Chapter 8: Sampling techniques

Chapter 8: Sampling techniques

Sampling terminology (expression)

1. Sample: Subset of a larger population
2. Population or universe (world) : Any complete group - People, sales territories, stores
3. Population element: individual member of the population
4. Census (survey) : Investigation of all individual elements that make up a population

1.      Target population

Ø  What is the relevant population?

–        Who can provide us the information we are seeking?  (mouse trap example)

Ø  Needs to be operationally defined

–        Men and women who have bought a mouse trap in the last 6 months

2.      Sampling fame (reputation or recognitation)

A list of elements from which the sample may be drawn (also called the working population)

Examples:

–        Customer mailing lists (internal data base)

–        Members lists (from industry bodies)

–        Phone book

–        Buy one from research companies like Dun and Bradstreet

Sampling frame error: entire population not adequately represented in the sample fra

Sampling unit

1.      Single element or group

 of elements (basic) selected for the sample

2.      For multi-stage sampling:

1. Primary Sampling Units (PSU)

                                                              i.      Example: selection of airline companies

2. Secondary Sampling Units

                                                              i.      Example: selection of airline flights from these companies

3. Tertiary Sampling Units

                                                              i.      Example: selection of passengers from these flights

Error associated with sampling

1.      Random sampling error: chance variations (different)

–        Reduced by increasing sample size

2.      Systematic (non sampling) error: errors in design that creates bias due to unrepresentative(misleading) sample

–        Sampling frame(structure) error: mismatch between sampling frame(structure ) and the population

–        Non response error: some fail to respond, creating a mismatch between actual respondents and the planned sample

Two major categorised of sampling

1.      Probability sampling

–        Known, nonzero probability for every thing (something that is possible or probable).

–        Random sampling error may be accounted for (describe) using confidence intervals.( For every time we feel confidence we may have the chance to make mistake)

2.      Non probability sampling

–        Probability of selecting any particular member is unknown - selection of sampling units may be quite random

 (You choose any product which u may know but that product u will not know which one you are going to choose.)It is an unknown factor we do not know which oen we are going to select.Since it is unknown we are not sure about it.

–        Making inferences about the population are statistically inappropriate. (having an opinion or idea about the population in number are not always correct because most of this counting are done at random.)

–        More suitable for exploratory research

Probability of sampling

 Simple random sampling

1.      A sampling procedure(method) that ensures that each element in the population will have an equal chance of being included in the sample.( When we sample any area for the population counting there is a chance that all the people will be included as a part of the population )

2.      Examples:

1. Pull names from hat.(Pick any name)
2. Rolling a roulette wheel. (taking a name as picking a winner from a lottery)
3. Random digit dialing.(any turning of number to pick the name)

 Systematic sampling

A simple process.Starting point is selected randomly, then every n^th name (sampling interval) from the list will be drawn. For examples:

1. List of students enrolled
2. Phone directory
3. Other innovative methods and systems:

                                                              i.      Drilling of phone book

                                                            ii.      Jelly bean method

Stratified (rank) sampling

Subsamples(subset of sample) are drawn within different level

Each level is more or less equal on some characteristic

Proportionate v. Disproportionate(fair verse unfair) Stratified Sampling

Example:

1. Ensuring adequate representation of different religions (or race)

Cluster (group) sampling

1. The purpose of cluster sampling is to sample economically while retaining the characteristics of a probability sample.
2. The primary sampling unit is no longer the individual element in the population, but a large cluster of elements
3. Involves two or more stages (multi-stage):
1. Random selection of clusters
2. Random selection or sample of elements within the chosen clusters

Nonprobability Sampling

Convenient sampling

1.      Also called random or accidental sampling

2.      The sampling procedure of obtaining the people or units that are most conveniently available

Examples:

–        Mall intercept

–        Lecturers using class for survey

–        Website surveys

Judgment sampling

Ø  Also called purposive sampling

Ø  An experienced individual selects the sample based on his or her judgment about some appropriate characteristics required  of the sample member

Ø  Examples:

o   Basket of products that make up the CPI

o   Basket of currencies in a managed float

o   Selection of customers for experience interviews

o   Selection of mock jurors

Quota sampling

Ø  Ensures  that the various subgroups in a population are represented on pertinent sample characteristics

Ø  To the exact extent that the investigators desire

Ø  It should not be confused with stratified sampling

Ø  Examples:

Ø  Equal numbers of males and females

Ø  Numbers of Telstra, Optus, Vodafone (or in Hong Kong: SmarTone, Peoples, Orange, CSL, New World, Sunday) customers based on current market share data

Snow ball sampling

Ø  Initial respondents are selected by probability methods

Ø  Additional respondents are obtained from information provided by the initial respondents

Ø  Economical way of locating rare populations by referrals

Ø  Example:

o   Locating golfing fanatics, audiophiles, etc. for use in focus groups

What is appropriate sampling design?

Ø  Degree of accuracy: representativeness (symbolized) required

Ø  Resources: costs

Ø  Time: time available to the researcher

Ø  Advanced knowledge of the population: availability of population lists/database

      National versus local: cluster sampling more suited for national projects

      Need for statistical analysis: ability to make inferences

After the sample design is selected

      Determine sample size

      Select actual sample units

      Conduct fieldwork

Determined sample size

What is statistic means?

Ø  Descriptive statistics

o   Used to describe the sample that you have at hand

Ø  Inferential statistics

o   Used on the sample to make conclusion or overview about the bigger population