Académique Documents
Professionnel Documents
Culture Documents
Hypothesis Development
and Testing
Dr. Rohit Joshi
IIM Shillong
Statistical inference
Statistical inference
Decision-making process:
Significance level
Assumptions
1.
2.
3.
of
4.
s
x z*
n
What is a Hypothesis?
A hypothesis is a
claim (assumption)
about the population
parameter
Examples of parameters
are population mean
or proportion
The parameter must
be identified before
analysis
Types of Hypotheses
Research Hypothesis
Statistical Hypotheses
Example Research
Hypotheses
Statistical Hypotheses
Two Parts
a null hypothesis
an alternative hypothesis
null: H0
alternative: Ha
Example:
H 0 : female = male
where is the average IQ of the
named gender
Alternative hypothesis
Ho : 40 Kg
Ha : 40 Kg
APPLICATION
APPLICATION
Is X 20 likely if ?
No, not likely!
REJECT
Null Hypothesis
X 20
Take a Sample
20
... Therefore,
we reject the
null
hypothesis
that m = 50.
Ho : 40 Kg
Ha : 40 Kg
Rejection Region
Critical Value
Critical Value
Ho : 40
Ha : 40
Two-tailed Test
Ho: 12
Ha: 12
Ho : P 0.18
Ha : P 0.18
Level of Significance,
Is designated by
significance)
, (level of
Level of Significance
and the Rejection Region
H0: =3
H1: < 3
H0: = 3
H1: > 3
H0: 3
H1: 3
Rejection Regions
Critical
Value(s)
/2
One-tailed Tests
Ho : 40
Ha : 40
Rejection Region
Non Rejection Region
=40 oz
Critical Value
Ho : 40
Ha : 40
Rejection Region
Non Rejection Region
=40 oz
Critical Value
Two-tailed Tests
Ho : 40
Ha : 40
Rejection
Region
Rejection
Region
Ho : Rs 74,914
Ha : Rs 74,914
.025
2
.025
2
Rejection
Region
Rejection
Region
Non Rejection Region
=0
14,530
Zc 196
.
Zc 196
.
Income Example:
Two-tailed Test
If Z Zc 196
. , reject Ho.
If Z Zc 196
. , do not reject Ho.
X 78,695 74,914
Z
2.75
14,530
n
112
Z = 2.75 Zc = 1.96, reject Ho
X Z n
Ha : Rs 74,914
c
.025
2
.025
2
Rejection
Region
Rejection
Region
Non Rejection Region
72,223
Zc 196
.
14,530
74,914 196
.
112
72,223
77,605
=0
Upper
Zc 196
.
Zc
14,530
74,914 196
.
112
77,605
.025
2
.025
2
Rejection
Region
Rejection
Region
Non Rejection Region
72,223
Zc 196
.
77,605
=0
Zc 196
.
Since X 78,695
4.3
Ha : <
4.3
n=
=COUNT(A4:H7)
0.05
Mean =
S=
Std Error =
=AVERAGE(A4:H7)
=STDEV(A4:H7)
=B12/SQRT(B9)
An exercise
SAT score distribution for India:
Population mean = 500,
Population st.dev. = 100
Sample mean of 35 students = 541
Develop and test Hypothesis
Application
Each day, the Indian Custom Service (ICS)
has historically intercepted about Rs. 28
Lacs in the contraband goods being
smuggled into the country, with the
standard deviation of Rs. 16 Lacs per day.
On 64 randomly chosen days in 2010, the
ICS intercepted an average of Rs. 30.3 Lacs
in the contraband goods. What would you
suggest the Commissioner about the
smuggling as compared to historic level.
22.2
26.6
25.3
30.4
23.2
28.1
23.1
28.6
27.4
26.9
24.2
23.5
24.5
24.9
26.1
23.6
Ho: 25
Ha: 25
df n 1 19
Rejection Regions
.025
2
.025
2
2.093
Critical Values
2.093
Rejection Regions
.025
2
.025
2
Non Rejection Region
2.093
Critical Values
2.093
X 25.51 25.0
t
1.04
S
2.1933
n
20
Since t 104
. 2.093, do not reject Ho.
Example
Size in Acres of 23 Farms
445
463
466
561
489
466
477
560
474
557
557
505
502
433
553
449
545
477
545
438
500
511590
Example
Ho : 471
Ha : 471
Rejection Region
.05
df n 1 22
t
Critical Value
1717
.
If t 1717
. , reject Ho.
Rejection Region
If t 1717
. , do not reject Ho.
.05
Non Rejection Region
t
Critical Value
X 498.78 471
t
2.84
S
46.94
n
23
1717
.
n P 5, and
nQ 5
A case
Audio Sounds runs a chain of stores selling
stereo systems and components. It has been
very successful in many university towns but
it has had some failures. Analysis of its
failures has led it to adopt a policy of not
opening a store unless it can be reasonably
certain that more than 15% of the students in
the town own stereo system costing at least
$1,000 or more. A survey of 300 of 2400
students at a small art college in the Midwest
has discovered that 57 of them own stereo
system costing at least $1000. If Audio
Sounds is willing to run a 5% risk of failure,
Application
XCO, a multinational manufacturer, uses a
batch process to produce widgets. Each
batch of widgets takes 8 hrs to produce and
has material and labor cost of $ 8,476.
Because of variation in machine efficiency
and raw material purity, the number of
widgets per batch is random. All widgets
made can be sold @ $2.50 each, and widget
production is profitable so long as the
batches sell for more than $ 12,500 on
average. XCO sampled 16 batches and
found 5040 widgets per batch on average,
with a standard deviation of 41.3 widgets.
