EC404 - Monsoon 2016 - Introduction to Statistics and Econometrics

Problem Set 2 Answers 7, 8 9, 13.

Archana Aggarwal
307, SSS-II

7.
a) H0: = 10 vs Ha: 10. The t statistic is {(11.3-10)(10)'3.1} = 1.326. From the t-table we
can read off the critical value from the tn-1 distribution for a two-tailed test at a 5% significance level.
For the t distribution with 9 degrees of freedom, that value is 2.26. For a two-tailed test, if the absolute
value of calculated t statistic is larger than the critical value, we can reject the null hypothesis. Since the
t statistic is smaller than the critical value in this case, we cannot reject the null hypothesis. That is, there
is no evidence that the average purchase is different from 10 liters.
b)
The p-value for the test is the probability that we get a sample statistic at least as adverse to the
null hypothesis as the mean in our sample. Once we have standardized the sample mean by subtracting
off its hypothesized mean and dividing by its standard error, we have a variable that is distributed tn-1.
(Because the population standard deviation is unknown, and n is small.) So, we want to know the
probability that a variable with a t9 distribution takes on an absolute value greater than 1.326.
H0: = 10 vs Ha: 10.

c)
The t statistic is {(10.3-10)(10)'3.1} = 0.306. The critical value is the same as in part a. Since
the calculated t-statistic is less than the critical value, we cannot reject the null hypothesis.

8.
Ans: (a) A one-sided t-test was performed.
H0: The salt contained in a certain brand of potato chips is what label claimed.
Ha: The salt contained in a certain brand of potato chips is more than the label claimed.
Or
H0: = 0 vs Ha: > 0
(b) For a one-sided t-test, the critical value for significance level at 0.1 and 19 degrees of freedom is
1.328. Since 1.6 > 1.328, the result is significant at the 0.1 level.
Similarly, for one-sided t-test, the critical value for significance level at 0.01 and 19 degrees of freedom is
2.539. Since 1.6 < 2.539, the result is not significant at the 0.01 level.

9.

(a) c = 15 / 4 .
(b)

Page 1 of 2

( !4< x, y < 4),

13.
Ans: (a) (i) Let X (ml) be the volume contained in a can. The sample size (500) is very large,
so the Central Limit Theorem assures us that, to a very good approximation, the underlying distribution
of the sample mean can be taken as Normal (alternatively, it does not seem unreasonable to assume
underlying Normality for X itself, though this is not stated in the question).
We have x = 302 and s = 0.5. Let be the true (long-term) mean of X. A 95% confidence interval for
is given by 302 (1.96 0.5/%500), i.e. 302 0.044, so the interval is (301.96, 302.04). For a 99%
interval replace 1.96 by 2.576 to give 302 0.058, so the interval is (301.94, 302.06).
(ii) Let n be the sample size. It is required that 1.96 0.5/%n < 0.05, i.e.

So take n = 385. [Note that n is still very large, so the Central Limit Theorem still comfortably applies. Had
n turned out to be small, consideration would have had to be given to using a t value rather than 1.96 from
N(0, 1) and that would have required an assumption of Normality for the underlying distribution.]
(b)

Let p be the true proportion of the population having defective ring pulls. The sample estimate is
. The variance of
is estimated as

so the standard deviation is estimated by %0.00004685 = 0.00684.

Hence an approximate 95% confidence interval for p is given by (0.024 1.960.00684), i.e. it is (0.011,
0.037). With 95% probability of being right, we can say that the range 1.1% to 3.7% covers the true
percentage of cans having defective ring pulls.

Page 2 of 2