Chapter 8

Probabilitytheoryprovides
a foundation for statistical internet

about population data

⑧ - is the universal set-denotes the set of all elements under consideration .

·
N

·
Venn used between
diagrams can be to
display sets and the relationships sets visually .

A is a subset of f
Union of A B
and

A C .
B
A UB

A B
A

Intersection of A and B Complement of A

A 1B AC

note . AUAC-1
A
A B

AC

Mutually
S

exclusive or disjoint sets A and B

A 1B = 0

A B

Experiment -

the process of making an observation or

taking a measurement that leads to collection of outcomes

Outcome space-collection of all possible outcomes of an experiment (1).

 Tree diagram

A bag contains 10 marbles . Three of the marbles and the 7 blue Kate draws one marble
remaining
are orange are .

places it back and then draws a second marble .

Draw illustrate above information .

a tree diagram to the

blue outcomes 0 .
49

-
blue Fo orange
3
10
BO 0 21 .

Kate

Fo
-D blue O 0 21 .

orange
to orange 00 0 09.

and X

OR

Determine Kate ?
the
probability that draws z orange marbles

33
10 X 10

= T
=
0 .
09

Determine the
probability that Kate draws blue marble ?

+ += = 0 9
.

OR

Determine the that Kate draws leastI orange marble ?

pobability at

0 .
09 + 0 .
21 + 0 .
21 =
0 .
91

-(an((mm)
Distributive : De :
law Morgan's law

NB#
-n (Buc) = (anB) =
nub V = OR

n = and

A v (B1C) = (AUB) 1 (Aud (ArB] = A

-
1B

Probability
·
is a numerical measure of the chance that an event will occur
Types of probability

subjective probability
-
the that will is based educated intuition .
probability event occur and guess export opinion
an on or
,

-
Can't be verified
statistically
-

.
e weather conditions .
g
.

objective probability :

Random predicted
or Stochastic event that cannot be with
certainty
-
the relative frequency with which they occur in a
long series of trials is stable

-
can be verified
statistically
-
.
e
.
g occurrence of number on the roll of die is near 116
any a

characteristics of probability
0 the event is to occur .
very unlikely
=

.
0 S =
the occurrence the it
of event is just as likely as is
unlikely.

1 =
the event is almost certain to occur .

Conditions of probabilities
·
the I
probability of each outcome must be between O and

·
the sum of the
probabilities of all the outcomes in the sample space = 1

Probability of events

equally likely events

Probability
·
of an event :

P(a) =
Where :

I
A = an event of a finite 1 n (A) =
number of outcome in event A

P(A) =

probability of event A ↑ (1) =

number
of outcomes in th
occurring

e .

g
.

Ten people four and six women in a meeting At random they select to preside .
men are . person
a
,

wome a

What is
selected

bility
that is

the p
the person a

Multiplication rule

sl
>
2x2Xz
= 8
me
= 2 x 3X2 = 12
There pictures
are
many have to
arrange from 1st position will 6th.

Spicydhave spisSe
↓ choose

6 x5x4X3x2x1

= 720
ways

OR

6 ! = 720

Only 3 picture can be chosen for positions

6 x 9 x4 = 120 Formula :

=
OR Cr
"

)!
C
=

*
2
! XS ! X2 :
↑ ↑ ↑
No . of units No .

of group No .
of group
pink blue

ENGI ENG2 ENG 3 MTHS 1 MTHS 2

① ② ③ S

Rearrange on bookshelf and order doesn't matter ?

5 ! =
120
no of items =

English book and maths book have to be together

-
2 ! x3 ! x2 ! =
24

Noof
ot

E E
We have 7 items line . What
to
arrange in a is the
probability that table and ball will be place

next to each other ?

+table
-all
Total

7 ! = 5040

&
 ENGI ENG2 ENG 3 #* MTHS2 MTHS 3 Eng 4

① ② ③ ⑤ 6

We have book (3 English) . What

to 7
arrange on a shelf maths and 4 is the
probability that the

maths books will be next to each other ?

7! = 5048

=> =

Codes
7
We are creating a character code :

the (1-9)
>
-

first 2 characters must be

(A-2)
-

the characters must letter

remaining be

Repetition is allowed

y
a
choices
no

choices &1 26 +
26x26x26x26

code : 67 BFT 92 x 265 =

962391456

words

with words :
working

MATHS

Repetition is not allowed

How 5 letter "words" can be formed ?

many
5 ! =
120

Repetition is allowed

g5 = 3125

The word must start with a M and end with an A

Repetition is allowed

1 x 93x1 = 125

Repetition is not allowed .

