Research Methods

(Objective, Methods, and Result sections with the context

of proposal or thesis writing)

Lemma Derseh
Major components of research proposal
Title
Acknowledgment, Table of content, List of tables, List of graphs,
and Acronym/abbreviation
Summary
Problem statement
 literature review
Objectives
Methods
Work plan
Cost of the project
References
Annex
Major sections of a scientific paper/Thesis

 Title Need to be accurate and informative

for effective indexing and searching
 Abstract
 Introduction
 Methods Main body, IMRaD
Introduction  Why?
 Results Methods  When, where, how, how much?
Results  What?
 Discussion Discussion (Conclusion)  So what?
 Acknowledgement
 Reference

NB Table of content, List of tables, List of graphs, and Acronym/Abbreviation would come next to Title. The
Objectives should also be stated after the Introduction section that contains subsections like Problem
statement, Literature review, Justification of the study , etc.
Research objectives and methods
(For both proposals and thesis)

However, presented with a context of a proposal writing

 Research Objectives
Objectives:
 The objectives should describe what you intend to achieve by doing
this piece of work.

 Or they are the steps you are going to take to test your hypotheses or
answer your research question.

 The objectives should be realistic, consistent and linked with

methods, timetable, and outcomes. They must also be SMART
(Specific, Measurable, Achievable, Realistic and Timelines)
 Research Objectives
 Research objectives should be written after the statement of the
problem is developed but before the methodology for the study is
planned

 This is because the objectives follow from the statement of the

problem and show the precise goal of the study, help guide the
development of the research methodology
 Types of Objectives- Based on the scope
 Commonly, research objectives are classified into general objectives
and specific objectives.

 The general objective is closely related to the statement of the

problem and is derived from it.

 Specific objectives are commonly considered as smaller portions of

the general objectives.

 Specific objectives should systematically address the various aspects

of the problem as defined under ‘Statement of the Problem’ and the
key factors that are assumed to influence or cause the problem.
 Types of Objectives- Based on the scope
Example
 The general objective of this study is to determine the prevalence
and associated factors of unwanted pregnancy among first year
undergraduate female students in Wollo University
Specific objective
 To determine the prevalence of unwanted pregnancy among first
year undergraduate female students in Wollo University
 To examine factors associated with unwanted pregnancy among
first year undergraduate female students in Wollo University
 Note that too many objectives should be avoided; even just one
clearly stated relevant objective for a study would be good enough

 If there is more than one objective the objectives can be presented in

the appropriate order of importance
Types of Objectives- Based on the framework of
stating objectives

Research objectives can take either of the following three

derivatives of a purpose/aim:

Positive sentence
 A clear statement of the specific objectives of the study, which
identifies the key study variables and their possible
interrelationships and the nature of the population of interest.

Example
 The objective of this study is to determine which operative
method of treating a fractured pelvis is associated with a lower
risk of postoperative infection
 Stating objectives
 When developing a specific objective, a researcher should use unbiased
verbs such as:
Determine
Investigate
Compare
Describe
Develop
 Discover
Explore
Test
Example: The objective of this study is to explore the effects of music
therapy on blood pressure in adult BP patients.

 The researcher should avoid biased verbs such as:

Demonstrate, prove, show, understand, appreciate etc.
 Types of Objectives cont...

Research Question
 It is used when a descriptive/exploratory research is required or
when little is known about the problem.

 These are the specific objectives stated in the form of a

questions

E.g. Does the administration of analgesic by nurses vs. by patients

themselves affect pain intensity during the first postoperative
recovery day in older adults?
 Types of Objectives cont...
Research Hypotheses
 It is used in an explanatory research i.e. they are predictive
statements as enough is known about the problem.

 There is a prediction of the answer to the research question

Examples
 Hypothesis: There will be a difference in knee injury occurrence
during ovulation.

 Null Hypothesis: There will be no difference in knee injury

occurrence during ovulation.
 Research methods

 Research methods is a way to systematically solve the research problem.

