Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
13 views

Exercises Lecture 5 Including Solutions

This document contains sample exercises and solutions for a lecture on hypothesis testing and confidence intervals. [1] A test of a hypothesis about a population mean against a sample mean yields a p-value of 7%. This means the 90% confidence interval does not contain the hypothesized mean value. [2] A 95% confidence interval is constructed for the population mean test score based on a sample of 1500 high school seniors. The interval range is given. [3] Several hypothesis tests are presented with hypotheses, sample data, conclusions, and p-value calculations shown.

Uploaded by

Asad Shahbaz
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
13 views

Exercises Lecture 5 Including Solutions

This document contains sample exercises and solutions for a lecture on hypothesis testing and confidence intervals. [1] A test of a hypothesis about a population mean against a sample mean yields a p-value of 7%. This means the 90% confidence interval does not contain the hypothesized mean value. [2] A 95% confidence interval is constructed for the population mean test score based on a sample of 1500 high school seniors. The interval range is given. [3] Several hypothesis tests are presented with hypotheses, sample data, conclusions, and p-value calculations shown.

Uploaded by

Asad Shahbaz
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

ECON1151: Exercises Lecture 5∗

1. Let y1 , . . . , yn be iid draws from a distribution with mean µ. A test of H0 : µ = 10 against


HA : µ 6= 10 using the usual t-statistic yields a p-value of 7%.

(a) Does the 90% confidence interval contain µ = 10?


(b) Can you determine if µ = 8 is contained in the 95% confidence interval? Briefly explain
your answers.

Solution:

(a) No, because the null hypothesis that the mean is 10 is rejected at the 10% level of
significance. This means that the 90% confidence interval does not contain µ = 10. See
the graph at the end of the document.
(b) No we cannot as this would require to calculate the test statistic under the null µ = 8.
This would require knowledge of the specific sample average obtained y and the standard
error of y (SE(y)).

2. A new version of the SAT is given to 1500 randomly selected high school seniors. The sample
mean test score is 1230 and the sample standard deviation is 145. Construct a 95% confidence
interval for the population mean test score for high school seniors.
Solution:
We simply use the corresponding formula for the 95% confidence interval which is:
 
y ± 1.96SE[y]
 
σY
= 1230 ± 1.96 √
1500
= [1230 ± 1.96 ∗ 3.744]
= [1230 ± 1.96 ∗ 3.744]
= [1230 ± 7.338]

∗
I am continuously updating these exercises. Please let me know of any mistakes or typos at rw6216z@gre.ac.uk

1
3. State the decision rule that would be used to test the following hypotheses (using a p-value).
Calculate the p-value and state your conclusion.

(a) H0 : µ0 ≥ 120 versus HA : µ0 < 120; y = 114.2, n=25, σY = 18 and α = 0.08


(b) H0 : µ0 = 42.9 versus HA : µ0 6= 42.9; y = 45.1, n=16, σY = 3.2 and α = 0.01
(c) H0 : µ0 ≤ 14.2 versus HA : µ0 > 14.2; y = 15.8, n=9, σY = 4.1 and α = 0.13

Solution:

(a) Decision rule: Reject H0 if p-value is less than α = 8%;


The p-value is the probability to obtain a sample draw which is more extreme (less in
line with the null hypothesis), than the current sample (i.e. the y = 114.2). Values which
are even less in line with the null of µ0 ≥ 120 are any values smaller than 114.2 and thus
we want to compute P (y < 114.2).
We rely on the CLT to conclude that y follows a normal distribution and thus we can
compute that probability by standardizing:
y−120
Z = 18/ √
25
= −1.61
P (Z < −1.61) = Φ(−1.61) = 0.0537
Since the p-value (5.37%) is less than then significance level (α = 8%) we reject the null
hypothesis.
(b) Decision rule: Reject H0 if p-value is less than α = 1%;
Equivalent reasoning as in the previous example but now we have a two sided test. So
observing a draw which is less in line with the null hypothesis is a draw which is farther
away from the mean under the null (42.9) than the current draw (45.1). So that means
observations which are larger than 45.1 or observations which are equally far away from
the mean but which lie below he mean (i.e. 45.1-42.9=2.2 and thus observations which
are below 40.7). We can compute these two probabilities again by standardization. And
we can exploit the symmetry of the standard normal distribution to compute one of them
and multiply by two:
p-value = P (Z < 40.7−42.9
√
3.2/ 16
) + P (Z > 45.1−42.9
√
3.2/ 16
) = P (Z < −2.75) + P (Z > 2.75) =
2Φ(−2.75) = 2 ∗ 0.3% = 0.6%
Since the p-value (0.6%) is smaller than the significance level (α = 1%) we reject the null
hypothesis.
(c) Decision rule: Reject H0 if p-value is less than α = 13%;
Similar to example b the p-value is the probability P (y > 14.2)
y−14.2
√ ) = P (Z > 1.17) = 1 − Φ(1.17) = 0.12
We standardize: P (Z > 4.1/ 9
Since the p-value (12%) is smaller than the significance level (α = 13%) we reject the
null hypothesis.

4. Suppose H0 : µ = 120 is tested agains HA : µ =6 120. If σ = 10 and n = 16, what p-value


is associated with the sample mean y = 122.3? Under what circumstances would the null be
rejected?
Solution:
y−µ
The p-value is P r(Z < −z) + P r(Z > z) where z = σy .
So lets compute z = 122.3−120
√ = 0.92
10/16
Since Φ(−0.92) = 17.88% and thus p = 2 ∗ 0.1788 = 0.3576

2
3

You might also like