1x 3 ! x1 = G

Working with words :

Soccer

Repetition is not allowed

How
many
6 letter words can be formed ?

6 ! = 720
 if

there are
reapting letters you have to divide
by number of reaping letter factors a

I
Working with words :
Working with words :

CHOCOLATE CHOCOLATE

Repetition is not allowed ·

Repetition is not allowed

·
Word must start with 0 O
an
Word
·

must start and end with an

How a letters words be formed ?

many can
How
many
a letters words can be formed ?

1 x 8 ! =
40320

x7=O
·

10320 =
2016

Working with words :

REAPPEARS

Repetition is not allowed

How words can be formed ?

many

9 letters

= x21x2) =
22680

word must start and end with .

R

1x7 ! x 1 =
5046

-
O 2 ! = 636

Permutations

1)(r 2)
r ! =
r(r -
-

...
2xI

Only 3 picture can be chosen for positions

6 x 9 x4 = 120 Formula :

=
OR Tr

"Pr
=

Combinations

unordered
-
the number of possible subsets of objects chosen from n distinct

object
.

Formula : "Our !
 There are S qualified applicants for 2 editorial positions
university
on a

Two
newspaper . of these
applicant are men and t h re e women . If the positions

are filled by randomly selecting two of the five applicants ,

what is the

?
probability that netheir of the men are selected

=
n

S P(A)

10

=
10 = 0 .

n(a)
=-2)x!
= 3

Conditional probability
·
the conditional of event A event B occurred
probability an
, given that an is
equal to

P(AIB) =

PLAIB) = "probability of A
given
B

provided P(B) > 0

S balanced tossed Calculate al

uppose a die is once. the
probability of obtaining given that an

Odd number was obtained. A = observe a 1 .

B =
observe an odd number .

: P(AIB)
A 1B
A
= 1
=

P(AnB) =
P(B) = 5 = E
=
= 0 .
33

Independent events

When
·
events are dependent , they can be affected by previous events.

·
Thus ,
these events depend on what happened before

·
When events are dependent ,
the chances or
probability that an event will happen will change .

·
when events independent they are affected previous
are , not
by events.

e
Thus , these events do not depend on what happened before

Event A and
B are independent
.

P(A) = 0 .

P(B) =
0 .

Determine P(AUB) ?

P(AuB) =
P(A) + P(B) -

P(AnB)

=
0 4
. + 0 3
.
-
10 4)(0 3)
.
.

= 0 .
58
 Formula for independent :

Determine P(AUB)" ? PAMB) = P(A) P(B)

-

1 -

0 .
58 = 0 . 42

Are the event A and B ?

Complimentary

. 4
0 + 0 3 .
= 0 7 .

· No

Are the events A and B exclusive

?
mutually
Mutually exclusive
P(A1B) = 0 4X0 3
. .

P(A1B) =
0

= 0 . 12

: They are not

mutually exclusive .

·
two A and B said independent :
events are to be if any one of the
following holds

P(AIB) =
P(A)

P(BIA) =
P(B)

P(A1B) = P(A) P(B)

Otherwise , the events are said to be

dependent.

:
example I

A 1B = 0 P(A) x P(B) =
E x =
P(A1B) =
016 = 0

No
,
AnB are not
independent because PLA(B) # PIAXP(B)

A nC = (1) P(A) x P(c) =

E x = = 5
P(anc) =
5

Yes ,
A and C are independent because PCARC) = PCAXP(C)

Mutually P(A)
exclusive
P(AUB) = + P(B)
The complement law

P(A)) = 1 -

P(A) P(A) = 0 .
4

P(A) = 1 -
P(A)

P(a)) = 1 -
0 .
4

= 0 . 6

General additive law

P(AUB) = P(A) + P(B) -

P(A1B)

Students were random assigned to one of six different sections of an

introductory English course e .

Sections 1 ; 2 ; 3 ; 4 ; S or 6 .

A : Susan even-numbered section .

is
assigned to an

B : Susan is
assigned to a section numbered lower than .
3

A = (2 : 4i6) = 5 =
B = (1 : 2) = =

AnB =
(2)
P(AnB) =
5

PCAUB) = P(a) + P(b) -

P(AB)
I
T
2 + -

5
2
=
3

= 0 .
667

Multiplicative law of
probability
P(A -B) = P(a) P(BIA)

=
P(B)P(A / B)

= PBXPLBI
P(B , nB1)
Z

=
0 . 143

a
= E(X) = ni N
= ni

o
" = var(X) = nir(1 -

i) Ov = Mi (1 -
il)

o = sd(f) = + m)1 -

π) or
1 - i
Random and
variables
probability distributions

note the difference in the

symbols

me I standar deviation variana

#
mean

J g2
I

Normal probability distribution

·
the normal distribution distribution It describes
is common
probability used in statistics . how the

values of continuous random variable are distributed in a

symmetric , bell-shaped curve.