It may be understood as a science of studying how research is done
scientifically

 The guiding principle for writing the Methods section is that it should
contain sufficient information for the reader to determine whether the
method is sound

 However, the methods do not have to be explain in great detail, but

should justify its significance over other methods.
 Research Methods cont…
Components of research methodology
 Study setting and period
 Study design
 Target population
 Study population
 Sample size and sampling procedure
 Variables
 Operational definitions of key terms
 Tools/instrument.
 Data collection procedures
 Plan for analysis of data and interpretation of the results
 Ethical considerations
 Research Methods cont…
Study setting and period

 It is related to the study area and other characteristics which are

more related to the study; write as a brief background
(1-2 Paragraphs)

 Include map if possible

 Write the time to which the data belongs to or the time where the
data represents the characteristics studied
 Research Methods cont…

Study design
 A study design is a specific plan or protocol for conducting the
study, which allows the investigator to translate the conceptual
hypothesis into an operational one

 The study design to be used should be appropriate for achieving

the objective of the study and should also be clearly stated
 Research Methods cont…

There are two broad groups of research designs

 Experimental

 Non-experimental (observational)

 Experiment - Set of observations, conducted under controlled

circumstances, in which the scientist (researcher) manipulates the
condition to ascertain what effect such manipulation has on the
outcome

 All experimental studies are longitudinal since subjects are

followed from exposure of intervention to outcome
 Research Methods cont…
Observational designs
 Cross-sectional/Surveys: the whole population or a sample is
studied at point a point in time

 Cohort Studies: subjects are selected with reference to their

exposure status and followed up

 Case Control Studies: Subjects are selected in reference to

their disease status.
 Research Methods cont…
Population
 Broadly, population can be categorized as target population and
study population

 Target population is a collection of items that have something in

common for which we wish to draw conclusions at a particular
time.

 In other words, there could be some elements of the target

population which are systematically excluded from the study
population and hence might not be represented in the sample

 Study(sampled) Population: is a part of target population and

actually accessible and legitimate for data collection
 Research Methods cont…
Inclusion criteria
 Use inclusion criteria that would give you a larger study
population, but
 legitimate to answer the research question
Example: For a study to assess contraceptive utilization (family
planning): mothers aged 15-45 from a study area can be included
into the study
Exclusion criteria
Think critically to find people who fulfill the inclusion criteria but
are not eligible
Example: Women who are infertile can be excluded from the study

NB: non-respondents cannot be part of exclusion criteria

 Sample size and sampling procedure
 Whether the study design to be employed is quantitative or
qualitative, a standard sampling technique is important in most
situations (i.e. except census)

 However, the sampling techniques in qualitative and quantitative

studies are different. This is because, in qualitative study, the aim is
not generalization of the population parameter (rather it is a
discovery), whereas this is true in quantitative study.

 As a result, in qualitative study we employ non-probability

sampling techniques (convenience, quota, judgment, snowball
sampling etc) so that we can select simply individuals using the
above techniques who are rich in information that we want
 Sample size and sampling … cont’d

 On contrary to this, in quantitative study, we have different

probability sampling techniques (simple random sampling,
stratified sampling, cluster sampling, systematic sampling, multi-
stage sampling etc) to get a representative sample.

 This is because it is only probability sampling technique that would

enable us to predict population parameters with certain level of
confidence (e.g. 95%) and level of precision (e.g. 100 gram for
mean weight estimation of newborns)
 Sample size and sampling … cont’d

 Therefore, in quantitative study, to take a sound generalization

about a population parameter, we must have a representative
sample.

 In turn, a representative sample could be secured if two things

have been fulfilled:
We must have a minimum but adequate sample size drawn
from the population

The sampling technique employed must be in line with the

distribution of the characteristic we are to study
Sample size and sampling … cont’d
 In quantitative study, the larger the sample size, the more will be
the likelihood to represent the population. Of course, after a certain
limit the gain in accuracy will diminish as the sample size
increases

 The adequate minimum sample size can be determined using:

 Thumb rule,
 Precision approach (single population proportion or mean
approach)
 Power approach (two population proportion or mean difference
approach)

 In any case, consider the non-response rate (contingency), design

effect or the number of independent variables.
Sample size and sampling … cont’d

 For qualitative study, it is very difficult to set the sample size from the
outset.

 Of course, with a thorough look at of the resource and time one has,
and some reading of similar studies, one can give a reasonable
indication.

