Research design: Design of Experiment

STRIDE CLASSES (MODULE- I)
TOPIC- RESEARCH DESIGN
Prepared by
Nirmala Devi
Asst. Professor
Institute of Pharmacy
Bundelkhand University, JHANSI (U.P.)

 Research & its purpose
 Research Methodologies
 Process of research
 Structure of research design
 Design of Experiment (DOE)
 Design of Synopsis
 Thesis/Dissertation writing

“If you steal from one author,
its Plagiarism; if you steal from
many, its research” (Wilson
mizner, 1876-1933).
Research is an organized and
systematic way of finding answers
to a set of questions.
Quality
Research
Systematic
Logical &
Objective
Emperical
Generalized
Feasibility/Fal
siability
Replicable
Hypothesis
testing

A checklist, flow diagram, or explicit text to guide
authors in reporting a specific type of research,
developed using explicit methodology e.g. CONSORT,
STROBE, PRISMA, PERT, AMA etc.

Identification of Problem/Research Gap
Review of Literature
Framing objectives
Formulation of Hypothesis
Tools and Technique for data collection & organization
Data analysis & Hypothesis testing
Data interpretation & Generalzation
Data compilation & Report submission

Descriptive (ex post facto) vs Analytical
Applied (action) vs Fundamental (pure/basic)
Qualitative vs Quantitative
Conceptual vs Empirical
Exploratory (field/lab experiment)
Cross-sectional vs Longitudinal
Experimental vs Non-experimental
Other eg. Cohort study (Prospective & Retrospective) Case-
Control Study, Developmental research design , type of trails (Pilot
studies and feasibility studies,Screening trails &Prevention trails
and Trails looking at causes and patterns of disease, Case control
studies, Sequential trails etc.)

•Study type/study design- Fixed (quantitative, theory driven, experimental design
and non-experimental) vs. Flexible (qualitative, e.g. Case study, ethnographic
study, grounded theory study).
•Research problem
•Hypotheses (priori hypotheses – outcomes predicted before, α-level, type I error or
posteriori hypotheses- when direction and extent of relation between variables is
unclear, β-level, type II error)
•Independent and dependent variable
•Experimental design
• Data collection methods
•Statistical analysis plan
•Grouping- depends on research hypotheses and sampling manner which depends on
factors involved

Research
Methods
Exploratory Causal/EffectDescriptive
longitudi
nal
Cross-
sectional
Ideas/insig
ht
Methods
Experiment
in lab
Field /
Site
Population
Variables

Unit of analysis-country, company or individual
Principle of research design (universal theory or local knowledge)
Theory or data, which comes first (iterative process)
Variables involved (dependent, independent, controlled, attribute,
organismic, intervening, extraneous etc.)
Hypothesis (one tailed, two-tailed) and the 65, 95 and 99 percent rule
-Will you verify or falsify the theory (better to formulate the theory and
disapprove it through one wrong answer.
Based on above factors there are two broad classification of Research
Design Approach: Quantitaive & Qualitative

Factual data in the form of pictures, words, photos, videos, audio, recording,
field notes, generalities, peoples own words or votes/polls etc.
Tend to start with broad questions rather than a specific hypothesis.
Develop theory rather than inductive or deductive.
Data sources
primary data- interviews(structured, semi-structured or unstructured), focus
group, questionnaire or surveys.
Secondary data- include diaries, self-reporting, written accounts of past
events/archive data and company reports. Direct observations may also be
recorded(audio/video), ethnography.
Drawback- inaccurate or false information provided, ethical issues, research
objectives difficult to fulfill on hesitant issues, results cannot be generalized.

Works with numerical data, information and result, statistical methods ,
variables controlled with RCD (Reference Classification database). Any
inference can be eliminated and effect of change can be measured.
Randomization to reduce subjective bias is possible.
Data sources
primary- survey (likert scale), observations( count or score/scale/code
the count).
Secondary data (indirect way, govt, agencies, organizations, SATs score
etc.)
Precautions for quantitative analysis- sample size, sample source, correct
stat test, Reducing Type I or Type II errors, generalisable, reproducible,
right techniques and tools.

SAMPLING
Probability Sampling
Non-Probability Sampling
Random/
unrestricted
Stratified
Cluster/
Multistage
double
Accidental/
incidental
Quota
Systematic
Purposive
Sequential

A research design is a master plan specifying the methods/ procedures of
collecting, and analyzing data using specified methodology. It should provide
adequate information to approve or disapprove the hypothesis. Also the design
should be flexible, Adaptable, efficient and economical.
Principles of research Design-
Ontology
Epistomology
The way you group together the research technique to make a coherent
picture.,
Methods and Techniques- what actually done to investigate the problem
Principles broaden the vision towards the problem and orient the researcher
towards opting for unbiased, rational, optimized and effective study.

“EXPERIMENTS ARE OBSERVATIONS UNDER CONTROLLED CONDITIONS”.
Experiment is a question put to nature. It is a procedure to
support, refute, or validates a hypothesis. The independent
variable is manipulated by the experimenter and dependent
variable is tested.
Experimental design refers to the framework or structure of
an experiment. In experimental design the researcher is
active agent rather than a passive observer. Experimental
design is a strongest design with respect to internal validity
by testing hypotheses of causal relationship among
variables.

Experimental designs are gold standard for research in medicine,
biology, Pharmacology, soft sciences and so on. They are resource and
labor intensive and hard to justify generalizability of the results in a
very tightly controlled or artificial experimental setting
• Experimental research design is further classified as :
TRUE EXPERIMENTAL DESIGN
QUASI EXPERIMENTAL DESIGN
PRE EXPERIMENTAL DESIGN

In true experimental designs the researchers have complete control over
the extraneous variables and can predict confidently that the observed
effect on the dependent variable is only due to the manipulation of
independent variable.
Essential Characteristics /Principle of true experimental design:
1. MANIPULATION or TRIAL.
2. REPLICATION
3. BLOCKING / CONTROL
4. RANDOMIZATION.
• Manipulation refers to conscious control of the independent variable by
the researcher through treatment or intervention to observe its effect
on the dependent variable. Egg. Medication (independent variable) and
Pain level (dependent variable)
• Control refers to the use of control group and controlling the effects of
extraneous variables on the dependent variable in which the researcher
is interested.
• The control ensured by adopting one of the following measures:
Matching, counterbalancing, Homogeneity by statistical test.

• Randomization minimize the threats of internal validity of the study and
eliminates the effects of extraneous variables on the dependent variables.
METHODS
•Simple Random Sample- Every subset of a specified size n from the
population has an equal chance of being selected.
•Stratified Random Sample -The population is divided into two or more
groups called strata, according to some criterion, such as geographic
location, grade level, age, or income, and subsamples are randomly
selected from each strata.
•Cluster Sample -The population is divided into subgroups (clusters) like
families. A simple random sample is taken of the subgroups and then all
members of the cluster selected are surveyed.
•Systematic Sample -Every kith member ( for example: every 10th person) is
selected from a list of all population members.

Confounding variable- it is an”extra” variable showing hidden effect. They can cause
increase variance and introduce bias. Eg weight gain and low exercise (confounding
variable-diet, gender, other)
Control variable
Criterion variable- another name for dependent variable used in non-experimental
situations eg. Multiple regression and canonical correlation which use existing
experimental data to make predictions. Predictor variable is analogous to independent
variable- both attached to causation concept.
Endogenous variable- from within the system
Dependent variable,
Explanatory variable
Intervening variable/ mediator variable
Manipulated variable,
Outcome variable- involved in non-experimental studies and no-numeric techniques like
expert opinion, case reports, program evaluations, quality improvement methods, case
control studies, cohort studies.

•Adaptive design (e.g. clinical trials)
• Balanced Latin square design
Balanced and unbalanced designs
Between subjects design
Case study
Case –control study
Completely randomized designs
Cross lagged panel design
Cross-sectional design
Cross-sequential design
Factorial design
Flexible design
Group sequential design
Matched pairs design
Parallel design
Plackett-Burman design
Pretest-Posttest Design
Randomized block design
Randomized control trial
Repeated Measures Design
Retrospective study
Split pilot- strip pilot design
Stepped wedge design

1. POST TEST ONLY CONTROL DESIGN • has two randomly assigned
group - experimental & control groups. • Both the groups are not
tested previous to the introduction of an intervention. • While
treatment is implemented on the experimental group only, post test
observations are made on both the groups.• This design is helpful in
situations where it is not possible to pre treat the subjects. • E.g., A
study on educational intervention related to contraception among
couples.
2. PRETEST-POST-TEST-ONLY DESIGN • In this design, subjects are
randomly assigned to either the experimental or control group. • The
effect of the dependent variable on both the groups is seen before
the treatment (pre test).• Following this the treatment is carried out
on experimental group only. • After treatment observation of
dependent variable is made on both the groups to examine the effect
of the manipulation of independent variable on dependent variable.

SOLOMON FOUR GROUP DESIGN
• Has two experimental and two control group of which only exp grp I &
control grp I receives the pre test followed by the treatment to the
experimental grp I & II.
• Finally all the four groups receive post test, where the effects of the
dependent variables of the study are observed and comparison is made
of the four groups to assess the effect of independent variable
(experimental variable) on the dependent variable.
• The experimental group II is observed at one occasion.
• To estimate the amount of change in experimental & control group II
the average test scores of experimental & control groups I are used as
baseline.
• It is most prestigious experimental research design, because it
minimizes the threat to internal and external validity. The test
effectively presents the reactive effects of the pre test.
• Any difference between the experimental and control group can be
more confidently attributed to the experimental treatment.
• The disadvantage of this design is that it requires a large sample and
statistical analysis, and therefore not commonly used in health care
researches.

FACTORIAL DESIGN
• Here the researcher manipulates two or more independent variables
simultaneously to observe their effects on the dependent variables.
• This design is particularly useful when there are more than two
independent variables to be tested e.g. researcher wants to test the
efficacy of two different medications.
• The design facilitates the testing of several hypotheses at a single
time.
• Typically factorial design incorporates 2x2 or 2x3 factorial etc.
The first number (alpha - A) refers to the independent variables or the
types of experimental treatments and the second number (beta -B)
refers to the level or frequency of the treatment.

RANDOMIZED BLOCK DESIGN • used when the researcher desires
to bring homogeneity among selected groups. • This is a simple
method to reduce the variability among the treatment groups by a
more homogenous combination of the subjects through randomized
block design.
• For example if the researcher wants to test the efficacy of three
different medications in reducing hypertension, to ensure homogeneity
among subjects under treatment, researcher randomly places the
subjects in homogenous groups (blocks). • like patients with
hypertension, diabetic patients with hypertension and hypertensive
patients with heart diseases. The design looks similar to that of
factorial design in structure, but out of two factors one factor is not
experimentally manipulated.
CROSS OVER DESIGN • the study subjects are exposed to more than
one treatment, also known as “repeat measure design. it establishes
the highest possible similarity among subjects exposed to different
conditions where groups compared obviously have equal distribution of
characteristics. • Sometimes this design is not effective because, when
subjects are exposed to two different conditions, their responses of the
second condition may be influenced by their experience in the first
condition.

Research design: Design of Experiment

ADVANTAGES OF TRUE EXPERIMENTAL DESIGN • Most powerful design to
establish the causal relationship between independent and dependent
variable. • Since the study is conducted under controlled environment,
it can yield a greater degree of purity in observation.Conditions that are
not found in natural setting can be created in experimental setting in a
short period of time that may take years to naturally occur (therefore
very useful in genetic studies). • Because the experiment is carried out
in experimental setting the problems of real life situations and the
personal problems of the researcher is eliminated.
DISADVANTAGES OF TRUE EXPERIMENTAL DESIGN
Replication of studies in animals and human demands ethical clearance.

Threats of internal validity
History :Some event beside the experimental treatment occurs during the course of the study , and this
event even influence dependent variable.
Maturation of subjects eg. Nutritional protocol on height & weight of malnourished children.
Testing Effect of taking a pretest of subjects’ performance of post test may sensitize an individual and
improve the score of the post test. Individuals generally score higher during second test regardless of
treatment.
Instrument change, calibration, observers or scorers error
Mortality/Loss or dropout of the subject during course of the study. Eg. longitudinal study
Selection bias if subjects are not selected randomly for participation in groups , there is a possibility of
comparison may not equivalent.
External validity (Hawthorne effect-Subjects may behave in particular manner because they are aware
that they are being observed)
Experimental effect-Threat to study results when researcher’s characteristic , mannerism, behavior may
influence subject matter.
Reactive effect of pretest Effect of pretest occurs when subjects have been sensitized to the treatment
because of taking pretest. Eg – pretest may sensitize to learn about HIV/ AIDS irrespective of health
education is provided
Novelty effect: Treatment is new, the subjects and researchers act different ways People : Generalization
depending upon the race, Place, Time etc may differ.

