confidence level = in 95% of all possible samples of 2253 US adults, the resulting c.i. would capture the actual population proportion of Us adults who use Twitter/others
sets a finite value - so you know when to reject ---- advantage = define rejection region in terms of sample mean & make conclusion in field you are collecting ---- disadvantage = stuck w/ fixed level for the test
confidence interval calculation
statistic ± (critical value) × (SD of statistic)
(critical value) × (SD of statistic)
when does ME get smaller?
confidence level decreases & sample size n increases
3 conditions to check before calc C.I.
1) Random → 2) Normal - means = at least 30; proportions = at least 10 → 3) Independent - sample size < 10% N