See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/314131855

Basic Approach to Data Analysis and Writing of Results and Discussion

Sections

Article  in  MAMC Journal of Medical Sciences · January 2017

DOI: 10.4103/mamcjms.mamcjms_50_16

CITATION READS

1 7,022

3 authors:

Satyanarayana Labani Komal Wadhwa

Indian Council of Medical Research, Dept of Health Research PVS Hospital
142 PUBLICATIONS   1,646 CITATIONS    3 PUBLICATIONS   1 CITATION   

SEE PROFILE SEE PROFILE

Smita Asthana
Indian Council of Medical Research
83 PUBLICATIONS   568 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Strengthening the role of ASHAs in addressing adolescents health issues published in Indian journal of Youth and Adolescent Health. View project

Remodelling of the existing data View project

All content following this page was uploaded by Satyanarayana Labani on 04 March 2017.

The user has requested enhancement of the downloaded file.

 ISSN 2394-7438

Volume 3 | Issue 1 | January-April 2017

MAMC
Journal of

Medical Sciences
Journal of MAMC Journal of Medical Sciences • Volume 3 • Issue 1 • January-April 2017 • Pages 1-***

Official Publication
Maulana Azad Medical College
 MAMC Journal of Medical Sciences

Review Article

Basic Approach to Data Analysis and Writing of Results and

Discussion Sections
Satyanarayana Labani, Komal Wadhwa, Smita Asthana
Division of Epidemiology and Biostatistics, National Institute of Cancer Prevention Research, Indian Council of Medical Research,
Department of Health Research, MOHFW, Noida, Uttar Pradesh, India

Abstract
A research paper or thesis writing is considered hard and very difficult process of intense concentration and brain work. Readers generally are
first interested in new results of the paper. Writing results section contains new results from research investigation but is difficult in
comparison to writing methods section as the latter section is already written at the proposal writing stage and requires only language change.
Results section is heart of the paper and its completion with methods section already written; implies more than 50% of paper or thesis writing
work and it become 70% paper writing work with writing of discussion section. Results writing section should be organized into different
segments of text and visuals such as tables, figures, algorithms, etc. In order to start writing results section, we make a beginning with data
analysis and its presentation of key findings as summarized results to yield an answer to the research question that study attempted. Answers to
the questions and interpretations are presented in the discussion section. Data analysis is primarily linked with writing text part of the results
and discussion of results. This is a desired sequence to work with in paper writing. The attempt of working in such a sequence provides a
convenient approach to young researchers or post graduate/under graduate medical students to gain confidence in writing a research paper or
thesis or a research report. While basic knowledge of study design and analysis is needed, the involvement of a qualified bio-statistician is
recommended in various stages up to publication. In this communication, we describe the basic approach of data analysis required for initiating
writing results and discussion while quoting the required rules for the purpose.

Keywords: basic data analysis, IMRAD, write discussion, write results

INTRODUCTION primary work in this direction to prepare required tables

and figures which serve as ingredients for paper writing.
Introduction, methods, results, and discussion (IMRAD)
Research manuscripts are presented in IMRAD format, yet
are sections of the format generally adopted in writing
the order of writing work generally begins with results
an original research article/paper.[1] Bio-medical post
section. This section provides new information related to
graduate (PG) or doctoral thesis[2-4] and other scientific
answering the research question. The desired answers to the
reports are also basically and broadly written in IMRAD
research questions and interpretations of results are
format with inclusion of some additional sections. There
are number of articles available in literature[5-19] on writing
of scientific manuscripts. A scientific manuscript tells the Address for correspondence: Dr. Satyanarayana Labani, MSc, PhD, FSMS,
story of the study from framing of research question, data Scientist G and Head, Division of Epidemiology and Biostatistics, National
collection, statistical analysis, and discussion of results. Institute of Cancer Prevention Research, Indian Council of Medical Research,
Results are a text-based presentation of key findings with Department of Health Research, MOHFW, I-7, Sector-39, Noida 201301, Uttar
Pradesh, India.
reference to tables and figures. Thus, data analysis is a E-mail: satyanarayanalabani@yahoo.com

Access this article online This is an open access article distributed under the terms of the Creative Commons
Quick Response Code: Attribution-NonCommercial-ShareAlike 3.0 License, which allows others to remix,
Website: tweak, and build upon the work noncommercially, as long as the author is
www.mamcjms.in credited and the new creations are licensed under the identical terms.

