Professional Documents
Culture Documents
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 1
Some Important Points
1. The farther the value of the true mean from 0 , the smaller for a
given and n. That is, for a specific sample size and level of
significance, larger differences in the mean are easier to detect than
smaller ones (i.e., larger power of test).
2. For a given mean difference and , will decrease as increases.
That is, to detect a specific difference in the mean, we can make the test
more powerful by increasing the sample size.
3. A decrease in the probability of one type of error always results in an
increase in the probability of the other provided that sample size n and
mean difference are fixed.
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 2
Two Ways to Draw Conclusions for Hypothesis Testing
P-value approach
o Can avoid above disadvantages.
o More commonly used and provided in published literature.
Both approaches lead to exactly the same conclusion for same value!
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 3
Comments on the Procedure
Hypothesis test aims to assess strength of evidence
against null hypothesis. Either we have enough
evidence to reject the null hypothesis or we do not
have enough evidence to reject.
Test statistic measures how much the data deviates
from the null hypothesis.
The critical region is determined assuming the null
hypothesis is true.
State a conclusion either by comparing the test
statistic to a critical region determined by the
decision maker or comparing the P-value to a
desired level of significance .
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 4
Close Connection between HT and CI
A close relationship exists between the hypothesis test (HT)
about any parameter, say , and the confidence interval (CI)
for . Specifically,
o If , is a 100(1-)% CI for , the test level of significance of
the two-sided hypothesis of 0 : = 0 vs 1 : 0 will lead to
rejection of 0 if and only if 0 is not in the 100(1-)% CI [, ].
o See the example on next two slides to confirm this!
HT Approach Solution:
H0: = 50 cm/sec vs H1: 50 cm/sec
0
51.350
Test statistic: 0 = = = 3.25
/ 2/ 25
Because significant level = 0.05, the critical region in
z-scale is z0.025 = 1.96 and z0.025 = 1.96
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 6
Example: CI and HT arrive the same conclusion
Because test statistic 0 0.025 = 1.96, i. e., the test statistic
falls in the rejection region, then we reject 0 : = 50 at 0.05
level of significant.
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 8
Test of the Mean of a Normal Distribution with Unknown Variance
Reject H0 if
t0<-t/2,n-1 or t0>t/2,n-1
-t/2,n-1 t/2,n-1 t-scale
Fail to reject H0 if
Rejection region Fail-to-reject region Rejection region
-t/2,n-1 t0 t/2,n-1
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 9
Test of the Mean of a Normal Distribution with Unknown Variance
-t,n-1 t,n-1
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 10
Test of Mean of a Normal Distribution with Unknown Variance
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 11
Example 9-6: Golf Club Design
An experiment was performed in which 15 drivers produced by
a particular club maker were selected at random and their
coefficients of restitution measured. It is of interest to determine
if there is evidence (for = 0.05) to support a claim that the
mean coefficient of restitution exceeds 0.82.
The observations are:
0.8411 0.8191 0.8182 0.8125 0.8750
0.8580 0.8532 0.8483 0.8276 0.7983
0.8042 0.8730 0.8282 0.8359 0.8660
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 14
Type II Error and Choice of Sample Size using OC Curve
Operating characteristic (OC) Curves plot for the t-test against a
parameter d for various sample size n. OC curves charts are provided
in Appendix of textbook for both two-sided and one-sided t-test.
0
= =
*Note that we replace with sample standard deviation s
because 2 is unknown in this case.
d = / = 0.02/0.02456 = 0.81.
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 16
OC Curve for one-sided t-test for = .
d=0.81, n =15
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 17
CASE III
Click Tests
Hypothesis to editonMaster title style
the Variance of a
Normal Distribution
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 18
Hypothesis Tests on the Variance of a Normal Distribution
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 19
Test of the Variance of a Normal Distribution
2
where/2,1
2
and1/1,1 are the
upper and lower
/2 /2 100 /2 percentage
points of the chi-square
distribution with n 1
degrees of freedom,
respectively.