Experimental control attempts to predict events that will occur in the
experimental setting by neutralizing the effects of other factors.
Physical Control Gives all subjects equal exposure to the independent
variable. Controls non-experimental variables that effect the dependent variable.
Selective Control Indirectly manipulate by selecting in or out variables that
cannot be controlled.
Statistical Control Variables not conducive to physical or selective
manipulation may be controlled by statistical techniques.

Synopsis is the blue-print of your research work that warrants
meticulous planning and professional assistance. It helps
verbalizing the idea of the paper and at the same time makes
it more concrete. It is a tool for thinking the subject and
argument of the paper. It helps you to focus and structure
the paper. It can be of maximum 3-4,000 words, excluding
appendices. Brief outline of future research.

•Preliminary pages (Title, Preface, Acknowledgement, table of content, list of table
and list of figures, abbreviations)
1. Introduction: Brief background/introduction to the subject.
2. Problem Statement and Formulation of Hypothesis
3. Aims and Objectives
4. Review of Literature: Consult the literature articles, books, reports, cases,
monographs, data bases) on the broad theme to highlight, General development in
the field [Ascending order in terms of year of Publications].
5. Present Study
A. Significance of the Problem:
Make a clear statement highlighting the exact coverage and purview of the problem
under investigation.
B. Conceptual Framework: theory base, define concepts, definitions, assumptions,
limitations and delimitations
the proposed study, which need to be defined for the purpose of the study in hand.
Make operational definitions of all such usages.

C. Methodology: ( To be applied as per research design)
a. Hypothesis/ Research questions: Make crisp statements ( e.g. one each for
all objectives) entailing not more than two variables describing the presumed
relationship or influence on each other.
b. Research design (sample, tool, technique, procedure , time, budget)
c. Sample: Mention the sampling method used.
d. Data Collection: Primary Data collection &Secondary Data collection : library
resources/reports consulted/internet resources used, Case studies, Case laws/judgments,
Content analysis/document analysis/Meta analysis
Statistical consideration/ software applications
6. Data Analysis / Discussion:
7. Chapter Scheme
8. Bibliography/References: Follow the standard referencing and citation.

Data analysis- hypothesis testing, correlations and cluster analysis
Hypothesis testing falsifiability or refutability of a statement, hypothesis,
or theory. Starts with null hypothesis (H0- there will be no difference) and
conclude either rejecting/ accepting alternative hypotheses (H1, H2, H3
etc.).
Analyzing quant data- group data, outliers or extreme cleaning- improves
central tendency (mean, median, mode, quartiles etc.), what average you
choose, mean, median or mode, spread of data: skewness/distribution,
range, variance and standard deviation. Choose statistical tests for either
difference (between/within gps) or correlation, Parametric vs on-
parametric (interval/ratio vs nominal/ordinal). It is to be noted that
correlation does not imply causation.
Data interpretation- importance
Significance level- a fixed probability of wrongly rejecting the null
hypothesis , if it is in fact true. Usually set it to 0.05 (5%).
P value-
Power- ability to detect a difference if there is one (degree of freedom, F
value)
Effect size- numerical way of expressing the strength or magnitude of a
reported relationship be it causal or not.

There are several research designs and the researcher must decide in
advance of collection and analysis of data as to which design would
prove to be more appropriate for his research project.
Optimum design, appropriate tools and techniques, exhaustive data
analysis and justification or interpretation of result in terms of the
assumption or research question’s answers are the soul and body of the
whole research.
Otherwise everything is nothing but a huge MESS!

William M.K. Trochin, Research Methods, 2nd edition, reprint 2009, Biztantra publication, Darya Ganj, New
Delhi.
C.K. Kothari, Research Methodology: Methods and Techniques, 2nd edition (1990), reprint 2002; Wiswa
Prakashan, Darya Ganj, New Delhi.
G.N. Prakash Srivastava. Advanced Research Methodology, 1st edition, 1994. Radha publications, New
Delhi.
Edusat websites (Pathshaala)
www.datasciencecentral.com
Alan C. Elliott, Wayne A. Woodward. In; Statistical Analysis Quick Reference Guidebook: With SPSS Examples
1st edition; 2006. ISBN: 1412925606,9781412925600
Library genesis()

Research design: Design of Experiment

More Related Content

What's hot (20)

Similar to Research design: Design of Experiment (20)

Recently uploaded (20)

Research design: Design of Experiment