 However, it is good to leave it for saturation of the study

 Sample size and sampling … cont’d

Rules of thumb
1. If the population size is small (N ‹ 100), there is little point in
sampling. Survey the entire population.
2. If the population size is around 500, 50% should be sampled.
3. If the population size is around 1500, 20% should be
sampled.
4. Beyond a certain point (N = 5000), the population size is
almost irrelevant and a sample size of 400 may be adequate.
5. Statistician – maximalist – at least 500

26
 Sample size and sampling … cont’d
Precision approach
Hence the absolute precision denoted by d is given as (N > 10,000)

d  z  s.e Where s.e is the standard error of the

2 estimator of the parameter of interest.

For mean estimation For proportion estimation

d = (Zα/2) Se d = (Zα/2) Se
d= (Zα/2) σ/√n d= (Zα/2) √p(1-p)/n
√n = (Zα/2) σ/d √n = (Zα/2) √p(1-p)/d
n = (Zα/2)2 σ2/d2 n = (Zα/2)2 p(1-p)/d2
 Precision approach (single population)
 The previous sample size formulas assumes that you are using
random sampling and the sampled population is large enough
that you do not need the finite population correction

 Finite population correction adjusts the sample size for

situations in which the target population is small enough that
the sample is more than 5% of the target population (or if N <
10,000). The formula is:
Nz 2 pq n0 n0
n 2  
d ( N  1)  z 2 pq N  1 n0 n
 1 0
N N N
 If the population size is known, we can use directly this
formula to determine the sample size; and it has more
significance when the population size is finite

 When the Epi Info StatCalc have a field to include the size of
the target population so that this may be taken into account
when calculating the sample size.
 Precision approach (Single population)
• We use single population formula if the aim is only to conduct
descriptive study like proportion estimation, mean estimation, etc.

• Most maser students have two specific objectives; the first is usually
about prevalence and the second is usually about associated factors.
Therefore, the first (the prevalence) study will require single population
formula

• In proportion estimation, p can be obtained from previous studies or

conduct pilot study or take 50%,

• In mean estimation, use either previous study, conduct pilot study to

get S, or if the target population is thought to be normally distributed,
you can use the fact that the range (highest to lowest value) is
approximately equal to 6 standard deviations, thus S = range/6 — but
you would still need to know something (maximum and minimum
whose z-value is large like 3 and -3 respectively) about the population
values.
Precision approach (Single proportion)
• Example: If a researcher is interested to estimate the proportion of
patients who are satisfied by the health service they got in a certain
institution, then he/she can use a literature containing the result
indicated in the next slide to determine the sample size. (Assume the
researcher needs to have a precision of 5%)

• This gives us a proportion = p = 303/(303+171) = 63.9%

 Thus using a single population formula,

n = (Zα/2)2 p(1-p)/d2 = 1.96 2 * (0.639)*(0.361)/(0.005)2 = 354

Suppose that you planed to use two stage sampling and expect that
about 10% of participants would be non-respondents. Determine the
sample size. Which one should be considered first? The design effect or
the contingency?
 How to use results of other researches for sample size determination

Unsatisfied
Characteristics Yes No Crude OR Adjusted OR P- Value
(95% CI) (95% CI)
Cost of treatment 0.522
Very cheap 59 69 1.0 1.0
Cheap 70 42 1.95 (1.16, 3.27) 1.36 (0.66 2.81) 0.400
Moderate 30 8 4.39(1.87, 10.30) 2.12 (0.64, 7.05) 0.220
Expensive 97 41 2.77(1.67, 4.58) 1.54(0.73, 3.26) 0.255
Highly expensive 47 11 5.00(2.38, 10.50) 2.35 (0.74, 7.53) 0.150
Residence
Urban 205 154 1.0 1.0
Rural 98 17 4.33 (2.48, 7.55) 2.71 (1.19, 6.16) 0.017*
Extra job < 0.001**
Goven’t worker 19 13 1.0 1.0
Need partime 210 58 2.48 (1.16, 5.31) 3.18(1.09, 9.06) 0.034
Partimer 60 96 0.43 (0.20, 0.93) 1.04(0.35, 3.15) 0.94
Has his own firm 14 4 2.40 (0.64, 8.93) 6.26 (3.21, 32.27) 0.028
Diagnosis Type
Complex 108 154 1.0 1.0
Simple 195 17 16.36 (9.41, 28.44) 13.55 (6.96, 26.37) <0.001**

Total 303 171

 = 0.05, * shows significant, * *shows highly significant, Underlined figures are overall p-values
Power approach (analytic or two population studies)
For proportion estimation use the formula:
2
z (r  1) p q  z1  rp1 q1  p 2 q 2 
  
n1   2 This is according to Fleiss
r ( p1  p 2 ) 2

Z  /2  Z1   p q(r  1)
2

Or n1  , This is according to Kelsey

r ( p1  p 2 ) 2

Variable Cohort Case-control

n1 number of exposed number of cases n2  r  n1
n2 number of unexposed number of controls
zα/2 z-score for two-tailed test based on α level
z1-β z-score for one-tailed test based on β level
p2 (OR ) p1  p 2 (RR)
r unexposed : exposed controls : cases p1 
1  [p2 (OR  1)]
p1 proportion of diseased proportion of exposure
among exposed among cases
q1 1 – p1
p2 proportion of diseased proportion of exposure p1  r  p 2
among unexposed among controls p q  1 p
q2 1 – p2 r 1
Assumptions: The sample is independent and randomly selected.
 Power approach ….
• The inputs to calculate sample size are confidence level,
power, p1, p2, or measure of association (OR or RR)
• Measures of associations are based on previous studies or
reports and should reflect the minimum effect that the
investigator considers worth detecting.
• Two examples are the relative risk and the odds ratio.

• If n1 = n2 then the first (Fleiss) formula simplifies to:

2
z 2 p q  z1  p1 q1  p 2 q 2 
  
n1   2
( p1  p 2 ) 2

Note that the result found in Epi Info of versions up to 3.4.5 which are
reported as Fleiss has some difference from hand calculations as it takes
continuity correction
 Power approach ….
• When the input to calculate sample size is provided as an odds
ratio rather than the proportion of diseased among exposed, the
proportion of diseased among exposed is calculated as (if the
proportion of diseased among non-exposed is available)
p 2 (OR)
p1 
1  [ p 2 (OR  1)]

• For a cross-sectional study where you want to calculate a

prevalence odds ratio, this is the formula one would use.

• However, when the input is provided as a risk ratio or relative risk

(RR) rather than the proportion of diseased among exposed, the
proportion of diseased among exposed is calculated as:
p1  p 2 (RR)
• This is also the formula used for a cross-sectional study where you
want to calculate a prevalence ratio.
 Power approach
For mean estimation
• If the sample size in each group to be taken is different, we use
the following formula for mean estimation
( z / 2  z  ) 2  2 (1  1 / k )
n1  kn2 and n2 
2
1 ( z / 2  z  ) 2  2 (k  1)
Or, n2  n1 and n1 
k 2

• Where k is the ratio of n1 to n2; and ε is a clinically meaningful mean

difference between two groups (μ1-μ2); σ2 is the variance of either of the two
groups which is assumed to be equal and if it is not known we take the pooled
variance of the two samples as:
2 ni
1
s 
2

n1  n2  2 i 1
 ij
( x
j 1
 x i .) 2
 Power approach
Example
A study is being planned to test whether a dietary supplement for pregnant
women will bring change to the birth weight of babies. One group of women will
receive the new supplement and the other group which has a double size of the
first group will receive the usual nutritional consultation. From a pilot study, the
standard deviation in birth weight is estimated as 500g and is assumed to be the
same in both groups. The hypothesis of no difference is to be tested at 5% level
of significance. It is desired to have 80% power of detecting a increase of 100g
n1 = 1/2 n2
( z / 2  z  ) 2  2 (1  1 / k )
n1  kn2 and n2 
2
n2 = (1.96 +0.842)25002(1+1/1/2)/ 1002 = 588

Thus a sample of 294 and 588 will be taken from group one and two, respectively
 Example for power approach
Example: If researchers wanted to determine the sample size that would help
them to examine the risk factors of patient dissatisfaction in another study area,
then he/she can use the research output shown before.

 If the planned design is either cross-sectional or cohort study, then we can

determine different sample sizes by using the three statistically significantly
associated variables which are residence, extra job, and diagnosis type.