·
Bell-shaped and
symmetrical : the curve is
symmetric about its mean
(M) , meaning
that the left and right sides of curve are mirror images . Most data points are clustered around the

mean with fewer and fewer further from

,
occurring as you move the mean .

·
Mean = Median = Mode : in a normal distribution , the mean , median ,
and mode

are all the same and located at the peak of the curve .

·
Standard Deviation (0) : This measures the spread or dispersion of data . A small standard deviation

means the data points are close to the mean , while a large standard deviation indicates they are spread .
out

The curve become wider or narrower based on the value of .

o

·
Asymptotic : The tails of the curve approach but never touch the horizontal axis , meaning the

extreme possible but less further from the

values are
likely as
you
move mean .

Empirical rule

· I 68 %
of the data lies within I standard deviation of the mean (between i-o and m + o

· 95 % of the data lies within 2 standard deviations of the mean (between M-20 and n + 20) .

·
99 7 % .

of data lies within 3 standard deviations of the mean

(between M-30 and M + 30)

Example :

So
follow normal distribution with

population
a Melo a n d a

and 110
 Density
for
distributionit

normalprobability
curve

=
the = for -
<
Ld
+ (x)
-

Where :

M
= mean

o Standard deviations

TV = 3 .
14159

e natural
is logarithm
=
e
e.g .

Notation : x -N(M 02) ,

2- score and -table

standard data
a z-score tells you how
many deviations a
point is from

the standard normal distribution It standardize

mean in a .

helps values so
you can

Compare them to the standard normal distribution , where the mean is 0 and the standard

deviation is 1 .

2 score formula :

z =
Let :

X the value data point

=
or

M =
the population mean

& standard
the population deviation
=

Example
200 50
;
200-100
100
M x o . 0
2
=
= =
; =

X = R2CO is 2standard deviations increments of RSO units about the mean Roo

The 2-score formula transforms a normal distribution into a standard normal distribution which allows
you to use

the Z-table to find probabilities .

↑ Z-table gives the from

cumulative
probability the far left of the curve to a
particular

2 value .
The table helps you find the probability that a random variable is less than a

a certain value . If you know the score , you can look it

up in the 2-table to find the

corresponding probability
.
.g
. standard deviation IS
population 100
E
= =

Steps :

①2-score : #100 = 1 .

00

& look up z = 1 . 00 in 2-table · It equal 0 8413

.

meaning 84 13
.
% chance of

selecting a value below IS

 :
example

=20
and a =

contains more than 200 calories.

(1) 2-value :

=
=
208
200 -

= 1 .
60

(2)
Probability ?

.
8 9452P(X > 208) = P(zs1 .
5)

= 1 -

(0 9452) .

= 0 .
0548

· there is S 48 % more than 208

a .

probability that the salad that

you
select on the menu contains

Parameters vs Statistics

·
Parameter : This is a fixed , unknown value that describes a characteristic of the entire population .

The population mean (M) or

population standard deviation (c) are parameters.

·statistic :Thicalculated froma sample(

which a subset of the
population . Th sam see

key difference :

· a parameter is a number that describes the whole population , and it doesn't change .

·
Central limit theorem (CLT)

·
CLT one of most important It states
is the concepts in statistics . that regardless of the

population's distribution , the sampling distribution of the sample mean will approach a normal distribution

as the sample size increases , = 3

typically when n

key points :

T and normalityevenit thepopulationisnotnormay distributethe

T
guarantees that the distribution of thesame see

·
Implication : the LIT allows you to make inferences about the population mean
using the sample mean , even if the

original data is skewed or non-normal .

2
X =
sample mean

M =
population mean

o standard deviation
population
=

n =
sample size
 Calculating Probabilities using the CLT

The allows to calculate probabilities do

CLT
you involving sample means , just like
you with

individual data points z-scores .

using

:
Steps

① Standardize the sample mean :

the sample into the formula :

convert mean a z-score
using

=
2

this 2-score now corresponds to a position on the standard normal distribution .

② Use the z-table :

Look the 2-value in the z-table to find the cumulative probability.

up ,

Example :

Suppose that the population distribution of the

gripping strengths of industrial workers is known to have a mean

Of 110 and standard deviation of 10. for a random sample of 75 workers , what is the
probability that the sample mean
gripping strength

will be :

(a) between 109 and 112 ?

