Problem Set 3: seizures.csv

A group of researchers are testing a new drug that should reduce seizures by suppressing activity of a particular class of neurotransmitter receptors. The dataset seizures.csv contains the results of a pilot study. Assume the pilot study is accurate.

1. What is the expected effect size (Cohen's d)?

2. How many animals (in each group) are needed to have a 90% chance of detecting a statistically significant (p < 0.05) effect of the drug?

3. The researchers are considering a pre/post design where the same animals are observed for a period with and without the drug. Assuming the effect size stays the same, how many total animals would be needed for this kind of research design?

4. The company rep sends out an email that a new version of the drug is available. This version has a stronger binding affinity and blocks receptors at a higher rate. The claim is that this should increase the effect of the drug by 20%. Unfortunately, the drug costs 50% more than the original drug. Discuss how you could use power analysis to determine the relevant ethical and financial trade-offs in using this new drug (for the purposes of this question, let's assume the company rep is telling the truth about the effectiveness of the new drug).

group seizures Control 8 Control 7 Control 9 Control 6 Treated 8 Treated 2 Treated 5 Treated 7