Which side you stand for..
Session # 16 “Better to acquit 100 guilty
Inferential Decision Making
Inferential Data Analysis-I
1 2
Hypothesis Testing
(Decision making with Data)
3 4
Hypothesis Testing
 Hypothesis testing can be used to determine whether
a statement about the value of a population parameter
should or should not be rejected.
 The null hypothesis, denoted by H0 , is a tentative
assumption about a population parameter.
 The alternative hypothesis, denoted by Ha , is the
opposite of what is stated in the null hypothesis.
 The hypothesis testing procedure uses data from a
sample to test the two competing statements
indicated by H0 and Ha.
Note: Decision is based on rejection or non-
rejection of Null Hypothesis ( H0 )
5 6
than convict one innocent .”
-Justice Oliver Wendell Holmes,
Supreme Court of the United States (1882-1899)
Hypothesis Testing
 Developing Null and Alternative Hypotheses
 Type I and Type II Errors
 Population Mean: s Known
 Population Mean: s Unknown
 Population Proportion
 Hypothesis Testing and Decision Making
 Calculating the Probability of Type II Errors
 Determining the Sample Size for
a Hypothesis Test About a Population mean
Developing Null and Alternative Hypotheses
• It is not always obvious how the null and alternative
hypotheses should be formulated.
• Care must be taken to structure the hypotheses
appropriately so that the test conclusion provides
the information the researcher wants.
• The context of the situation is very important in
determining how the hypotheses should be stated.
• In some cases, it is easier to identify the alternative
hypothesis first. In other cases, the null is easier.
• Correct hypothesis formulation will take practice.
1
 Developing Null and Alternative Hypotheses Developing Null and Alternative Hypotheses

 Alternative Hypothesis as a Research Hypothesis  Alternative Hypothesis as a Research Hypothesis

• Many applications of hypothesis testing involve • Example:
an attempt to gather evidence in support of a A new teaching method is developed that is
research hypothesis. believed to be better than the current method.
• In such cases, it is often best to begin with the • Alternative Hypothesis:
alternative hypothesis and make it the conclusion The new teaching method is better.
that the researcher hopes to support.
• Null Hypothesis:
• The conclusion that the research hypothesis is true
The new method is no better than the old method.
is made if the sample data provide sufficient
evidence to show that the null hypothesis can be
rejected.

7 8

Developing Null and Alternative Hypotheses Developing Null and Alternative Hypotheses

 Alternative Hypothesis as a Research Hypothesis  Alternative Hypothesis as a Research Hypothesis

• Example: • Example:
A new sales force bonus plan is developed in an A new drug is developed with the goal of lowering
attempt to increase sales. blood pressure more than the existing drug.
• Alternative Hypothesis: • Alternative Hypothesis:
The new bonus plan increase sales. The new drug lowers blood pressure more than
the existing drug.
• Null Hypothesis:
The new bonus plan does not increase sales. • Null Hypothesis:
The new drug does not lower blood pressure more
than the existing drug.

9 10

Developing Null and Alternative Hypotheses Developing Null and Alternative Hypotheses

 Null Hypothesis as an Assumption to be Challenged  Null Hypothesis as an Assumption to be Challenged

• We might begin with a belief or assumption that • Example:
a statement about the value of a population The label on a soft drink bottle states that it
parameter is true. contains 67.6 fluid ounces.
• We then using a hypothesis test to challenge the • Null Hypothesis:
assumption and determine if there is statistical The label is correct. m > 67.6 ounces.
evidence to conclude that the assumption is • Alternative Hypothesis:
incorrect. The label is incorrect. m < 67.6 ounces.
• In these situations, it is helpful to develop the null
hypothesis first.

11 12

2
 Summary of Forms for Null and Alternative Null and Alternative Hypotheses
Hypotheses about a Population Mean
 The equality part of the hypotheses always appears  Example: Metro EMS
in the null hypothesis. A major west coast city provides one of the most
 In general, a hypothesis test about the value of a comprehensive emergency medical services in the
population mean m must take one of the following world. Operating in a multiple hospital system
three forms (where m0 is the hypothesized value of with approximately 20 mobile medical units, the
the population mean). service goal is to respond to medical emergencies
with a mean time of 12 minutes or less.
H 0 : m  m0 H 0 : m  m0 H 0 : m  m0 The director of medical services wants to
H a : m  m0 H a : m  m0 H a : m  m0 formulate a hypothesis test that could use a sample
of emergency response times to determine whether
One-tailed One-tailed Two-tailed or not the service goal of 12 minutes or less is being
(lower-tail) (upper-tail) achieved. This will help him to improve the service
quality.

13 14

Null and Alternative Hypotheses Type I Error

The emergency service is meeting  Because hypothesis tests are based on sample data,
H0: m we must allow for the possibility of errors.
the response goal; no follow-up
action is necessary.  A Type I error is rejecting H0 when it is true.
 The probability of making a Type I error when the
The emergency service is not
Ha:m null hypothesis is true as an equality is called the
meeting the response goal;
level of significance.
appropriate follow-up action is
necessary.  Applications of hypothesis testing that only control
the Type I error are often called significance tests.
where: m = mean response time for the population
of medical emergency requests

15 16

Type II Error Type I and Type II Errors

 A Type II error is NOT REJECTING H0 when it is false.

