Sampling Procedures in Marketing Research

1
Sampling Procedures in Marketing Research

• Chapter 12

2
Chapter Topics

• Process of sample planning
• Should we take a census (complete canvas) or a
  sample?
• What kind of sample should be taken?
• What size should the sample be?
• Telephone sampling

3
Sampling Objectives

• Two broad objectives to the use of samples
• Estimation
• Testing of hypotheses
• Each objective involves making inferences about a
  population on the basis of information from a
  sample

4
Precision vs. Accuracy

• Precision in sampling relates to sampling error
  and the SIZE of the confidence limits of an
  estimate (errors of DEGREE)
• Accuracy in sampling relates to non-response
  bias, memory error, misunderstood questions,
  problematic definition of terms, and processing
  errors (errors of KIND)

5
Process of Sample Selection
Define population
Census vs. sample
Sample design
Sample size
Estimate costs of sampling
Execute sampling process
6
Define Population

• Population all units or elements possessing
  particular relevant features or characteristics
  in common, to which one desires to generalize
  study results
• Defining a population requires specifying
• Which elements are included (WHAT)
• Location (WHERE)
• Time frame (WHEN)
• One of the most difficult tasks in sampling

7
Census or Sample?

• Census
• ALL members of the chosen population
• Pros
• Higher level of confidence
• Preferred when
• attribute of interest occurs rarely in the
  population
• Population is small
• Variance in characteristic being measured is high
• Cost of error is high
• Cons
• More timely to complete
• Oftentimes more costly when compared to sampling
• Relatively less effort to complete than sampling

• Sample
• A SUBSET of the chosen population
• Pros
• Less time to complete
• Better for controlling non-sampling errors
• More detailed information possible
• Sample more available than entire population
• Less costly than census
• Cons
• Relatively more effort to complete than census

8
Sample Design

• 1. Are the survey objectives stated precisely?
• 2. Are the eligibility criteria clear and
  definite?
• 3. Are rigorous sampling methods chosen?
• 4. What type of sample should be used?
• 5. What is the appropriate sampling unit?
• 6. What frame is available for the population and
  what problems might arise?
• 7. How are refusals and non-response to be
  handled?

9
Type of Sample

• Probability sampling
• Allows for bias-free selection of sample units
• Permits the measurement of sampling error
• Non-probability sampling
• Relies on the expertise of the person taking the
  sample
• For a given cost, one can normally select a
  larger non-probability sample than probability
  sample

10
Sampling Unit

• The basis of the sampling procedure
• A segment of the population actually chosen by
  the sampling process
• Can be individual elements (e.g. parent) or
  aggregates of individual elements (e.g. entire
  household)

11
Sample Frame

• Usually a physical listing of the sampled
  elements selected
• Incomplete frames or frames too comprehensive can
  lead to coverage error
• Most widely used frame in survey research
  telephone directory

12
Sampling Frame - Population
13
Sample Size

• 4 ways to determine sample size
• Arbitrarily or judgmentally determined
• Minimum cell size needed for analysis
• Budget-based
• Specifying a desired precision in advance

14
Costs of Sampling

• Overhead costs
• Relatively fixed
• Variable costs
• dependent on scope of study
• High costs often lead to changes in sample
  design and/or smaller-sized samples

15
Execute Sampling Process

• A sample must be
• Representative
• Mirrors the various patterns and subclasses of
  the population
• Adequate
• Sufficient size to provide confidence in
  stability of samples characteristics
• Requires precise measurement

16
Non-probability Sampling Procedures

• Quota
• Judgment
• Convenience
• Snowball (multiplicity)
• Definition sample elements do not have a known,
  nonzero chance of being selected for the sample

17
Quota Sampling

• Most commonly employed non-probability sampling
  procedure
• Sizes of various subclasses of the population are
  first estimated from some outside source (ex.
  Bureau of the Census data)

• Advantages
• Lower costs
• Great convenience in selecting respondents

• Disadvantages
• Potentially many sources of selection bias

18
(No Transcript)
19
Judgment Sampling

• Using sound judgment or expertise to carefully
  and consciously choose the elements
• Intent is to select elements in such a way that
  errors of judgment in the selection will cancel
  each other out

