RMS 9th djakdjkajd asdjajda (Sampling).ppt

sfdfdsfks l;sdf;lsf

1. SAMPLING OUTLINE
•DEFINITION
•WHY SAMPLING?
•NORMAL DISTRIBUTION
•SAMPLING DESIGN
•PROBABILITY SAMPLING
•NON PROBAB.SAMPLING
•PRECISION AND CONFIDENCE
•CALCULTION OF SAMPLE SIZE

2. DEFINITIONS
•POPULATION:GROUP,EVENTS TOBE
INVESTIGATED
•ELEMENT:AMEMBEROFPOPULATION
•POPULATION FRAME:LISTOFELEMENTS
INTHEPOPULATIONFROM WHICHTHE
SAMPLEISDRAWNE.G.DIRECTORY
•SAMPLE:SAMPLEISASUBSETOFPOPULATION.
ITCOMPRISESSOMEMEMBERSSELECTEDFROM
IT.SOMEBUTNOTALLELEMENTS OFTHE
POPULATIONWOULDFROMTHESAMPLE.
•SUBJECT: AN ELEMENT OF A SAMPLE OR A
SINGLE MEMBER OF THE SAMPLE

3. DEFINITIONS CON…
•SAMPLING :
ITISAPROCESSOFSELECTINGASUFFICIENT
NUMBEROFELEMENTSFROMTHEPOPULATION
TOUNDERSTAND, ANALYSEITSPROPERTIES,TO
GENERALIZE FORTHEWHOLE POPULATION.
SAMPLING REDUCES EFFORT ANDCOSTIF
POPULATIONISLARGE.
THEPARAMETERS OFTHEPOPULATION ARE
(MEAN,SDANDVARIANCE)ANDTHESAME
STATISTICSAREFORSAMPLEBUTSYMBOLSARE
DIFFERENT

4. DEFINITIONS CON…
REPRESENTATIVES
INAREPRESENTATIVE SAMPLEITS
CHARACTERISTICS ARETHESAMEAS
THOSEOFTHEPOPULATION
•SAMPLING DESIGN:
FORDESIGNOFSAMPLEONENEEDS
TARGET POPULATION, PARAMETER TO
STUDY,SAMPLINGFRAME,SAMPLESIZE,
TIMEANDRESOURCES REQUIRED

5. PROBABILITY AND NON -PROBABILITY
SAMPLING
PROBABILITY:
UNRESTRICTED (SIMPLE RANDOM)
RESTRICTED OR COMPLEX PROBABILTY SAMPLING
SYSTEMATIC
STRATIFIED RANDOM: (PROPORTIONATE AND
DISPROPORTIONATE)
CLUSTER:(SINGLE STAGE AND MULTISTAGE)
AREA
DOUBLE
NON-PROBABILITY:
CONENIENCE
JUDGEMENTAL
QUOTA

6. PROBABILITY SAMPLING
EACHELEMENT HASSAMECHANCE OF
BEING SELECTED, USED WHEN
REPRESENTATIVE SAMPLEISIMPORTANT
•UNRESTRICTED/ SIMPLE RANDOM SAMPL:
EACHELEMENTHASKNOWNANDEQUALCHANCE
OFBEINGSELECTED
•LEASTBIASANDMOSTGENERALIZABLE.

7. RESTRICTED PROBA..
•SYSTEMATIC:
DRAW Nth ITEM RANDOMLY IN THE POPULATION,
EFFICIENT AND USED FOR ATTITUDE SURVEYS
ETC.
•STRATIFIED RANDOM:
ITHELPSTOESTIMATEPOPULATIONPARAMETERS, THERE
MAYBEIDENTIFIABLESUBGROUPS OFELEMENTSWITHIN
THEPOPULATION.E.G.TRAININGPROGRAM

8. PROBABILITY S. CONT…
•PROPRTIONATE STRATIFIED RANDOM:
PROP.SELECTIONFROMEACHGROUPE.G.JOB
LEVELS
•DISPROPORTIONATE STRAT.RANDOM:
DISPROPORTIONATE SAMPLINGDECISIONSARE
MADEEITHERWHENGROUPS/SECTIONS ARETOO
SMALLORTOOLARGE.ITISALSOSOMETIMES
DONEWHENITISEASIER,SIMPLERANDLESS
EXPENSIVETOCOLLECTDATAFROMONEOR
MORESUBGROUPS.
•CLUSTER:
ITISUSEDWHENLISTOFPOPULATIONISNOT
AVAILABLE,ITISLEASTEXPENSIVEANDLEAST
DEPENDABLE

9. PROBA.SAMPLING CON..
•MULTISTAGE CLUSTER:
CLUSTERINEACHAREAANDSUBCLUSTERSAND
RANDOMSELECTION
•AREA SAMPLING:
POPULATION WITHINEACHGEOGRAPHICAL
CLUSTER,LESSCOSTLY
•DOUBLE SAMPLING:
1stSAMPLEFORPRELEMINARY INFORMATIONOF
INTEREST,2
nd
TIMESAMPLEUSEDFORFURTHER
DETAIL

10. NON PROBAB. SAMPLING
•ELEMETS PROBABILITY OF SELECTION NOT
KNOWN, FOR QUICK FINDINGS
•CONVENIENCE : EASILY AVAILABLE SAMPLE
ELEMENTS TAKEN
•PURPOSIVE: CONFINED TO SPECIFIC
GROUP WHO CAN PROVIDE DESIRED
INFORMATION
•JUDGEMENT: BEST PEOPLE TO PROVIDE
INFORMATION
•QUOTA: ENSURE CERTAIN PEOPLE ARE
REPRESENTED IN A STUDY BY QUOTA

11. PRECISION AND CONFIDENCE
•PRECISION:
HOW CLOSE OURESTIMATE ISTOTRUE
POPULATIONCHARACTERISTIC STATSSTANDARD
SAMPLINGERROR=Sx=S/SQUAREROOT[n]
•CONFIDENCE:
PRECISIONDENOTESHOWCLOSEWEESTIMATE
THEPOPULATIONPAPRAMETERS BASEDONTHE
SAMPLESTATISTIC.CONFIDENCEDENOTES HOW
CERTAINWEARETHATOURESTIMATEREALLY
TRUEFORPOPULATIONE.G95%.
LARGER THESAMPLE SIZEHIGHER THE
PRECISIONORSMALLERTHESAMPLINGERROR.

12. PRECISION AND CONFID…
•NO SAMPLE HAS EXACTLY SAME
CHARACTERISTICS AS POPULATION
•PROBABILITY SAMPLING COMES CLOSER
TO POPULATION STATISTICS
•X,S,S^2 MEAN,STANDARD DEV., VARIANCE
OF SAMPLE
•U,SIGMA,SIGMA^2 OF POPULATION
•n SAMPLE SIZE,N POPULATION

13. SAMPLE SIZE
•EFFECTED BY VARIABILITY OF
POPULATION
•PRECISION/ACCURACY NEEDED
•COST/BENEFITOFINVESTIGATION
•MOSTRESEARCH SAMPLESSIZE>30AND
<500
•FORSUBSAMPLES 3OINEACHCATEGORY
•FORMULTIPLEREGRESSION ANALYSIS
SAMPLE SIZE10TIMESNUMBER OF
VARIABLES
•EXPERIMENTAL RESEARCH SAMPLESIZE
10-20