Scribd
Upload
46 views

Citation: Michael J. Baker University of Strathclyde Sampling The Marketing Review 2002, 3, 103-120

This document discusses different sampling methodologies for research. There are two broad approaches - probability based samples and non-probability based samples. Probability based samples include random samples, stratified samples, cluster samples, and systematic samples. Random samples give each unit an equal chance of selection, while stratified samples divide the population into subgroups and sample proportionately from each. Cluster samples pick subgroups then sample from within, and systematic samples select units at regular intervals. Non-probability samples like quota and convenience sampling rely on the researcher's judgment in selecting units based on predetermined criteria rather than random selection. The key advantage of probability samples is they allow statistical inferences about the overall population.

Uploaded by

prabin ghimire
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
46 views

Citation: Michael J. Baker University of Strathclyde Sampling The Marketing Review 2002, 3, 103-120

This document discusses different sampling methodologies for research. There are two broad approaches - probability based samples and non-probability based samples. Probability based samples include random samples, stratified samples, cluster samples, and systematic samples. Random samples give each unit an equal chance of selection, while stratified samples divide the population into subgroups and sample proportionately from each. Cluster samples pick subgroups then sample from within, and systematic samples select units at regular intervals. Non-probability samples like quota and convenience sampling rely on the researcher's judgment in selecting units based on predetermined criteria rather than random selection. The key advantage of probability samples is they allow statistical inferences about the overall population.

Uploaded by

prabin ghimire
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Citation:

Michael J. Baker University of Strathclyde


Sampling
The Marketing Review 2002, 3, 103-120

Introduction:
There are three basic methodologies available to researchers observation, experimentation and survey. The acquisition of additional
information is a resource-hungry process which requires the decisionmaker to balance some element of risk or uncertainty against the time
and money which would be necessary to reduce this still further. Sampling
offers us a means of doing this in an acceptable way and in this paper we
shall look at the basic ideas underlying sampling, the different kinds of
sample available, their various advantages and disadvantages, and the
planning of a sampling operation. Such technical matters as calculating
sample size, variability, error and bias are outside the scope of an
overview of this kind but references to appropriate sources are provided
for those who wish to explore these issues further.
The Basis of Sampling
In the case of a manufacturing process the population to be sampled for
quality control purposes could be every unit of output within a given time
period, while for the provision of a social service for elderly persons it
could be every individual aged 65 and over in a defined geographical
area. With the result that in some circumstances the results from a
properly controlled sample may be more accurate than an attempted
census. In addition to reliability the other acid test of sample design is
validity by which is meant the degree to which the survey measures that
which it purports to measure.
The concepts of reliability and validity are a frequent source of confusion a
very clear and precise definition is offered by Martin and Bateson (1986).
1) Reliability concerns the extent to which measurement is repeatable
and consistent; that is, free from random errors. An unbiased
measurement consists of two parts: a systematic component,
representing the true value of the variable, and a random
component due to imperfections in the measurement process. The
smaller the error component, the more reliable the measurement
reliable measures, sometimes referred to as good measures, are
those which measure a variable precisely and consistently. At least
1 | Page

four related factors determine how 'good' a measure is:


a. Precision: How free are measurements from random errors?
This is denoted by the number of 'significant figures' in the
measurement. Note that accuracy and precision are not
synonymous: accuracy concerns systematic error (bias) and can
therefore be regarded as an aspect of validity (see below). A
clock may tell the time with great precision (to within a
millisecond), yet be inaccurate because it is set to the wrong
time.
2. Validity concerns the extent to which a measurement actually
measures those features the investigator wishes to measure, and
provides information that is relevant to the questions being asked.
(a) Accuracy: Is the measurement process unbiased, such that
measured values correspond with the true values? Measurements
are accurate if they are relatively free from systematic errors
(whereas precise measurements are relatively free from random
errors).
(b)Specificity: To what extent does the measure describe what it is
supposed to describe, and nothing else?
As noted, the sampling frame is a defined population from which the
sample is to be drawn and so must be accurate, adequate, up to date and
relevant to the purposes of the survey for which it is to be used. Given a
sampling frame, and having defined the sampling unit, the next step is to
determine the most cost-effective way of selecting specific sampling units
from the sampling frame. In the absence of a sampling frame one cannot
draw a probability based sample and so will have to resort to some
judgemental or non-probabilistic method. Such sampling is often referred
to as purposive and alternative approaches will be reviewed below.
Probability-based Samples
Random Samples:An unrestricted random sample is one in which every unit has an equal
chance of selection and where the selected unit is replaced before another
is drawn, i.e. the same unit could occur more than once in a sample. The
great advantage of random sampling is the ease with which the sampling
error may be calculated
Stratified Samples
As the term suggests, in the case of proportionate stratified sampling the
amount of the sample drawn from each stratum is proportionate to the
stratums share of the total population, whereas in the case of
2 | Page

disproportionate sampling one varies the proportion within each stratum


in accordance with a criterion or criteria which reflects the variability
within the strata. While a disproportionate sampling procedure increases
the complexity of the statistical calculations necessary to arrive at an
unbiased sample estimate, the improvement in efficiency (quality of
output) is well worth the effort, for in addition to being able to make
statements about the population as a whole; one can make statements
about each of the strata individually.
Cluster Sampling
Cluster sampling is similar to stratified sampling in that both techniques
require the researcher to sub-divide a population into a set of mutually
exclusive and exhaustive sub-groups. The methods differ in that in cluster
sampling one samples the sub-groups whereas in stratified sampling one
selects a sample from within each sub-group.
Systematic Sampling
Another form of random sampling which is widely used is the systematic
sample. This method has advantages over simple random sampling in that
it is not necessary to number every unit in the sampling frame. Instead
one divides the population by the required sample size to determine the
sampling interval {'k'} and then selects a random starting point where
after every item is selected systematically, e.g.
Population = 2000, sample = 200 2000
k = 200 = 10
Random start = 7, therefore sample = 7, 17, 27, 37, 47..., 1997
Non-Probability Based or Purposive Samples
The obvious benefit of probability-based sampling procedures is that they
allow one to draw inferences about the population from which the sample
was drawn and state these with a known degree of confidence that any
similarly chosen sample would yield the same results as that given by the
present sample. The need for simpler and less expensive sampling
procedures is largely met by judgemental approaches in which a sample
is selected for a particular purpose (hence purposive sampling) with the
two main techniques being quota and convenience sampling. The
essential difference between quota and random sampling is that in the
case of quota sampling the interviewer selects respondents in accordance
with some predetermined criteria such as age, sex and occupation, while
in the random sample the respondents are selected (using the same
criteria) by an objective methodology independent of the interviewer.
Quota Sampling

3 | Page

This is used when we want to obtain a given proportion of respondents


in all social classes, sex and age groups. This is usually designed to
match the overall proportion of respondents in all classes, sex and age
groups. Your instructions will include a quota sheet stating how many
men and women in each class, sex and age group are to be
interviewed.
Summary:Two broad approaches to sampling are available probability based or
random samples and non- probability based or purposive samples. . In a
true random sample it must be possible to list every member of the
population to be surveyed and then select the sampling units (ESU) in
such a manner that each has a known and non-zero chance of being
included. A variety of techniques including simple random samples,
stratified samples, cluster-samples and systematic samples were
reviewed as were judgement, quota and convenience samples as nonprobability based approaches.

4 | Page

You might also like