In: Statistics and Probability
Consider common applications and misconceptions about a confidence interval of a proportion.
Explain how you arrived at your answers and conclusions, showing computations as appropriate.
Below is the scenario.
From a survey that Mika conducted for this veterinary medicine class, he found that 11 households out of a total of 30 households claimed to own at least one dog or cat. Mika constructed a 95% confidence interval in order to predict the population proportion, π, leading to 0.1867 ≤π ≤ 0.5466. (He used the Excel function CONFIDENCE.NORM.)
Mika concluded that:
1) It was Ok to use the sample proportion p = 11/30 = 0.3667 to construct this confidence interval;
2) the proportion of households in this whole state that would claim to own a dog or cat would be in the range of 36.67% +/- 5% = 31.67% - 41.67%; and
3) He was glad that he had not chosen a larger sample because a sample greater than n = 30 would have caused the confidence interval to become wider and less precise.
Solution1:
95% confidence interval for population proportion is
p^-z*sqrt(p^*(1-p^)/n,p^+z*sqrt(p^*(1-p^)/n
p^=x/n=11/30=0.3667
therefore
0.3667-1.96*sqrt(0.3667*(1-0.3667)/30,0.3667+1.96*sqrt(0.3667*(1-0.3667)/30
0.1943,0.5391
It was Ok to use the sample proportion p = 11/30 = 0.3667 to construct this confidence interval but
the 95% confidence interval constructed is wrong.
Solution2:
let u do hypothesis test for popualtion proportion
left tailed test
H0:p=0.50
Ha:p<0.50
alpha=0.05
z=p^-p/sqrt(p*(1-p)/n
=0.3667-0.5/sqrt(0.5*(1-0.5)/30)
=-1.460
p value=NORM.S.DIST(-1.46;TRUE)
p=0.072145
p>0.05
Fail to reject Null hypothesis
Accept null hypthesis
There is no sufficient statisitcal evidence at 5% level of significance to support the claim
he proportion of households in this whole state that would claim to own a dog or cat would be in the range of 36.67% +/- 5%
Solution3:
n>30 large samples
follows normal distribution according to the central limit theorem
the greater the sample size ,better is the approximation
As sample size increases,confidence interval becomes wider
and more precise.