Download INTRODUCTION

Sampling
Neuman and Robson Ch. 7
Qualitative and Quantitative Sampling
1
Content



Why Sample?
Qualitative vs. Quantitative Sampling
Non-Random Sampling



Non-probability
Not representative of population
Random sampling


Probability
Representative of population
2
Why Sample?: Cost



You want to keep the cost of data collection reasonable
Surprisingly required sample size is virtually
independent of the population size.
Samples of modest size can often give good results



at vastly lower cost than a full population census.
Sample of 1,000 drawn from a population of 40 million
Secondary data (National statistics), if available, may be
cheaper still
3
Why Sample?:
Relevance and Flexibility


Surveys can cover topics that are most
relevant to a particular research
question
With Survey can tailor data collection

to particular subgroups
4
Why Sample?: Speed

Surveys can often be set up and completed in
short period


deliver results in months or even weeks.
By contrast a Census need long lead-times


administrative statistics are produced to inflexible
administrative schedules.
Typically every 5 or 10 years
5
Basic terms and concepts in
Sampling: 1

Population:


Sample:


list of all units (e.g. names of people, addresses of businesses)
Representative sample:


the segment of population that is selected for investigation
Sampling frame:


the universe of units from which the sample is to be selected
a sample that reflects the population accurately
Sample bias:


distortion in the representativeness of the sample
Happens when your sample is not representative
6
Basic terms and concepts in
Sampling: 2

Non-probability sample:


Probability sample:


sample selected using random selection
Non-response:


sample selected not using random selection method
when members of sample are unable or refuse to take part
Census:

data collected from entire population
7
Non probability sampling
8
Types of non-probability sampling: 1
1. Convenience sampling


the most easily accessible individuals
useful when piloting a research questionnaire
2. Snowball sampling


researcher makes initial contact with a small group
these respondents provide names for additional people to
survey
9
Types of non-probability sampling: 2
3. Quota sampling






Often used in market research and opinion polls
Relatively cheap, quick and easy to manage
Often involves stopping people in the street
proportionately representative of a population’s social groups
interviewers select people to fit their quota for each category
Can be biased – friendly people are more likely to be willing to
participate
10
Sampling error


Difference between sample and population
Biased sample does not represent population



Sources of bias



some groups are over-represented;
others are under-represented
When you use non-probability sampling,
When non-response takes place (charateristics of those
who fail to responds, may differ from those that do
Probability sampling reduces sampling error

and allows for use of inferential statistics (t tests etc)
11
Probability Sampling
12
Probability Sampling

Used for quantitative research

Representative of population

Can generalize from sample to population

through use of sampling distribution
13
Logic Behind Probability Sampling

Problem:
The populations we
wish to study are
almost always so
large that we are
unable to gather
information from
every case.




















14
Logic Behind Probability Sampling
(cont.)








Solution:
We choose a sample
-- a carefully chosen
subset of the
population –

Use information
gathered from the
sample to generalize
to the population.

















 

15
Sample size

Absolute size matters


The larger the sample


the more precise and representative it is likely to be
As sample size increases


more than relative size
sampling error decreases
Important to be honest

about the limitations of your sample
25
Factors affecting sample size: 1

Time and cost



after a certain point (n=1000), increasing sample size
produces less noticeable gains in precision
very large samples are decreasingly cost-efficient
(Hazelrigg, 2004)
Non-response



response rate = % of sample who agree to participate
(or % who provide usable data)
responders and non-responders may differ on a crucial
variable (can introduce bias)
26
Factors affecting sample size: 2

Heterogeneity of the population


the more varied the population is, the larger
the sample will have to be
Kind of analysis to be carried out


some techniques require large sample
e.g. contingency table; inferential statistics
27
Limits to generalization

Findings can only be generalized to the
population from which the sample was
selected


be wary of over-generalizing in terms of locality
Time, historical events and cohort effects

results may no longer be relevant and so require
updating (replication)
28
Error in survey research

Sampling error


Sampling-related error



inadequate sampling frame; non-response
makes it difficult to generalize findings
Data collection error



unavoidable difference between sample and population
implementation of research instruments
e.g. poor question wording in surveys
Data processing error

faulty management of data, e.g. coding errors
29
Types of Probability
Sample
30
Types of Probability Sample
1. Simple random sample
2. Systematic sample
3. Stratified random sample
4. Multi-stage cluster sample
31
1. Simple random sampling

Each unit has an equal probability of selection

Sampling fraction: n/N

where n = sample size and N = population size

List all units and number them consecutively

Use random numbers table to select units
32
2. Systematic sampling



Select units directly from sampling frame
From a random starting point, choose
every nth unit (e.g. every 4th name)
Make sure sampling frame has no
inherent ordering

if it has, rearrange it to remove bias
33
3. Stratified random sampling
• Starting point is to categorise population into ‘strata’
• relevant divisions, or departments of companies for example
• Age groups, social class, or income levels are other
examples
• So the sample can be proportionately representative
of each stratum
• Then, randomly select within each category as for a
simple random sample
34
4. Multi-stage cluster sampling

Useful for widely dispersed populations

First, divide population into groups (clusters) of units,

like geographic areas, or industries, for example

Sub-clusters (sub-groups) can then be sampled from
these clusters, if appropriate

Now randomly select units from each (sub)cluster

Collect data from each cluster of units, consecutively
35
END
36