2. 22
– What is the study about?What is the study about?
– Why is the study being made?Why is the study being made?
– Where will the study be carried out?Where will the study be carried out?
– What type of data is required?What type of data is required?
– Where can the required data be found?Where can the required data be found?
Research design makingResearch design making
questionsquestions
3. 33
– What period of time will the study include?What period of time will the study include?
– What will be sample design?What will be sample design?
– What techniques of data collection will be used?What techniques of data collection will be used?
– How will the data be analyzed?How will the data be analyzed?
– In what style will the report be prepared?In what style will the report be prepared?
Research design makingResearch design making
questionsquestions
4. 44
RESEARCH DESIGNRESEARCH DESIGN
RESEARCH DESIGNRESEARCH DESIGN refers to the plan, structure, andrefers to the plan, structure, and
strategy of research--thestrategy of research--the blueprintblueprint that willthat will guide theguide the
research processresearch process..
Decision regarding what, where, when, how much,Decision regarding what, where, when, how much,
by what meansby what means
It is conceptual structure within which research isIt is conceptual structure within which research is
conducted.conducted.
6. 66
RESEARCH DESIGNRESEARCH DESIGN
Research design contains:Research design contains:
Plan for specifying sources and types ofPlan for specifying sources and types of
information relevant to research probleminformation relevant to research problem
Strategy for choosing the approach for gatheringStrategy for choosing the approach for gathering
and analyzing the dataand analyzing the data
The time and costThe time and cost
Operational designOperational design: techniques of specifying the procedure of: techniques of specifying the procedure of
sampling, statistical and observational design is carried out.sampling, statistical and observational design is carried out.
7. Need of Research DesignNeed of Research Design
For proper planning before dataFor proper planning before data
collection and data analysis.collection and data analysis.
For proper calculation of availabilityFor proper calculation of availability
of staff, time, and money.of staff, time, and money.
For organizing researcher’s ideas in aFor organizing researcher’s ideas in a
formed way which will help to findformed way which will help to find
out researcher’s flaws andout researcher’s flaws and
inadequacies.inadequacies.
77
8. 88
Effective Research DesignEffective Research Design
– MAXMAXimize Systematic Varianceimize Systematic Variance
– MINMINimize Error Varianceimize Error Variance
– CONCONtrol Variance of Nuisancetrol Variance of Nuisance
(harm)/Extraneous (not essential)/(harm)/Extraneous (not essential)/
Exogenous (out side part)/Confounding (toExogenous (out side part)/Confounding (to
surprise and confuse) variablessurprise and confuse) variables
Guiding principleGuiding principle for effective control offor effective control of
variances ( state of being variable) (and,variances ( state of being variable) (and,
thus, effective research design) is:thus, effective research design) is:
TheThe MAXMINCONMAXMINCON PrinciplePrinciple
9. 99
Feature of Good Research DesignFeature of Good Research Design
Effecting factors for research designEffecting factors for research design
– The means of getting informationThe means of getting information
– The availability and skills of researcher and teamThe availability and skills of researcher and team
– The objective of the problem to be studiedThe objective of the problem to be studied
– The nature of the problem to be studiedThe nature of the problem to be studied
– The availability of time and money for theThe availability of time and money for the
research work.research work.
Flexible, appropriate, efficient, economical andFlexible, appropriate, efficient, economical and
so on.so on.
Minimizes bias and maximizes the reliabilityMinimizes bias and maximizes the reliability
One single design cannot server the purpose of all
type of researches
10. Different Research Design forDifferent Research Design for
different kind of researchdifferent kind of research
Exploratory or a formulated researchExploratory or a formulated research
Flexible research design that willFlexible research design that will
consider many phenomenaconsider many phenomena
Descriptive researchDescriptive research
Accuracy becomes a majorAccuracy becomes a major
considerationconsideration
1010
11. 1111
Effective Research DesignEffective Research Design
Adequate (full range of) variability in values ofAdequate (full range of) variability in values of
research variables,research variables,
Precise and accurate measurement,Precise and accurate measurement,
Identifying and controlling the effects ofIdentifying and controlling the effects of
confounding variables, andconfounding variables, and
Appropriate subject selectionAppropriate subject selection
RECAP:RECAP:
Effective research design is a function of ?Effective research design is a function of ?
12. QuestionQuestion
Explain the meaning and significanceExplain the meaning and significance
of research design?of research design?
1212
14. 1414
EXPERIMENTAL DESIGNSEXPERIMENTAL DESIGNS
RESULT:RESULT: Significant Improvement from O1 to O2Significant Improvement from O1 to O2
(i.e., sig. O2 - O1 difference)(i.e., sig. O2 - O1 difference)
QUESTION:QUESTION: Did X (the drug) cause theDid X (the drug) cause the
improvement?improvement?
One of the simplest experimental designs is the ONE GROUP PRETEST-
POSTTEST DESIGN--EXAMPLE?
One way to examine Efficacy of a Drug:
O1 X O2
Measure DRUG Measure
Patients’ Condition Experimental Patients’ Condition
(Pretest) Condition/ (Posttest)
intervention
15. 1515
EXPERIMENTAL DESIGNSEXPERIMENTAL DESIGNS
– HaveHave only shownonly shown “X” is a SUFFICIENT“X” is a SUFFICIENT conditioncondition
for the changefor the change “Y”“Y” (i.e., presence of X is(i.e., presence of X is
associated with a change in Y).associated with a change in Y).
But,But, is “X” also a NECESSARYis “X” also a NECESSARY condition forcondition for
Y?Y?
– How do you verify the latter?How do you verify the latter?
By showing that the change would not haveBy showing that the change would not have
happened in the absence of X—using ahappened in the absence of X—using a
CONTROL GROUP.CONTROL GROUP.
David Hume would have been tempted to say “YES.”
He was a positivist and wanted to infer causality based
on high correlations between events.
But such an inference could be seriously flawed.
Why?
David Hume, 18th
Century Scottish
Philosopher
16. 1616
EXPERIMENTAL DESIGNSEXPERIMENTAL DESIGNS
CONTROL GROUPCONTROL GROUP simulates absence of Xsimulates absence of X
– Origin of using Control Groups (A tale fromOrigin of using Control Groups (A tale from ancient Egypt)ancient Egypt)
Pretest Post-TestPretest Post-Test Control GroupControl Group Design--Design--Suppose randomSuppose random
assignment (assignment (RR) was used to control confounding variables:) was used to control confounding variables:
RR Exp. Group O1E X O2EExp. Group O1E X O2E
RR Ctrl Group O1C O2CCtrl Group O1C O2C
RESULT:RESULT: O2E > O1E & O2C Not> O1CO2E > O1E & O2C Not> O1C
QUESTION:QUESTION: Did X cause the improvement in Exp.Did X cause the improvement in Exp.
Group?Group?
17. 1717
EXPERIMENTAL DESIGNSEXPERIMENTAL DESIGNS
– NeedNeed proper form of controlproper form of control—e.g.,—e.g., PlaceboPlacebo..
RR Exp. Group O1E X O2EExp. Group O1E X O2E
RR Ctrl Group O1C Placebo O2CCtrl Group O1C Placebo O2C
QUESTION:QUESTION: Can we now conclude X caused the improvementCan we now conclude X caused the improvement
in Exp. Group?in Exp. Group?
NOT NECESSARILY! Why not?
• Power of suggestibility (The Hawthorne Effect)
CONCLUSION?
• Maybe, but be aware of the Experimenter Effect (it tends to
prejudice the results especially in medical research).
• SOLUTION: Double Blind Experiments (neither the subjects
nor the experimenter knows who is getting the placebo/drug).
18. 1818
EXPERIMENTAL DESIGNSEXPERIMENTAL DESIGNS
Experimental studiesExperimental studies need toneed to controlcontrol forfor potentialpotential
confounding factorsconfounding factors that may threaten internal validitythat may threaten internal validity
of the experiment:of the experiment:
–Hawthorne EffectHawthorne Effect is only one potential confounding factoris only one potential confounding factor
in experimental studies.in experimental studies.
Other such factors areOther such factors are::
–History?History?
Biasing events that occur between pretest and post-testBiasing events that occur between pretest and post-test
–Maturation?Maturation?
Physical/biological/psychological changes in the subjectsPhysical/biological/psychological changes in the subjects
–Testing?Testing?
Exposure to pretest influences scores on post-testExposure to pretest influences scores on post-test
–Instrumentation?Instrumentation?
Flaws in measurement instrument/procedureFlaws in measurement instrument/procedure
19. 1919
EXPERIMENTAL DESIGNSEXPERIMENTAL DESIGNS
Experimental studiesExperimental studies need toneed to controlcontrol forfor potentialpotential
confounding factorsconfounding factors that may threaten internal validitythat may threaten internal validity
of the experiment (Continued):of the experiment (Continued):
–Selection?Selection?
Subjects in experimental & control groups different from the startSubjects in experimental & control groups different from the start
–Statistical Regression (regression toward the mean)?Statistical Regression (regression toward the mean)?
Subjects selected based on extreme pretest valuesSubjects selected based on extreme pretest values
Discovered byDiscovered by Francis GaltonFrancis Galton in 1877in 1877
–Experimental Mortality?Experimental Mortality?
Differential drop-out of subjects from experimental and controlDifferential drop-out of subjects from experimental and control
groups during the studygroups during the study
–EtcEtc..
Experimental designs mostly used inExperimental designs mostly used in natural and physicalnatural and physical
sciencessciences..
Generally, higher internal validity, lower externalGenerally, higher internal validity, lower external
validityvalidity
22. 2222
CORRELATIONAL DESIGNSCORRELATIONAL DESIGNS
The design of choice inThe design of choice in social sciencessocial sciences since thesince the phenomenonphenomenon
under study is usuallyunder study is usually not reproducible in a laboratory settingnot reproducible in a laboratory setting
Researcher hasResearcher has little or no controllittle or no control over study’s indep., dep.over study’s indep., dep.
and theand the numerous potentialnumerous potential confounding variablesconfounding variables,,
Often the researcherOften the researcher concomitantly measuresconcomitantly measures all the studyall the study
variables (e.g., independent, dependant, etc.),variables (e.g., independent, dependant, etc.),
Then examines the following types of relationships:Then examines the following types of relationships:
– correlations among variablescorrelations among variables oror
– differences among groupsdifferences among groups,,
Inability to controlInability to control for effects offor effects of confounding variablesconfounding variables makesmakes
causal inferencescausal inferences regarding relationships among variablesregarding relationships among variables
more difficultmore difficult and, thus:and, thus:
Generally, higher external validity, lower internal validityGenerally, higher external validity, lower internal validity
NON-EXPERIMENTAL/CORRELATIONAL DESIGNSNON-EXPERIMENTAL/CORRELATIONAL DESIGNS
23. 2323
CORRELATIONAL DESIGNSCORRELATIONAL DESIGNS
NOT NECESSARILY! EXAMPLES:NOT NECESSARILY! EXAMPLES:
– Water Fluoridation and AIDSWater Fluoridation and AIDS
((San Francisco ChronicleSan Francisco Chronicle, Sep. 6, 1984), Sep. 6, 1984)
– Armed store robberies and cold weatherArmed store robberies and cold weather
– Longevity and PollutionLongevity and Pollution
– In-vitro birth and likelihood of developing eyeIn-vitro birth and likelihood of developing eye
tumorstumors
– Hearing problem and blood pressureHearing problem and blood pressure
What can a significant correlation mean then?What can a significant correlation mean then?
Non-experimental designs rely on correlational evidence.
QUESTION: Does a significant correlation between two
variables in a non-experimental study necessarily represent a
causal relationship between those variables?
24. 2424
CORRELATIONAL STUDIESCORRELATIONAL STUDIES
a.a. Both variables are effects of a common cause (orBoth variables are effects of a common cause (or bothboth
correlated with a third variable), i.e., spurious correlationcorrelated with a third variable), i.e., spurious correlation
(e.g., air pollution and life expectancy, hearing problem &(e.g., air pollution and life expectancy, hearing problem &
blood pressure, country’s annual ice cream sales andblood pressure, country’s annual ice cream sales and
frequency of hospital admissions for heat stroke)frequency of hospital admissions for heat stroke)
b.b. Both var.Both var. alternative indicatorsalternative indicators of same conceptof same concept
(e.g., Church attend. & Freq. of Praying--religiosity).(e.g., Church attend. & Freq. of Praying--religiosity).
c.c. Both parts ofBoth parts of a common "system" or "complex;" tend toa common "system" or "complex;" tend to
come as a packagecome as a package
(e.g., martini drinking and opera attendance--life style)(e.g., martini drinking and opera attendance--life style)
d.d. Fortuitous--Fortuitous--Coincidental correlationCoincidental correlation, no logical relationship, no logical relationship
(e.g., Outcome of super bowl games and movement of stock(e.g., Outcome of super bowl games and movement of stock
market)market)
AT LEAST FOUR OTHER POSSIBLE INTERPRETATIONS/REASONS
FOR CORRELATIONS BETWEEN TWO VARIABLES:
25. 2525
CORRELATIONAL STUDIESCORRELATIONAL STUDIES
Covariation RuleCovariation Rule (X and Y must be(X and Y must be
correlated)--Necessary but not sufficient condition.correlated)--Necessary but not sufficient condition.
Temporal Precedence RuleTemporal Precedence Rule (If X is the cause, Y(If X is the cause, Y
should not occur until after X).should not occur until after X).
Internal Validity RuleInternal Validity Rule (Alternative plausible(Alternative plausible
explanations of Y and X-Y relationships should beexplanations of Y and X-Y relationships should be
ruled out (i.e., eliminate other possible causes).ruled out (i.e., eliminate other possible causes).
– In practice, this means exercising caution byIn practice, this means exercising caution by
identifying potentialidentifying potential confounding variables andconfounding variables and
controlling for their effectscontrolling for their effects).).
WHEN IS IT SAFER TO INFER CAUSAL
LINKAGES FROM STRONG CORRELATIONS?
John Stuart Mill’s Rules for Inferring Causal Links:
John Stuart Mill
1806-1873