Brian L. answered  11/25/19
Board Certified Oncology Pharmacist and Pharmacotherapy Specialist
Let n1 = the number (sample size) of Europeans surveyed = 12083
Let n2 = the number (sample size) of Americans surveyed = 862
Let x1 = the number of Europeans who thought genetic engineering was risky = 7628
Let x2 = the number of Americans who thought genetic engineering was risky = 448
Let p1 = the proportion of Europeans who thought genetic engineering was risky
p1 = (x1)÷(n1) = 7628 ÷ 12083 = 0.631
Let p2 = the proportion of Americans who thought genetic engineering was risky
p2 = (x2)÷(n2) = 448 ÷ 862 = 0.520
Even though this is considered a binomial random variable, the normal distribution may be used as a sampling distribution because n1p1 and n2p2 are both greater than 5.
The standard error (SE) is going to be [(p1q1 ÷ n1)+((p2q2 ÷ n2)]1/2 = √(p1q1 ÷ n1)+((p2q2 ÷ n2)
Note that q1 = (1-p1) and q2 = (1-p2). Stated another way, q1 is the proportion of Europeans who did NOT find genetic engineering risky and q2 is the proportion of Americans who did NOT find genetic engineering risky.
The estimator for the difference in proportions of persons who found genetic engineering risky is going to be (p1 - p2).
For a 97% confidence interval, the critical Z value is 2.17.
The 97% CI for difference in proportions is
(p1 - p2) ± 2.17√(p1q1 ÷ n1)+((p2q2 ÷ n2)=(0.631-0.520)±√[ (0.631 x 0.369 ÷ 12083) + (0.520 x 0.480 ÷ 862)]
=0.111 ± 0.038
97% CI for difference in proportions (Europe - United States) is 0.073 to 0.149
97% CI for difference in PERCENTAGE is 7.3% to 14.9%
        Brian L.
11/26/19
    
            
                    
                    
                    
Jacqueline A.
Thank you so much!11/26/19