Introduction To Topics in Hypothesis Testing

TEACHING BASIC STATISTICS ….
Research Problem: How effective is

Minoxidil in treating male pattern baldness?
This can be answered
Specific Objectives: by ESTIMATION
Introduction to Topics 1. To estimate the population proportion of
patients who will show new Thishair
cangrowth after
be answered by
in Hypothesis Testing being treated with Minoxidil.
HYPOTHESIS TESTING
2. To determine whether treatment using Minoxidil
is better than the existing treatment that is
known to stimulate hair growth among 40% of
patients with male pattern baldness.
Training on Teaching
Basic Statistics for Note: Most of the Slides were prepared by Prof.
Question: How do we achieve these objectives
Tertiary Level Teachers Josefina Almeda of the School of Statistics, UP
Diliman
using inferential statistics?
Summer 2008
Session 7.3
TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….
Two Areas of Inferential Statistics What is Hypothesis Testing?
 Estimation  Hypothesis testing is an area of

statistical inference in which one
Point Estimation evaluates a conjecture about some
Interval Estimation characteristic of the parent population
based upon the information contained
in the random sample.
 Hypothesis Testing  Usually the conjecture concerns one
of the unknown parameters of the
population.
Session 7.2 Session 7.4

What is a Hypothesis? Example of a Hypothesis

This drug is guaranteed
 A hypothesis is a to change in cholesterol  The proportion of registered voters in
claim or statement levels (on the average) Quezon City favoring Candidate A
by more than 30%!
about the population exceeds 0.60.
parameter
 Examples of parameters
Note: The parameter of interest here
are population mean
or proportion is p which is the proportion of registered
 The parameter must voters in Quezon City favoring
be identified before analysis Candidate A.
© 1984-1994 T/Maker Co.

Example of Hypothesis Things to keep in mind
 The mean body temperature for patients  Analyze a sample in an attempt to

admitted to elective surgery is not equal to distinguish between results that can
37.0oC. easily occur and results that are
highly unlikely
Note: The parameter of interest here is We can explain the occurrence of highly
 which is the mean body unlikely results by saying either that a rare
event has indeed occurred or that things
temperature for patients admitted to aren’t as they are assumed to be.
elective surgery.
Example to illustrate the basic

Solution to Example:
approach in testing hypothesis
A product called “Gender Choice” a. We normally expect around 50 girls in
claims that couples “increase their 100 births. The result of 52 girls is
chances of having a boy up to 85%, a close to 50, so we should conclude
girl up to 80%.” Suppose you conduct that Gender Choice is effective. If the
an experiment that includes 100
couples who want to have baby girls, 100 couples used no special methods
and they all follow the Gender Choice of gender selection, the result of 52
to have a baby girl. girls could easily occur by chance.
Con’t of Example Solution to Example
Using common sense and no real b. The result of 97 girls in 100 births is
formal statistical methods, what extremely unlikely to occur by chance. It
should you conclude about the could be explained in one of two ways:
Either an extremely rare event has occurred
effectiveness of Gender Choice if by chance, or Gender Choice is effective
the 100 babies include because of the extremely low probability of
a. 52 girls getting 97 girls by chance, the more likely
explanation is that the product is effective.
b. 97 girls

Explanation of Example Null Hypothesis

When actually conducting the
test, we operate under the
We should conclude that the product is assumption that the
  denoted by Ho parameter equals some
effective only if we get significantly more
 the statement being tested specific value.
girls that we would expect under normal
circumstances.  it represents what the experimenter doubts to
 Although the outcomes of 52 girls and 97 be true
girls are both “above average,” the result of  must contain the condition of equality and
52 girls is not significant, whereas 97 girls must be written with the symbol =,  , or 
does constitute a significant result.
For the mean, the null hypothesis will be

stated in one of these three possible forms:
 Ho:  = some value

Components of a Formal  Ho:   some value
Hypothesis Test  Ho:   some value
Note: the value of  can be obtained

from previous studies or from
knowledge of the population

For the mean, the alternative hypothesis will be

Example of Null Hypothesis stated in only one of three possible forms:
 The null hypothesis corresponding to  Ha:   some value

the common belief that the mean body  Ha:  > some value
temperature is 37oC is expressed as
 Ha:  < some value
Ho:   37 o C
 We test the null hypothesis directly in
the sense that we assume it is true Note: Ha is the opposite of Ho. For example,
and reach a conclusion to either reject if Ho is given as  = 37.0, then it follows
that the alternative hypothesis is given
Ho or fail to reject Ho.
by Ha:   37.0.
Alternative Hypothesis Note About Using  or  in Ho:
 denoted by Ha  Even though we sometimes express Ho

