the proportion of people in each sample who experience side effects is recorded. are the sample sizes large enough to assume that the sampling distribution of the difference in sample proportions is approximately normal?