research methodology class
TRANSCRIPT
-
8/8/2019 Research Methodology Class
1/28
Chapter ObjectivesChapter Objectives
12/12/201012/12/2010 Biju K ThapaliaBiju K Thapalia 11
Introduction
Sampling Techniques
Sampling Design
Sample Size
Sampling Errors Sampling Distribution
-
8/8/2019 Research Methodology Class
2/28
To show how estimation is done.
-
8/8/2019 Research Methodology Class
3/28
Sampling is that part of statistical
practice concerned with the selection of
an unbiased or random subset of
individual observations within a
population of individuals intended to
yield some knowledge about thepopulation of concern, especially for the
purposes of making predictions based on
statistical inference.
Sampling is the act, process, or technique of
selecting a suitable sample, or arepresentative part of a population for the
purpose of determining parameters or
characteristics of the whole population.
-
8/8/2019 Research Methodology Class
4/28
A census studydraws inferences from the
entire bodyofunits of interest (thepopulation)
A sample studydraws inferences from a
sample drawnfrom the population
-
8/8/2019 Research Methodology Class
5/28
Basic reasons-
Cost is too highPopulation is dynamic, it may change
Time limitation
Improve qualityofdata
-
8/8/2019 Research Methodology Class
6/28
Defining Population
Specifying a samplingframe, a setof items or
events possible to measure
Specifying a sampling methodfor selectingitems or events from the frame
Determiningthe sample size
Implementingthe sampling plan
Sampling anddata collecting
-
8/8/2019 Research Methodology Class
7/28
ProbabilitySampling
Probability sampling is anobjective procedure
inwhich the probabilityofselection is known
in advance for each population unit
Non-probabilitySampling
Nonprobability sampling is a subjective
procedure inwhich the probabilityofselection
for each population unit is unknown
beforehand
-
8/8/2019 Research Methodology Class
8/28
Non-probability
Quota
Snowball
Convenience
Purposive
y Probability
y Simple random
y
Systematicy Stratifiedrandom
y Cluster
y Multi-stage
-
8/8/2019 Research Methodology Class
9/28
Each elementofthe population have equal and
independentopportunityto be selected
Entire population should be enumeratedUse ofLoterry or Random numbers table
Useful for homogeneous populationw.r.t
specific characteristics
Sampling error is more for same sample size-
so large sample size needed
-
8/8/2019 Research Methodology Class
10/28
For population is large anddifficultto list all
Arrangementofelements on some pattern
w.r.t studyfeatureChoose anyfirst element andthenrepeatnext
kth element and soon
kdepends on sample size(n) and population
size (N); k=N/n
Easyto administer
-
8/8/2019 Research Methodology Class
11/28
For hetrogenous population
Population is divided into strataSimple random samplingorSystematic
sampling in each strata
Proportionate ordisproportionate allocation
-
8/8/2019 Research Methodology Class
12/28
When population is widely spreadortoo large
Lookfornatural Clusters
Randomly choose clustersIn each cluster select elementrandomly
Also known as multi-stage samplingtechnique
Less accurate thanfrom otherrandom
sampling
-
8/8/2019 Research Methodology Class
13/28
Think and lets find it-
-
8/8/2019 Research Methodology Class
14/28
Used in multi-stage sampling
Probabilityofselecting a Cluster is dependentonthe size ofthe cluster
Insures representative sample ofpopulation
elements
Used in multi-stage sampling
Probabilityofselecting a Cluster is dependentonthe size ofthe cluster
Insures representative sample ofpopulation
elements
-
8/8/2019 Research Methodology Class
15/28
Atthe convenience ofresearcher
Useful for idea testingor
groundwork
Cheap and easyto administer
Biased and unrepresentative
Non-probability Sampling
Convenience sampling
-
8/8/2019 Research Methodology Class
16/28
Selectionofsubjects are basedon
some judgmentor purpose
It is guided, sonotrandom
Depends onresearchers abilityto
analyze the subject matterofresearch
Easyto use
Results cannot be generalized
Non-probability Sampling
Judgment sampling
-
8/8/2019 Research Methodology Class
17/28
Mix ofstratified and convenience
sampling
Less costly and easyto implement Mostly use in marketing surveys,
opinion polls, etc.
Non-probability Sampling
Quota sampling
-
8/8/2019 Research Methodology Class
18/28
Subjectguides next subjectofthe
population
Easy ifall population is not known
or samplingframe is not available
Non-probability Sampling
Snowball sampling
-
8/8/2019 Research Methodology Class
19/28
What should be our sample size?
Whatdetermines it?
Research Design- Descriptive or
Experimental
Precisionofmeasure-Validity and Reliability
Approaches todetermine the size-
Via Statistical Significance
Via Confidence Interval
Sample Size-A million dollar question
-
8/8/2019 Research Methodology Class
20/28
What should be the sample size?What should be the sample size?
Three factors indetermining sample size
The variabilityofthe population
The level ofconfidence desired inthe estimation
The degree ofprecisiondesired in estimatingthe
population characteristics
-
8/8/2019 Research Methodology Class
21/28
Sample size for estimating a MeanSample size for estimating a Mean
The sample size n is given by
n= W>/e)2
Where
W= estimationofpopulation standard
deviation
Z = the standardized z-value associated
with the level ofconfidencee =Acceptable tolerance level oferror
-
8/8/2019 Research Methodology Class
22/28
Sample size for estimating a ProportionSample size for estimating a Proportion
The sample size n is given by
n= Z2 pq/e2
Where
p = estimationofexpected population
proportion holdingthe characteristic
of intereston intuitionor prior
information, q = 1-p
Z = the standardized z-value associated
with the level ofconfidence
e =Acceptable tolerance level oferror
-
8/8/2019 Research Methodology Class
23/28
NumericalNumerical
A survey proposed to be conducted toknow the annual earnings of old MBAstudents of PU. How large should be
sample size be taken in order toestimate the mean annual earningswithin plus and minus Rs. 1000 at 95%confidence level? The standard deviation
of the annual earnings of the entirepopulation is known to be 15000.
Answer n =732
-
8/8/2019 Research Methodology Class
24/28
NumericalNumerical
Assume you are a manger of a bank inKathmandu. When you were posted inDharan you remember that among 200customers 45 % of them availed anattractive loan scheme. Now your bosswants you to know how the same schemewill be taken in Kathmandu. He needs yourestimation to be precise of +3 % with a 99
% confidence level. So what will be yoursample size?
Answern= 1831
-
8/8/2019 Research Methodology Class
25/28
Sampling ErrorsSampling Errors
Errors which are made in selecting samples
that are not representative of population dueto wrong definition of sample frame or
excluding some population Bound to present however we are
careful.
Will reduce if increase the size of thesample- but not in same proportion
Can reduce (but not eliminate) if usedrandom sampling
-
8/8/2019 Research Methodology Class
26/28
Sampling DistributionSampling Distribution
Sampling distribution of a statistics is thefrequency distribution which is formed withvarious values of statistic computed fromdifferent samples of the same size drawnfrom the same population.
This distribution approaches normal shapeeven when the population is notdistributed as such.
The mean and standard deviatio n o f thesample distribution holds some relationwith true parameter value
-
8/8/2019 Research Methodology Class
27/28
Sampling DistributionSampling Distribution-- Why important?Why important?
Characteristics of its helps us to
To estimate the unknown population
parameter
To setthe confidence limits ofthe
parametervalues estimated
Totest hypothesis anddraw statistical
inference from it
-
8/8/2019 Research Methodology Class
28/28
Reference forSampling chapterReference forSampling chapter
y Social Science Research and ThesisWriting Wolffand Pant
y MethodologyofResearch inSocial Science Krishnaswami and Ranganatham
y Marketing Research within a ChangingInformationEnvironment
Hair, Bush and Ortinau
y Statistics (CAFoundation Course)
Arora andAroray Statistics forManagement
Levin