For reprints contact: reprints@medknow.com

DOI: How to cite this article: Labani S, Wadhwa K, Asthana S. Basic

10.4103/mamcjms.mamcjms_50_16 Approach to Data Analysis and Writing of Results and Discussion
Sections. MAMC J Med Sci 2017;3:6-15.

6 © 2017 MAMC Journal of Medical Sciences | Published by Wolters Kluwer - Medknow

 Labani, et al.: Data analysis and writing results and discussion

provided in the discussion section. Results section attempts
to present the results of the analyses obtained on data
collected on the medical investigation of interest after
application of statistical analysis methods of descriptive
and inferential types. Thus the data analysis and writing
of results and discussion sections are interrelated. There
is much literature and published guidelines available
exclusively for writing results and discussion sections of
a paper[20-26] but the presentation of their interrelationship
with data analysis is attempted in this comprehension. Such
a presentation along with gist of rules of writing is likely to
provide a brief in sight of writing results and discussion
sections of a paper to young researchers, under graduate
(UG), and PG bio-medical students. We begin with
some essential basics of handing data by summarization
illustrations that help identifying the direction for
generating required material for results section. In
addition, we explain the rules as observed from
literature for writing results and discussing the results.
DATA
use. Sometimes information on a particular variable may
be missing. The missing information of an item may be
coded as “9” or “99” instead of leaving cell blank. Excel
spreadsheet with information of each individual in a row
may be formed as a data file for several individuals data.
Such data files can be transferred to or made on statistical
software such as SPSS or SAS or STATA or EPI. Statistical
packages are widely available for tasks such as one-way
tabulation or two-way tabulations. For data summarization,
apart from tabular method, the other important methods
are graphical and numerical.
By using data, tabular presentations may be in one dimension
or two dimensions. For example, a one-dimensional or one-
way classification of information on menstrual history of
Statistical Methods of Analysis
Statistical Methods
of Analysis

Data analysis has two broad approaches − obtaining

descriptive features as estimates and performing tests of
hypothesis to obtain inference on the data collected Descriptive Methods of
[Figure 1]. Data can be classified as either numerical or Methods – Statistical
proportion, Inference-
categorical. Numerical or quantitative data can be further mean and basis of
SD theoretical
classified as discrete or continuous and categorical data can distributions
be nominal or ordinal in nature [Figure 2]. For example,
individual’s height and weight, etc., are continuous.
Categorical data are further classified as nominal or
ordinal. In nominal, categories have no ordering (e.g., sex: Hypothesis
Estimation-
male/female). In ordinal, categories are ordered (e.g., grade: 95%confidence Testing-
A/B/C/D, rating: high/medium/low). Any measurement of intervals chisquare, t,
F tests (p values)
quantitative type or categories of a quality observed from
individual’s measurements are called variables. Quantitative Figure 1: Showing approach of statistical methods of analysis
variables as against qualitative categories are those which are
usually either measurements or counts in a discrete or
Table 1: Specimen data form for cervical screening
continuous form. Continuous variable assumes
proforma
uninterrupted range of values and on the other hand,
counts are positive whole numbers. Quantitative variable
1 Name of Woman
such as age is numerical information and measured in
years or months. Education and marital status will have 2 Woman ID □□
Age
non-numeric categories of qualities and whose categories 3 □□
Education:
are coded by assigning numerical numbers for the purpose 4 1. Illiterate 2. Primary 3. High School
5. Graduate and above 6. Others
4. Intermediate □□
of counting category responses. Quantities also can be Marital Status:
converted sometimes into a qualitative variable by 5 1. Never Married 2. Married
Separated/Divorced
3. Widowed 4. □□
different groups of categories such as high and low blood Religion:
pressure for a quantitative variable. 6 1. Hindu 2. Muslim 3. Christian 4. Sikh 5. Others_____ □
Menstrual History:

Data are summarized in the form of tables, graphs or

7 1. Regular 2. Irregular 3.Menopause □
Clinical Sign:
numbers. The tables can be one dimension or two 8 1. Normal Cervix 2. Unhealthy Cervix □
3. Cervical Erosion 4. Others____
dimensions depending on the type of objective of analysis
VIA Result:
on a set of observed variables. Data are basically collected on 9 1. Positive 2. Negative 3. Not Done □
a tool. Data collection tools include questionnaire, schedule VILI Result:
10 1. Positive 2. Negative 3. Not Done □
or a proforma [Table 1]. Data from a questionnaire can be Cytology Result:
entered in a computer on an Excel spreadsheet, for example, 11 1. Positive 2. Negative 3. Not Done □
as un-coded [Table 2] and coded [Table 3] form for further

MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017 7

 Labani, et al.: Data analysis and writing results and discussion

woman may have categories such as regular, irregular, and
menopause may be performed by counting the codes
corresponding to each response. Then codes are decoded
and present the results. Similarly, a two-dimensional or a
two-way table would have various headings of categories in
horizontal rows and vertical columns. Such a cross tabulation
is done when the interest is in assessing relationship between
two variables. The values or numbers related to a combination
of row and column is called a cell in a two-way table. If there
are only two rows and two columns in a two-way table, such a
table is called 2 × 2 table.

Variable
Consider the hypothetical data on cervical cancer screening
shown in Table 2 as it is entered in the Excel sheet. The
coding for different variables is performed to make statistical
Qualitative Quantitative analyses easier. Post coding for the above hypothetical
or categorical measurement
example is illustrated in Table 3.

Nominal Ordinal Discrete Continuous DATA ANALYSIS

(not ordered) (ordered) (count data) (real-valued)
e.g. gender, e.g. degree of e.g. number of e.g. weight, blood There are different methods of summarizing data depending
ethnicity, blood pain, severity of cigarettes pressure,
type accident, tumor smoked/day, cholesterol, fasting on the type of data and the dimension (univariate, paired, etc.)
grade children in family blood sugar
of the variable. Univariate is one measurement or variable
Figure 2: Showing type of variables in data analysis taken at a time for analysis, for example, height.

Table 2: Hypothetical sample data of 6 women on cervical screening

W_ID Age Edu Mart_Status Religion Mens_His Cli_Sign VIA_Res VILI_Res Cyto_Res
13-001 35 Illiterate Married Hindu Regular Normal cervix Negative Positive Negative
13-002 50 Illiterate Married Hindu Menopause Unhealthy cervix Negative Negative Negative
13-003 40 Primary Married Hindu Irregular Cervical erosion Negative Negative Negative
13-004 31 Illiterate Married Hindu Regular Others Negative Negative Negative
13-005 46 Primary Married Hindu Menopause Unhealthy cervix Positive Positive Negative
13-006 35 Illiterate Married Hindu Regular Normal cervix Negative Negative Positive
W_ID = women ID, Edu = education, Mart_Status = marital status, Reli = religion, Men_His = menstrual history, Cli_Sign = clinical sign, VIA_Res = VIA
result, VILI_Res = VILI result, Cyto_Res = cytology result.

Table 3: Post coding of the hypothetical data of 20 women on cervical screening

W_ID Age Edu Mart_Status Reli Mens_His Cli_Sign VIA_Res VILI_Res Cyto_Res
13-001 35 1 2 1 1 1 2 1 2
13-002 50 1 2 1 3 2 2 2 2
13-003 40 2 2 1 2 3 2 2 2
13-004 31 1 2 1 1 4 2 2 2
13-005 46 2 2 1 3 2 1 1 2
13-006 35 1 2 1 1 1 2 2 1
13-007 42 1 2 1 2 3 2 2 2
13-008 40 1 2 1 2 2 2 2 2
13-009 50 1 2 1 3 2 1 1 1
13-010 30 1 2 1 2 1 2 2 2
13-011 40 3 3 2 2 1 2 2 2
13-012 30 1 2 1 1 1 2 2 2
13-013 45 1 2 1 2 2 2 2 2
13-014 50 3 2 1 3 3 2 2 1
13-015 50 1 3 1 3 2 2 2 2
13-016 55 2 2 1 3 2 2 2 2
13-017 30 4 2 1 1 1 2 2 2
13-018 49 4 2 1 2 1 1 1 2
13-019 40 1 2 2 1 2 2 2 2
13-020 45 1 2 1 2 3 2 2 2
W_ID = women ID, Edu = education, Mart_Status = marital status, Reli = religion, Men_His = menstrual history, Cli_Sign = clinical sign, VIA_Res = VIA
result, VILI_Res = VILI result, Cyto_Res = cytology result.

8 MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017

 Labani, et al.: Data analysis and writing results and discussion

Multivariables or multiple measurements are taken two or
more at a time, for example, height, weight, sex. A
Summarization can be in the form of graphs, as histograms
which is a visual summary of the sample distribution,
quantile–quantile plot which compares a sample to a
known distribution, or scatter plot which compares a pair
of points in (X, Y) axis.
The three broad steps to summarize data are:
(1) Classify categorical data types into different tabular
forms − one-way or two-way, etc.
(2) Use the appropriate numerical summaries for quantitative
type of data − mean, standard deviation (SD), etc.
(3) Use appropriate visual summaries for different types of
data − for categorical, bar, pie, etc., and for qualitative −
histogram, scatter plot, box plot, etc.
For categorical data, frequency or counts are used to know
how frequent different categories are. Tables are used to
count the frequency or proportions from the total. Examples
of tables are given in Tables 4–6. One-way tables may be
manually computed for variables such as menstrual history
and clinical signs as shown in Tables 4 and 5 by counting the
frequency of codes for menstrual history (code 1 or 2 or 3) and
clinical signs (code 1, 2, 3 or 4) given in columns 4 and 6 of
Table 3. Two variables such as menstrual history and clinical
signs were used as an example for studying relationship
between two variables with the help of a two-way table
[Table 6]. Then relevant percent for each category will be
calculated and presented in the tables. The data can then be
exported to statistical software such as SPSS for data analysis.
These tables can also be generated in the computer using a
statistical package for any sample size easily after a data file is
created for the purpose.
For numerical data, the center or location measures the “center”
of the data, for example, sample mean and sample median. For
quantitative variable such as age [2nd column of Table 3], mean
and SD or other locations can be computed. Similarly, the other
commonly used presentations for various other locations are
percentiles, deciles, tertiles, etc. Percentiles divide data and
present in one hundred while deciles and tertiles are ten and
three parts, respectively. The spread or dispersion measures the
“spread” of the data, for example, sample variance, inter-
quartile range. In multiple variable numerical data, each
dimension is uni-variate. Hence, numerical summaries from
uni-variate data can be used for two-group comparison for a
characteristic under study [e.g., hemoglobin (Hb) levels in
diabetic and non-diabetics in terms of sample mean and SD].
To study two measurements or variables at a time to assess their
relationship for example mother’s Hb level and newborn’s
birth weight, numerical summaries such as sample
correlation and regressions are used. Sample correlations
and regressions measure the linear relationship between two
variables.
The third type of summarization is graphical for both
numerical and categorical type of data. Histograms are
for numerical data, histogram is a method to show the
“shape” of the data by using frequencies of the
measurements in the sample. The distribution of the
variable under study can be assessed whether it is
symmetric (no skew) or bell shape, right or positive-
skewed, or left or negative-skewed. Graphical
presentations are commonly used in almost all disciplines.
Graphical are used for displaying trends and contrasts
hidden in the statistical information, and this method is
often used in the presentation of statistical data in
scientific articles and reports. There are various types of
diagrams for displaying of different types of data. The
frequently used types of diagrams are bar, pie, line,
scatter diagram, etc. It is assumed that the reader is
familiar with such presentations and advised to use Excel
spreadsheet or any package for preparing good graphs.
Table 4: One-way table for menstrual history in women on
cervical screening
Menstrual history N (%)
Regular 6 (30)
Irregular 8 (40)
Menopause 6 (30)
Table 5: One-way table for clinical signs in women on
cervical screening
Clinical signs N (%)
Normal cervix 7 (35)
Unhealthy cervix 8 (40)
Cervical Erosion 4 (20)
Others 1 (5)