At = 0.025, can XCO conclude that its
widget operation is profitable?
Errors
Type I Error
Rejects a true null hypothesis
Has serious consequences
Type II Error
Result Probabilities
H0: Innocent
Jury Trial
Hypothesis
The Truth
Verdict
Innocent
The Truth
Guilty
Decision
H 0 True
Do Not
Innocent
Correct
Error
Reject
1-
H0
Guilty
Error
Test
Correct
Reject
H0
Type I
Error
( )
H 0 False
Type II
Error ( )
Power
(1 - )
(continued)
Type II error
H0 is incorrect
H0 is accepted
correct decision
Type II error ()
H0 is rejected
Type I error ()
correct decision
Where does
this number
come from?
Type II error
Valueof
of Power
Power==(1
(1--)
)
Value
0.8739
0.8739
0.7405
0.7405
0.5577
0.5577
0.3613
0.3613
0.1963
0.1963
0.0877
0.0877
0.0318
0.0318
0.0092
0.0092
0.0021
0.0021
0.1261
0.1261
0.2695
0.2695
0.4423
0.4423
0.6387
0.6387
0.8037
0.8037
0.9123
0.9123
0.9682
0.9682
0.9908
0.9908
0.9972
0.9972
1.0
0.9
0.8
Power
61
61
62
62
63
63
64
64
65
65
66
66
67
67
68
68
69
69
Powerofa One-TailedTest:=60,=0.05
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
60 61 62 63 64 65 66 67 68 69 70
Significance level
Increases when
Sample size
decreases
Zc 1.645
Zc
010
.
12 ( 1645
. )
60
11979
.
If X 11979
. , reject Ho.
If X 11979
. , do not reject Ho.
Do Not Reject Ho
Type I
Error
95%
=.05
Ho is True
Correct
Decision
Z0
Ho is False
Correct
Decision
19.77%
=.8023
Type II
Error
Z1
Do Not Reject Ho
Type I
Error
95%
Correct
Decision
=.05
Ho is True
Ho is False
Correct
Decision
92.92%
=.0708
Z0
Type II
Error
Z1
Critical Values
Approach to Testing
of obtaining a test
p-value: Probability
statistic more extreme (
or
) than
the observed sample value given H0 is true
Called observedlevel of significance
Smallest value of
that an H0 can be
rejected
Compare the
p-value with
If p-value
, do not reject H
General Steps in
Hypothesis Testing
e.g.: Test the assumption that the true mean
number of of TV sets in U.S.
homes is three
(
Known)
1. State the H0
H0 : 3
2. State the H1
H1 : 3
4. Choose n
=.05
n 100
Z test
3. Choose
5. Choose Test
General Steps in
Hypothesis Testing (continued)
6. Set up critical
value(s)
Reject H0
-1.645
100 households surveyed
7. Collect data
8. Compute test
statistic and p-value
9. Make statistical
decision
Assumptions
Z test statistic
X X
/ n
Rejection Region
H0: 0
H1: > 0
H0: 0
H1: < 0
Reject H0
Reject H0
Z Must Be
Significantly Below 0
to reject H0
368 gm.
Z 1
.95
= .05
0 1.645 Z
Critical Value =
1.645
.04
.05
.06
.9738
.9744
.9750
Test Statistic:
X
Z
1.50
Reject
Do
Not Reject at = .
Decision:
05
.05
1.645
1.50
No evidence that
Conclusion:
true mean is more
p -Value Solution
p-Value is P(Z 1.50) = 0.0668
Use the
alternative
hypothesis to
find the
direction of the
rejection region.
P-Value =.0668
1.0000
- .9332
.0668
1.50
Z
Z Value of Sample Statistic
p -Value Solution
(continued)
= 0.05
1.50
1.645
368 gm.
Test Statistic:
X 372.5 368
Z
1.50
15
n
25
Decision:
= 0.05
n = 25
Critical Value: 1.96
Reject
.025
.025
-1.96
0
1.50
1.96
Do Not Reject at = .
Conclusion:
05
No Evidence that
True Mean is Not 368
p-Value Solution
(p Value = 0.1336) ( = 0.05)
Do Not Reject.
p Value = 2 x 0.0668
Reject
Reject
= 0.05
1.50
1.96
Connection to
Confidence Intervals
For X 372.5, 15 and n 25,
the 95% confidence interval is:
372.5 1.96 15 / 25 372.5 1.96 15 / 25
or
366.62 378.38
If this interval contains the hypothesized mean (368),
we do not reject the null hypothesis.
It does. Do not reject.
t Test: Unknown
Assumption
X
t
S/ n
368 gm.
Test Statistic:
X 372.5 368
t
1.80
S
15
n
36
= 0.01
n = 36, df = 35
Critical Value: 2.4377
Decision:
Reject
.01
0
1.80
2.437
t35
p -Value Solution
(p Value is between .025 and .05) ( = 0.01).
Do Not Reject.
p Value = [.025, .05]
Reject
= 0.01
1.80
2.4377
t35
Proportion
Proportion
(continued)
X Number of Successes
ps
n
Sample Size
ps p
ps
Check:
np 500 .04 20
5
n 1 p 500 1 .04
480 5
H0: p .04
H1: p .04
= .05
n = 500
Reject
.025
1.96
1.14
.04 1 .04
500
1.14
.025
-1.96
p 1 p
n
.05 .04
Decision:
pS p
Conclusion:
p -Value Solution
(p Value = 0.2542) ( = 0.05).
Do Not Reject.
p Value = 2 x .1271
Reject
Reject
= 0.05
1.14
1.96