Study Designs and Biases:

Case-control study:

Enroll children with a learning disability and those without and ask their parents whether they were raised on a vegetarian diet or omnivorous diet.

Prospective cohort study:

Enroll infants whose parents have chosen to follow a vegetarian diet in their families and some whose parents have chosen an omnivorous diet for their families. Follow the children until age 10 and ask parents about diagnosis of learning disabilities.

Consider how different types of bias might affect the validity of these studies.  Please answer the following questions with 1 to 2 sentences.

A. How could detection bias occur in your prospective cohort study?

B. Do you think this detection bias would strengthen or attenuate the association between vegetarian diet and learning disabilities? Defend your choice briefly.

C. In reality, some parents may start their child on a vegetarian diet and then switch to an omnivorous diet or vice versa. If conducting a cohort study, which measure of association, would you prefer to use in this situation? What specific exposure information would you need in order to calculate this measure of association?

D. Is selection bias more likely to be a problem in the cohort study or the case-control study? Briefly describe how selection bias could occur in that study design.

E. In the case-control study, suppose you asked parents about past diet. How could this result in differential misclassification of the exposure? What is another name for this bias?

F. After conducting your case-control study, you estimate the odds ratio for the association between vegetarian diet and development of learning disabilities to be 0.7. You suspect that all parents are under-reporting vegetarian diet to the same extent, regardless of the learning disability status of their children, due to controversies about the impact of such a diet on very young children. If you are correct, what might be a possible value for the true(unbiased) odds ratio? Briefly defend your answer.