STATISTICAL INFERENCE
TEST OF HYPOTHESIS
STATISTICAL INFERENCE
Is a procedure whereby, on the
basis of observed data in a
sample, we generalize
particular
and drawconclusions about the
population from which the sample
isdrawn.
Test of Hypothesis
Statistical Hypotheses A
A statistical hypothesis is :
a statement about
the parameter(s)of one or more populations.
Test of Hypothesis...
Test of a Hypothesis
• A procedure leading to a decision about a particular
hypothesis
• Hypothesis-testing procedures relyon using the
information in a random sample from the
population of interest.
• If this information is consistent with the hypothesis,
then we will conclude that thehypothesis is true; if
this infornmation is inconsistent with the hypothesis,
we will conclude that the hypothesis is false.
•Hypothesis testing is a procedure for making
inferences about a population.
Population
Sample
Inference
Statistic
Parameter
• Hypothesis testing allows us to determine whether
enough statistical evidence exists toconclude that a
belief (i.e. hypothesis) about aparameter is supported
bythe data.
Concepts of Hypothesis Testing
1. There are two hypotheses,the null and the alternative
hypothesis.
2. The procedure begins with the assumption that the null
hypothesis istrue.
3. The goal is to determine whether there is enough
evidence to infer that the alternative hypotheses is
true, or the null is not likely to be true.
4.There are two possible decisions: Conclude that there
is enough evidence to the alternative hypothesis.
Reject the null.
Conclude that there is notenough evidence to support
the alternative hypothesis. Fail to reject thenull.
The Null Hypothesis
Is denoted by Ho
It isthe statement that is under investigation or being
tested. It asserts there is no change.
Thenull hypothesis (H) will always state that the
parameter equals the value specified in the
alternative hypothesis (H)
H, is the hypothesis we hope to reject
-The null hypothesis could be:
H: H= 15 minutes
The Alternative Hypothesis
•Is the research hypothesis, denoted by H,
This is the statement you will adopt in the situation
the evidence (data) is sostrong that you reject the
null hypothesis.
• H is the hypothesis we hope to accept
The alternative hypothesis could be:
H;:u>15 minutes, or
H,: u<15 minutes, or
H,: u15 minutes
Types of Tests
One-Tailed Tests
Left-tailed Tests: H: H =k; H: H<k
Right-tailed Tests: H: u =k; H,: >k
Two-Tailed Tests
. Two-tailed Tests: H: H =k;H,:u#k
Type of Test to Use
This depends on what you suspect. If you
suspect the mean value to be greater than
claimed, so youwould lean to a right-tailed
test.
If you suspect the mean value to be less than
claimed, you woulduse a left-tailed test
Or perhaps you wish totest whether the mean
value is different from the target (twotailed
test).
Concepts of Hypothesis Testing...
• Two possible errors can be made in any test:
-A Type I error occurswhen we reject a true null
hypothesis and
- A Type IIerror occurs when we do not reject a false null
hypothesis.
• There are probabilities associated with cach type of error:
P(Type Ierror) =a
P(Type II error )= B
Errors in Hypothesis Testing
Tests of Statistical Hypotheses
Table 9-1 Decisions in Hypothesis Testing
Decision H,IsTrue H, Is Falsc
Fail to rejcct Ho nocrror type Ilcrror
Reject H, typc Icrror no crror
a =Ptype Ierror) =P(reject H when H, is true)
Sometimes the Type error probability is called the
I
significance level, or the a-error, or the size of the test.
Types of Errors...
•A Type lerror occurs when wereject a true null
hypothesis (i.e.Reject H, when it is TRUE)
ICRT- LS
• A Type II error occurs when we don't reject a false
null hypothesis (i.e. Do NOT reject H, when it is
FALSE)
Relation Between Type I and Type II Errors
The two probabilities, Type Ierror a and Type
IIerror is B, are inversely related. Decreasing
one increases the other.
1-a= 95
C =05
Z CRIT- L6S
1- - 99
ZCRIT233
Power of Test
Definition
The pover of a statistical test is the probability of rejecting the null hypothesis H,
when the alternative bypothesis is true.
• The power is computed as I- ß. and power can be interpreted as
theprobability of correcthy rejecting a falsemll hypothesis. We often
compare statistical tests by comparing their power properties.
Statistical Hypothesis...
H: u = 50 centimeters per second
H;: u#50 centimeters per second
Reject H Fail to Reject Ho Reject Ho
l 50cm/s l= 50cm/s l 50cm/s
48.5 50 51.5
Figure 9-1 Decision criteria fortesting Ho: =50 centimeters
per second versus H:u50 centimeters per second.
Formulating the Hypothesis
Example 9-1
Consider the propcllant burning rate problem.Suppose that if the burning rate is less than
50centimeters per second, we wish to show this with a strong conclusion. The hypotheses
should be stated as
H: u = 50 centimeters per second
H;p< 50 centimeters per second
Here the critical region lies in the lower tail ofthe distribution of T. Since the rejcction ofH,
is always a strong conclusion, this statement of the hypotheses will produce the desired out
come ifH, is rejected. Notice that, although the null lypothesis is stated with an equal sign,it
is understood to include any value of not specificd by the alternative hypothesis. Therefore,
failing torejct H, does not mean that =50 centimeters per second exactly, but only that we
do not have strong evidence in support of H:
Formulating the Hypothesis...
