Expenditure on one category of food is estimated on the basis of a family expenditure survey. The amounts spent are surpassed to vary around an unknown mean, with a standard deviation of 6 dollars. For a sample of 80 families, the average expenditure is 43 dollars.
a) Compute a 95% confidence interval for the mean family expenditure?
b) How large of a sample is required to estimated the mean expenditure to within 50 cents?
c) Do we need to assume that the individual expenditure follow a normal distribution?
Compute a 95% confidence interval for the mean family expenditure?
In this case n = 80, x-bar = 43, s = 6
Standard error of the mean = 6 /√80 = 0.67
The 95% confidence interval = (μ - 1.96 x SE) - (μ + 1.96 x SE), and 1.96 x 0.67 = 1.32
Hence (43 - 1.32) %26lt; μ %26lt; (43 + 1.32), i.e. 41.68 %26lt; μ %26lt; 44.32
b. (This corrects an error in my original answer to this question.)
To get an estimate to within 50 cents with a 95% confidence interval it is necessary to apply the z-value as determined above to the standard error of the mean.
In other words, we require that 1.96 x 6 / √n = 0.5
So √n = (1.96 x 6) / 0.5 = 23.52 or n = 553
So with the same sample mean, for example, this would give the confidence interval as
42.50 %26lt; μ %26lt; 43.50
--------------------------------------...
c. No. The sample means are normally distributed in all cases, irrespective of the shape of the population distribution.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment