Lecture 9 SAMPLING DESIGN AND PROCEDURE. Population and Sample Population –The entire group that the researcher wishes to investigate Element –A single.

Lecture 9Lecture 9

SAMPLING DESIGNSAMPLING DESIGN AND PROCEDUREAND PROCEDURE

Population and SamplePopulation and Sample

PopulationPopulation–The entire group that the The entire group that the

researcher wishes to investigateresearcher wishes to investigateElementElement–A single member of the A single member of the

populationpopulation

Population (Sampling) FramePopulation (Sampling) Frame

– A listing of all the elements in the A listing of all the elements in the population from which the sample is drawnpopulation from which the sample is drawn

SampleSample

– A subset of the populationA subset of the population

SubjectSubject

– A single member of the sampleA single member of the sample

Population and SamplePopulation and Sample

CENSUSCENSUS

INVESTIGATION OF ALL INDIVIDUAL INVESTIGATION OF ALL INDIVIDUAL ELEMENTS THAT MAKE UP A ELEMENTS THAT MAKE UP A POPULATIONPOPULATION

When Is A Census Appropriate?When Is A Census Appropriate?

NecessaryFeasible

TARGET POPULATIONTARGET POPULATION

RELEVANT POPULATIONRELEVANT POPULATION

OPERATIONALLY DEFINEOPERATIONALLY DEFINE

COMIC BOOK READER?COMIC BOOK READER?

Why Sample?Why Sample?

Greater accuracy

Availability of elementsAvailability of elements

Greater speed

Greater speed

Sampling provides

Sampling provides

Lower costLower cost

SAMPLING FRAMESAMPLING FRAME

A LIST OF ELEMENTS FROM WHICH A LIST OF ELEMENTS FROM WHICH THE SAMPLE MAY BE DRAWNTHE SAMPLE MAY BE DRAWN

WORKING POPULATIONWORKING POPULATION

MAILING LISTS - DATA BASE MAILING LISTS - DATA BASE MARKETERSMARKETERS

SAMPLING FRAME ERRORSAMPLING FRAME ERROR

SamplingSampling

Process of selecting a sufficient Process of selecting a sufficient number of elements from the number of elements from the populationpopulation

Reasons for Sampling: practicality Reasons for Sampling: practicality (time and resources), destructive (time and resources), destructive samplingsampling

Need for a representative sample Need for a representative sample

SAMPLING UNITSSAMPLING UNITS

GROUP SELECTED FOR THE SAMPLEGROUP SELECTED FOR THE SAMPLE

PRIMARY SAMPLING UNITS (PSU)PRIMARY SAMPLING UNITS (PSU)

SECONDARY SAMPLING UNITSSECONDARY SAMPLING UNITS

TERTIARY SAMPLING UNITSTERTIARY SAMPLING UNITS

TWO MAJOR CATEGORIES OF TWO MAJOR CATEGORIES OF SAMPLINGSAMPLING

PROBABILITY SAMPLINGPROBABILITY SAMPLINGKNOWN, NONZERO PROBABLITY FOR KNOWN, NONZERO PROBABLITY FOR EVERY ELEMENTEVERY ELEMENT

NONPROBABLITY SAMPLINGNONPROBABLITY SAMPLINGPROBABLITY OF SELECTING ANY PROBABLITY OF SELECTING ANY PARTICULAR MEMBER IS UNKNOWNPARTICULAR MEMBER IS UNKNOWN

Probability and NonprobabilityProbability and NonprobabilitySamplingSampling

Probability SamplingProbability Sampling– Elements in the population have known Elements in the population have known

chance of being chosenchance of being chosen– Used when the representativeness of the Used when the representativeness of the

sample is of importancesample is of importance

Nonprobability SamplingNonprobability Sampling– The elements do not have a known or The elements do not have a known or

predetermined chance of being selected as predetermined chance of being selected as subjectssubjects

Probability SamplingProbability Sampling

Unrestricted/Simple Random SamplingUnrestricted/Simple Random Sampling– Every element in the population has a known and equal Every element in the population has a known and equal

chance of being selected as a subjectchance of being selected as a subject– Has the least bias and offers the most generalizabilityHas the least bias and offers the most generalizability

Restricted/Complex Probability SamplingRestricted/Complex Probability Sampling • Systematic SamplingSystematic Sampling• Stratified Random SamplingStratified Random Sampling• Cluster Sampling (USM, UM, etc)Cluster Sampling (USM, UM, etc)• Area SamplingArea Sampling• Double Sampling (USM and then grad students)Double Sampling (USM and then grad students)

PROBABLITY SAMPLINGPROBABLITY SAMPLING

SIMPLE RANDOM SAMPLESIMPLE RANDOM SAMPLE

SYSTEMATIC SAMPLESYSTEMATIC SAMPLE

STRATIFIED SAMPLESTRATIFIED SAMPLE

CLUSTER SAMPLECLUSTER SAMPLE

MULTISTAGE AREA SAMPLEMULTISTAGE AREA SAMPLE

SIMPLE RANDOM SIMPLE RANDOM SAMPLING SAMPLING

a sampling procedure that ensures that a sampling procedure that ensures that each element in the population will have each element in the population will have an equal chance of being included in an equal chance of being included in the samplethe sample

Simple RandomSimple Random

AdvantagesAdvantages Easy to implement Easy to implement

with random with random dialingdialing

DisadvantagesDisadvantages Requires list of Requires list of

population population elementselements

Time consumingTime consuming Uses larger sample Uses larger sample

sizessizes Produces larger Produces larger

errorserrors High costHigh cost

SYSTEMATIC SAMPLING SYSTEMATIC SAMPLING

A simple processA simple process

every every nnth name from the list will be drawnth name from the list will be drawn

SystematicSystematic

AdvantagesAdvantages Simple to designSimple to design Easier than simple Easier than simple

randomrandom Easy to determine Easy to determine

sampling sampling distribution of distribution of mean or proportionmean or proportion

DisadvantagesDisadvantages Periodicity within Periodicity within

population may population may skew sample and skew sample and resultsresults

Trends in list may Trends in list may bias resultsbias results

Moderate costModerate cost

STRATIFIED SAMPLINGSTRATIFIED SAMPLING

Probability sampleProbability sample

Subsamples are drawn within different Subsamples are drawn within different stratastrata

Each stratum is more or less equal on Each stratum is more or less equal on some characteristicsome characteristic

Do not confuse with quota sampleDo not confuse with quota sample

StratifiedStratified

AdvantagesAdvantages Control of sample size Control of sample size

in stratain strata Increased statistical Increased statistical

efficiencyefficiency Provides data to Provides data to

represent and analyze represent and analyze subgroupssubgroups

Enables use of Enables use of different methods in different methods in stratastrata

DisadvantagesDisadvantages Increased error will Increased error will

result if subgroups are result if subgroups are selected at different selected at different ratesrates

Especially expensive if Especially expensive if strata on population strata on population must be created must be created

High costHigh cost

CLUSTERCLUSTER SAMPLINGSAMPLINGThe purpose of cluster sampling is to The purpose of cluster sampling is to sample economically while retaining sample economically while retaining the characteristics of a probability the characteristics of a probability sample.sample.The primary sampling unit is no longer The primary sampling unit is no longer the individual element in the the individual element in the population. population. The primary sampling unit is a larger The primary sampling unit is a larger cluster of elements located in proximity cluster of elements located in proximity to one another.to one another.

Population Element Possible Clusters in Malaysia

Malaysian adult population StatesDistrictsMetropolitan Statistical AreaCensus tractsBlocksHouseholds

EXAMPLES OF CLUSTERS

Population Element Possible Clusters in Malaysia

College seniors CollegesManufacturing firms Districts

Metropolitan Statistical AreasLocalitiesPlants

EXAMPLES OF CLUSTERS

Population Element Possible Clusters in Malaysia

Airline travelers AirportsPlanes

Sports fans Football stadiaBasketball arenasBaseball parks

EXAMPLES OF CLUSTERS

Cluster Cluster

AdvantagesAdvantages Provides an unbiased Provides an unbiased

estimate of population estimate of population parameters if properly parameters if properly donedone

Economically more Economically more efficient than simple efficient than simple randomrandom

Lowest cost per Lowest cost per samplesample

Easy to do without listEasy to do without list

DisadvantagesDisadvantages Often lower statistical Often lower statistical

efficiency due to efficiency due to subgroups being subgroups being homogeneous rather homogeneous rather than heterogeneousthan heterogeneous

Moderate costModerate cost

Stratified and Cluster SamplingStratified and Cluster Sampling

StratifiedStratified Population divided Population divided

into few subgroupsinto few subgroups Homogeneity Homogeneity

within subgroupswithin subgroups Heterogeneity Heterogeneity

between subgroupsbetween subgroups Choice of elements Choice of elements

from within each from within each subgroupsubgroup

ClusterCluster Population divided Population divided

into many into many subgroupssubgroups

Heterogeneity Heterogeneity within subgroupswithin subgroups

Homogeneity Homogeneity between subgroupsbetween subgroups

Random choice of Random choice of subgroups subgroups

DoubleDouble

AdvantagesAdvantages May reduce costs if May reduce costs if

first stage results first stage results in enough data to in enough data to stratify or cluster stratify or cluster the populationthe population

DisadvantagesDisadvantages Increased costs if Increased costs if

discriminately useddiscriminately used

Nonprobability SamplesNonprobability Samples

Cost

FeasibilityFeasibility

TimeTime

IssuesIssues

No need to generalize

Limited objectivesLimited

objectives

Nonprobability Nonprobability Sampling MethodsSampling Methods

ConvenienceConvenience

JudgmentJudgment

QuotaQuota

SnowballSnowball

NONPROBABLITY SAMPLINGNONPROBABLITY SAMPLING

CONVENIENCECONVENIENCE

JUDGMENTJUDGMENT

QUOTAQUOTA

SNOWBALLSNOWBALL

Nonprobability SamplingNonprobability Sampling

Convenience SamplingConvenience Sampling– Based on availability, e.g. students in a Based on availability, e.g. students in a

classroomclassroom

Purposive SamplingPurposive Sampling– Specific targets, because they posses the Specific targets, because they posses the

desired infodesired infoJudgement samplingJudgement samplingQuota samplingQuota sampling

CONVENIENCE SAMPLINGCONVENIENCE SAMPLING

also called haphazard or accidental also called haphazard or accidental samplingsampling

the sampling procedure of obtaining the the sampling procedure of obtaining the people or units that are most people or units that are most conveniently availableconveniently available

QUOTA SAMPLING QUOTA SAMPLING

ensures that the various subgroups in a ensures that the various subgroups in a population are represented on pertinent population are represented on pertinent sample characteristicssample characteristics

to the exact extent that the investigators to the exact extent that the investigators desiredesire

it should not be confused with stratified it should not be confused with stratified samplingsampling

JUDGMENT SAMPLING JUDGMENT SAMPLING

also called purposive sampling also called purposive sampling

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

SNOWBALL SAMPLING SNOWBALL SAMPLING

a variety of procedures a variety of procedures

initial respondents are selected by initial respondents are selected by probability methods probability methods

additional respondents are obtained from additional respondents are obtained from information provided by the initial information provided by the initial respondentsrespondents

Area SamplingArea Sampling

Sample SizeSample Size

Factors Determining Sample SizeFactors Determining Sample Size

– Homogeneity of populationHomogeneity of population

– Level of confidenceLevel of confidence

– PrecisionPrecision

– Cost, Time and Resources Cost, Time and Resources

Larger Sample SizesLarger Sample Sizes

Small error range

Number of subgroupsNumber of subgroups

Confidence level

Confidence level

WhenWhen

Population variance

Desired precisionDesired

precision

Roscoe’s Rule of ThumbRoscoe’s Rule of Thumb

>30 and <500 appropriate for most >30 and <500 appropriate for most researchresearch

Not less than 30 for each sub-sampleNot less than 30 for each sub-sample

In multivariate analysis, 10 times or more In multivariate analysis, 10 times or more the number of variablesthe number of variables

Simple experiment with tight controls, 10-Simple experiment with tight controls, 10-20 quite sufficient20 quite sufficient

WHAT IS THE APPROPRIATE WHAT IS THE APPROPRIATE SAMPLE DESIGNSAMPLE DESIGN

DEGREE OF ACCURACYDEGREE OF ACCURACY

RESOURCESRESOURCES

TIMETIME

ADVANCED KNOWLEDGE OF THE ADVANCED KNOWLEDGE OF THE POPULATIONPOPULATION

NATIONAL VERSUS LOCALNATIONAL VERSUS LOCAL

NEED FOR STATISTICAL ANALYSISNEED FOR STATISTICAL ANALYSIS

What Is A Good Sample?What Is A Good Sample?

PreciseAccurate

AFTER THE SAMPLE DESIGN AFTER THE SAMPLE DESIGN IS SELECTEDIS SELECTED

DETERMINE SAMPLE SIZEDETERMINE SAMPLE SIZE

SELECT ACTUAL SAMPLE UNITSSELECT ACTUAL SAMPLE UNITS

CONDUCT FIELDWORKCONDUCT FIELDWORK

SYSTEMATIC ERRORSSYSTEMATIC ERRORS

NONSAMPLING ERRORSNONSAMPLING ERRORS

UNREPRESENTATIVE SAMPLE UNREPRESENTATIVE SAMPLE RESULTSRESULTS

NOT DUE TO CHANCENOT DUE TO CHANCE

DUE TO STUDY DESIGN OR DUE TO STUDY DESIGN OR IMPERFECTIONS IN EXECUTIONIMPERFECTIONS IN EXECUTION

ERRORS ASSOCIATED ERRORS ASSOCIATED WITH SAMPLINGWITH SAMPLING

SAMPLING FRAME ERRORSAMPLING FRAME ERROR

RANDOM SAMPLING ERROR RANDOM SAMPLING ERROR

NONRESPONSE ERRORNONRESPONSE ERROR

RANDOM SAMPLING ERRORRANDOM SAMPLING ERROR

THE DIFFERENCE BETWEEN THE THE DIFFERENCE BETWEEN THE SAMPLE RESULTS AND THE RESULT SAMPLE RESULTS AND THE RESULT OF A CENSUS CONDUCTED USING OF A CENSUS CONDUCTED USING IDENTICAL PROCEDURESIDENTICAL PROCEDURES

STATISTICAL FLUCTUATION DUE TO STATISTICAL FLUCTUATION DUE TO CHANCE VARIATIONSCHANCE VARIATIONS

Define the target population

Select a sampling frame

Conduct fieldwork

Determine if a probability or nonprobability sampling method

will be chosen

Plan procedure for selecting sampling units

Determine sample size

Select actual sampling units

Stages in the Selectionof a Sample

End of lessonEnd of lesson