You are on page 1of 24

Sampling

Using a few employees to understand


the whole organization

Topics for this Chapter

Good sampling practices


Non-probability sampling
Sampling concepts & terminology
Probability sampling
Populations & sampling frames
Sampling designs
Multistage cluster sampling

Good Sampling

Good sampling implies:

A well-defined population
An adequately chosen sample
An estimate of how representative the
sample is (Sample statistics Population
parameters)

Non-probability Sampling

Reliance on available subjects


Purposive/judgemental
Snowball
Quota
Selecting informants
Generalization?

Representative?

Probability Sampling

Conscious & unconscious sampling


bias

Not typical/representative of populations


they have been chosen from

Representativeness & probability

Equal chance of being selected


More representative than non-probability
Can estimate
accuracy/representativeness

Probability Sampling Theory

Use random selection methods:

Serves as a check on conscious and


unconscious bias
Offers access to the body of probability
theory, which provides the basis for
estimates of population parameters and
estimates of error

Probability Sampling Theory (cont.)

Important rules regarding


distribution:

Many independent random samples will


be distributed around a population
parameter in a known way
It provides a formula for estimating how
closely the sample statistics are
clustered around the true value

Probability Sampling Theory


(cont.)

Important rules regarding distribution


(cont.):

Sampling error for binomial

P Q
n

Probability Sampling Theory


(cont.)

Important rules regarding distribution


(cont.):

Standard error gives an indication of the


extent to which the sample estimates
will be distributed around the population
parameter
Will fall within specified increments (each
equal to one standard error) from the
population parameter

Probability Sampling Theory


(cont.)

Important rules regarding distribution


(cont.):

34% = 1 standard error above


34% = 1 standard error below
Together = 68%
95% = 2 standard errors
99,9% = 3 standard errors
Confidence interval vs Confidence level

Sampling Concepts & Terminology

Element (UoA)
Population
Study population
Sampling unit
Sampling frame
Observation unit

Variable
Parameter
Statistic
Sampling error
Confidence level
Confidence interval

Populations and Sampling Frames

Findings from sample


Representative of sampling frame

Is sampling frame complete?

To generalize

Representative
Probability

Review of Populations and


Sampling Frames

Findings based on a sample can be


taken as representing the elements
comprising the sampling frame
Sampling frames do not always
include all elements
Determine the extent of omissions
To be generalized, even to the
population, all elements must have an
equal representation in the frame

Types of Sampling Designs

Simple Random Sampling


Systematic Sampling
Stratified Sampling

Multistage Cluster Sampling

Multistage cluster sampling,


stratification

Farmworkers

Probability Proportionate to Size


(PPS)

Systematic Sampling

N
n
N Population
n Sample

Systematic Sampling (cont.)


Draw a sample of 25 employees from a
division comprising 100 employees. Use
systematic sampling.

N ?
?
n ?

Stratified Sampling (cont.)

Sampling error is reduced by:

A large sample produces a smaller


sampling error
A homogeneous population produces
samples with smaller sampling errors
than does a heterogeneous population.

Stratified sampling ensures that


appropriate numbers of elements are
drawn from homogeneous subsets of
that population

Stratified Sampling (cont.)

Aim of stratification is to:


Organize the population into
homogeneous subsets and
To select the appropriate number of
elements from each

Stratified Sampling

n
N
N Population
n Sample

Stratified Sampling (cont.)


English = 224
Afrikaans = 300
isiZulu = 410
isiSotho = 147
Use stratified sampling to get a sample
of 330 from the organization.

Stratified Sampling (cont.)

n
N

330

0,31
1081

English = 224 x 0,31


Afrikaans = 300 x 0,31
isiZulu = 410 x 0,31
isiSotho = 147 x 0,31

Stratified Random Sampling


Age

Females

Males

20-40

2000

1500

Over 40

2500

2000

Total

4500

3500

Draw a random sample of


500 from 8000

Snowball Sampling

Subjects are difficult to locate


Used for explorative purposes

You might also like