• Disadvantages
• No way of knowing if typical cases are, in fact,
  typical without external check or objective basis
• Its value depends entirely upon expert judgment
  of researcher

• Advantages
• Inexpensive
• Convenient to use
• Less time-consuming
• Results as good as probability sampling

20
Judgment Sampling Examples
A company wants to know why its new products
failed and conducts surveys on competitors for
products similar or related to those it produced.
Should a Company begin marketing in a foreign
market? Choose specific international banking
personnel and government trade specialists as a
judgment sample to provide the information.
21
Convenience Sampling

• Generic term for a variety of ad hoc procedures
  for selecting respondents
• Example
• typical cities chosen as test markets due to
  their demographic make-up (New York, Fort Worth)
• intercept interviews in shopping malls
• Tear-out questionnaire in magazine/newspaper

• Disadvantages
• May lead to length-biased data

• Advantages
• Sampling units are accessible, convenient, and
  cooperative

22
Snowball Sampling

• Generic term for a variety of ad hoc procedures
  for selecting respondents
• Example
• Random telephone call and regardless of who
  answers, ask if they know anyone else who meets
  the qualifications of the survey.
• Focus on a list of qualified individuals then
  ask for two referrals.
• Focus on rare groups identified by. Experiences
  (medical history), Interests (expert mountain
  bikers), or Opinions (group membership)

• Advantages
• Sampling units are described as rare, but
  accessible, convenient, and cooperative. Low cost

• Disadvantages
• May lead to sub-classification

23
Probability Sampling Designs

• Simple Random Sampling
• Systematic Sampling
• Stratified Sampling
• Cluster Sampling
• Area Sampling

24
Simple Random Sampling

• Best known type of probability sample
• Each sample element has a known and equal
  probability of selection, and each possible
  sample of n elements has a known and equal
  probability of being the sample actually selected
• It is drawn by a random procedure from a sample
  frame
• Difficult to obtain a sampling frame that permits
  a simple random sample to be drawn
• Not widely used in marketing research

25
Systematic Sampling

• Each sample element has a known and equal
  probability of selection
• If the population contains N ordered elements and
  a sample size n is desired, find the ratio of N/n
  and round to nearest integer to obtain sampling
  interval
• Example population of 600, sample of 60 ?
  sampling interval is 10 ? random number of 4
  selected between 1-10. Sample elements become 4,
  14, 24, etc.
• Assumes that population elements are ordered in
  some fashion (ex. telephone directory, card
  index)
• Can be more reliable (i.e. lower sampling error)
  than simple random sampling

26
Stratified Sampling

• A simple random sample taken from each stratum of
  interest in the population
• In proportionate stratified sampling, sample
  drawn from each stratum is made proportionate in
  size to the relative size of that stratum in the
  total population
• In disproportionate stratified sampling, one
  departs from preceding proportionality by taking
  other circumstances into account (e.g. relative
  size of stratum variances)
• Sample size needs to be calculated for each
  stratum
• Sample selection can be time-consuming and costly
  if many stratums are used

27
Cluster Sampling

• A simple random or stratified sample is selected
  of all primary sample units, each containing more
  than one sample element all elements within the
  selected primary units are then sampled
• Example If researcher chooses to sample city
  blocks and interviews families residing in the
  blocks, the blocks, not the individual families,
  would be selected at random
• Lower interviewing costs

28
Area Sampling

• Sampling of geographical areas (ex. counties,
  blocks)
• If only one level of sampling takes place before
  the elements are sampled, it is a single-stage
  area sample
• If one or more successive samples within the
  larger area are taken before settling on the
  final clusters, it is a multistage area sample

29
Telephone Surveys

• Directory Assisted vs. Random Digit Dialing (RDD)
• Best-known method is the Mitofsky-Waksberg method
  of random digit dialing
• Answering Machines lower chance, but higher
  participation rate
• Caller ID allows people to screen calls
• Kish technique

30
Telephone Sampling Procedures
31
International Research

• International sampling involves the need to
  balance within-country representativeness with
  cross-national comparability
• Well-defined sample frames do not exist in all
  foreign markets
• Comparative and theoretical types of
  international research allows for non-probability
  sampling