In: Statistics and Probability
Some years, the Gallup Poll asks respondents how much confidence they have in various American institutions. You may assume the results are based on a simple random sample of 1000 persons each year; the samples are independent from year to year.
a) In 2005, only 41% of the respondents had “a great deal or quite a lot” of confidence in the Supreme Court, compared to 50% in 2000. Is the difference real?
b) In 2005, only 22% of the respondents had “a great deal or quite a lot” of confidence in Congress, whereas 24% of the respondents had “a great deal or quite a lot” of confidence in labor. Is the difference between 24% and 25% real?
Solution
The test procedure for both parts is the same, which is explained in brief below.
Back-up Theory
To test if observed difference in sample proportions reflects real difference, i.e., difference in population proportions
Hypotheses:
Null H0 : p1 = p2 Vs HA : p1≠ p2
Test Statistic:
Z = (p1cap – p2cap)/√[pcap(1 - pcap){(1/n1) + (1/n2)} where p1cap and p2cap are sample proportions, n1, n2 are sample sizes and pcap = {(n1 x p1cap) + (n2 x p2cap)}/(n1 + n2).
Distribution, Significance Level, α, Critical Value and p-value:
Under H0, distribution of Z can be approximated by Standard Normal Distribution
So, given a level of significance of α%, Critical Value = upper (α/2)% of N(0, 1), and
p-value = P(Z > | Zcal |)
Using Excel Functions Statistical NORMSINV and NORMSDIS, Critical Value and p-value can be found.
Decision:
If | Zcal | > Zcrit, or equivalently, p-value < α, H0 is rejected.
Now, to work out the solution,
Given n1 = n2= 1000 for both parts
Part (a)
Calculations:
| 
 n1 =  | 
 1000  | 
| 
 n2 =  | 
 1000  | 
| 
 x =  | 
 410  | 
| 
 y =  | 
 500  | 
| 
 p1cap =  | 
 0.41  | 
| 
 p2cap =  | 
 0.5  | 
| 
 pcap =  | 
 0.455  | 
| 
 Zcal =  | 
 -4.041322996  | 
| 
 α =  | 
 0.05  | 
| 
 Zcrit  | 
 1.959963985  | 
| 
 p-value  | 
 0.00005315  | 
Decision:
Since | Zcal | > Zcrit, or equivalently, p-value < α, H0 is rejected.
Conclusion :
There is enough evidence to conclude that the difference in proportion is real. Answer 1
Part (b)
| 
 n1 =  | 
 1000  | 
| 
 n2 =  | 
 1000  | 
| 
 x =  | 
 220  | 
| 
 y =  | 
 240  | 
| 
 p1cap =  | 
 0.22  | 
| 
 p2cap =  | 
 0.24  | 
| 
 pcap =  | 
 0.23  | 
| 
 Zcal =  | 
 -1.062687855  | 
| 
 α =  | 
 0.05  | 
| 
 Zcrit  | 
 1.959963985  | 
| 
 p-value  | 
 0.2879235303  | 
Decision:
Since | Zcal | < Zcrit, or equivalently, p-value > α, H0 is accepted.
Conclusion :
There is not enough evidence to conclude that the difference in proportion is real, i.e., the observed difference is not real. Answer 2
DONE