2 -087 and =1 . 73

P[10 <* 112] =

P[-087(2 < 1 .

75]
=

p[2 <1 .
73] -
P[z -0 .

87]

= 0 9582
.
-
0 .
1972

=
0 .
766

% 76 6 % .

of the industrial workers have a mean gripping strength between 109 and 112

(b)
2
=
=
0 87
.

P[X > 111] = P[z > 0 .

87]

= 1
-

[20 .

87]

=
1 - 0 8078 .

= 0 . 1922

: 19 22 % of the industrial workers have strength then 111 .

mean
gripping
a
greater
.
 Definitions :

·
Normal distribution : a
probability distribution that is
symmetric and bell-shaped , characterized

deviation .
by its mean and standard

· Z-Score : the number of standard deviations a data point is from the mean ,
used

to standardize values.

·
Parameter : A fixed , unknown value that describes the entire population .

·
Statistic : A value calculated from a sample , used to estimate a population parameterr

·
Sampling distribution : the
probability distribution of a sample statistic .

·
Standard Error (SE) : the standard deviation of the
sampling distribution ,

showing how much the

sample mean varies from the
population mean ,

·
Central limit theorem (CLT) : the theorem that states distribution of
the
sampling
the sample mean will be
approximately normal for
large sample sizes.

Using the t-distribution to calculate probabilities

The t-distribution is used instead of the normal distribution when :

·
the
sample size is small (n < 30)

· the standard (S) is used instead .

population deviation

The t-distribution has heavier tails than the normal distribution , which means accounts for more

variability As sample -distribution

, especially with smaller sample sizes .
the size increases ,

approaches the normal distribution .

Formula :

Degrees of freedom (df) .

The t-distribution defined freedom (df) equal

is
by its degrees of which is to n -

Using the -distribution table :

To calculate probabilities the -distribution :

using

① find the t-score

using the formula :

& Determint the

degrees of freedom (df) : df =
n -1

③ use the -distribution table to find cumulative probability for the calculated t-score

and degrees of freedom .

Example

youwattotestifanew teachingmethodimprovesstudent performanceAsampe sstudents asa of

① calculate the +-score :

6 Degrees of freedom :
af = 25-1 =
24

() t = 2 . S ; df =
2) can use linear interpolation it
necessary to get the precise probability
you

value .

Calcutting
probabilitie- T) + p
,

Example :

Calculate thefollowing probably 24 degrees of freedom

from the t-table :

#
: PITC

1873 o
P2 = 0 965
.

+
= 1 -

0 .

963

Pl = 0 . 960 = 0 .
037

Using the
Chi-squared distribution to calculate probabilities

The
chi-squared distribution is
commonly used in two situations :

O To test
hypotheses about the variance of a
population .

&
In -

goodness-of-fit-tests and tests of independence

keyproperties" distribution is skewed to the right especially for small

degree of freedomh e

·
It is defined by its of ,
which depends on the problem [e g . .
df = n-1 when
testing variance or df = (r -1) (c -
1)

for
contingency
tables]

·
Random variable this distribution take
with can
only on
positive values and its

density is skewed to the

right ,
but becomes more symmetric as It increases.
Steps to calculate
probability using the Chi-square distribution

O determine the degree of freedom (df) : the degrees of freedom are

typically determined by the problem

context
. In cases for test of independence or
goodness-of-fits of is related to the number
categories
many
a
,

minus
1.

② Formulate the problem : determine whether

you are
looking for P(X > K)
,

P(X(k) or P(a < X <

b) , where X is the Chi-square random variable ·

③ Use a
Chi-square table or software : the
chi-square table provides the critical values for different

probabilities and degrees of freedom . If the table does not

give you the probability .
exactly you want

Example 1
:

Finding P(X(K) for a

given value of K

Let's the P(XX10) where X follows

say you want to calculate probability , a

Chi-square distribution with 6 degrees of freedom .

O Find the cumulative

probability for P(X < 10) :

Lock the table for 6

at Chi-square degree of freedom
·

for X 10 and df 6 P(X (10) is 0 8013

the cumulative
probability approximately
-

= =
,
.

.
2 Find P(X > 10) :

P(X (10) = 1 -

P(X(10)

= 1 -0 8013 .

= 0 .
1987

Interpretation : the
probability that the Chi-square random variable X with 6 degrees of freedom is greater
than 10 is

approximately 0 1987
.
(19 87 . % )
 Example 2 : Find P(X(K) for a
given value of K

·
Interpretation : the probability that X is less than S when X follows a
Chi-square distribution with 8

degrees of freedom is
approximately , So2
0 or 15 02 %
.

