Sie sind auf Seite 1von 3

Sampling Summary

Sampling Methods

Population : The complete set of observation we are concerned

Sample : A subset selected from the population

In general we are interested to calculate the population mean and population variance
2
.

Reason why it is difficult find and
2

1. Population size is too large
2. Time and cost

Hence the need to select a sample from the population to estimate and
2


Types of sampling methods for sample size n

Methods Process Advantages Disadvantages
Random 1) Assign unique numbers to all
members of the population.
2) Generate n unique numbers.
3) Select members from the list.
1) Unbiased.
2) Every possible
sample has equal
chance of being
selected.
1) Sample may not be
representative of the
population.
2) Need access to the
whole population.
Stratified 1) Divide the population into strata
(groups).
2) Calculate the required number
for each strata.
number in strata
population size
n
3) Do random sampling for each
strata.
1) Unbiased.
2) Sample is
representative of the
population.

1) More time
consuming as
compared to simple
random sampling.
2) Strata may not be
clearly defined.
3) Need access to the
whole population.
Systematic 1) Assign unique number to all
members of the population in
ascending number.
2)
population size
k
n

3) Generate random offset between
1 and k
4) Start from offset in the list,
select member.
5) Select every
th
k members down
the list.
1) Unbiased.
2) Easier to
implemented as
compared to random
and stratified
3) May be
representative of
population depending
how the members are
lined up.
1) May not be
representation of the
population if there is a
cycle in the list.
2) Need access to the
whole population.
Quota Similar to stratified. Except
1) The required number for each
strata needs not be proportional to
the population ratio.
2) Selection from each strata is not
random.
1) Easy
2) Used in situation
where access to the
whole population is
not available.
1) Biased


Sample mean and variance
From n observation
1 2 3 n
x x x x + + + +

Sample mean :
1
x x
n
=


Sample variance :
( ) ( )
2 2
2 2
1 1
x
x x x x
n n
= =



Estimating the population and
2

From n observation
1 2 3 n
x x x x + + + + , we can estimate the population mean and variance.

1
n
x x = =

2
2
1
x
n
n
=

( )
( )
2
2
2
1 1
1 1
x
x x x
n n n
(
(
= =
(




For data that are given in the form ( ) x a

and ( )
2
x a

, you should do a substitution of


y x a = .
Then y x a = and

1
x y a y a
n
= = + = +


Variance is unchanged by translation, so

( )
2
2
2
1
1
y
y
n n

(
(
=
(

.

In certain questions you may need to use properties of summation.
( ) ( ) ( ) ( ) ( ) f f g g r r r r =


k nk =



Sample mean distribution
Given that
( )
2
, ~ N X , The sample mean distribution
2
1 1 3
~ , N
n
X X X X
X
n n

| | + + +
=
|
\



Central Limit Theorem
Given any random variable X where X follows any known or unknown distribution
(exception Normal). The mean and variance of X are and
2
respectively, then if n is
large ( 50 n > ),

( )
2
1 2 3
, ~ N
n
X X X X n n + + + + approximately by CLT
and
2
, ~ N
n
X

| |
|
\
approximately by CLT
Note: Continuity correction is not needed at H2 Level

Das könnte Ihnen auch gefallen