Question
Suppose you are an investigator interested in demonstrating that a certain exposure variable is associated with a health outcome of interest. You conduct a case-control study and find that your estimated odds ratio is 3.5. If you can justifiably reason that whatever possible bias that exists is towards the null, why should you feel good about the results of your study? Explain.
Suppose you are a government scientist reviewing a study your agency has funded to evaluate whether a certain pharmaceutical drug already on the public market causes harmful health consequences. The study finds a statistically significant estimated effect of 3.5 that indicates that the drug is harmful. However, based on your review, you can convincingly argue that any bias that may exist is away from the null. Should you recommend that the drug be taken off the market? Explain.
Source:
