Research

1
RESEARCH METHODS
BY
SAHILU ASSEGID, MD, MPHE,
ASSISTANT PROFESSOR
RESEARCH METHODOLOGY
BY
SAHILU ASSEGID,
MD, MPHE, ASSISTANT PROFESSOR
JANUARY/2009
2
Contents
1. Chapter one-------------------------------- Health research
2. Chapter two----------------Overview of research process
3. Chapter three---------------------- The research proposal
4. Chapter four-------Research question/ survey research
5. Chapter five-------------Study designs/ survey research
6. Chapter six---------------------Data sources for research
7. Chapter seven-----------------Sampling and sample size
8. Chapter eight---------------------------------------------Data
9. Chapter nine------------------------------ Data processing

BY
SAHILU ASSEGID,
JANUARY/2009
3
General Description of the
Course
Research is a systematic search for
information and new knowledge. It is an
essential and powerful tool for accelerating
health development. It is necessary to the
process of identifying priority problems
and to designing and evaluating polices
and program that will be of the greatest
health benefit, using existing knowledge
and available resources.
4
General Description of the
Course
This course is intended to give to
the learner the proper knowledge
and skills in problem identification,
planning implementation and
evaluation of research undertaking

5
Core tasks performed in occupational
reality, covered by this course:
Conceptualize and identify a problem in the
health system
Describe the contribution of research toward
solving priority problem
Prepare a research proposal by involving all
stakeholders
Implement the study proposal
Analyze and write a report
Communicate and implement research findings.

6
1. HEALTH RESEARCH

Is a systematic collection, analysis and
interpretation of data to solve health
problem
BY
SAHILU ASSEGID,
JANUARY/2009
7
2. PURPOSE OF HEALTH RESEARCH

To generate knowledge essential to effectively
promote the health of the population

Without this knowledge, effective action is
impossible because it has no logical or empirical
basis
BY
SAHILU ASSEGID,
JANUARY/2009
8
3. MAKING PROGRESS IN HEALTH

Progress in health (medicine) can only be
achieved if good quality information are
generated through research to guide
interventions and to inform the general public.
BY
SAHILU ASSEGID,
JANUARY/2009
9
4. FUNDAMENTAL EPIDEMIOLOGICAL
ASSUMPTIONS

1. Human disease does not occur at
random.

2. Human disease has causal and
preventive factors that can be identified
through systematic investigation.
BY
SAHILU ASSEGID,
JANUARY/2009
10
5. IMPORTANCE OF RESEARCH
1. Guiding action
2. Developing new tool
3. Saving resources
4. Multiplying benefits
5. Enhancing capacity development
process
BY
SAHILU ASSEGID,
JANUARY/2009
11
6. SOURCES OF INFORMATION
Standard scientific/research methods
- Epidemiology
- Demography
- Health service research
Other methods
- Expert assessment
- Community assessment
BY
SAHILU ASSEGID,
JANUARY/2009
12
7. THE ICE BERG PHENOMENON
Routine information comes mainly from people who attend the health services
Proper and unbiased assessment of health conditions is important for good
public health practice!!!

CURATIVE CARE PREVENTIVE CARE

_

PEOPLE SEEN

People not
seen

People not seen
BY
SAHILU ASSEGID,
JANUARY/2009
13
8. RESEARCH
Research= identifying problem and designing
appropriate action
1. Health facility
History physical examination Tests Diagnosis

2. Community
Talk with Records Surveys Community
Community Diagnosis

=> Established methods must be used in both!!!
BY
SAHILU ASSEGID,
JANUARY/2009
14
9. RESEARCH MUST BE

1. Purposeful: what do you want to contribute?
2. Targeted: who are the audiences?
3. Credible: consider sources of information,
method of data collection, personnel involved...
4. Timely: is the information needed?
* Research is done to find solutions to health
problems.

BY
SAHILU ASSEGID,
JANUARY/2009
15
10. CHARACTERISTICS OF A GOOD
RESEARCH PROPOSAL

Have a clear statement of the problem
Have clear plans- objectives & methods
Have clear vision of the outcome
Builds on existing data

BY
SAHILU ASSEGID,
JANUARY/2009
16
11. HOW DO WE CHOOSE A RESEARCH
TOPIC?
Career development
Priority research areas
- Relevance
- Avoidance of duplication
- Feasibility
- Political acceptability
- Applicability
- Urgency of data needed
- Ethical acceptability
Resource availability

BY
SAHILU ASSEGID,
JANUARY/2009
17
SCIENTIFIC INQUIRY

INTRODUCTION
BY
SAHILU ASSEGID,
JANUARY/2009
18
SESSION OBJECTIVES ON:
Scientific inquiry
Positivism
Scientific theory and empiricism
Objectivity
The process of theory construction
Specify the topic
Specify the assumptions
Specify the range of phenomena
Specify the major concepts and variables
Specify the propositions, hypotheses, and relationships
Specify the theory

BY
SAHILU ASSEGID,
JANUARY/2009
19
OVERVIEW
Paradigms

Theories

Methodologies

Methods
BY
SAHILU ASSEGID,
JANUARY/2009
20
PARADIGMS
Paradigm: is a basic set of beliefs that
guide action.
Interrelated beliefs about:
1. Ontology
2. Epistemology
3. Methodology
4. Axiology

BY
SAHILU ASSEGID,
JANUARY/2009
21
PARADIGM.
Particular configurations of answers to
ontological, epistemological,
methodological and axiological questions
are organized into paradigms:
1. Positivist paradigm
2. Interpretive paradigm
3. Critical paradigm
BY
SAHILU ASSEGID,
JANUARY/2009
22
ONTOLOGICAL QUESTIONS
What is the nature of reality?
Is there a way that things really work?
Are there laws that can explain the
relationship between things?
Is there a Truth that can be known?
Is there a reality out there that good
research can discover?
BY
SAHILU ASSEGID,
JANUARY/2009
23
EPISTEMIOLOGY
Refers to the relationship between the
knower and the known.
Is what is learned independent of the
researcher?
Is about social reality
BY
SAHILU ASSEGID,
JANUARY/2009
24
METHODOLOGY
Focuses on the procedures for
understanding the world
Encompasses the entire research design:
- Qualitative or quantitative?
- What kind of sampling procedure?
- Logic used
BY
SAHILU ASSEGID,
JANUARY/2009
25
AXIOLOGY
Concerned with values and ethics
What is the role of values in the research
process?
Should researchers be disinterested
scientists or emotionally engaged in the
research process?
Status characteristics
BY
SAHILU ASSEGID,
JANUARY/2009
26
POSITIVIST PARADIGM
Ontological assumptions:
An objective reality exists
Social reality is stable and patterned, so it
can be known through rigorous
investigation
BY
SAHILU ASSEGID,
JANUARY/2009
27
POSITIVIST PARADIGM
Epistemological assumption: what can be
learned about social world exists
independently of the researcher
Axiological assumptions: research should
be objective and value free
Methodological: Reliability, validity and
generalizability are key concerns
BY
SAHILU ASSEGID,
JANUARY/2009
28
THEORY
Set of interlinked propositions that explain some
social phenomena
Ideas we have about how the world works and
why people do what they do
Orienting frameworks
Conceptual lenses
Causal models
Worldviews
Distinct theories etc.
BY
SAHILU ASSEGID,
JANUARY/2009
29
CHAPTER TWO

OVERVIEW OF RESEARCH PROCESS
Selection of a
BY
SAHILU ASSEGID,
JANUARY/2009
30
1. PLANNING RESEARCH
=>Developing proposal
A written proposal is important:

1. To clearly define the problem under study
2. To avoid reinventing the wheel
3. To clearly depict the methodology to be used
4. To be cost and time conscious
5. To be clear about what to expect at the end
BY
SAHILU ASSEGID,
JANUARY/2009
31
2. COMPONENTS OF RESEARCH PROPOSAL
A. Statement of the problem/introduction
B. Literature review
C. Objectives
D. Methodology
E. Ethical considerations
F. Work plan
G. Dissemination and utilization of results
H. Cost of the project
I. Assurance of the investigator
J. Advisor (approval)

BY
SAHILU ASSEGID,
JANUARY/2009
32
A. STATEMENT OF THE
PROBLEM/INTRODUCTION
1. Concise description of the nature of the problem:
- What is it?
- Magnitude
- Distribution (who, where, when)
- Severity and consequences
2. Systematically elucidate why the proposed research should be
undertaken:
- Brief description of any attempts to solve the
problem in the past-successes, failures and
challenges
- Provide convincing argument that available knowledge
is insufficient to solve the problem under study

BY
SAHILU ASSEGID,
JANUARY/2009
33
STATEMENT OF THE
PROBLEM/INTRODUCTION...

3. Describe the significance of the proposed study:
-what do you hope to achieve with the study results?
BY
SAHILU ASSEGID,
JANUARY/2009
34
B. LITERATURE REVIEW
Search widely:
- Library
- Electronic search engines
- Grey literature

Evaluate the relevance of the literature to your study
Include information directly relevant to your study
Be concise
BY
SAHILU ASSEGID,
JANUARY/2009
35
LITERATURE REVIEW...
Include:
- Detailed account of the subject matter you want to
research
- Entertain confounding factors and other related matter
only pertaining to the current research
- Avoid repetitions and lengthy statements

(Maximum 5-7 pages)
BY
SAHILU ASSEGID,
JANUARY/2009
36
LITERATURE REVIEW...
Avoid reinventing the wheel/avoidance of
duplicating previous work
Learn the gaps
Learn the various methods used

=> An opportunity to develop professional
confidence in the field of study!!!

BY
SAHILU ASSEGID,
JANUARY/2009
37
C. OBJECTIVES
General objective:
- Summarizes what is to be achieved by the study
- Should be clearly related to the statement of the
problem
Specific objectives:
- Logically connected parts of the general objective
- Focus the study on the essentials
- Direct the design of the investigation
- Orient collection, analysis and interpretation of the
data
BY
SAHILU ASSEGID,
JANUARY/2009
38
D. FORMULATING OBJECTIVES: RESEARCH
QUESTIONS VS. HYPOTHESIS
A hypothesis requires sufficient knowledge of
the problem to be able to predict relationships
among factors which can then explicitly tested.

Research questions are formulated when the
investigators do not have enough insight into the
problem being studied.
BY
SAHILU ASSEGID,
JANUARY/2009
39
E. CRITERIA FOR SETTING RESEARCH
QUESTIONS
1. Focused- each covering a single point
2. Ordered- in logical sequence
3. Realistic- and feasible to answer
4. Operational- using action verbs such as:
- Determine - Verify - Identify
- Describe - Assess - Compare
- Calculate - establish - Explore
5. Measurable- outcomes at the end of the research

BY
SAHILU ASSEGID,
JANUARY/2009
40
F. METHODOLOGY
What is the choice of the study design?
What would be the study population?
What would be the required size of the study subjects?
How would the study subjects selected/sampled?
Data collection: What, How, Who, Where, When?
Data analysis-Coding, entering, cleaning, storing,
recording, choice of statistical methods
Operational definitions of crucial concepts
Selection and training of field workers/research staff
Field testing of the research methods and tools
Supervision and quality control
BY
SAHILU ASSEGID,
JANUARY/2009
41
G. ETHICAL CONSIDERATIONS
Professional obligation to safeguard the safety of study
subjects

Refer to national and international guidelines

Describe potential ethical concerns and mechanisms to
minimize harms and maximize benefits

Every research can potentially cause ethical concerns!!!
BY
SAHILU ASSEGID,
JANUARY/2009
42
H. DISSEMINATION AND UTILIZATION OF
RESULTS
Briefly describe the dissemination plan
- Feedback to the community
- Feedback to local communities
- Identify relevant authorities that need to be informed
- Scientific publication
- Presentation in meetings/conferences

Briefly describe how the study results can be best
translated into application.
BY
SAHILU ASSEGID,
JANUARY/2009
43
I. BUDGET- COST OF THE PROJECT

Clearly identify the resource requirements
Be realistic in costing/budgeting
The work-plan is a good starting point for preparing
budget-budget for activities
Prepare budget justification
Include 5-10% contingency
Identify funding resources & prepare budget according to
the required format.
BY
SAHILU ASSEGID,
JANUARY/2009
44
J. YOUR PROPOSAL

Simple and clear
Good statement of the problem: why do you
want to study?
Pertinent literature review
Few objectives
Clear & detailed methodological description
BY
SAHILU ASSEGID,
JANUARY/2009
45
K. SEE TO WHOLE PICTURE
=>

-----------------
Dummy tables,
conclusions and
recommendations
Always focus on your objectives!!!
Process
_
Selection of study population
-Use appropriate methods
-- Use appropriate tools

Outcome
_
What would be the result?
-What can be concluded?
-What can be recommended?

Objectives
BY
SAHILU ASSEGID,
JANUARY/2009
46
L. YOUR RESEARCH: IS IT IMPORTANT?
=>
Literature
Research
Production

Science is a sum of cumulative knowledge: Previous
researches are the basis for yours and your work will be
the basis for future research.

BY
SAHILU ASSEGID,
JANUARY/2009
47
M. CONSULTATION

Use advisors as much as possible

Timely

Be selective: whom to consult
BY
SAHILU ASSEGID,
JANUARY/2009
48
N. VISUALIZE YOUR THESIS
Cover pages
Acknowledgement
Abstract (1)
Introduction (1-2)
Literature review (5-7)
Methods (4-7)
Results (7-10)
Discussion (5-7)
References (1-2)
Annexes
24-36 pages
Only the main body
text
BY
SAHILU ASSEGID,
JANUARY/2009
49
O.MANAGING YOUR RESEARCH PROJECT
Keep good note
Establish good communication with your advisor (s)
Keep time and promises: motivates your advisor (s)
Keep your document to a manageable size: readability
and clear message
Lead your research work: be ready to manage crisis
Advisors: advisors, assessors but not dictators
Do your research heartily!!!

BY
SAHILU ASSEGID,
JANUARY/2009
50
BE REALISTIC
Do not be overambitious!!

Good Luck!!!
BY
SAHILU ASSEGID,
JANUARY/2009
51
THE END

OF
OVERVIEW OF THE RESEARCH
PROCESS

BY
SAHILU ASSEGID,
JANUARY/2009
52
CHAPTER THREE

THE RESEARCH PROPOSAL
BY
SAHILU ASSEGID,
JANUARY/2009
53
1. STEPS IN UNDERSTANDING A RESEARCH
1. Problem identification and definition
2. Review relevant literature
3. Choose the appropriate research design
4. Development of proposals
5. Data collection
6. Data analysis
7. Writing a report
BY
SAHILU ASSEGID,
JANUARY/2009
54
2. RESEARCH PROPOSAL
Although the beginning investigator might have a plan of
what he wants to do in mind, it is important to set it down
in writing.
This systematically presented research scheme is known
as protocol
A written protocol serves major functions:
- Ideas and procedures are clearly defined and spelled out
- Gaps and errors are easier to recognize and correct when the plan
is seen on paper than the plan is in ones mind
- A written protocol can be studied by anyone whose advice is
desired or whose approval is required.
BY
SAHILU ASSEGID,
JANUARY/2009
55
RESEARCH PROPOSAL...
- The written protocol constitutes a permanent record of the
procedures that should be followed and referred in the course
of the study.
Some people have hate of writing a protocol and final
report
A way out for a non writer with a good idea for a
research project is to team-up with a co-investigator who
is willing and able to write the proposal and the final
report.
BY
SAHILU ASSEGID,
JANUARY/2009
56
3. CONTENTS OF A PROPOSAL
A. TITLE
- Remember that the title will be read by thousands of
people
- Perhaps few people will read the entire paper
- All words in the title should be chosen with great care
and, association with one another must be carefully
managed
- The title page identifies the proposal and provides the
endorsement of appropriate organizational (e.g.
University) officials

BY
SAHILU ASSEGID,
JANUARY/2009
57
TITLE PAGE...
Length of a title:
- Occasionally titles are too short
- Much more often, titles are too long
- Long titles are often less meaningful than short ones
- Words such as Studies on, Investigations on and
Observations on are waste words
A good title is defined as the fewest possible
words that can adequately describe the contents
of the study
BY
SAHILU ASSEGID,
JANUARY/2009
58
TITLE PAGE...
Title is a label, it is not a sentence
Titles should almost never contain abbreviations
BY
SAHILU ASSEGID,
JANUARY/2009
59
B. SUMMARY
A concise summary of the material presented in the
proposal, it is a mini version of the paper
Though it appears at the front of the proposal, it is
written last
These materials are condensed to a page or less (less
than 250 words)
A well prepared summary enables reader to:
- Identify the basic content of a document quickly & accurately
- Determines its relevance to their interest, and
- Decide whether they need to read the document in its entirely
Summary should be typed as a single paragraph
BY
SAHILU ASSEGID,
JANUARY/2009
60
SUMMARY
The summary should:
- State the significance and need for the study
- State the objectives and scopes of the investigation
- Describe the methods employed
- Summarize important findings (In case of final report)
- State the principal conclusions (In case of final report)
Summary should never give any information or
conclusion that is not stated in the paper
Reference to the literature must not be cited
Summary should be written in the past tense because it
refers to work done (in case of final report)
BY
SAHILU ASSEGID,
JANUARY/2009
61
C. INTRODUCTION
A brief statement of the problem to be addressed
Need for a background and rationale of the envisaged
study
- What have other researchers concluded about this topic?
- What theories can be used to shed light on it and what are their
major components ?
- What relevant empirical research has been done previously?
- Are there consistent findings, or do past studies disagree?
- Are there limitations and gaps in the body of existing research, that
the proposed study can resolve?

BY
SAHILU ASSEGID,
JANUARY/2009
62
INTRODUCTION...
Introduction generally contains information
showing:

- What has been accomplished in the field
- That the investigator is well acquainted with the past and
current work and with the literature in the field, and
- That the proposed project will advance or add to the
present store of knowledge in the field or be important to
the solution of the problem.
BY
SAHILU ASSEGID,
JANUARY/2009
63
D. OBJECTIVES
General objective:
- What exactly will be studied
- General statement specifying the desired outcomes of
the proposed project
Specific objectives:
- Specific statements summarizing the proposed activities
and including description of the outcomes and their
assessment in measurable terms

BY
SAHILU ASSEGID,
JANUARY/2009
64
E. MATERIALS AND METHODS
The most important part of the proposal!
The proposal describes in detail the general
research plan!
- Description of study area
- Description of study design
- Description of study subjects
- Eligibility criteria
- Determination of sample size
- Description of selection process (sampling method)
- Description of the type of data to be collected

BY
SAHILU ASSEGID,
JANUARY/2009
65
MATERIALS AND METHODS...
- Methods of data collection
- Description of the outcomes and explanatory
variables...if any
- How data quality is ensured
- Description of the criteria for case definition (operational
definition)
- Presentation of data analysis methods
BY
SAHILU ASSEGID,
JANUARY/2009
66
F. SIGNIFICANCE OF THE PROPOSED WORK

- Needs to be stated in sufficient details
- Why is the topic worth studying?
- What is the practical significance of the proposed
research?
- Does it contribute to improved understanding of health
services delivery, problem solving or the refinement of
existing theory?
BY
SAHILU ASSEGID,
JANUARY/2009
67
G. AVAILABLE FACILITIES

- Facilities available to promote the aspired research
should be stated
BY
SAHILU ASSEGID,
JANUARY/2009
68
H. ETHICAL CONSIDERATIONS
- Plans for safeguarding the safety and welfare of the
study subjects and the method of obtaining their consent
need to be explained
- Does consenting to be subject lead to additional costs in
tests, medical care etc. for the subject? If so, who is
responsible for the cost?
- What are the potential risks for the subjects? In your
estimation, do the explained procedures involve any
potential risk for the subject-physically, psychologically,
socially or legally?
BY
SAHILU ASSEGID,
JANUARY/2009
69
ETHICAL CONSIDERATIONS...
- If any of your procedures create potential risks for any of
the subjects, describe:
- Other methods, if any, that were considered and why they will not
be used and

- What precautions you plan to take to reduce the possibility of such
risks
BY
SAHILU ASSEGID,
JANUARY/2009
70
Who will benefit from this study? What is the significance
of potential benefits to be gained by:

- The subjects used in the investigation
- Persons similarly situated
- The scientific community and
- The humankind in general

BY
SAHILU ASSEGID,
JANUARY/2009
71
- How will the subjects privacy be protected? What are
your procedures for safeguarding each subjects right
with respect to:
- Safety and security of the individual?
- Privacy and confidentiality (including protection and anonymity of
the data)?
- Embarrassment?
- Discomfort and
- Harassment (i.e. would there be stigma or repercussion form having
participated in the study)?

BY
SAHILU ASSEGID,
JANUARY/2009
72
I. COMMUNICATION PLAN
- It indicates how to make research findings available to
others, particularly those interested in the study
outcomes.

- It provides research results to a regional, national or
international audience.

- Researchers should negotiate with the funding agency,
iff there is any, on publication rights prior to submitting
papers for professional publication.

BY
SAHILU ASSEGID,
JANUARY/2009
73
J. TIME PLAN
- It describes the plan of assessing the ongoing progress
toward achieving the research objectives.

- The plan specifies how each project activity is to be
measured in terms of completion, the time line for its
completion.

- A good work time plan enables both the investigators
and the advisors to monitor project progress and provide
timely feedback for research modification or adjustment.

BY
SAHILU ASSEGID,
JANUARY/2009
74
TIME PLAN...
- In the time plan:
- Different components/phases/stages of the study
should be stated
- Description of activities in each phase
- Time required to accomplish the various aspects of the
study should also be indicated
BY
SAHILU ASSEGID,
JANUARY/2009
75
K. BUDGET
- Budget items need to be explicitly stated
- Cost for every budget item should be
quantitatively shown
- There might be a need for budget justification of
certain costs whose requirement is not obvious.
- Typically, proposal budget reflects direct and
indirect costs
BY
SAHILU ASSEGID,
JANUARY/2009
76
BUDGET...
1. Direct cost:
- Personnel salaries and wages of all participants of the
study: PI, supervisors, data collectors, training drivers,
guards, data entry clerks, data analysis, report writing
etc.
- Consumable supplies: office supplies (stationeries),
computers, chemicals, and educational materials
- Equipments: properties which are expensive
- Travel; cost of project-related travel
- Communications: postage, telephone, telegram, fax, e-
mail channels associated with the project

BY
SAHILU ASSEGID,
JANUARY/2009
77
BUDGET...
- Publication: the cost incurred of preparing and publishing
the results of the research. It includes: technical reports,
manuscripts, illustrations, graphics, photocopy, slides
and overheads.
- Consultant services: the budget should identify each
consultant
- Other direct costs: costs of all items that do not fit into
any of the above direct costs
BY
SAHILU ASSEGID,
JANUARY/2009
78
BUDGET...
2. Indirect costs: those costs incurred in the support and
management of the proposed activities that can not be
readily determined by direct measurement. Examples
include,
- Overhead costs for institutions or associations
- General administrative costs
- Operational and maintenance
- Depreciation and use allowances

BY
SAHILU ASSEGID,
JANUARY/2009
79
OTHERS
L. Collaborative arrangements... If any
M. PIs assurance of study
N. Appendix
- References
- Questionnaires and other collection forms
- Dummy tables

THE END
BY
SAHILU ASSEGID,
JANUARY/2009
80
CHAPTER FOUR

RESEARCH QUESTION/SURVEY
RESEARCH
BY
SAHILU ASSEGID,
JANUARY/2009
81
OUTLINE OF THE PRESENTATION

Formulation of a research question
Survey research- cross sectional studies
Overview of sampling
Tips on questionnaire
Overview of rapid assessment
BY
SAHILU ASSEGID,
JANUARY/2009
82
1. FORMULATION OF A RESEARCH QUESTION
Defining a problem:
- First step & the most difficult in research
- Tendency for the beginner to ask diffuse or vague
question
- New researchers wish to make important discoveries
- Impractical desire leads to failure
- Sciences often progress in small steps
- Thus, research problems must be stated in terms of
clear, simple and answerable questions

BY
SAHILU ASSEGID,
JANUARY/2009
83
FORMULATION OF A RESEARCH QUESTION...
=>Need to review relevant literature, why?
- To learn about what is already known pertaining to the
study question:
- Research should bridge the gap of information
- Research should be need based
- Research should be problem oriented/problem solving
- To be familiar with problems that others have faced
- To learn about the study methods employed
- We have to use standard methods for comparison of our
findings
BY
SAHILU ASSEGID,
JANUARY/2009
84
- To be familiar with pertinent variables to be measured or
controlled

- To realize how little is known about the particular topic
one wishes to investigate

- To get encouragement to proceed with the proposed
study or be convinced to switch to unexplored related
problems
- Need for enough justification to conduct research
BY
SAHILU ASSEGID,
JANUARY/2009
85
- Although the beginning investigator might have a plan of
what he/she wants to do in mind, it is important to set it
in writing

- Systematically presented scheme- Protocol

BY
SAHILU ASSEGID,
JANUARY/2009
86
- A written protocol serves major functions:
- Ideas and procedures are clearly defined and spelled out
- Gaps and errors are easier to recognize and correct when the plan
is seen on paper than the plan is in ones mind
- A written protocol can be studied by anyone whose advice is
desired or whose approval is required.
- The written protocol constitutes a permanent record of the
procedures that should be followed and referred in the course
of the study.

BY
SAHILU ASSEGID,
JANUARY/2009
87
2. PARTS OF A PROPOSAL
Title
- Where problem being worked on is judicated precisely
Abstract/summary
- Brief presentation of the envisaged study
- Precise statement of the problem
- Highlight the objectives
- Mention of the methodology/procedures to be employed
- Expected outcomes
BY
SAHILU ASSEGID,
JANUARY/2009
88
PARTS OF A PROPOSAL...
Introduction/Background & rationale...

- Statement of why the topic is of interest
- Importance of the problem
- Current status of the issue
- Gaps in knowledge
- Rationale behind the envisaged study

BY
SAHILU ASSEGID,
JANUARY/2009
89
Objectives:
- General- contribution to the overall understanding of the issue
under investigation
- Specific- indication of facts envisaged to be established
- indication of profiles expected to be elucidated
Methodology:
- What is expected to be done and how
- Study design, sample size and sampling techniques
- Study population, study area etc.
- Variables, data sources, data collection methods
- Data analysis, techniques, software and estimates
BY
SAHILU ASSEGID,
JANUARY/2009
90
Ethical considerations:
- Safety ( information, consent, benefit, feedback...)
- Respect for human right ( culture, confidentiality...)
- Need to state ethical concerns/ issues and provisions
Work plan:
- Components/phases/stages of the project
- Description of activities in each phase
- Time required for each phase
BY
SAHILU ASSEGID,
JANUARY/2009
91
Budget:
- Quantify budget items
- Indicate unit costs
- Indicate sub-totals and grand total costs
References:
- List of literatures referred to
Appendices:
- Any other information the reader might need to know
BY
SAHILU ASSEGID,
JANUARY/2009
92
CHAPTER FIVE

STUDY DSIGNS/ SURVEY RESEARCH
BY
SAHILU ASSEGID,
JANUARY/2009
93
1. STUDY DESIGNS
STUDY DESIGNS

DESCRIPTIVE ANALYTIC

DEALING DEALING OBSERVATIONAL EXPERIMENTAL
WITH WITH - case control - intervention
POPULATION INDIVIDUALS - cohort studies
-correational/ - case report
Ecological - case series
- cross sectional
94
2. COMPARISON OF DESCRIPTIVE AND
ANALYTIC STUDY DESIGNS

Descriptive studies:
Describe events by person, place and time.
Used by health manager to allocate and to plan
effectively
Generate epidemiologic hypothesis
Inexpensive, less time consuming
Are the most common studies(80%)
Most(80%) of medical literatures are descriptive

95
COMPARISON
Analytic studies:
Have comparative groups and apply hypothesis
testing
In analytic studies, data are more likely gathered
for specific study.
. In analytic studies information on controls is as
good as that on the cases.
Analytic studies provide information on outcome
at individual level.
96
3. CLASSIFICATION OF STUDY DESIGNS
I. DESCRIPTIVE:
A. Dealing with population
- Correlational/ ecological
B. Dealing with individuals
- Case reports
- Case series
- Cross sectional surveys
97
CLASSIFICATION
II. ANALYTIC
A. Observational
- Case control studies
- Cohort studies
B. Experimental
- Intervention studies

98
CLASSIFICATION
I. Descriptive:
- often routine sources e.g. hospital records
studies that have investigated health and
health related events by person, place and
time and suggest reasons or generate
hypothesis for the observed events.
- Their characteristics are that the numerator
data are

99
CLASSIFICATION

- Denominator data are often from census
- Relatively little information over controls.

100
3.1 DESCRIPTIVE STUDIES
ECOLOGICAL/ CORRELATIONAL STUDIES

Units of analysis are populations or groups of
people rather than individuals.

101
DESCRIPTIVE STUDIES

DRAWBACKS - are features inherent in the design
of correlational studies
- Failure to link exposure with disease in particular
individuals
- Inability to control confounders
- Show only average exposure level, not actual
individual values
- Non differential/ differential misclassification
102
3.2. DESCRIPTIVE STUDIES
Case report/ case series
- Refer to one and more than one cases
respectively
- Unusual
- Link between clinical medicine and epidemiology
- Give clue in identification of new disease or
adverse effect of exposure
103
3.2. DESCRIPTIVE STUDIES
- Single case of thrombo-embolism, led to
generation of hypothesis that use of oral
contraceptive has no association with thrombo-
embolism.
- Case series gives clue as occurrence of
epidemic? New disease?
- Drawbacks are that, there is no comparison
group.

104
3.3 CROSS SECTIONAL STUDIES
Status of disease/exposure is assessed
Fixed point-time window is same calendar time
Variation in real time from person to person
e.g. pre employment retirement
Is snap shot
Used to assess health status / health care needs
Used to assess prevalence of diseases or health
outcome in certain occupation
105

106
CROSS SECTIONAL
- Show only determinants of survival and
etiology
- Used only to generate hypothesis after
describing person, place and time very well.
Hypothesis generation
1. Method of difference between and within PPT
=> Different diseases caused by a single factor
107
CROSS SECTIONAL
2. Method of agreement between and within PPT
=> Different factors cause same disease e.g. HIV-
homosexuals, drug users and hemophiliacs.
3. Concomitant variation- homogeneity of
frequency of factor, is it proportional to
frequency among diseased?
=> Increased proportion of factor leading to
increased proportion of diseases (correlation
studies)

108
CROSS SECTIONAL
Drawbacks are:
A. Temporal relationship egg or chicken?
phenomena e.g.
Q1 Do you know any modern contraceptive
method? Y/ N
Q2 Are you using modern contraceptive? Y/ N

109
CROSS SECTIONAL
Suppose, you find association between the
two. How would you interpret it?

- Knowledge use? Or
- Use knowledge?
110
CROSS SECTIONAL
Suppose, you find association between the
two. How would you interpret it?

- Knowledge use? Or
- Use knowledge?
111
CROSS SECTIONAL
B. Need for follow up
Job A 100 80 well 80 well
20 ill 10 ill
10 ill 10 ill
moved
Job B 100 95 well ----------------95 well
5 ill 15 ill
X Y

112
4. ANALYTIC STUDIES

A. OBSERVATIONAL STUDIES

CASE CONTROL STUDIES
COHORT STUDIES

113
4.1 CASE CONTROL STUDIES
A. Uses of case control studies:
- Establish disease- exposure association
E.g. behavioral factor vs. HIV seroconversion
- Investigate consequence of diseases
E.g. psychiatric disorder vs. malnutrition
- Evaluate intervention or screening program e.g.
pap smear
- For post marketing drug surveillance
E.g. OCP vs. thrombo embolism

114
CASE CONTROL
B. Characteristics:
- Suited for diseases with long latency
- Are efficient in terms of time and cost
- Are suited for rare disease
- Used to study many potential etiologic
exposure vs. outcome/disease
- Helps in early knowledge about
diseases/outcomes.
115
1. STUDY DESIGNS
Broadly divided into experimental and
observational studies:
1. Experimental studies
- Investigator intervenes and makes one variable change
- The condition under which the study is conducted and the method of
assigning the subjects is controlled
- Usually randomly allocate similar individuals to exposed or non-
exposed group ...clinical trial
- Difficulties of performing well on human populations
- Availability of abundance observational data
- Epidemiological studies ten to concentrate on observational
studies RESEARCH METHODOLOGY
BY
SAHILU ASSEGID,
JANUARY/2009
116
STUDY...
2. Observational studies:
- Nature allowed to take its course
- Changes/differences in one characteristic are studied in relation to
changes/differences in the other
BY
SAHILU ASSEGID,
JANUARY/2009
117
2. MAJOR TYPES OF OBSERVATIONAL
STUDIES
1. Cross sectional study
2. Case control study
3. Cohort study
- Range from the weakest to the strongest
- Each study type has its own merits
- Study types have a well evident limitations
- Choice of study methods depend on several
factors
BY
SAHILU ASSEGID,
JANUARY/2009
118
3. CROSS-SECTIONAL STUDIES
Alternative terminologies:
- Survey research
- Prevalence study
- Survey
- Sample survey (partial enumeration)
Methodological issues:
- From whom to collect information
- What methods to use for collection
- How to process & interpret the generated information
BY
SAHILU ASSEGID,
JANUARY/2009
119
CROSS-SECTIONAL STUDIES...

Census and sample survey:
- Surveys similar to census
- Census implies enumeration of entire population
- Surveys examine a sample of a population
- Census dates back to Egyptian civilization
- Political function of survey research continue to the
present day
BY
SAHILU ASSEGID,
JANUARY/2009
120
Some characteristics of survey research:
- One of many research tools
- Not appropriate to certain research areas
- Can be used properly in certain cases
- Could be effective when used with other methods
- Other methods in social research include:
- Case study
- Participant observation
- Analysis of existing data etc. (see evaluation research designs)
BY
SAHILU ASSEGID,
JANUARY/2009
121
Factors dictating sample surveys:
- Population being large or dispersed
- Shortage of financial resources
- Limited time- results demanded urgently
- Shortage of trained manpower
- Consideration for a small population:
- School children
- Workers in a small factory
- Inhabitants of a village
BY
SAHILU ASSEGID,
JANUARY/2009
122
Factors dictating sample surveys cont...
- Probability sampling methods give result of the required
precision;
- Wrong to arbitrarily select cases
- Difficult to rely on volunteers
- Difficult to count on people at hand
- Situations where arbitrarily selected population or families are
intensively studied: Sociological surveys

BY
SAHILU ASSEGID,
JANUARY/2009
123

Initial steps:
- Define study question clearly
- Identify study population e.g. school population
- If population is small, study total population
- If target population is large, representative sample is
selected
BY
SAHILU ASSEGID,
JANUARY/2009
124
Surveys examine certain characteristic and
proceed as follows:
- Consider a sample
- Construct a questionnaire to elicit information
- Administer questionnaire to a sample of study subjects:
- Face-to-face interviews
- Self administered formats
- Mailed questionnaires
BY
SAHILU ASSEGID,
JANUARY/2009
125

- Responses coded into a standardized form
- Coded responses analyzed
- Standardized responses subjected to aggregated
analysis to provoke description of the sample
- The descriptive conclusions then generalized to the
population from which the sample is selected
BY
SAHILU ASSEGID,
JANUARY/2009
126
4. USES OF CROSS-SECTIONAL STUDIES
Measures of prevalence:
- Describe the burden of disease
- Describe the distribution (person, place & time)
Migrant studies:
- Gives clues on effect of genetic background and
environmental exposure on risk of disease
KAP studies:
- Purely descriptive with out necessarily relating
population characteristics to any health outcomes
BY
SAHILU ASSEGID,
JANUARY/2009
127
USES OF CROSS-SECTIONAL STUDIES...
Public health planning:
- Information used to plan service
- To assess utilization of health service
- To assess effectiveness of health service
Studies of disease determinants:
- Gives clue of etiology pertaining to diseases of
slow onset and long duration
- Cross-sectional studies have certain advantages over
case-control and cohort studies
- They have a well known limitations

BY
SAHILU ASSEGID,
JANUARY/2009
128
5. RESEARCH DESIGN FOR SURVEY
Two major types:
A. Single cross-sectional design;

- Non-directional design
- Provides information about;
- Frequency of disease, and
- Relationship between disease and other variables of
interest
BY
SAHILU ASSEGID,
JANUARY/2009
129
RESEARCH DESIGN FOR SURVEY...
B. Design for assessing change:
1. Time series design;
- Successive independent samples checked, trend design
- Involves the selection and measurement of two or more different
samples at different times from the same
2. Panel design;
- Information collected from the same sample on more than one
occasion
3. Combined designs;
- Provides a way of overcoming some of the weaknesses of the
previous designs (assesses the learning effect)
BY
SAHILU ASSEGID,
JANUARY/2009
130
6. ADVANTAGES OF CROSS-SECTIONAL
STUDIES
Over case-control studies:
- Generalizability, since based on a sample of the
general population
- Not just those seeking medical care, unlike
case-control studies
Over cohort studies:
- No need for large sample size
- No need for long period of follow up
- No need to define incident cases
BY
SAHILU ASSEGID,
JANUARY/2009
131
7. DATA COLLECTION IN PREVALENCE
STUDIES
- Once study population is defined, necessary data is
collected
- Presence of disease may be determined by
- Interview of residents
- Examination of residents
- Reviewing of hospital records
- Contacting practicing physicians
- Presence of exposure (to possible causative factors) should be
determined by appropriate tests e.g. cigarette smoking, by
interview or air pollution levels - by appropriate measuring
devices.
BY
SAHILU ASSEGID,
JANUARY/2009
132
8. DATA ANALYSIS
- Manually, calculator, computer
- Tabulate the study data by dividing population into sub
groups
- Calculate the necessary frequencies and estimates etc.

=> Point prevalence
= Number of persons with disease at the time of survey
Total population surveyed
BY
SAHILU ASSEGID,
JANUARY/2009
133
9. FACTORS INFLUENCING PREVALENCE
RATE

Prevalence rate is increased by:
- Long duration of the disease
- Prolongation of life of cases without cure
- Increase in new cases
- In-migration of cases
- Migration of healthy people
- In-migration of susceptible individuals e.g. aged

BY
SAHILU ASSEGID,
JANUARY/2009
134
FACTORS INFLUENCING PREVALENCE RATE...

Prevalence rate is decreased by:
- Short duration of diseases
- High case fatality from disease
- Decrease in new cases
- In-migration of healthy individuals
- Migration of cases
- Improved cure rate of cases

BY
SAHILU ASSEGID,
JANUARY/2009
135
10. INTERPRETATION OF FINDINGS

- Caution- existing cases may not be all cases
- Diseases of short duration may be missed
- Cases of long duration may be over represented
- Prevalence depends on incidence and duration (P=ID)
- Institutionalization of some affected person in specialist
centers could also underestimate prevalence
BY
SAHILU ASSEGID,
JANUARY/2009
136
11. SELECTING THE SAMPLE
Importance of sampling:
- To save time and money
- Measurements more accurate on samples than entire
population (census)
The sampling process: definitions
- Sampling unit- the basic unit around which a sampling
procedure is planned e.g. person, household
- Sampling frame- listing of all sampling units
Defining the population:
- Target population
- Study population

BY
SAHILU ASSEGID,
JANUARY/2009
137
12. TYPES OF SAMPLING

Probability sampling
Non-probability sampling
BY
SAHILU ASSEGID,
JANUARY/2009
138
A. PROBABILITY SAMPLING
1. Simple random sampling:
- Uses a random process in which each person has
equal chance of being selected

2. Systematic sampling:
- Involves first deciding what fraction of the population to
be studied- one-half/tenth
BY
SAHILU ASSEGID,
JANUARY/2009
139
PROBABILITY SAMPLING
3. Stratified sampling:
- Involves dividing the population into distinct sub groups
(strata) according to some important characteristics such
as sex, age socio-economic status-and taking a random
sample from each strata.

4. Cluster sampling:
- A process of taking a random sample of natural groupings
(clusters) of individuals in the population

BY
SAHILU ASSEGID,
JANUARY/2009
140
B. NON-PROBABILITY SAMPLING
1. Consecutive sampling:
- Involves taking every subject who meets the selection
criteria over a specified time interval or number of
patients.
2. Convenience sampling:
- A process of taking those members of the accessible
population who are easily available.
BY
SAHILU ASSEGID,
JANUARY/2009
141
NON-PROBABILITY SAMPLING...
3. Judgment sampling:
- Involves selecting from the study population, those
individuals judged by the sampler to be most appropriate
for the study
- Problem- lack of representative ness of the target
population.
4. Snow-ball sampling:
- Used when the population members are hard to identify
- Relies upon currently identified sample members to
identify other members of the population to be contacted.

BY
SAHILU ASSEGID,
JANUARY/2009
142
13. TIPS ON QUESTIONNAIRE ADMINISTRTION
A. General guidelines:
- Questions should be easy and simple
- Maintain level of interest of respondents
- Fewer questions than more
- Shorter and clearer questions
- Avoid going on fishing expedition- focus only on relevant
information
- Make use of questions from existing surveys
BY
SAHILU ASSEGID,
JANUARY/2009
143
TIPS ON QUESTIONNAIRE ADMINISTRTION...
B. Types of questions:
- Open-ended questions:
- Respondents asked to provide their own answers
- Closed-ended questions:
- Respondents asked to select their answers
- Provide greater uniformity of responses
C. Question order:
- Group related questions together
- Begin the questionnaire with easy questions

BY
SAHILU ASSEGID,
JANUARY/2009
144
TIPS ON QUESTIONNAIRE ADMINISTRTION...
D. Pre-testing a questionnaire:
- Valid to make the questionnaire works
- Temptations to omit pre-testing
-Expert review to identify content omissions
- Non-expert reviews to assess difficult questions & examine
logical flow of items.
E. Administration of questionnaire:
- Self administered- in a supervised setting
- Interviewer administered
=> Face-to-face- communicating directly
=> Over telephone-especially in short questions
- Convincing respondents is a key to data quality-----------RAP- read
BY
SAHILU ASSEGID,
JANUARY/2009
145
THE END

OF

SURVEY RESEARCH
BY
SAHILU ASSEGID,
JANUARY/2009
146
CHAPTER SIX

DATA SOURCES FOR RESEARCH
BY
SAHILU ASSEGID,
JANUARY/2009
147
1. SCIENTIFIC METHOD

=> The pursuit of systematic interrelation of facts by logical
arguments from accepted postulates, observation, and
experimentation and a combination of these three in
varying proportions.
BY
SAHILU ASSEGID,
JANUARY/2009
148
2. LOGICAL ARGUMENTS
- Deductive argument: conclusion follows with logical necessity or
certainty from the premises. Nothing new is revealed because we
are arguing from the general to the specific.
- Specialization: moving from a large set of objects, postulates, or
events, to consideration of a smaller set of objects or events.
- Inductive argument: The premises provide some evidence for the
truth of the conclusions. Discovering general laws by the
observation and combination of particular instances.
- Generalization: Passing from the consideration of one object,
postulate or occurrence, to the consideration of a set of objects,
postulates or occurrences.
=> In statistics, we attempt to formalize and use these concepts in a
quantitative way.
BY
SAHILU ASSEGID,
JANUARY/2009
149
3. SCIENTIFIC PROGRESS
- We gain knowledge by iterating between models and data:

Hypothesis
Model, conjecture New hypothesis, New model

Purposes and
Understanding
Deductive Inductive
Argument Argument

Data, Measurements New data
BY
SAHILU ASSEGID,
JANUARY/2009
150
4. BASIC STUDY STEPS
- State the problem: What are the questions?
- Devise a plan of solution: What will I do?
- Implement the plan: This is how I do it?
- Analysis of data: What happened?
- Interpretation of results: What does this mean?
- Re-examination: Is my logic correct? What next?

=> Study design and study implementation may require
iteration.
BY
SAHILU ASSEGID,
JANUARY/2009
151
5. GRAPHICAL DEPICTION OF SCIENTIFIC
STUDY
Experimental
Observational

How to
Measure?

Problem
Constraints
Objectives
And
Hypothesis
Knowledge
Base
Interpretation
Design
Data
STATISTICAL ANAYSIS
Graphics & visualizations
- Modeling
-Estimation & confidence intervals
-Formal Statistical Tests
Conclusions
BY
SAHILU ASSEGID,
JANUARY/2009
152
6. DATA
Any record, descriptive account, or symbolic
representation of an attitude, event or process,
may constitute a data point
Data is the key component of a research
- It is a capital mistake to theorize before one has data
- The key to a good research is good data!
=>We must not draw any conclusion from the data before
we ascertain whether or not any problem exists in the
data and, if so, their possible effects.

BY
SAHILU ASSEGID,
JANUARY/2009
153
DATA...
=>Guarding against misuse of data is as important as
learning how to make effective use of data.
- We need to learn to use numbers carefully the meaning of the
numbers to distinguish FACT from FICTION.

Data:
- Primary data: data that one has collected oneself
- Secondary data: data that has already been collected by
some body else
BY
SAHILU ASSEGID,
JANUARY/2009
154
DATA...
Type Advantages Disadvantages
Primary data
We collected it.
So, we understand
the curiosities that
it contains.
We collected it.
So, there were
probably not very
many
observations
Secondary data
There will often be
a very large
number of
observations
We did not collect
the data, so it may
not really answer
the question that
interests
BY
SAHILU ASSEGID,
JANUARY/2009
155
DATA...

- Researchers must decide on what data to collect and
how to collect it, all the while keeping in mind the
requirements of the data analysis.

=> Researchers must understand the strengths and weaknesses of
their data. No data will perfectly reflect reality, but researchers have
the responsibility to understand the sources of data imprecision and
potential for bias, so that they may temper conclusions accordingly.
BY
SAHILU ASSEGID,
JANUARY/2009
156
7. DATA (OBJECTIVE OF THE STUDY)

- The starting point of any investigation must be
defining clearly its objectives, since these will
determine the appropriate study design and type
of data needed.
- Objectives may be categorized into one of three
main types.
- An investigation usually has several objectives,
which can be, of course, be of different types.
BY
SAHILU ASSEGID,
JANUARY/2009
157
DATA (OBJECTIVE OF THE STUDY)...

1. Estimation- of certain features of a population.
2. Investigation of association between a factor
of interest and a particular outcome, such as
disease or death.
3. Evaluation- of drug or therapy or of an
intervention aimed at reducing the incidence
(severity) of disease.
BY
SAHILU ASSEGID,
JANUARY/2009
158
8. WHAT DATA TO COLLECT
- The development of data collection instruments is a
crucial point in the conduct of research.
- The information needed to address the primary study
hypothesis determines what data need to be collected.
- In order to be certain that the necessary data are going
to be collected, it is a useful practice to dummy out
tables of planned manuscript before drafting the data
collection document.
- These tables can serve as a checklist for the data
collection instrument.
BY
SAHILU ASSEGID,
JANUARY/2009
159
WHAT DATA TO COLLECT...
- Two broad categories of data collection:

A. Rapid appraisal method (qualitative)
B. Formal method (quantitative)
BY
SAHILU ASSEGID,
JANUARY/2009
160
A. RAPID APPRAISAL- WHAT DATA...
- Used to:

- Obtain non-quantitative information rapidly
- When little time and resources are available
- Limited information is still useful
- No data are initially available and quick preliminary
inquiry is necessary to provide direction for further
study
BY
SAHILU ASSEGID,
JANUARY/2009
161
RAPID APPRAISAL- WHAT DATA...

- The methods constitute:
- Key-informant interview
- Focus group discussion
- Community/group interview
- Structured direct observation
- Record review
BY
SAHILU ASSEGID,
JANUARY/2009
162
- Limitations:
- The reliability and validity of the information generated
can be questionable in many instances
- The methods do not generate quantitative data from
which generalization can be made for a whole population
- The general credibility of these methods is low
compared to formal survey methods

BY
SAHILU ASSEGID,
JANUARY/2009
163

- The rapid appraisal methods should be used
selectively, depending on:
- The purpose of the study
- Availability of resources
- Above all, the nature of the information required
BY
SAHILU ASSEGID,
JANUARY/2009
164
B. FORMAL METHOD-WHAT DATA TO
COLLECT

- The formal method requires standard statistical
procedures and techniques to generate
quantitative data that can be analyzed to draw
conclusions within specified confidence margins.
BY
SAHILU ASSEGID,
JANUARY/2009
165
FORMAL METHOD-WHAT DATA...

- Formal methods include:
- Cross-sectional survey
- Case-control study
- Longitudinal survey (cohort study)
- Censuses
- Experiments
BY
SAHILU ASSEGID,
JANUARY/2009
166

- Four kinds of data are usually collected in
quantitative study:
A. Exposure
B. Disease/outcome
C. Confounders/Risk modifiers
D. Bias indicators
BY
SAHILU ASSEGID,
JANUARY/2009
167
A. Exposure: The goal of exposure measurement
is to document differences in the experience of
cases and compare to non-cases.
B. Disease/Outcome: Researchers may wish to
collect information from several points in the
process. At a minimum, the researchers must
collect the data required to determine whether
study participants meet the criteria set out in
the studys case definition (outcome).
BY
SAHILU ASSEGID,
JANUARY/2009
168
C. Confounders/Risk modifiers:
- Confounders are causal risk factors for the outcome
of interest other than the exposure under consideration.
- Data on confounders must be collected, therefore, to allow
appropriate statistical adjustments to be made in the design
or analysis stage.
- Risk modifiers: Are factors that influence the strength of
an exposure-disease relationship.
- Age, gender, social class could be risk modifiers in certain
investigations.
BY
SAHILU ASSEGID,
JANUARY/2009
169
D. Bias indicators: Are questions that help to
assess systematic differences in data quality
between cases and non-cases.
BY
SAHILU ASSEGID,
JANUARY/2009
170
9. COLLECTING THE DATA
- After deciding what data need to be collected,
the investigator has to decide how to collect it:
A. Decide on mode of data collection
- Rapid appraisal
- Formal survey
B. Developing the forms to guide data collection
C. Identify the data sources
- Individuals, patients, proxies, medical records, direct
observations
BY
SAHILU ASSEGID,
JANUARY/2009
171
COLLECTING THE DATA...
D. Identify data collectors
E. Timing of data collectors
F. Pre-testing of data collection instruments
G. Data management process
=> Decision in each of these areas affect the
others, and investigators should understand the
trade-offs between them in order to make good
decision when planning a research.
BY
SAHILU ASSEGID,
JANUARY/2009
172
10. MAINTAINING DATA QUALITY
Sources of errors:
- It is important to be aware of the various ways errors
may creep into the design, conduct and analysis of a
study in order to be able to take steps to minimize their
occurrence.
=>Two sources of errors:
A. Random
B. Systematic
BY
SAHILU ASSEGID,
JANUARY/2009
173
11. ERRORS IN EPIDEMIOLOGICAL STUDIES
Error

Random error (chance)

Systematic error (bias)

Study size

BY
SAHILU ASSEGID,
JANUARY/2009
174
ERRORS IN EPIDEMIOLOGICAL STUDIES...
A. Random errors:
- Give rise to reduce precision, but not in general to
invalidity.
B. Systematic errors:
- Potentially serious, since the bias they cause may lead
to the invalid conclusions being drawn
- Systematic errors occur during data collection through:
- Questionnaire or equipment faults
- Observer errors
- Respondent mistakes
BY
SAHILU ASSEGID,
JANUARY/2009
175
ERRORS IN EPIDEMIOLOGICAL STUDIES...
- During data processing through:
- Coding
- Copying
- Programming
- Errors also arise through inappropriate choice of methods
of analysis and incorrect interpretation of results.
BY
SAHILU ASSEGID,
JANUARY/2009
176
12. BIAS

EVERY EPIDEMIOLOGICAL STUDY SHOULD BE
VIEWED AS A MEASUREMENT EXERCISE
K. ROTHMAN
BY
SAHILU ASSEGID,
JANUARY/2009
177
SHOULD I BELIEVE MY MEASUREMENT?
Soya beans Asthma ED
RR=23
True association
-
Causal
-Non-causal
Chance?
Confounding?
Bias?
BY
SAHILU ASSEGID,
JANUARY/2009
178
SOURCES OF BIAS

Bias may be classified into three main
categories:
1. Selection bias
2. Confounding bias
3. Information bias
BY
SAHILU ASSEGID,
JANUARY/2009
179
DEFINITION OF BIAS

Any systematic error in an epidemiological
study that results in an incorrect estimation
of the association between exposure and
risk of disease.
BY
SAHILU ASSEGID,
JANUARY/2009
180
SOURCES OF BIAS

Bias may be classified into three main
categories:
1. Selection bias
2. Confounding bias
3. Information bias
BY
SAHILU ASSEGID,
JANUARY/2009
181
SOURCES OF BIAS...
I. Selection bias: may result from one (or both) of
two different design inadequacies in the way
subjects are selected for study.
=>The first is if selected subjects for study differ in some
way from those not selected. Main causes are:
- High non- response rate
- Loss to follow-up
- Inappropriate choice of sampling frame (technique)
=>The second type is selection bias in a comparative study
resulting from an inappropriate choice of comparison
groups.

BY
SAHILU ASSEGID,
JANUARY/2009
182
1. SELECTION BIAS
A. Sampling bias
B. Ascertainment bias
- Surveillance
- Referral, admission
- Diagnostic
C. Participation bias
- Self selection (volunteerism)
- Non-response, refusal
- Healthy worker effect, survival
BY
SAHILU ASSEGID,
JANUARY/2009
183
1.1 SELECTION BIAS IN CASE-CONTROL
STUDIES
- Selection bias:
Cases Control A
Liver cirrhosis Trauma ward
Heavy alcohol
use 80 40

Light/no alcohol 20 60
use
OR= 6
How representative are hospitalized trauma patients of the population
which gave rise to these cases?
BY
SAHILU ASSEGID,
JANUARY/2009
184
SELECTION BIAS IN CASE-CONTROL
STUDIES...
- Selection bias:
Cases Control A Control B
Liver cirrhosis Trauma ward Non trauma ward

Heavy alcohol
use 80 40

Light/no alcohol 20 60
use

OR=6 OR= 36

10
90
BY
SAHILU ASSEGID,
JANUARY/2009
185
STUDIES...
- Diagnostic/ascertainment bias:
- Diagnostic approach related to knowing exposure status
Cases Controls
Uterine cancer
Takes OC

Does not take OC

- OC use => breakthrough bleeding => increased chance of detecting
uterine cancer
- Over estimation of a => over estimation of OR

a b
c d
BY
SAHILU ASSEGID,
JANUARY/2009
186
STUDIES...
- Admission bias:
- Exposed cases have more chance of admission than controls

Cases of lung Controls from
Cancer Surgical wards
Contact with
Asbestos

No contact with
Asbestos

- Lung cancer cases exposed to asbestos not representative of lung cancer
cases
- Over estimation of a => over estimation of OR

a b
c d
BY
SAHILU ASSEGID,
JANUARY/2009
187
STUDIES...
- Survival bias:
- Only survivors of a highly lethal disease enter study

Cases of
Hemorrhagic fever Controls

Exposed to lethal

Not exposed to lethal

- Contact with risk factor lethal leads to rapid death
- Under estimation of a => under estimation of OR

a b
c d
BY
SAHILU ASSEGID,
JANUARY/2009
188
STUDIES...
- Non-response bias:
- Papanecolau test
Cases of cervical Controls
Cancer

Did not have a test

Had test

Total 1000 1060
- Controls chosen among women at their homes 1300 homes
contacted=> 1060 controls. Controls mainly house wives with lower
chance of having test than gainfully employed
- Under estimation of d => under estimation of OR

a b
c d
BY
SAHILU ASSEGID,
JANUARY/2009
189
1.2 SELECION BIAS IN COHORT STUDIES
-Non-response bias:

Lung cancer
Yes No Total

Smoker

Non smoker

RR= 9

90 910 1000
10 990 1000
BY
SAHILU ASSEGID,
JANUARY/2009
190
SELECION BIAS IN COHORT STUDIES...
-Non-response bias:

Lung cancer
10% of smokers Yes No Total
Dare to respond
Smoker

Non smoker

RR= 9

9 91 100
10 990 1000
BY
SAHILU ASSEGID,
JANUARY/2009
191
- Non response bias:

Lung cancer
Yes No Total

Smoker

Non smoker

RR= 4.7 50% of cases that smoked lost to
Follow-up

45 910 955
10 990 1000
BY
SAHILU ASSEGID,
JANUARY/2009
192
=> Loss to follow-up:
- Bias due to differences in completeness of
follow-up between comparison groups
- Example:
- Study of disease risk in migrants
- Migrants more likely to return to place of origin
when having disease
=> Lost to follow-up
=> Lower disease rate among exposed (migrants)
BY
SAHILU ASSEGID,
JANUARY/2009
193
2. MINIMIZING SELECTON BIAS

- Clear definition of study population
- Explicit case and control definitions
- Cases and controls from same population
- Selection of exposed and non exposed without
knowing disease status (retrospective cohort)
BY
SAHILU ASSEGID,
JANUARY/2009
194
II. Confounding bias: Occurs whenever there are
important differences between the groups being
compared that are also related to the variable of
interest.
III. Information bias: Results from systematically
incorrect measurements of responses or from
differential misclassification of disease or
exposure status of participants.
- It has variety of causes: questionnaire faults, observer errors,
respondent errors, and instrumental errors.
BY
SAHILU ASSEGID,
JANUARY/2009
195
MAINTAINING DATA QUALITY...
- Questionnaire faults may result from:
- Culturally inappropriate questions
- Ambiguous wording, or
- Too many questions

- Observer errors may be due to:
- Misunderstanding of procedures
- Misinterpretation of responses, or
- Simply mistakes

BY
SAHILU ASSEGID,
JANUARY/2009
196
- Respondent errors may arise through:
- Misunderstanding
- Faulty recall
- giving the perceived correct answer or
- Lack of interest (income tax, HIV/STI)
- Instrumental errors
- Faulty calibration
- Contaminated reagents
- Incorrect dilution or mixing of reagents, or
- Inaccurate diagnostic tests

BY
SAHILU ASSEGID,
JANUARY/2009
197
4. MINIMIZING INFORMATION BIAS
- Standardize measurement instruments
- Administer instruments equally to cases and
controls (exposed/unexposed)
- Use multiple sources of information
- Questionnaire
- Direct measurements
- Registers
- Case records
- Use multiple controls
BY
SAHILU ASSEGID,
JANUARY/2009
198
Clues to poor quality data:
1. Many missing values
2. Impossible or unlikely values
3. Inconsistencies
4. Irregular patterns
5. Suspicious regularity
BY
SAHILU ASSEGID,
JANUARY/2009
199
1. Missing values: data with too many missing values will
be less useful in the analysis and may indicate that
some thing went wrong with the data collection
process.
2. Impossible or unlikely values: sometimes data may
contain extreme values that are seemingly
unreasonable. For example- a persons age 12o would
be suspicious and 200 would be impossible.
- Missing values are often coded as 99 or 999 in the data file
and these may be mistakenly interpreted as valid age (value)

BY
SAHILU ASSEGID,
JANUARY/2009
200
3. Inconsistencies: are often present in the data set. For
example
- A man died because of pregnancy
- In 1840 population census report in USA, the number of African-
American insane idiots were larger than the total number of
African-Americans in some towns.
BY
SAHILU ASSEGID,
JANUARY/2009
201

4. Irregular patterns: a careful inspection of data sometimes
reveals irregular patterns. For example, census of
Turkey have a much greater frequency of multiple of 5
than numbers ending in 4 or 6 and more even-numbered
ages than odd-number ages.
- The tendency to digit preference in age reporting is quite common

BY
SAHILU ASSEGID,
JANUARY/2009
202
-Suspicious regularities: large and consistent difference in
the value of a variable may indicate that there was a
change in the measurement process that should be
investigated.
- Of course lack of inconsistencies and irregularities does not mean
there is no problem with the data. Too much consistency and
regularity sometimes are grounds for special inquire into the cause.
BY
SAHILU ASSEGID,
JANUARY/2009
203
=>Steps to have a good quality of data:
- Researchers should consider a number of steps
to maintain the quality of collected data once the
investigator begins in earnest.
- Standardized training of data collectors:
- A manual of study procedures should be proved to
each data collector and should be included.
- A review of the general procedure
- A detailed discussion of every data item to be collected and
- A discussion of how to resolve potential problems
BY
SAHILU ASSEGID,
JANUARY/2009
204
- During the study, the quality of each data
collectors work can be checked by recollecting
data on a sub sample of the study population
(5%) by trained supervisors.
- Crucial to maintaining data quality is the timely
editing and processing of the collected data.

BY
SAHILU ASSEGID,
JANUARY/2009
205

Data editing entails:
- Checking the form over completeness
- Noting any component provided by the data collector
- Coding open-ended questions
- Identifying illegal or suspicious data values and
- Identifying inconsistencies in the data set
BY
SAHILU ASSEGID,
JANUARY/2009
206
The editing should be done by the data collectors,
supervisors and principal investigator in the field.

It is best to resolve any problems while the data source
is still at hand
BY
SAHILU ASSEGID,
JANUARY/2009
207
If there is a problem with a particular data
collector, the problem needs to be resolved
quickly or data he/she collected might be
unusable
Further editing can be done by computer as the
data is being processed

BY
SAHILU ASSEGID,
JANUARY/2009
208
Each time the data is transcribed, error can occur:
- When entering the data from a paper onto computer,
enter the data twice if feasible
- Keying errors can be identified by comparing the two
data files that are generated
- At the very least, a sub-sample of forms should be re-
entered to document the quality of data entry
BY
SAHILU ASSEGID,
JANUARY/2009
209
6. CONCLUSION
Hierhoizer (Am. J. Med 1991; 21-26) has called data the
Researcher's sand. A lens maker takes sand, refines
it, melts it, and through a long process of grinding and
smoothing, fashions a lens with which to see the world
more clearly. Similarly, a Researcher takes data, refines
it and smoothes it until a clearer picture of nature is
revealed. If the sand is dirty o impure, the lens will be
cloudy and distorted. If data is unreliable or invalid, the
Researchers understanding of nature will be cloudy and
distorted.
BY
SAHILU ASSEGID,
JANUARY/2009
210
CONCLUSION...
By paying close attention to the data collection
process from the conception of the data
collection document through the editing of the
data after it is collected, the epidemiologists help
keep his sand pure so that, the end, nature
may be viewed with much clarity and possible.

No amount of sophisticated analysis can salvage either
a poorly designed or badly carried out study.
BY
SAHILU ASSEGID,
JANUARY/2009
211
THE END

OF

DATA SOURCES FOR RESEARCH
BY
SAHILU ASSEGID,
JANUARY/2009
212
CHAPTER SEVEN

SAMPLING AND SAMPLE SIZE
BY
SAHILU ASSEGID,
JANUARY/2009
213
I. OBJECTIVES
Identify and define the population to be studied
Identify and describe common methods of sampling
Discuss problems of bias that should be avoided when
selecting a sample
List the factors to consider when deciding on sample
size
Decide on the sampling methods and sample size most
appropriate for the research design they are developing.
BY
SAHILU ASSEGID,
JANUARY/2009
214
1. WHAT IS SAMPLING?

Sampling is a process of choosing a section of
the population for observation and study
BY
SAHILU ASSEGID,
JANUARY/2009
215
2. WHY SAMPLING?

Cost in terms of money, time and manpower

Accessibility

Utility
BY
SAHILU ASSEGID,
JANUARY/2009
216
3. PURPOSES OF SAMPLING
Representation:
- Sample surveys are almost never conducted for the purpose of
describing the particular sample under study. Rather they are
conducted for purposes of understanding the larger population from
which the sample was initially selected.
- A great deal of work has been done over the years in developing
sampling methods that provide representative samples for the
general population.
E.g. International survey programs such as the DHS series, EPI
coverage surveys have perfected the art of household sampling.
BY
SAHILU ASSEGID,
JANUARY/2009
217
4. CHALLENGES

Devise sampling plan that are both feasible and
capable of producing unbiased estimates for the
target population.
- If strategies are not systematic and replicable over
time, then any changes observed over time or any
difference reported by different surveys may simply be
the result of different sampling strategies.
BY
SAHILU ASSEGID,
JANUARY/2009
218
5. THE HIERARCHY OF SAMPLING
=> Study subjects
The actual participants in the study
Sample
Subjects who are selected
Sampling frame
The list of potential subjects from which the sample is drawn
Source population
The population from whom the study subjects would be obtained
Target population
The population to whom the results would be applied
BY
SAHILU ASSEGID,
JANUARY/2009
219
6. HOW SAMPLE?
There are two broad categories of sampling procedures:
probability methods and non-probability methods.

In probability sampling, every individual (element) in the
defined source of population may be selected into the
sample with a known (non-zero) probability.

Non-probability sampling is not based upon the statistical
principles which govern probability sampling.
BY
SAHILU ASSEGID,
JANUARY/2009
220
7. WHY IS THE SUB-CLASSIFICATION
NEEDED?
Homogeneity: If all members of a population were
identical to one another in all respects, there would be
no need for careful sampling procedures;
E.g. To study properties of a hydrogen atom, a chemist need not
enumerate all hydrogen atoms in the world and apply probability
sampling method to select sample of hydrogen molecules for the
study.
Heterogeneity: Faced with variation, in the population
under study, however, the researcher must use more
controlled sampling procedures.
BY
SAHILU ASSEGID,
JANUARY/2009
221
8. PROBABILITY SAMPLING
Tends in practice to be characterized by the use of lists
or sampling frame and clearly defined sampling
selection procedures.

Estimates of population characteristics derived from
studies based upon probability sampling methods may
be expected to approximate the true population value
(parameters like proportion, mean, OR, RR etc.) within a
specified margin of error.

SRS, Systematic, Stratified and Multistage sampling
BY
SAHILU ASSEGID,
JANUARY/2009
222
8.A. SIMPLE RANDOM SAMPLING

Each individual in the frame ha equal chance of
being selected.
- Use of lottery method
- Use of random numbers
- Use of computer programs to generate
random numbers
BY
SAHILU ASSEGID,
JANUARY/2009
223
8.B. SYSTEMATC SAMPLING

Every K
th
individual in the frame is selected for inclusion
in the sample.

K is sampling interval (standard distance between
individuals) obtained by dividing the number in the frame
by the sample size.

Systematic sampling involves a danger if the list of
individuals has some periodicity or some pattern.
BY
SAHILU ASSEGID,
JANUARY/2009
224
8.C. STRATIFIED SAMPLING
Sampling error is reduced by two factors:
1. Large sample size produces smaller error than do
small samples
2. Homogeneous population produce smaller errors
than heterogeneous population
Stratified sampling is based on the second factor ensure
samples are drawn from homogeneous population
The choice of stratifying variable depends on the
investigator (variables you want to represent accurately)
BY
SAHILU ASSEGID,
JANUARY/2009
225
8.D. CLUSTER SAMPLING

Involves selection of groups called clusters followed by
selection of individuals within each selected cluster.

Can be used when it is either impossible or impractical
to compile exhaustive list of individuals of the target
population.

Cluster sampling is recommended for its efficiency,
however accuracy is less because it is subject to more
than one sampling error unlike SRS.

BY
SAHILU ASSEGID,
JANUARY/2009
226
8.E. MULTISTAGE CLUSTER SAMPLING
Multistage cluster sampling then involves, the
repetitions of two basic steps: the listing and
sampling.
- List primary sampling units
- The sample of those units is selected
- List secondary sampling units
- Then sample of those units is selected
- And so forth
BY
SAHILU ASSEGID,
JANUARY/2009
227
9. SAMPLING FRAME
Sampling frames are an integral part of probability
sampling
Developing sampling frame:
- Is determining whether and where members of the
particular group tend to gather
- Identify location and estimate sizes of these potential target
group that serve as a sampling frame for the main survey
Household surveys and institutional based surveys are
simple or readily available for developing sampling frame
However in looking at sub-populations that are hard to
reach, sampling frame development requires preliminary
qualitative survey.
BY
SAHILU ASSEGID,
JANUARY/2009
228
10. DEVELOPING SAMPLING FRAME
Creating lists of sites
Sketching specific geographic areas
Creating lists of members of the target group
Estimating sizes of these potential target groups
Interviewing key informants
Spending time or observation by walking through the
community
Reviewing records such as census, DHS data, health
services reports, school registers, geographical data etc.
Discussion with experts & program managers involved in
the target group.
BY
SAHILU ASSEGID,
JANUARY/2009
229
11. NON-PROBABILITY SAMPLING
In general, they do not produce representative data for
larger population
They are cheaper and easier and are used where:
- Probability sampling methods are not feasible, because
it is not possible to construct adequate sampling frame
(e.g. animal studies).
- Good for pre-tests, pilot studies, in-depth interviews
- Precise representative ness is not necessary (e.g. to
develop sampling frame at initial design of the survey,
pilot studies, epidemic investigation, political polls etc.)
BY
SAHILU ASSEGID,
JANUARY/2009
230
12. TYPES OF NON-PROBABILITY SAMPLING

1. Snowball sampling: Key informants are contacted and these key
informants in turn identify other members of the target group. The
process goes on until the required sample is achieved.
2. Purposive (Judgmental) sampling: Selection of subjects on the
basis of your knowledge of the population, its elements and the
nature of the research objective.
3. Convenience (Reliance) sampling: Selection of a sample based
on easy accessibility e.g. friends, workmates.
4. Quota: Selection of samples based form groups cased on a fixed
quota.
BY
SAHILU ASSEGID,
JANUARY/2009
231
13. ARE SAMPLE DATA REALLY ACCURATE?

Probability samples, while never perfectly
representative, are typically more representative
They need not however be representative in all
aspects
Representation in sampling is limited to
those characteristics that are relevant to
the substantive interests of the study.
BY
SAHILU ASSEGID,
JANUARY/2009
232
14. HOW LARGE A SAMPLE?
1. Too large:
- Unnecessary involvement of extra subjects
- High cost
- Time constraints
2. Too small:
- Not able to show biological/social effect
- False conclusion

BY
SAHILU ASSEGID,
JANUARY/2009
233
15. CONSIDERATIONS
The study design and the sampling method
p = the level of the magnitude
d = the magnitude of change you want to be
able to detect reliably (marginal error)
= how sure you want to be that a change of
the magnitude would have occurred by
chance (level of significance)
1- = how sure you want to be that you will
observe a change of the magnitude, if it did
in fact occur (power)
BY
SAHILU ASSEGID,
JANUARY/2009
234
END
OF

CHAPTER SEVEN
(SAMPLING AND SAMPLE SIZE)
BY
SAHILU ASSEGID,
JANUARY/2009
235
CHAPTER EIGHT

DATA
BY
SAHILU ASSEGID,
JANUARY/2009
236
OBJECTIVES

Define what data mean and describe their
selection
Describe how data are measured
Identify the sources of data
Describe various data collection methods and
tools and challenges involved in collecting data
BY
SAHILU ASSEGID,
JANUARY/2009
237
1. WHAT IS DATA?
A variable is measurable characteristic of a person,
object or phenomenon, which can take on different
values.
Measurement is the assignment of number or name or
events according to a set of rules.
Data can be defined as an aggregate of variables as a
result of measurement or counting.
Data are facts i.e. items of information that the
respondent believes to represent the truth for what was
asked.
BY
SAHILU ASSEGID,
JANUARY/2009
238
2. WHY DO WE COLLECT DATA?
Data are collected for the purpose of describing
individual units
The many descriptions are then aggregated and
manipulated in order to describe the whole sample and,
by extension, the target population.
The units of analysis could be:
- Individuals
- Families
- Clubs
- Organizations
- Villages, towns, cities etc.

BY
SAHILU ASSEGID,
JANUARY/2009
239
WHY DO WE COLLECT DATA?...

A given research may involve more than one
unit:
E.g. a household survey may include questions;
- About the household (family size, housing income...)
- About the individual members (age, sex...)
BY
SAHILU ASSEGID,
JANUARY/2009
240
3. WHAT DATA DO WE COLLECT?
How do we select the indicators?
- Measure the aspects
- Relevant to the objective
- Easy to interpret
- Feasible to collect
- Program planning (e.g. intervention)
BY
SAHILU ASSEGID,
JANUARY/2009
241
4. HOW DO WE COLLECT DATA?

A. What method of collection we use
B. What tools of collection we use
C. Who collects the data
D. How do we maintain the quality
BY
SAHILU ASSEGID,
JANUARY/2009
242
A. WHAT METHOD OF COLLECTION WE USE?
Quantitative method: Generally rely on
structured and standardized approaches;
- Census, DHS, BSS, health facility surveys are
some common standardized methods
Qualitative method: Generally rely on a variety
of semi-structured or open ended methods;
- FGD, In-depth interview are some common
methods
BY
SAHILU ASSEGID,
JANUARY/2009
243
B. WHAT TOOLS OF DATA COLLECTION
METHOD WE USE?

Sources of data:
1. Existing data (secondary data)
2. Sample survey (primary data)
BY
SAHILU ASSEGID,
JANUARY/2009
244
5. ANALYSIS OF EXISTING DATA

Scientific research should not be equated with
the collection and analysis of original data

Some research topics can be examined through
analysis of data already collected and compiled
(e.g. DHS data, hospital records etc.)
BY
SAHILU ASSEGID,
JANUARY/2009
245
ANALYSIS OF EXISTING DATA...
Advantages:
- Minimize cost
- Examination of trends over time
Disadvantages:
- Data is not always accessible
- Ethical issues concerning confidentiality may arise
- Information may be imprecise, limited to some
variables or incomplete
- Data collection may not be standardized
- Ecological fallacy

BY
SAHILU ASSEGID,
JANUARY/2009
246
6. AVAILABLE DATA
Mortality reports
Morbidity reports
Epidemic reports
Reports of laboratory utilization (including laboratory test
results)
Reports of individual case investigations
Social surveys (e.g. hospital admissions, disease
registers, and serologic surveys)
Demographic data

BY
SAHILU ASSEGID,
JANUARY/2009
247
7. SAMPLE SURVEY (PRIMARY DATA)
Primary data collection can be done using:
- Observation
- Interview
- Administering written questionnaires
- Focus group discussions
- Other data collection techniques

BY
SAHILU ASSEGID,
JANUARY/2009
248
A. OBSERVATION
Systematically selecting, watching and recording
behaviors and characteristics of living beings, objects
or phenomena.
Observation of human behaviors is a much used data
collection technique. It can be undertaken in two different
ways:
- Participant observation: the observer takes part in the
situation he/she observes
- Non-participant observation: the observer watches the
situation, openly or concealed, but does not participate
BY
SAHILU ASSEGID,
JANUARY/2009
249
B. INTERVIEW

An interview is a data collection technique that
involves oral questioning of respondents, either
individually or in group.
Answers to the questions posed during an
interview can be recorded by writing them down.
BY
SAHILU ASSEGID,
JANUARY/2009
250
C. QUESTIONNAIRE
Questionnaire development is a difficult process
including:
I-Defining concepts (validity and reliability)
II- Formulation of questions
III- Types of questions
IV- Selection and training of interviewers
V- Guideline for interview
VI- Pre-testing
VII- Conducting the actual interview
BY
SAHILU ASSEGID,
JANUARY/2009
251
QUESTIONNAIRE...
I. Conceptualization:
- Most often there are unspecified concepts rich in
meaning (e.g. marital satisfaction, socio economic status
etc.). Such concepts must be reduced to oversimplified.

- Operationalization is the process where by researchers
specify empirical observations that can be taken as
indicators of the attributes contained within a given
concept.
BY
SAHILU ASSEGID,
JANUARY/2009
252
QUESTIONNAIRE...
- Operationalization are ad hoc summaries of experience
and observation, they may not have real meaning:

- Begin with the general notion
- Enumerate all the different sub-meanings and opposite meanings
- Refer previous research on the topic and the conceptualization of
the variable

BY
SAHILU ASSEGID,
JANUARY/2009
253
QUESTIONNAIRE...
- To ensure that everyone understands actually what has
been measured (validity) and to ensure that there will
be consistency in the measurement (reliability), it is
necessary to clearly define the variables (and indicators
of variables).

=> Remember: that the ultimate goal is to collect data,
not to create data!

BY
SAHILU ASSEGID,
JANUARY/2009
254
QUESTIONNAIRE...
- Validity and reliability:
Validity- a measurement is said to be valid if it
measures what it is supposed to measure e.g. you might
ask respondent about his/her marital satisfaction. How
valid answer do you get?
Reliability of measurements- if repeated measurements
of a characteristic in the same individual under identical
conditions, produce similar results, we would say that the
measurement is reliable e.g. you might ask respondents
how many bottles of beer he/she drink. How reliable
answer do you get?
BY
SAHILU ASSEGID,
JANUARY/2009
255
QUESTIONNAIRE...
II. Formulation of questions:
- Clear and unambiguous (interviewer, respondent)
- Avoid double barreled questions e.g. USA should
abandon space war and spend the money on domestic
program, (agree, disagree).
- Ensure the respondents competency to answer the
question e.g. at what age did your child talked back to
you?
- Relevant questions e.g. for a married person asking a
question have you had sex in the last months?
BY
SAHILU ASSEGID,
JANUARY/2009
256
QUESTIONNAIRE...

- Short statements and order of questions
- Skip patterns
- Avoid negative questions e.g. USA should not avoid
nuclear weapons, dont you agree that condom should
be used in any sexual contact?
- Use standardized questionnaire
BY
SAHILU ASSEGID,
JANUARY/2009
257
QUESTIONNAIRE...
Standardized questionnaire:
- Standardized questionnaires have been used in
numerous settings and their continued use will allow
results to be compared.
- They will provide guidelines in developing questions
- But small changes in wording and the order of questions
greatly affect responses e.g. on HIV/AIDS the
organizations such as FHI, MEASURE EVALUATION,
UNAIDS, WHO etc. have develop standardized
questionnaires.

BY
SAHILU ASSEGID,
JANUARY/2009
258
QUESTIONNAIRE...
III. Types of questions:
- Asking questions
- Open ended
- Close ended (exhaustive, mutually exclusive,
questions with multiple answers)
- Reading statements
- Yes/no type
- Likert scale (strongly/agree/neutral/disagree/strongly
disagree format)
BY
SAHILU ASSEGID,
JANUARY/2009
259
QUESTIONNAIRE...
IV. Who collects data
- Actual interview should begin only after training
interviewers and supervisors
- You may be lucky to find intelligent, experienced
interviewers. But interviewers must be trained for each
new survey, because every survey differs from every
other one.
BY
SAHILU ASSEGID,
JANUARY/2009
260
QUESTIONNAIRE...
V. Guideline for interview:
- Interviewer training:
- General instructions
- Administrative issues
- Questionnaire specification
- Practice interviews in class
- Practice interviews in the field
- Supervision:
- Remain vigilant throughout the survey
- Spot-checks for questionnaires for completeness & accuracy
- Begin the work of coding open-ended responses

BY
SAHILU ASSEGID,
JANUARY/2009
261
QUESTIONNAIRE...
- Self administered questionnaire:
- A self administered questionnaire is a data collection tool
in which written questions are presented that are to be
answered by the respondents in a written form.

- A self administered questionnaire can be administered in
different ways e.g.:
- Through mailing to respondents who should mail their responses
back
BY
SAHILU ASSEGID,
JANUARY/2009
262
QUESTIONNAIRE...
- Gathering all or part of the respondents in one place at one time,
giving oral or written instructions, and letting the respondents fill
out the questionnaires, or
- Hand delivering questionnaires to the respondents and collecting
them back later
- The questions can be either open ended or closed (with pre
categorized answers)

BY
SAHILU ASSEGID,
JANUARY/2009
263
QUESTIONNAIRE...
VI. Conducting pre-tests:
- Pre-tests are initial tests of one or more aspects
of research
- Research tools (questions, interviews etc.)
- Sample design (source, frame etc.)
- Data collection procedures (feasibility, administrative
issues etc.)
BY
SAHILU ASSEGID,
JANUARY/2009
264
QUESTIONNAIRE...
VII. Conducting the actual interview:
=> What are some of the challenges in
collecting data?
- Source bias
- Selection of respondents
- Refusal bias
- Measurement bias
- Interviewer bias
BY
SAHILU ASSEGID,
JANUARY/2009
265
8. SCALES OF MEADUREMENT
Scales of measurement: Information is measured
by assigning numbers or names to items or
events according to a set of rules.
E.g. Measured by assigning numbers:
- Age
- Weight
- Distance
- Income
E.g. Measured by assigning names or events:
- Sex
- Color
- Outcomes of disease
BY
SAHILU ASSEGID,
JANUARY/2009
266
9. DEPENDENT AND INDEPENDENT
VARIABLES
The variable that is used to describe or measure
the problem under study is called dependent
variable. These are also called outcome
variables.
The variables that are used to describe or
measure the factors that are assumed to cause
or at least influence the problem are called the
independent variable. These are also known as
exposure variables.

BY
SAHILU ASSEGID,
JANUARY/2009
267
10. CONFOUNDING VARIABLE
A variable that is associated with the problem
and with possible cause of the problem is a
potential confounding variable.
Confounding is a mixing of effect of the
exposure under study on the disease with that of
third factor. This third factor must be associated
with the exposure and, independent of that
exposure, be a risk for the disease.
A confounding variable may either strengthen or
weaken the apparent relationship between the
problem and possible cause.
BY
SAHILU ASSEGID,
JANUARY/2009
268
CONFOUNDING VARIABLE...
Relationships:

Cause/exposure
(independent variable)
Effect/outcome
(dependent variable)
Other factors
(confounding variables)
BY
SAHILU ASSEGID,
JANUARY/2009
269
EXAMPLE
Mothers education Malnutrition
(Independent variable) (Dependent variable)

Family income
( Confounding variable)
BY
SAHILU ASSEGID,
JANUARY/2009
270
EXERCISE
1. A health researcher believes that in a certain region,
anemia, malaria and malnutrition are serious problems
among adult males and in particular adult females. He
wishes to study the prevalence of these diseases
among adults of various ages, occupational groups
and educational backgrounds to determine how
serious problems these diseases are for this
population.
- What are the dependent and independent variables in
the study?
-Which of these are categorical and which are numerical
variables?
BY
SAHILU ASSEGID,
JANUARY/2009
271
EXERCISE

2. A Zonal health manager receives a complaint that a
particular wereda, that one health center often runs out
of anti TB drugs. In a preliminary investigation, this
shortage of anti TB drugs is confirmed. The zonal
manager decides to investigate why this shortage of anti
TB drugs in the health center.
BY
SAHILU ASSEGID,
JANUARY/2009
272
EXERCISE...

- What is the dependent variable in the study?
- What would be a meaningful indicator for the dependent
variable?
- How would you define shortage of anti TB drugs?
- Can you think of some independent variable?
- Which variables are measurable as they are and which
ones need indicators?

BY
SAHILU ASSEGID,
JANUARY/2009
273
END

OF
DATA
BY
SAHILU ASSEGID,
JANUARY/2009
274
CHAPTER NINE

DATA PROCESSING

BY
SAHILU ASSEGID,
JANUARY/2009
275
OBJECTIVES
Understanding the process involved in:
- Data entry
- Data coding
- Data cleaning
- Examine data errors:
- Checking outliers
- Handling missing values
Perform exploratory data analysis:
- Gain familiarity with & understanding of the data
- Determine the approach to take and methods to use for
further statistical analysis

BY
SAHILU ASSEGID,
JANUARY/2009
276
WHAT IS DATA PROCESSING?

Data processing refers to:
- Data entry onto a computer
- Data coding
- Data checks and correction
The aim of this process is to produce a relatively
clean data set which may be imported into a
statistical package.
BY
SAHILU ASSEGID,
JANUARY/2009
277
1. DATA ENTRY
Data entry concerns the transfer of data from a
questionnaire to a computer file.
What the microscope is to biology and what the
telescope is to astronomy, the computer is to survey
research.
Computers are reliable, portable, computationally
powerful and easy to use.
However, they give you what you give them. Computer
scientists refer to this principle as GIGO (Garbage in
Garbage out).
There are many database computer programs that can
be used for data entry purposes.
BY
SAHILU ASSEGID,
JANUARY/2009
278
DATA ENTRY...
Who does data entry?
- Data often entered to a computer by a clerk who may not
be familiar with how the research was designed & how
the data was collected.
- To facilitate data entry and minimizing errors, the data
entry person should not make guesses, calculations,
coding etc.
- Data entry is quick and easy for the data entry person if
he/she simply type the information which is seen on the
answer sheet (i.e. direct data entry).

BY
SAHILU ASSEGID,
JANUARY/2009
279
DATA ENTRY...
- If the questionnaire is adequately designed, direct data
entry is possible if:
- Answers are put in separate column or separate answer sheet
- Documents are edited before data entry
- Closed ended questions are pre-coded etc.

- When working with computers, note to:
- Saving your work frequently
- Keep back-ups (more than one copy)
- Share time with other users etc.
BY
SAHILU ASSEGID,
JANUARY/2009
280
2. DATA CODING

- For computers to work their magic they must be
able to read your data. In general computers are
at best with numbers.
- Alphabetic codes and open ended responses
must be translated to numbers through the
process called coding.
BY
SAHILU ASSEGID,
JANUARY/2009
281
DATA CODING...
- Coding:
- is assigning a separate (non-overlapping)
numerical code for separate answers and
missing values
- E.g. instead of using Male and Female for the
variable sex, it can be indicated as:
1= Male, 2= Female

BY
SAHILU ASSEGID,
JANUARY/2009
282
DATA CODING...
Coding missing values:
- Missing values occur when measurements were
not taken, or respondents did not answers etc.
- In general, missing values should not be entered
as a blank because some statistical packages
interpret blanks as zeros
- Ideally, a code should be chosen to denote a
missing value (e.g. code 9 or 99 or 999 is
often used missing values).
BY
SAHILU ASSEGID,
JANUARY/2009
283
DATA CODING...
Who does the coding?
- The PI should coordinate the coding process and ideally
all the coding should be done by one person.
- Certainly, no more than three different people should be
involved in this process.
- If the work is done by more than one person, they
- Code book provides:
- A guide used in the coding process
- Locating the variables
- Loss of the code assignments of the values of the variable
- Decoding back to original variables when reorting
BY
SAHILU ASSEGID,
JANUARY/2009
284
3. DATA CLEANING
- Checking for errors, impossible or implausible values
and inconsistencies that may be due to coding or data
entry process.
- The aim of this process is therefore to produce a clean
set of data for statistical analysis.
- No matter how carefully the data have been entered,
some errors are inevitable.
BY
SAHILU ASSEGID,
JANUARY/2009
285
DATA CLEANING...
- Errors can result from:
- Incorrect reading
- Incorrect reporting
- Incorrect filling
- Incorrect sensing
- Incorrect coding
- Incorrect typing
- Incorrect etc.
BY
SAHILU ASSEGID,
JANUARY/2009
286
DATA CLEANING...

Some people think that data cleaning is data consuming,
and ignore doing it.
The failure to exercise care of this stage of the research
can cause difficulties at the analysis stage or even
subject to the danger of lying/cheating.

BY
SAHILU ASSEGID,
JANUARY/2009
287
4. EXPLORING DATA

- Exploratory data analysis involves:
- Examination of data errors
- Describing data using frequency distributions,
summary statistics and graphical techniques
- As data errors are often detected at this stage,
the process of exploratory data analysis and
data cleaning are typically iterative.

BY
SAHILU ASSEGID,
JANUARY/2009
288
EXPLORING DATA...

- The aim of this process is for the researcher to gain
familiarity with and understanding of the data, in order to
determine the approach to take and methods to use in
further statistical analyses.

- The methods used differ depending on the type of data,
it is helpful to know what type of data is involved.
BY
SAHILU ASSEGID,
JANUARY/2009
289
EXPLORING DATA...
=> Types of data:
- Information is measured by assigning numbers or
names to items (events) according to a set of rules.
- This results of different types of data (scales of
measurement) such as Nominal, Ordinal, Ranked,
Interval and Ratio.
- To minimize complexity from statistical analysis point of
view, it helpful to distinguish between two broad types of
data, namely:
A. Categorical and B. Numeric variables.
BY
SAHILU ASSEGID,
JANUARY/2009
290
EXPLORING DATA...
A. Categorical variables:
- Measured by assigning names to the items
- Categorical variables consists of naming observations
or classifying them into various mutually exclusive and
exhaustive categories.
E.g. the variable sex has two categories, Male and
Female, such variables with only two categories are
called binary or dichotomous.
The variable Blood group has more than two (four)
categories namely, A, B, AB and O.

BY
SAHILU ASSEGID,
JANUARY/2009
291
EXPLORING DATA...
B. Numeric variables:
- Measured by assigning numbers to items
E.g. - Number of AIDS cases reported during one year period
- Number of beds available in a particular hospital
- Weight, cholesterol level, time, temperature...

- At times, we might require lesser degree of detail than
afforded by numeric data, hence we occasionally
transform into categorical.

BY
SAHILU ASSEGID,
JANUARY/2009
292
EXPLORING DATA...

- Consider the variable age. Age might be transformed to
form a new variable , say age group, which consists of
categories.
- E.g. age group may consist of four categories <30, 30-
44, 45-60 and >60 where we are interested in the
number of subjects falling into each.

BY
SAHILU ASSEGID,
JANUARY/2009
293
EXPLORING DATA...
1. What is categorization?
- Categorization is the process of grouping numeric variables into a
series of non-overlapping, mutually exclusive intervals.
- Categorization is achieved by dividing the range of values, which the
continuous variable takes into intervals.
- By categorization, the information is summarized in a way making it
easier to determine the nature of the data
- However, there is a consequential loss of information
- Common questions arise during categorization are:
- Why categorization?
- How categorization?
- When categorization?
=> No straight forward answers are available for the above questions

BY
SAHILU ASSEGID,
JANUARY/2009
294
EXPLORING DATA...
2. Why categorization?
- No problem of extreme values
- Easier to handle and analyze
- The methods are conceptually simple
- The methods are with few assumptions
- Easier to explain and interpret
- Easier when adjusting confounding
- Easier to use standard software
BY
SAHILU ASSEGID,
JANUARY/2009
295
EXPLORING DATA...
3. How categorization?
- Using the knowledge of biological mechanism
- Equally spaced
- Equally sized (quintiles used in nutritional epidemiology)
- Combinations of the above two
- Use of same categories as previous studies
- Too few categories results much loss of information
- Too many categories results that the objective of
summarization not met

BY
SAHILU ASSEGID,
JANUARY/2009
296
EXPLORING DATA...
4. When categorization?
- Some authors (e.g. Kramer) suggest that categories should be
defined without looking the data at the design stage
- Others suggest that we should examine the nature of the data by
running frequency (which is what most of us do currently)
- But we should not (never) refine the category after analyzing the
data
- If the analysis is based refining categories, there will be a bias to
find significant association when in fact it is not true (this approach is
a data generated hypothesis).
BY
SAHILU ASSEGID,
JANUARY/2009
297
5. CHECKING THE ACCURACY OF THE DATA

- Verifying the accuracy of a sample of completed
data (e.g. questionnaire) should be checked at
data entry.

- If resources permit, the data should be entered
twice and the two entered data sets compared to
detect data entry errors.

BY
SAHILU ASSEGID,
JANUARY/2009
298
6. CHECKING OUTLIERS
- Data should be checked for values that are out of
range i.e. values that are implausible, impossible and/or
are inconsistent with other information gathered in the
survey interview.
- Outliers are values that are considerably larger or
smaller than other values of the variable
- For continuous data, values outside the expected
range would be a suspect. For e.g., I would be unlikely
for an adult mans height to be below 140 cm. or above
205 cm.

BY
SAHILU ASSEGID,
JANUARY/2009
299
CHECKING OUTLIERS...
- With categorical data, a value outside the response list
would constitute an error e.g. males reporting having
become pregnant could not be a valid response.
- Decision is needed as to whether the item is to be
changed, regarded as a missing or as an outlier.
BY
SAHILU ASSEGID,
JANUARY/2009
300
7. HANDLING MISSING VALUES
- The data should also checked for missing items i.e.,
items that should have been completed during the
course of the survey interview, but were not, either
because of respondent inability, refusal or interviewer
error.
- Decision will need to be made as to whether to insert
values for missing data items ( a process known as
imputation) or simply to disregard missing data items
during analysis.
- A decision to change or regard missing implausible
values should be consistent and documented.
BY
SAHILU ASSEGID,
JANUARY/2009
301
8. RUNNING FREQUENCY DISTRIBUTION

- When the data has been clean and checked, the next
explanatory step is to examine the distribution of each
variable.
- The term probability distribution or just distribution
refers to the way data are distributed, in order to draw
conclusions about a set of data.
BY
SAHILU ASSEGID,
JANUARY/2009
302
RUNNING FREQUENCY DISTRIBUTION...
- With numeric variables, the aim is to determine whether
or not normality may not be assumed.
- If not, we may consider transforming the variable,
or wish to categorize the variable for analysis (e.g.
age groups)
- With categorical variables, we obtain the frequency
distribution of each variable.
- If some categories have too few values, based on the
distribution, we may wish to amalgamate certain categories
for analysis.
BY
SAHILU ASSEGID,
JANUARY/2009
303
9. DISTRIBUTION OF CATEGORICAL
VARIABLES
- The probability distribution of a categorical variable tells
us with what probability the variable will take on the
different possible values. That is, it specifies all possible
outcomes of the categorical variable along with the
probability that each will occur.
- E.g. consider the value on the face showing up from
tossing a die. The probability distribution of this variable
is:
Value on face 1 2 3 4 5 6
Probability 1/6 1/6 1/6 1/6 1/6 1/6
=> Notice that the total probability is 1
BY
SAHILU ASSEGID,
JANUARY/2009
304
10. APPLICATION
- By calculating the probability of an event in
epidemiological studies, we can estimate:

1. Prevalence of certain diseases in a certain community
(e.g. TB, diabetes, heart disease)
2. The prevalence of certain characteristics (e.g. high
blood pressure, low birth weight) or
3. The prevalence of certain behavior (e.g. smoking, drug
use, condom use).
BY
SAHILU ASSEGID,
JANUARY/2009
305
10. DISTRIBUTION OF NUMERICAL VARIABLES
- Under different circumstances, the outcome of a random
variable may not be limited to categories or counts.

- E.g. suppose, X represents the continuous variable
height, rarely is an individual exactly equal to 170 cm.
tall, X can assume an infinite number of intermediate
values 170.1, 170.2, 170.3 etc.

- Because a continuous random variable X can take on
uncountably infinite number of values, the probability
associated with any particular one value is almost equal
to zero.

BY
SAHILU ASSEGID,
JANUARY/2009
306
DISTRIBUTION OF NUMERICAL VARIABLES...

- However, the probability that X will assume some value
in the interval enclosed by two ranges say X
1
and X
2
.

- As a continuous variable can take an infinite number of
values, it helps to visualize the probability distribution as
a curve and probabilities as area under the curve.

BY
SAHILU ASSEGID,
JANUARY/2009
307
11. CHARACTERISTICS OF A DISTRIBUTION
- Features commonly used to describe a distribution are
location, dispersion, modality and skewness.
- Location: tells us something about the average value of
the variable.
- Dispersion: tells us something about how spread out,
the values of the variable are.
- Modality: refers to the number of peaks in the
distribution.
- Skewness: refers to whether or not the distribution is
symmetric.
=> A distribution is said to be symmetric if it is
symmetrically distributed about the mode.

BY
SAHILU ASSEGID,
JANUARY/2009
308
12. THE NORMAL DISTRIBUTION
- The ND is used extensively in the analysis of continuous
variables and has an especially important role in
statistics.
- It has been found to be a good approximation for many
distributions that arise in practice.
- The ND is unimodal and symmetric.
- The ND is completely described by two parameters,
referred as the mean (read mu) and SD (read
sigma)
- The mean defines the location of the distribution can
be any number (negative, positive or zero)
BY
SAHILU ASSEGID,
JANUARY/2009
309
THE NORMAL DISTRIBUTION...
- The SD defines the dispersion of the distribution about
the mean, must be positive.
BY
SAHILU ASSEGID,
JANUARY/2009
310
13. SOME BASIC PROPERTIES
- For any ND, about 68% (most) of the
observation is contained one SD of the mean.

- Likewise about 95% (majority) of the probability
is contained within two SD, and

- 99% (almost all) within three SD of the mean.
BY
SAHILU ASSEGID,
JANUARY/2009
311
13. APPLICATION OF THE ND
- A woman gave birth 49 weeks after her husband left
military service. What is the chance of the husband
being a father?
- In many studies it is almost established the distribution of
gestational week is normal.
- Suppose we take a birth weight study at TAH of 10,000
newborns (Enqusellasie et. al.). The results showed that
gestational week was ND with mean about 40 weeks
and SD 2 weeks
- Applying the data, it is extremely unlikely for the husband
to be a father, because 49 weeks is more than 4 SD
above the mean.

BY
SAHILU ASSEGID,
JANUARY/2009
312
END

OF

DATA PROCESSING

BY
SAHILU ASSEGID,
JANUARY/2009

Research

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Research

Uploaded by

Copyright:

Available Formats

1

You might also like