2
2
1/2,1 /2,1
Critical region Fail-to-reject Critical region
region
2 2
Reject H0 if 02 < 1/2 ,n-1 or 0
2
> /2,1
2 2
Fail to reject H0 if 1/2,1 2 /2,1
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 20
Test of the Variance of a Normal Distribution
One sided hypotheses:
H0: 2 = 02 H0: 2 = 02
H1: 2 < 02 H1: 2 > 02
2 2
1,1 ,1
Fail to reject H0 if 02 2
1 Fail to reject H0 if 02 ,1
2
,n-1
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 21
Example 9-8 Automated Filling
An automated filling machine is used to fill bottles with liquid
detergent. A random sample of 20 bottles results in a sample
variance of fill volume of s2 = 0.0153 (fluid ounces)2. If the
variance of fill volume exceeds 0.01 (fluid ounces)2, an
unacceptable proportion of bottles will be underfilled or
overfilled. Is there evidence in the sample data to suggest that
the manufacturer has a problem with underfilled or overfilled
bottles? Use = 0.05, and assume that fill volume has a
normal distribution.
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 24
Large-sample Hypothesis Tests on a Population Proportion
Let X be the number of observations in a random sample of size
that belongs to a class of interest. Then, = / is a point
estimator of the proportion of the population that belong to
this class.
If the null hypothesis 0 : = 0 is true, we have
~ 0 , 0 1 0 approximately
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 25
Large-sample Hypothesis Tests on a Population Proportion
Two-sided hypothesis:
H0: p = p0 H 1 : p p0
-Z/2 +Z/2
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 26
Large-sample Hypothesis Tests on a Population Proportion
One-sided hypotheses:
H0: = 0 H0: = 0
H1: < 0 H1: > 0
-Z Z-scale Z Z-scale
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 27
P-value, Type II Error and Choice of Sample Size
P-value: same as the hypothesis test in Case I
2[1 (| z0 |)] two tailed test
P = 1 ( z0 ) upper tailed test
( z )
0 lower tailed test
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 28
Example 9-10 Automobile Engine Controller
A semiconductor manufacturer produces controllers used in
automobile engine applications. The customer requires that the
process fallout or fraction defective at a critical manufacturer step
not exceed 0.05 and that the manufacturer demonstrate process
capability at this level of quality using = 0.05. The semiconductor
manufacturer takes a random sample of 200 devices and finds that
four of them are defective. Can the manufacturer demonstrate
process capability for the customer?
Parameter of interest: the process fraction defective .
Test hypotheses: 0 : = 0.05 vs 1 : < 0.05
0 4200 0.05
Test statistic is 0 = = = 1.95
0 10 200(0.05)(0.95)
Use P-value approach: Since P-value= 0 = 0.0256, which is less
than 0.05, we reject 0 and conclude that there is sufficient evidence
showing that the process fraction defective < 0.05.
Alternatively, use the fixed level of significance approach: since z0 =
1.95 < 0.05 = 1.64, we reject 0 .
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 29
Typical Steps to Perform a Hypothesis Test
1. Identify the parameter of interest .
2. State the null hypothesis, H0 and the alternative hypothesis
H1 about .
3. Select the appropriate test statistic and compute the value of
the test statistic from the data.
4. Find either the critical values/region from the fixed level of
significance or the P-value based on the computed test
statistics from Step 3.
5. Make decision
1. Reject H0 with a specified level of significance if the value of the test
statistic is in the critical region or fail to reject H0 otherwise.
2. Reject H0 if the P-value is less than a specified level of significance,
or fail to reject H0 otherwise.
6. Interpret the decision
Based on the decision made in Step 5, we state a conclusion in the
context of the original problem.
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 30
Dr. Yaping Wang, Industrial Engineering Dept. INDE 2333 Lecture, Spring 2017 31