 For factors having only two categories, we take the reference categories ( in this
case the first category) as unexposed and the last categories as exposed groups. If
a variable has more than two categories ( e.g. extra job), we take as exposed group
a category having a p-value of less than 5% and has a proportion closest to the
proportion of the unexposed category ( as it gives us a largest sample size)

 To find the proportion in exposed and unexposed groups, we will look into the
counts horizontally. For instance, for diagnosis type the proportion for diseased
among exposed and non exposed groups will be 195/(195+17) and 108/(108+154),
respectively

 The final sample size will be the largest of all the three estimates (provided that
there is no other objective). But if there is ,for example, a prevalence objective we
take the largest of the four sample sizes.
 Power approach
Variable Diseased/ Total Proportion Sample size
dissatisfaction (for two groups)
Yes NO
Residence
Urban 205 154 359 0.57 94
Rural 98 17 115 0.85
Extra job
Gov’t worker 19 13 32 0.59 206
Has his firm 14 4 18 0.78
Diagnosis type
Complex 108 154 262 0.41 32
Simple 195 17 212 0.92
Total 303 171

The maximum sample size will be 206 i.e.103 in each group. This is true if our objective is only
factor study; otherwise we take 354 if there is also prevalence objective. Design effect and then
contingency should also be considered accordingly
 Power approach
For researchers who are to conduct case control study, we can calculate
percent exposed in diseased and non-diseased groups vertically as follows
Variable Diseased/ dissatisfaction Sample size
Yes (diseased) NO(not diseased) (for two
groups)
Residence
Urban 205 154 122
Rural 98 (32%) 17 (10%)
Extra job
Gov’t worker 19 13 1306
Has his firm 14 (5%) 4 (2%)
Diagnosis type
Complex 108 154 30
Simple 195 (64%) 17 (10%)
Total 303 171

The maximum sample size, 1306, will be taken (653 in each group). On top of
that, design effect and contingencies should be considered accordingly.
 Sample size and sampling …cont’d
 However, size by itself can not be a guarantee to get a
representative sample. Rather, for example, if the characteristic we
are going to study is uniformly distributed throughout the larger
population, we can use simple random sampling or cluster
sampling provided that there is a means to clustering the
population

 Otherwise, if there is another variable by which the distribution of

the characteristic could importantly vary, we can stratify the
population by that variable, and in that case stratified sampling
would be appropriate
 Sample size and … cont’d
 Sometimes the population that we want to study could be distributed in
a very large geographic area or large population. In that case, it will be
costly covering all the study area.

 Therefore, multistage sampling could be more appropriate in such

situations
Example
To study the prevalence of common cold in Gondar town during the last
1 month period, we may assume that the distribution of common cold is
uniform in all the 10 kebeles of the city. Therefore, we can select 3
kebeles randomly from which only some of the households and then
some of the household members will be included in the sample.

Questions: In this example, we used a three stage sampling. Please,

identify the three stages .The study population for this example will be
all individuals in Gondar town, but not only all individuals in the 3
Kebeles! Why? What do you think is the target population?
Operational definition of variables
 An operational definition describes or defines a variable in terms of the
operations or techniques used to make it happen or measure it.

 When quantitative researchers describe the variables in the study, they

specify what they mean by demonstrating how they measured the
variable

 Demographic variables like age, sex & ethnicity are usually measured
simply by asking the participants to choose the appropriate category from
the list

 However, abstract concepts like knowledge about how to prevent

HIV/AIDS infection need to be defined operationally. For example, the
investigator may provide item questions and change the responses into a
score that could show the level of knowledge in HIV/AIDS.
 Research instrument/Tools.
 One of the most important components of a research design is a
set of research instruments because they are the tools we use to
collect the data

 Some of them are questionnaires, measuring instruments (e.g

anthropometric measuring instruments like meter tape, balance
scale), etc for quantitative study, and

 interview guideline (check list), tape recorder , etc. for

qualitative study
 Research Tools cont…

 Questionnaires are the most frequently used data collection tools

in quantitative research.

 If there are standard questionnaires already designed by other

organizations, use it as one should not re-invent the wheel.

 Otherwise, developing a valid and reliable questionnaire is a must

to reduce measurement error which is the "discrepancy between
respondents' actual attributes and their survey responses
 Pre-Test and Pilot Study
 To find a valid and reliable questionnaire, pre-test and pilot studies are
important. However, we can also employ them (especially pilot study)
for other additional purposes.

 Generally, a pretest is used on isolated problems of the research design,

mainly to evaluate the general approachability and feasibility of the
questionnaire, and identify the specific problems of communication
between the interviewer and the respondent in terms of specific questions
or items of information sought

 Whereas, pilot study is mainly with the objective of checking

organization and arrangement of the whole research process
(administrative and organizational arrangements in general, supplies,
equipments, field work operation, questionnaires, sample size
determination and data processing).Thus, it is a full scale dress rehearsal
of the survey
 Pre-Test and Pilot Study cont…

With the pilot study, we can benefit:

 The adequacy of the sampling frame from which it is proposed

to select the samples
 The estimates needed to estimate the sample size i.e. SD and P
 The none–response rate
 Adequacy of the questionnaire
 Making a sensible choice from the alternative methods of
collecting the data
 To estimate probable cost needed to conduct the research, etc
 Data Collection procedures
Write clearly:
 The data collection technique that you plan to use (face to face
interview, self-administered questionnaire, telephone, post, web, in-
depth interview, FGD, observation, …….)

 Who will be the data collectors and supervisors ( educational

background, sex, experience, from where they are i.e out of the
study area or not, etc.)
 The data collection tools (questionnaire, interview guide, tape
recorder, measuring instruments, etc. )
 The time and setting/place of data collection
 What data quality controlling mechanism you plan to use (pre-test,
pilot, training, field exercise, supervision and spot checking,
editing, verification, etc)
Planning Data Analysis and Interpretation
of the Results
A plan for data analysis and interpretation of the results will
be prepared because it helps the researcher to make sure that:

 The information needed will be collected

 All the information collected will be analyzed

 The data will be collected in such a way that it will be analyzed

based on what was planned
 Planning Data Analysis and …. cont’d
In the plan for data analysis and interpretation, write clearly
what will be done in the following three stages:
i. Data processing stage
 Write that you will edit, verify, and clean the data and code it as necessary
before analysis. Write also that you will merge or re-categorize the levels of
variables in such a way that they will be suitable for analysis
ii. Analysis stage
 Write clearly that you will present descriptive findings based on the type of
variables (e.g. for categorical variables you will use chards, graphs, tables
etc)
 Write also that you will present analytical results using a certain standard
statistical method (e.g. linear regression, ANOVA, logistic regression,
survival analysis, etc) and indicate the corresponding measures of
association (e. g. OR, RR, HR, coefficient or β, etc)
 Indicate how you are going to select variables from univariable to
multivariable analysis. Fix also the level of significance , α , that you plan to
use
 Plan for data analysis and interpretation

iii. Interpretation
 Write clearly that you will review the objectives which are often
used as a basis for organizing the report on the analysis of data
and interpretation of the results

 Indicate also that you will discuss your findings with known
theories on the topic and other recent research findings done

 You should also write that you will conclude and recommend to
the relevant body or organization based on the research findings
and discussions.
Writing the Result section
(For Thesis writing only)
 Result
 Primary functions:
 Objectively describe the findings (data interpretation and
conclusions belong in Discussion)
 Direct readers to figures or tables that present the supporting
data.

 In general results should be presented starting from descriptive to

inferential

 In the beginning, give a general descriptive analysis including the

number of cases, mean and standard deviation for each
dependent variable given each level of each of the independent
variables in your study.
 Result … cont’d
 Statements made in the text must be supported by the results
contained in figures and tables.

 Figures should “stand-alone”- a reader should be able to look at a

figure and it’s legend and be able to understand the results
(without reference to the text).

 Give each figure (at the bottom) and table (at the top) a simple,
self-explanatory title in the form of a sentence. (Most of the
content of paper should be evident from reading these titles
alone).

 The order of figures and tables should follow the discussion

themes (based on their importance)
 Result cont…

 All results should be presented, including those that are

unexpected or insignificant or

 The presence or absence of a significant influence or relationship

between or among the research variables would be also presented

 The results of statistical tests can be resented in parentheses

following a verbal description

 Summarize all the findings in a paragraph or two, and with it, lead
the readers into the discussion part.