 Is the statement that must be true if the with the symbol  or  as in Ho:  37.0
null hypothesis is false
or Ho:   37.0, we conduct the test by
 the operational statement of the theory assuming that  = 37.0 is true.
that the experimenter believes to be true
and wishes to prove  We must have a single fixed value for 
 Is sometimes referred to as the research so that we can work with a single

hypothesis distribution having a specific mean.

Note About Stating Your Own

Some Notes:
Hypotheses:
 In this context of trying to support the
goal of the research, the alternative
 If you are conducting a research study hypothesis is sometimes referred to as
and you want to use a hypothesis test the research hypothesis.
to support your claim, the claim must  Also in this context, the null hypothesis
be stated in such a way that it is assumed true for the purpose of
conducting the hypothesis test, but it is
becomes the alternative hypothesis, so hoped that the conclusion will be
it cannot contain the condition of rejection of the null hypothesis so that
equality. the research hypothesis is supported.
Example in Stating your Hypothesis Example of a Research Problem

Suppose that the government is deciding
If you believe that your brand of whether or not to approve the manufacturing of a
new drug. A drug is to be tested to find out if it
refrigerator lasts longer than the mean
can dissolve cholesterol deposits in the heart’s
of 14 years for other brands, state the arteries. A major cause of heart diseases is the
claim that  > 14, where  is the hardening of the arteries caused by the
mean life of your refrigerators. accumulation of cholesterol. The Bureau of
Food and Drug (BFaD) will not allow the
Ho:  = 14 vs. Ha:  > 14 marketing of the drug unless there is strong
evidence that it is effective.

Con’t of Research Problem Con’t of Research Problem

A random sample of 98 middle-aged men has been To measure the effectiveness of the drug for each
selected for the experiment. Each man is given a middle-aged man, we can look at the percent
standard daily dosage of the drug for 2 consecutive change in cholesterol levels experienced by all
weeks. Their cholesterol levels are measured at the middle-aged men who took the drug before and
beginning and at the end of the test. The interest is to after they took the drug.
determine if the intake of the drug lead to reduced We summarize effectiveness in terms of the
cholesterol levels; that is, a hypothesis test will have to population mean.
be performed to determine if the drug is effective or not.
Based on the results of the experiment, the director of Let  be the population mean of the percent
BFaD will decide whether to release the drug change in cholesterol levels. BFaD decides to
to the public or postpone its release and classify the drug as effective only if, on the
request for more research. average, it reduces the cholesterol levels by
more than 30% (  30%).
Con’t of Research Problem Stating the Null Hypothesis

To perform a statistical hypothesis test, we
must firstly identify the parameter of interest,
and have some educated guess about the true The null hypothesis represents no practical
value of the parameter. In the case of the BFaD change in cholesterol levels before and after
example, the possible states of the drug’s the drug use. In terms of , we say
effectiveness are referred to as hypotheses.
Because the director wants only to know Ho:   30%
whether it is effective or not, either of the
following hypotheses applies.
The drug is ineffective.

