Question: Suppose a national polling agency conducted 100 polls in a year, using proper random sampling, and reported a 95% confidence interval for each poll. About how many of those confidence intervals would we expect not to cover the true population value?