Example 3 : Pla < x (b)

finding

Let's say you want to calculate P(scX(15) for Chi-square distribution with 7
degrees of

freedom .

D
findPX are table for of 7 we find PXEos

② Find P(x < 5) :

·
again the table for df 7 P(X(5)
using
= = 0 2887
, .

③ find P(scX(1s) = P(X(1s) -

P(X(S)

= 0 8650 .
-

0 2887
.

= 0 5763
.

interpretation : the
probability that X lies between 5 and IS for a
chi-square distribution with 7
degree

of freedom is
approximately 0 5763
.
(or St .

63 )
+

The F-distribution

It
hypothesis testing . compares the variances of two populations

to determine if
they are
equal . The shape of the f-distribution is
asymmetrical and skewed to the
right
,

values concentrated left side but extend right

meaning that most of the are on the
indefinitely towards the .

Key properties :

a
· itisdefinedbytwodegrees otdenotedas Vunerator/aum andVe(denominator
o

negative .

·
As VI and V2 increase ,
the F-distribution becomes more symmetric and approaches the normal

distributio n.
 Degree of Freedom (df)

i
-

V2 (denominator degrees of freedom) · This is the

degrees of freedom for the variance in the denominator .

This is
usually associated with the residuals or error terms and equals the total number of observations

minus the number of groups.

F-table

An f-table provides critical values for f-distribution certain levels (often

the
given significance & 0 05 or
= .

a = 0 .
01) and degrees of freedom for numerator and denominator .

How read the +-table :

vz).
·
the denominator freedom (df
rows correspond to the
degrees of =

·
the columns correspond to the numerator degrees of freedom (dt = V, .
)

·
The intersection of and column critical -value for particular
a row a
gives the a
significance

level .

Calculating probabilities Using the F-table .

To calculate the observed f-value critical

probability that an is less than or
greater than a value ,
you

can follow these steps :

Determine freedom
1 .
the
degrees of :

number of total observation

-

VI : the of associated with the variance in the numerator e number of groups

.

.
g

V2 : the of associated with the variance the denominator. number of total observation
in e .

g .
-
number of groups

.
2 Select the significance level /X) : common choices areo s (5%) or 0 01
. (1 %)

. Look
3 the Critical value the F-table and and
up in
using the
corresponding V , V2 values chosen

significance level .

4
. Compare the observed f-value to the critical value :

if
·

the observed f-value is

greater than the critical value from the f-table , the result is

Statistically significant
.

·
if the observed f-value less critical the result
than not
statistically
is the value , is

significant

&

of :

Vi = 3 -
1 = 2

Vz =
1S -
3 = 12
For would f-table and
a
significance level oos,
you refer to an for the critical F-value
corresponding v1 2
=
,

12 Let's the table value 3 89 If calculated F-value is

greater than 3 89,
V2 = .

say give your

. . .

the result is
statistically significant
-

Note :

·
the f-distribution used
is
only in
right-tailed tests because it test for differences in variances .

·
it assumes that the from which samples drawn distributed.
populations the are are
normally

Definitions"

·
-distribution : a distribution used for small samples where the population standard deviation is unknown .

·
t-score : standardized score used in the -distribution .

·
Degree of freedom (df) : the number of independent of information used to estimate
pieces a parameter.

·
Chi-squared distribution (x2) : A distribution and
used for
testing independence , goodness-of-fit ,

population variance.

·
f-distribution : a distribution used in variance comparisons and ANOVA
.

·
F-statistic : the ratio of variances used in
testing hypotheses
Chapter 9

Point Estimators vs interval Estimators

Pointestimatorsoint estimator provides a

single value as an estimate of a population parameter based on sample

data .

i
-

sample mean (52) : Estimate of the population mean (M)

-sample variance Estimate ofthe population varianceon a

Interval Estimators

: which parameter expected lie

definition an interval estimator
provides a range of values within the population is to ,

with a specified level of confidence.

Purpose : To
quantify the
uncertainty
to the estimate and provide a
range where the true parameter is
likely to be

found

·
confidence interval for the population proportion : provides for the true proportion based on
a
range the

sample proportion .

Example

Certain computer components temperatures.

A manufacture of computer components
can be
very sensitive to
high

conducts experiments to determine the temperature at which the components will fail

let mean
temperature at fail .
M which the components
=

Point So ° 90 % fo a

confidence interval
estimate :

J -

Calculating point Estimators

·
Point Estimator : sample Mean (5)

·
formula : c =
in S " =,
i

-
e :: each observation in the
sample

sample size
② Population variance (a2) :

·
Point estimator :
sample variance (s2)
~
formula : S
2
- Zi = (xi -
<]2
=
Sample mean

n =
sample size

③ Population proportion (p) :

(p)
· :
point estimator sample proportion

·
formula : p =
:
x number of successes in the sample

n : sample size

Characteristic of Estimators

unbiasedness

·
definition : unbiased expected the
an estimator is if its value equal true population parameter .

: (M) .
·
example the sample mean (c) is an unbiased estimator of the
population mean

Consistency
:

·
An estimator is
consistent if , as the sample size increases , it
converges in
probability to

the true population parameter .

·
the sample becomes
e .

.
g mean closer to the population mean as the sample size increases

Efficiency
·
An estimator is efficient If smallest all estimators
it has unbiased
variance
among

·
Among different estimators of the population mean , the sample mean typically has the smallest variance .

Sufficiency :

·
An estimator is sufficient if it captures all the information about the parameter contained in the

sample .

Calculating and
interpreting confidence intervals for population Mean

formula : 5 = z( Example ① standard error

JC : sample mean 5 = S St = =0 = 1

2 : z-value o = 10

O : standard deviation n 100

population
=

: 2 =
aSt
n sample size .
②

·
③ interval

lower bound : 50 - 1 .
96 = 48 04.

upper bound : So + 1 96 .
=
51 .
96

: the 95 % confidence interval to

population is [48 .
04 ; 51 .
95]

large samples : >30

CI =
p = 22

where :

↑ =
sample proportion

2/2 =
2-distribution

n =
sample size

Example

A of 500 . Compute the

survey finds brand
320 prefer a particular
people that people

9 Sy Confidence interval for

·

the population proportion .

Step 1 :

sample proportion
-

p =0 = 0 .

64

Stepaner = 0004608 = 0 . 2 15
.

Step 3 201 = 1 .

96

:
Step 5 confidence interval

CI = 0 64
.
10 0421 . =
[0 .
5979 ; 0 .
6821)

& We are as % confident that the true proportion of people who prefer the brand 59 . 79 % and 68 21 %
.
 Small samples

n130 S is unknow , use -distribution

Formula :

(I = c = + x(x

t 2/2 = + -

distribution (df = n-)

S =
sample standard dreiation

Example :

a random sample of I5 students' test scores has a mean of 78 and a standard deviation

of 10 . Calculate the 95%· Confidence interval for the population mean

si =
78

S2SE = = 2 .
582

53 tal :

= 2 .
145

S4 ME =
2 . 145 x 2 .
582 = S .

S4

55 (1 = 78 1 s S4 . =
[72 .
46 183 .
54]

: aSX. Confident that true mean test score for the population is between 72 .
46 and

83 54 . .

kell points :

·
For
large sample proportions ,
use the 2-distribution and normal approximation .

·
For small sample means , use the -distribution,

·
the confidence interval within population
gives a
range which the true parameter is
likely to

fall with a
given level of confidence .
Hypothesis testing
·

Hypothesis = claim or statement about a

population parameter or in

Hypothesis
·
= A standard procedure claim population parameter
testing
test for a about a -

decision
regarding the validity of the claim

·
A manufacturer turnover exceeds R1 million
of car
tyres claims : His annual mean

Test a
single population mean

·
Panasonic Claims : that one out of three households is in
possession of Panasonic

(Test it
product. single population proportion :

steps of hypothesis testing

E
.
g
.

Step 1 : State the hypotheses

left-side (less than)
·
Null
hypothesis (Ho) :
Ho: M = 5

Represents
-

the status quo or no effect

.

Il
" Ha:M 5
-
E The is R50 000
.
g mean
salary of
employees
.

Right-side (more than 3

Ho : M = 50 , 000

Ho : M = 5

Ha : m = 5
·
Alternative Hypothesis (H , orHa) :
Two-side (not equal)
Represents the

HoHe
-
research claim effect want to .
test
or
you

"The R5000" (Him 30000

-
E not
.

g
. mean salary of employees
is

Step 2 : choose the level (a)

significance
·
The
significance level (a) is the probability of rejecting the null hypothesis when it is
actually true (Type I .
error)

Common value of a are :

·
a 0 05 :
= .

5% chance of rejecting. Ho when it's true .

·
a 0 01 : 1% Chance of rejectingHo it's true
when
=
.
.

·
The lower the significance level , the
stronger the evidence required to
rejectHo.