The drug is effective. Session 7.26 Session 7.28
Note About Testing the Validity of

Stating the Alternative Hypothesis
Someone Else’s Claim
Ha:   30%
Sometimes we test the validity of someone else’s claim,
Whenever sample results fail to support the such as the claim of the Coca Cola Bottling Company that
“the mean amount of Coke in cans is at least 355 ml,”
null hypothesis, the conclusion we accept is
which becomes the null hypothesis of Ho:   355
the alternative hypothesis. In our illustration,
if results from the sample of percentage In this context of testing the validity of someone else’s
change in cholesterol levels fail to support Ho, claim, their original claim sometimes becomes the null
hypothesis (because it contains equality), and
then the director concludes Ha and says that it sometimes becomes the alternative
the drug is effective. hypothesis (because it does not contain
the equality).
What is a Test of Significance? Two Types of Errors

• A test of significance is a problem of
deciding between the null and the alternative
hypotheses on the basis of the information
contained in a random sample.  Type I Error
• The goal will be to reject Ho in favor of Ha,

because the alternative is the hypothesis that  Type II Error
the researcher believes to be true. If we are
successful in rejecting Ho, we then declare
the results to be “significant”.

Type I Error Type II Error

 The mistake of rejecting the null hypothesis
when it is true.
 It is not a miscalculation or procedural  The mistake of failing to reject the
misstep; it is an actual error that can occur null hypothesis when it is false.
when a rare event happens by chance.
 The probability of rejecting the null  The symbol  (beta) is used to
hypothesis when it is true is called the represent the probability of a type II
significance level (  ).
 The value of  is typically predetermined,
error.
and very common choices are  = 0.05
and  = 0.01.
Examples of Type I Error Examples of Type II Errors
1. The mistake of rejecting the null 1. The mistake of failing to reject the
hypothesis that the mean body null hypothesis (  = 37.0) when it
temperature is 37.0 when that mean is actually false (that is, the mean
is really 37.0. is not 37.0).
2. BFaD allows the release of an 2. BFaD does not allow the release
ineffective medicine of an effective drug.

Summary of Possible Decisions in Controlling Type I and Type II

Hypothesis Testing Errors
True Situation  The experimenter is free to determine . If the
test leads to the rejection of Ho, the researcher
The null The null can then conclude that there is sufficient evidence
hypothesis is hypothesis is supporting Ha at  level of significance.
true. false.  Usually,  is unknown because it’s hard to
Decision We decide TYPE I error CORRECT calculate it. The common solution to this difficulty
to reject the (rejecting a true decision is to “withhold judgment” if the test leads to the
null null hypothesis) failure to reject Ho.
hypothesis.
  and  are inversely related. For a fixed
We fail to CORRECT TYPE II error sample size n, as  decreases  increases.
reject the decision (failing to reject
null a false null
hypothesis. hypothesis)
Analogy to Decisions in Hypothesis Controlling Type I and Type II

Testing Errors
Trial  In almost all statistical tests, both  and 
The Truth can be reduced by increasing the sample
size.
Verdict Innocent Guilty
 Because of the inverse relationship of 
Innocent Correct Error and , setting a very small  should also
be avoided if the researcher cannot
afford a very large risk of committing a
Guilty Error Correct Type II error.
)

Controlling Type I and Type II Example to illustrate Type I and

Errors Type II Errors
 The choice of  usually depends on the
consequences associated with making a
Consider M&Ms (produced by Mars,
Type I error. Inc.) and Bufferin brand aspirin tablets
(produced by Bristol-Myers Products).
Common Choices Consequences of
of  Type I error The M&M package contains 1498
candies. The mean weight of the
0.01 or smaller very serious individual candies should be at least
0.05 moderately serious 0.9085 g., because the M&M package
0.10 not too serious is labeled as containing 1361 g.
Controlling Type I and Type II Example to illustrate Type I and

Errors Type II Errors
 The usual practice in research and industry
is to determine in advance the values of  The Bufferin package is labeled as
and n, so the value of  is determined. holding 30 tablets, each of which contains
 Depending on the seriousness of a type I 325 mg of aspirin.
error, try to use the largest  that you can Because M&Ms are candies used for
tolerate.
enjoyment whereas Bufferin tablets are
 For type I errors with more serious drugs used for treatment of health
consequences, select smaller values of  .
Then choose a sample size n as large as is problems, we are dealing with two very
reasonable, based on considerations of different levels of seriousness.
time, cost, and other such relevant factors.
Example to illustrate Type I and Example to illustrate Type I and