The bottler wants to be sure that the bottles meet the
specification on mean internal pressure or bursting
strength, which for 10-ounce bottles is a minimum
strength of 200 psi. The bottler has decided to formulate
thedecision procedure for a specific of bottles as a
lot
hypothesis testing problem. There are two possible
formulations for this problem, either
Ho: p =200 psi H: u =200 psi
H:p> 200 psi H:p< 200 psi
Stepsin Hypothesis Testing
1. From the problem context, identify the parameter of
interest.
2. State the null bypothesis, H, .
3. Specifyan appropriate alternative hypothesis, H,.:
4. Choose a significance level, a.
5. Determine an appropriate test statistic.
6. Establish the rejection region for the statistic.
7. Compute any necessary sample quantities, substitute
these intothe equation for the test statistic, and
compute that value.
8. Decide whether or not Ho should be rejected and
report that in the problem context.
Totestour hypotheses, we can use
twodifferent approaches:
•The rejection region or critical region
approach (typically used when computing
statistics manually), and
•The p-value approach (which is generally
used with a computer and statistical
software).
The Rejection/Critical Region Method
Deternmine H, H, and a.
Predetermine therejection/critical region. The
rejection region is a range of values such that if
the test statistic falls into that range, we decide to
reject the null hypothesis Ho, in favor of the
alternative hypothesis.
- Then compute the test statistic, and ifit is in the
critical region, reject H,otherwise do not reject
H,.
Totest our hypotheses, we can use
two different approaches:
Therejection region or critical region
approach (typically used when computing
statistics manually), and
•The p-valueapproach (which is generally
used with a computer and statistical
software).
One tailed-test with rejection region on right
D This is a one tailed-test, because the rejection region is located in
only one tail of the sampling distribution:
J=170
Rejection Region T>I,
O More correctly, this is an example ofa right- tailed test.
H: = 170
Ha: n > 170
One tailed-test with rejection region on left
o The rejection region will be in the left tail.
Reject Ho
Hypothesis to test:
H: <Mo
Mo
Two tailed-test with rejection region in both tails
o The rejection region is split equally between the two tails.
7
Reject Ho Reject H,
a/2 Hypothesis totest: a/2
Mo
Bo-al2
The P-value Method
Determine H. H, and a.
Compute for the P-value and compare to .
We willreject the null hypothesis if P-value <a
IfP-value > a, wedo not reject the null
hypothesis.
The P-value is the smallest level of significance that would lead to rejection of the
null hypothesisHo with the given data.
Interpreting the p-value
The smaller the p-value,the more statistical evidence
exists to support the alternative hypothesis.
If the p-value is less than 1%,there is overwhelming
evidence that supports the alternative hypothesis.
Ifthep-value is between 1% and 5%,there isastrong
evidence that supports the alternative hypothesis.
If thep-value is between 5% and 10% there is a weak
evidence that supports the alternative hypothesis.
If the p-value exceeds 10%, there is noevidence that
supports the alternative hypothesis.
Interpreting the p-value...
Overwhelming Evidence
(Highly Significant)
Strong Evidence
(Significant)
Weak Evidence
(Not Significant)
NoEvidence
(Not Significant)
0 .01 .05 10
Conclusions of a Test of Hypothesis
o If we reject the nullhypothesis, we conclude that
there is enough evidence toinfer that the alternative
hypothesis is true.
o If wefail to reject the null hypothesis, we conclude
that there is not enough statistical evidence to infer that
the alternative hypothesis is true. This does not mean
that we have proven that the null hypothesis is true!
Table 10.3. Tests Concerniug Meas
Ho Value of Test Statistic H, Critical Region
z<-Za
a known z> a
z<-ol2 0r Z> Za/2
U=n -1, t<-ta
t>ta
unknown o t<-ta2 or t > taj2
|
(1-2)–do 41-2< do z<-a
|1 -2=do L1-2 > do
di and a known 1-2t do <-Za/2 Or> a/2
t= (21 -2)– do
V=
Spv1/n1 + 1/n2
-2, 1 -2< do t<-ta
41 -2= do
o =
+ n2
but unknown,
1-2 > do t>ta
t<-ta/2 t>
(n -1)s; (n-l)s3 or tal2
+
n + n2 -2
t'= (1–#2) do –
VSi/n, +s/nz t'<-la
|!1-2 = do U= (si/n,+s37u4)? 1 -2 > do '>ta
(si/n)2 (6/n2)2"
12-1 11-2#do t'<-ta/2 Or l> la/2
i #o2 nnd unknown
Lp =do t= 1-do 4p < do t<-ta
paired sa/yn D> do t> ta
observations V=n-1 1D do t<-ta/2 or t> taj2