Population Condition in reality
 It is difficult to control for the probability of making H0 True H0 False
a Type II error. (Conclude that the person is NOT (Conclude that the person is
Conclusion GUILTY) GUILTY)
 Statisticians avoid the risk of making a Type II
Do not reject H0
error by using “do not reject H0” and not “accept H0”. Correct
(Conclude that the person is not Type II Error
Decision
GUILTY )
Reject H0 Correct
(Conclude that the person is not Type I Error Decision
GUILTY)

H0 : The person is Not Guilty

17 18

3
 One Population
Scenario
 If population standard
---------------------- deviation known, then
our test statistics is a z-
Inference test.
about • Sample(s):
collected from
 If population S.D is
unknown, our test
Single Population one population
only. statistics is t-test with n-1
Parameter • Hypothesis Test
for: sample mean.
degrees of freedom.
 Hypothesis test about
population proportion
use z-test.

19 20

Critical Value Approach to Lower-Tailed Test About a Population Mean:

One-Tailed Hypothesis Testing s Known
 The test statistic z has a standard normal probability  Critical Value Approach
distribution.
 We can use the standard normal probability Sampling
distribution table to find the z-value with an area distribution
of a in the lower (or upper) tail of the distribution. x - m0
of z 
 The value of the test statistic that established the Reject H0 s/ n
boundary of the rejection region is called the
critical value for the test. a 
Do Not Reject H0
 The rejection rule is:
• Lower tail: Reject H0 if z < -za z
• Upper tail: Reject H0 if z > za -za = -1.28 0

21 22

Upper-Tailed Test About a Population Mean: p-Value Approach to

s Known One-Tailed Hypothesis Testing
 Critical Value Approach  The p-value is the probability, computed using the
test statistic, that measures the support (or lack of
support) provided by the sample for the null
Sampling
hypothesis.
distribution
x - m0  If the p-value is less than or equal to the level of
of z 
s/ n Reject H0 significance a, the value of the test statistic is in the
rejection region.
a
Do Not Reject H0  Reject H0 if the p-value < a .

z
0 za = 1.645

23 24

4
 Suggested Guidelines for Interpreting p-Values Lower-Tailed Test About a Population Mean:
s Known
 Less than .01  p-Value Approach p-Value < a ,
Overwhelming evidence to conclude Ha is true. so reject H0.

 Between .01 and .05

a = .10 Sampling
Strong evidence to conclude Ha is true. distribution
x - m0
 Between .05 and .10 of z 
s/ n
Weak evidence to conclude Ha is true. p-value
7
 Greater than .10
Insufficient evidence to conclude Ha is true.
z
z = -za = 0
-1.46 -1.28

25 26

Upper-Tailed Test About a Population Mean: Steps of Hypothesis Testing

s Known
p-Value Approach p-Value < a ,
 Step 1. Develop the null and alternative hypotheses.
so reject H0.
Sampling Step 2. Specify the level of significance a.
distribution Step 3. Collect the sample data and compute the
x - m0 a = .04
of z  value of the test statistic.
s/ n
p-Value Approach
p-Value
Step 4. Use the value of the test statistic to compute the
 p-value.
Step 5. Reject H0 if p-value < a.
z
0 za = z=
1.75 2.29

27 28

Steps of Hypothesis Testing One-Tailed Tests About a Population Mean:

Critical Value Approach
Step 4. Use the level of significanceto determine the
critical value and the rejection rule.
Step 5. Use the value of the test statistic and the rejection
rule to determine whether to reject H0.
s Known
 Example: Metro EMS
The response times for a random sample of 40
medical emergencies were tabulated. The sample
mean is 13.25 minutes. The population standard
deviation is believed to be 3.2 minutes.
The EMS director wants to perform a hypothesis
test, with a .05 level of significance, to determine
whether the service goal of 12 minutes or less is
being achieved.

29 30

5
 One-Tailed Tests About a Population Mean: One-Tailed Tests About a Population Mean:
s Known s Known
 p -Value and Critical Value Approaches  Critical Value Approach

1. Develop the hypotheses. H0: m 4. Determine the critical value and rejection rule.
Ha:m
For a = .05, z.05 = 1.645
2. Specify the level of significance. a = .05 Reject H0 if z > 1.645

3. Compute the value of the test statistic. 5. Determine whether to reject H0.

x - m 13.25 - 12 Because 2.47 > 1.645, we reject H0.

z   2.47
s / n 3.2 / 40 There is sufficient statistical evidence
to infer that Metro EMS is not meeting
the response goal of 12 minutes.

31 32

Upper-Tailed Test About a Population Mean:

s Known
 Critical Value Approach

Sampling
distribution
x -m
of z  s / n0
=2.47 Reject H0

a
Do Not Reject H0

z
2.47
0 za = 1.645

33