SAMPLING

A method by which some units/items of a given population/occurrence are selected as representatives of the entire population.

Term used in Sampling

Population Total number of units/people/ occurrences under study.

Element Individual member/unit of population

Sample frameA known list of elements from which the sample is actually drawn.

SampleRepresentative part of the whole/population under study.

Subset of the population that represents the entire population.

They have similar characteristics of population.

Subject Individual member of sample.

The Sampling Design Process

a. Reduces costs, labour and time Quality Management/Supervision Accuracy and Reliability of Results Sampling may be the only way (bulbs)

Deming argues that the quality of study is often better with sampling than with census.

Further he says that it is good to interview few in a nicer way than to cover everybody in population.Why sampling?

Sampling MethodsProbability:In which each and every member of the population gets equal/non zero chance to become the part of the sample.

Used when we know our elements OR population frame

Non probability:In which every member/unit from the population does not get equal chance of being selected in the sample.

Used when we do not know our elements OR population frame

SamplingProbability Simple Random stratified cluster nonsystematicsystematic Non Probability quotasnowballjudgmentconvenience

Probability sampling methods

Random Sampling Method

It is divided in to : A. Systematic and B. Non systematic

Systematic Random Sampling:In which an initial starting point is selected by a random process and than every nth number is selected.

Example: If we want to have a sample size of 50 houses from the population of 500, then we can have sample from every 10th house.

Simple Random SamplingThis implies that every number is selected independently of every other element. This method is equivalent to a lottery system.

Stratified Sampling MethodA probability sampling technique that uses two step process to partition into subpopulation or strata .

Divide sampling frame into homogeneous subgroups (strata) e.g. age-group, occupation etc.

Draw random sample in each strata.Used for large population without distance e.g Study of Students of Diff Departments of Karachi University.

Steps Involved in Stratified Sampling1. Divide the population into stratas or groups.

2. Identify the population in each strata.

3. Select the number of respondents either proportionately or disproportionately.

4. Select final respondents by applying simple random sampling method

Total PopulationMale 60Female 4010% = 610% = 4100 students: 10%Selecting Numbers of Respondents by Proportionate ( Size )

Larger the size of the group the more we select, the smaller the size of strata the less we select.Strata-1Strata-2

Cluster SamplingSame as stratified, but used when the population is large and dispersed, e.g study of Faculty Members in Universities of Pakistan or study of the farmers of Pakistan who are cultivating wheat.

Cluster Sampling

Faculty Members in Pak UniversitiesPunjabSindhNWFPBalochistanMaleFemaleMaleFemaleMaleFemaleMaleFemaleSeniorJuniorQualifiedNon qualified

Non probability Sampling.Each and every member from the population does not get the equal chance of being selected in the sample.

Convenience Here the samples are drawn on the convenience of the researcher.

According to most convenient location, time, etc. respondents are selected.

Convenience sampling may misrepresent the population.

JudgmentIn judgment sampling researcher uses his/ her own educated guess or judgment to identify who will be in the sample.

Snow ballIt is commonly used when it is difficult to identify members of the desired population.

Make contact with one or two respondents in the population. Ask these respondents to identify further new respondents and so on. And this process of obtaining data by initial respondent , and then from referral to referral is called as snow ball.

Quota The quota sample establishes a specific quota or percentage for various types of individuals to be interviewed.

This can be included in prob and non prob sampling.

Quota sampling may be viewed as two-stage restricted judgmental sampling. The first stage consists of developing control categories, or quotas, of population elements (male and female). In the second stage, sample elements are selected based on convenience or judgment (if it is non prob sampling).

Population Composition Sample Control CharacteristicNumber Percentage 200 Male 60060 120 Female 40040 80 ________ ____ 1000100 200

Strength and weakness of sampling techniquesConvenience

Judgmental

Quota

Snow Ball

strengthweaknessLeast expensive, least time consuming, most convenient Selection biasness, sample is not representative of (P)Low cost, convenient , less time consumingDoesnt allow generalization, subjective instead of objectiveSample can be controlled from certain characteristics.Selection bias, no assurance of representative.Can estimate rare characteristicsTime consuming

Strength and weakness of sampling techniquesSimple

Random

Systematic

Stratified

Cluster

Strength Weakness Easily understood, results are projectableDifficult to construct sampling frame, expensive, lower precision, no assurance of representativeCan increase representative ness, easier to implement, than Srs, Sampling frame not necessary.Can decrease representativeIncludes all important subpopulation, precision.Difficult to select relevant stratification variable, expensive,not feasible to verify many variables.Cost effective , easy implementLow statistical efficiency

Sample Size:Factors to determine sample size1. Cost2. Time3. Importance of decision4. Reliability requirements5. Population size6. Nature of the problem7. Diversity of population

Sample size:It is believed that larger the sample size, greater the extent of the reliability of data.

The size of sample depends on:The characteristics of population the type of info requiredThe cost involved etc

Roscoe (1975) proposes the following rule of thumb:

Sample size larger than 30 and less than 500 are appropriate for most of the research.

Having a sample size of 5000 is not necessarily better than having a sample size of 500.

In UK, national surveys of house wives buying habits, a sample size of 2000 was used and same done in Europe.

Example:In UK, more than 10 mln ballots were mailed, of which 3 million were returned. Of these, 41% supported Theature and 55% favored opponent .

But in actual Theature won.

Sample Sizes Used in Research Studies

Type of Study

Minimum Size

Typical Range

Problem identification research (e.g. market potential)

500

1,000-2,500

Problem-solving research (e.g. pricing)

200

300-500

Product tests

200

300-500

Test marketing studies

200

300-500

TV, radio, or print advertising (per commercial or ad tested)

150

200-300

Test-market audits

10 stores

10-20 stores

Focus groups

2 groups

4-12 groups

Type I and type II Error If we have the sample size too small, the sampling error might be so large (hypotheses which is actually true will be rejected) It is called type II error.

The other error the researcher makes is to accept the hypothesis, when it is actually false. This is known as type II error.

THE

END

*************************************