Type II Errors Type II Errors
Consequently, in testing the claim that
If the M&Ms don’t have a population  = 0.9085 g for M&Ms, we might choose
mean weight of 0.9085 g, the  = 0.05 and a sample size of n = 100.
consequences are not very serious, In testing the claim of  = 325 mg for
Bufferin tablets, we might choose  = 0.01
but if the Bufferin tablets don’t have a and a sample size of n = 500.
mean of 32.5 mg of aspirin, the
The smaller significance level  and large sample
consequences could be very serious. size n are chosen because of the more serious
consequences associated with the
commercial drug.
Example to illustrate Type I and The Test Statistic - a statistic computed from the
sample data that is especially sensitive to the differences
Type II Errors between Ho and Ha
• The test statistic should tend to take on certain
If the M&Ms have a mean that is too values when Ho is true and different values when
large, Mars will lose some money but Ha is true.
consumers will not complain. • The decision to reject Ho depends on the value of
In contrast, if the Bufferin tablets have the test statistic
too much aspirin, Bristol-Myers could • A decision rule based on the value of the test
be faced with consumer lawsuits. statistic:
Reject Ho if the computed value of the
test statistic falls in the region of rejection.
Region of Rejection or Critical Region- the set Types of Tests

of all values of the test statistic which will lead to the
rejection of Ho  Two-tailed Test. If we are primarily concerned with
deciding whether the true value of a population
parameter is different from a specified value, then
Factors that Determine the Region of the test should be two-tailed. For the case of the
Rejection mean, we say Ha:   0.
 the behavior of the test statistic if the null hypotheses  Left-tailed Test. If we are primarily concerned with
deciding whether the true value of a parameter is
were true less than a specified value, then the test should be
left-tailed. For the case of the proportion, we say
 the alternative hypothesis: the location of the region Ha: P  P0.
of rejection depends on the form of Ha  Right-tailed Test. If we are primarily concerned
 level of significance (): the smaller  is, with deciding whether the true value of a parameter
is greater than a specified value, then we should
the smaller the region of rejection use the right-tailed test. For the case of the
standard deviation, we say Ha:   0.
Level of Significance
Critical Value/s
and the Rejection Region
 the value or values that separate the 
critical region from the values of the Ho:   30 Critical
Value(s)
test statistic that would not lead to Ha:  < 30
rejection of the null hypothesis. Rejection 0
Regions 
Ho:   30
 It depends on the nature of the null Ha:  > 30
hypothesis, the relevant sampling 0
/2
distribution, and the level of Ho:   30
significance. Ha:   30
0
US1
The p-value - the smallest level of significance at

which Ho will be rejected based on the
Conclusions in Hypothesis Testing
information contained in the sample 1. Fail to reject the null hypothesis Ho.
2. Reject the null hypothesis Ho.
Notes:
An Alternative Form of Decision Rule  Some texts say “accept the null hypothesis”
(based on the p-value) instead of “fail to reject the null hypothesis.”
 Whether we use the term accept or fail to
reject, we should recognize that we are not
Reject Ho if the p-value is less than or proving the null hypothesis; we are merely
equal to the level of significance (). saying that the sample evidence is not strong
enough to warrant rejection of the null
hypothesis.
Example of Making Decisions

Wording of Final Conclusion
Using the p-value
Start
If the level of significance =0.05, Yes
Does (Reject (This is the
the Ho) “The sample data only case in
original Do you reject support the claim which the
p-value Decision claim contain
the condition
Ho? that….(original original claim
is
of claim).” supported.)
equality
No
“There is not sufficient
0.01 Reject Ho.
No (Original claim (Fail to
does not contain Reject sample evidence to
equality and Ho) support the claim
0.05 Reject Ho. becomes Ha) that….(original claim).”
0.10 Do not reject Ho

Slide 55
US1 The term accept is somewhat misleading because it seems to imply incorrectly that the null hypothesis has been proved. The phrase fail to reject says more
correctly that the available evidence isn't strong enough to warrant rejection of the null hypothesis.
UP Stat, 27 Jan 2005

Example in Making Final

Wording of Final Conclusion
Conclusion
Start
“There is sufficient (This is the
Yes only case in
evidence to which the
Does (Reject
the
original
Ho)
warrant rejection of
the claim
original claim
is rejected.)
 If, on the other hand, you claim that the
Do you reject
claim contain
Ho? that….(original mean body temperature is 37.0oC, that
is  = 37.0, you will either reject or fail
the condition claim).”
of
equality
No
Yes (Original claim (Fail to
“There is not sufficient to reject the claim; in either case, you
contains equality sample evidence to
and becomes Ho)
Reject
Ho) warrant rejection of the will not support the original claim.
claim that….(original
claim).”