:
Step 3 collect data and compute the test statistic

·
for small sample (nc30) , use a .
t-test

·
for
large sample (n = 30) or known population variance , use a 2-test

·
2-test formula (for large samples)
-

.
2 = F
·
T-test formula (for small samples) : T =

Step 4 : determine the P-value or Critical value

P-value
·

Compare the test statistic to a critical value or the to the

significance level .
 ·Dave approachthe pataret. .
·
Critical value : rejectH o
approach if the test statistic exceeds the critical value ,

Step S : Make a decision

RejectHo : if the evidence is

strong enough (P-value La or test statistic exceeds critical value)
!

·
fail to Ho
reject If the evidence not
is
strong enough.
.

Example 10 . 1

A manufacturer of a new cell

phone battery claims that the mean lifetime of

30 hours . An claim less than 30 hours

the
battery is
employee that the mean is

A random sample of 81 batteries renders a

sample mean of 28 7
.

hours with a &

sample standard deviation 8 hours. Test the

hypothesis at 5 % significance

level .

I
③z -
①
Mo = 30

Ho : 30
M =

n = 81

Ha : Mc 30 Y = 28 7.

2 = 1 463
S 8
.

z b
P = p(z -
1 . 46)
= 0 .
0721

a = 0 0.
.
5

⑤ p
> a 2 = -
1 .
46 critical v : -
1 .
645

. Do not
reject Ho value > critical
- V alue

: the population mean lifetime is not less than so hours

The
employees' s statement is incorrect
 z -

value = -
1 .
46 critical value = -1 .
64S

2 value > Critical value

··
L

Example 10 2 .

showed that the mean number of tries per match was 62

.
with a standard deviation 3 . .
2 Does the sample contain

Sufficient information to contradict the

rugby enthusiast's statement ? Assume X = 0 1
. .
Assume that the

population distributed .
is
normally

① M =
S

Ho : M = S n =
25
T

Ha MS i 6 2
=
·
.

S =
3 2 .

T
=
X = 0 .
1

T
= 1 . 875
-
# ·
-

1.87s 0 1 . 875

2 P(T 879)

Tot
> 1
p = .

= 2 (1 - P( + <1 .
875))

T2 =
1 . 896

P , =09,

pLx

:. Reject Ho
Example 10 3 .

A sample of size 6 is available to test if the mean mass of horses in a

The The
large hard exceeds 450kg. significance level Sy . sample values are as

follows : 499 ; 521 ; 452 ; 459 ; 538 ; 588 .

M = 450 n =6

* =

E
-

s
= 501 .

197

S = 56 .
433

Ho : M 450kg
=

Ha : M> 450

T
=
SI : X = 0 05 .

⑤
501 .
167
-

450
56 .
933

T =
5

. 201
2

# p = P(T2 201) .

=
1[0960)
x (2 20-2a
.
o a

J
=
1 -0 .

960

0 0
4
= .

PLX
:
Reject Ho

Mass of horses is
larger herd exceeds
456kg
Example 10 4 .

A manufacturer clarms that S% of goods manufactured are defective . Out of 200

items Is
randomly chosen from the
supply
store 14 items were defective ·
there

sufficient evidence to show that more than 5 % of items are defective ? Assume that

X = 0 .
1

T = 0 .
0S

① n =
200

Ho : Y 0 0 s

e
= .

Ha : < 0 .
08

P(z s 30) = 0 07
p
= .
.

② = 1 -

p(z(1 30) .

p -
Th = 1 -0 9032
.

[L (1 -

mu)
z = u =
0 .
0968

a = 0 .
1 ⑤

pa

③
· Reject Ho

0 07 0 05 More than 5 % manufactured Items are defective

of the
-

. .

0 05(1 -
0 05)
2
. .

=
200

2 = 1 .
298

-u
8 1 30
.
Example 10 .
5

A manufacturer claims that his market share is 60 % However , a random sample

of 500 clients show that 275 supports his product. Test the manufacturer
only
Statement at 1 % level
significance ,

① T =
0 6 .

Ho : Y 6 (P(z > 2 28)

p
= 0 .

=
.

n = 500

28))
M

Ha .
Yo 0 .
6 = 2(1 -

p(z2 .
x =
275

1
=
2 -

2(0 9887) .

p
= 0 55
.

② =
0 .

0226

p -
Th

[L (1 -

mu)
z = u ⑤

>L
p

X = 0 . 01

: Do not reject Ho

⑤ The manufacturer's market share is not different than

0 55-0 6
. . 60 %

z =

2 =
-

2 .

