure on cognitive ability, and the Grandjean et al. (1998) study found a spurious relation. Why might Grandjean et al. have seen an apparent effect, even if there really was no such effect? One possibility is that the cases (those exposed to high levels of methylmercury) were also exposed to high levels of PCBs, and that PCB exposure during pregnancy decreases cognitive ability.

1e. Is it possible that methylymercury exposure during pregnancy really does decrease cognitive ability, even though Myers et al (2003) found no such effect? (one paragraph answer)

Answer. Yes, it is possible. Myers et al. (2003) studied several hundred individuals exposed to low levels of methylmercury. If methylymercury has a very small but real impact on cognitive ability, it is entirely possible that Myers et al. would have been unable to reject the null hypothesis of no effect with their sample sizes of several hundred, even though a larger sample size might have enabled them to reject their null hypothesis. In this situation, Myers et al. would be unable to detect an effect of methylmercury exposure on cognitive ability that really was present. By analogy, suppose your bathroom scale reads out pounds, but not ounzes. You go on a diet and lose 1 ounze the first day, but your bathroom scale cannot detect this small change in your weight. There is a real change in your weight that is too small for your equipment to detect.