A Summary of the Steps in Hypothesis Testing
Example in Making Final
Determine the objectives of the experiment
Conclusion (responsibility of the experimenter).
 If you want to justify the claim that the 1. State the null and alternative hypotheses.
mean body temperature is different 2. Decide on a level of significance, .
from 37.0oC, then make the claim that Determine the testing procedure and methods of
  37.0. This claim will be an analysis (responsibility of the statistician).
alternative hypothesis that will be 3. Decide on the type of data to be collected and
supported if you reject the null choose an appropriate test statistic and testing
hypothesis of Ho:  = 37.0. procedure.

TEACHING BASIC STATISTICS ….
Con’t of Steps in Hypothesis

Testing
4. State the decision rule.
5. Collect the data and compute for the value of
the test statistic using the sample data.
6. a) If decision rule is based on region of
rejection: Check if the test statistic falls in the
region of rejection. If yes, reject Ho.
b) If decision rule is based on p-value:
Determine the p-value. If the p-value is less
than or equal to , reject Ho.
7. Interpret results.
Session 7.61

Introduction To Topics in Hypothesis Testing

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Introduction To Topics in Hypothesis Testing

Uploaded by

Copyright:

Available Formats

TEACHING BASIC STATISTICS ….

Research Problem: How effective is

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Two Areas of Inferential Statistics What is Hypothesis Testing?

 Estimation  Hypothesis testing is an area of

Session 7.2 Session 7.4

What is a Hypothesis? Example of a Hypothesis

Session 7.5 Session 7.7

Example of Hypothesis Things to keep in mind

 The mean body temperature for patients  Analyze a sample in an attempt to

Example to illustrate the basic

Session 7.9 Session 7.11

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Con’t of Example Solution to Example

Session 7.10 Session 7.12

Explanation of Example Null Hypothesis

Session 7.13 Session 7.15

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

For the mean, the null hypothesis will be

 Ho:  = some value

Note: the value of  can be obtained

Session 7.14 Session 7.16

For the mean, the alternative hypothesis will be

 The null hypothesis corresponding to  Ha:   some value

Session 7.17 Session 7.19

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Alternative Hypothesis Note About Using  or  in Ho:

 denoted by Ha  Even though we sometimes express Ho

 Is sometimes referred to as the research so that we can work with a single

Session 7.18 Session 7.20

Note About Stating Your Own

Session 7.21 Session 7.23

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Example in Stating your Hypothesis Example of a Research Problem

Session 7.22 Session 7.24

Con’t of Research Problem Con’t of Research Problem

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Con’t of Research Problem Stating the Null Hypothesis

The drug is ineffective.

Note About Testing the Validity of

Session 7.29 Session 7.31

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

What is a Test of Significance? Two Types of Errors

• The goal will be to reject Ho in favor of Ha,

Session 7.30 Session 7.32

Type I Error Type II Error

Session 7.33 Session 7.35

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Examples of Type I Error Examples of Type II Errors

Session 7.34 Session 7.36

Summary of Possible Decisions in Controlling Type I and Type II

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Analogy to Decisions in Hypothesis Controlling Type I and Type II

Session 7.38 Session 7.40

Controlling Type I and Type II Example to illustrate Type I and

Session 7.41 Session 7.43

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Controlling Type I and Type II Example to illustrate Type I and

Example to illustrate Type I and Example to illustrate Type I and

Session 7.45 Session 7.47

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….

Region of Rejection or Critical Region- the set Types of Tests

TEACHING BASIC STATISTICS …. TEACHING BASIC STATISTICS ….