282

S
-
t·
-

2 .
28 O 2 28
.
The difference between two
population means

·
⑨

·
Alternative
hyphotheses .

left-sided : Hai Mi -

M2 < - D Ha : u + M2

: Ha :
right-sided : M1-M20 Ha W , > M2

Hypothesis testing
for the difference between two
population means

large samples (both n , and n2

> 30)

2
Example 11 .

A done difference and

study is to estimate the in salaries of teachers in private government

schools in a
specific education community A random sample of 100 teachers at schools
. private

showed annual of R10 S 000 with standard deviation R15200. A random

an mean
salary a
of

sample of 200 realised annual of 2000

teachers at
government school a mean
salary
standard deviation R13500 Test the hypothesis of
with a of .
that the mean
salary teachers

schools higher of schools. Test at 5%

at
private is than that teachers at
government a

significance level .

ni 100 200
=
n2 =

51 105 000 88000

x1
= =

S1 = 19200 =
135 00

Ho : ul =
M2
p
=

= 1
P(z

p(zc9 47)
> 9 .

47)

.
-s
2/2

as
Ha : M , >M2
< 1-0
.
9998

C 0 .
001

~ O
10 5 000 -

88 . 000
a
z =
45200 ↑ (12600
pa

z = 9 , 471
: Reject Ho

mean of private school is higher than public school .

Example 11 .
2

A transport in Helsinki Finland if

company &
,
wishes to know the mean travel time

A
to work differs between commuters
travelling by bus or
by train .
survey as

conducted bus and train commuters which resulted the

among , in
following statistics

&

Test the hypothesis that travel time for train commuters is shorter

than travel bus (c 1)

time for commuters
travelling by
= 0 .

He : Mi =

M2 ④
Hai M , <
M2 p-value

#1 T2 P(z( 2 40)
p
-
=
.
-

z
=r
l P(z) z 40
-
-
= .

=
0 .
0085

2 = 0 1
.

③
2 48
-

= p
:
Reject Ho

z =
-

2 , 40

small samples : either n , or ne 30

1 -
* z

T-Spanthe

where

sp = (n , -
1)S , + (nz -
15
ni + nz -

2
Example 11 3 .

Xi = 68 X2 = 60

Si =
7 8 .

S2 = 8 .

ni = 11 n2 =
14

Hoi Mr T
2 = 0 01 .

②
1 -
* z

T-Spanthe

where

sp = (n , -
1)S , + (nz -
15
ni + nz -

where

Sp = (1) -
1) (7 8]2 .
+ (14 -

18 43
.

11 +
14 2
-

=
8 .
145

60 -
60

T =
sins

= 2 .
438

① p-value

Tg = 2 .

438 p = 2p(T > 2 .

438)

If = 29 = 2(1-p(T <2 438) .

T1

2()P P) (Ty T
J
P
2 313 +
-
-

=
x
.

=
T2 2 500
-

= .

2-c[(90098s) J
985
P1 =
0 .

x (2 -

438-2 .

313) + so

P2 = 0 941 .
=

= 2 -
2(0 988)
.

= 0 024
.

p >

.. Do not Ho
reject
Example 11 .
4

x
= [2 =
M
x

=
#
+ 3 02 + 0 38 .
. . .
.
=
036 .

11
II

7 76

1474
= .

= 3 585
,

S2 =
703 .
529- (11) /7 .
76)2

= 18

·
= 2 .
028

&: m =
M ni = 11

Ha .
Mi
<
M2 n2 =
11

df = 22 -
2 =
20

(11-7(2 02872
2

Sp = ( 11 -
1) (1 628) .
+ .

3 885.
- 7 . 76

859)
11 + 11-2 += (1 .

1
839
=
+ 4 942
. -

= .

① p
=
P( + -

4 .

942)
=
1 -

P(T2-4942)
L 1 0
-
.

995

L0 005 .

The difference between two

population proportions

·
Alternative :
hypotheses
left-sided : Ha :it 1
-
The <O -Ha :, < Th2

: Ha Th - it > -* Ha : Th >Ttz
right-sided :

two-sided : Ha :t ,
-
The oD Ha : Ft2
 Test statistic for population proportions

↑, -P2
z = p(1-) + m

where p = X1 + X

Ni + nz

Example 11 .
5

= ↑ =
p

ni =
175 1 n2 = 150 2

X =
60 X 2 = 40 = 0 267
.

=
0 .
343

& = 0 .
1

& Ho : = The

Ha it ,
:
site

③
= 0 . 308

z-+
=
1 . 479

① P-value :

p(2> 479)
p
=
.

= 1 -

P(z <1 .
479)

=
1 -
0 .
9306

= 0 .

0694

⑤ : cornbix
PLE is more popular