Table 6: Comparison of clinical signs with menstrual history in women on cervical screening
Clinical signs Menstrual history
Regular, N (%) Irregular, N (%) Menopause, N (%) Total, N (%)
Normal cervix 5 (71.4) 3 (37.5) 0 (0.0) 8 (40.0)
Unhealthy cervix 1 (14.3) 2 (25.0) 4 (80.0) 7 (35.0)
Cervical erosion 0 (0.0) 3 (37.5) 1 (20.0) 4 (20.0)
Others 1 (14.3) 0 (0.0) 0 (0.0) 1 (5.0)
Total 7 (100.0) 8 (100.0) 5 (100.0) 20 (100.0)

MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017 9

 Labani, et al.: Data analysis and writing results and discussion

Inferential summaries of results presentation have the contents in the results section. Such contents in the form of
components of estimation and test of hypothesis. The basis comprehensive tables and figures that have potential to
of estimation of measures is dependent on whether the data answer the research question are part of results section. As
are qualitative or quantitative. The quantitative measures such a part of the team member of the investigation, authors must
as mean, SD, etc., and qualitative measures in frequency or have some basic knowledge or training in research
proportion or rate or ratio etc., are computed to summarize methodology, else it is advised to involve a qualified bio-
data. These measures on sample (small group) of statistician as a team member.
observations are called estimates of the population (larger
group). A measure of sampling variability called standard Involvement of a qualified bio-statistician as a team
error (SE) is used in estimating the confidence intervals (CIs). member
CI is measure of certainty as an interval estimate of the
population summary measures such as mean and proportion. It is advised that preparing to begin writing for publication of
research work begins with the data analysis that is required
For example, suppose we have data on cholesterol level in
200 children of 4–10 years of age with mean 130 and SD 25, for results section of research paper. Data analysis is
then what is the 95% CI of mean? The 95% CI is performed the investigators who are trained to that
requirement themselves or by a qualified bio-statistician,
(mean + 2 SE), (mean−2 SE) or (130 ± 2 × 25 × SQRT 300)
(130 + 2 × 1.45), (130−2 × 1.45) or (127, 133). The this depends on the depth of exploration of data needed. It
interpretation that there is 95% chance of population mean is a well known fact that if the quality of data collected by
following a research design suited for the research question
of cholesterol level in children of 4–10 years of age to be
included in the interval (127, 133). A threshold of 95% being investigated or objective being achieved is not met, the
confidence statement indicates that there remains an writing however best will not yield desired results for
publication. Therefore, it demands either efforts of the
uncertainty of 5% which could result into a critical region
that becomes the basis for hypothesis testing. The probability investigators trained on these aspects or involvement of a
of wrongly rejecting a true null hypothesis (no difference biostatistician from the design stage of the research study
(methodology section) through the subsequent stages of
between groups − proposition) is an error in statistical
decision-making. This is also referred to as P value. The analysis and interpretation of results. In all such stages,
value of this error is generally kept at 0.05. This threshold of
5% is also called the level of significance. A result is called as
“statistically significant” when P < 0.05. The approach of
analysis of quantitative data and qualitative data are shown in
Figures 3 and 4, respectively. There are situations for
comparison of means in hypothesis testing. This could be
a comparison of means in two independent groups such as the
comparison of mean Hb levels in a random sample of well-
nourished and undernourished categories or groups. The
situations where more than two means are to be compared,
the procedure is called analysis of variance with F-test for
assessing overall significance. When data on a group of the
same patients are obtained at two different time points such as
blood pressure or heart rate before and after a therapy, it is
said to be in a dependent or paired set-up. Both these settings Figure 3: Showing methods of analysis for quantitative comparison of
of unpaired and paired could be present with qualitative two groups
setting too. When groups of comparison are more than two
procedures, other than t-test are used. Along with results of
estimates, CIs and tests of hypothesis, assessing strength of Hypothesis Tesng for Qualitave data
relationship between two qualitative and two quantitative
variables, evaluators of diagnostic or screening test Qualitative
Data
performances, etc., along with multiple regressions
considering and evaluating the joint effect of several
Unpaired Paired
variables are also used in the presentation of results
section. The analysis and presentation of results from such
methods is out of scope of this introductory communication.
Reader may refer to other material[27-30] for more details on Chi- Mc-Nemar
Fisher’s Kappa
methods of analysis that obtain results that are presented in Square
Test
Exact Test Statistics
Test
results section. The computation of tables or quantities and
their estimates in terms of CIs or tests of hypotheses are Figure 4: Showing methods of analysis for qualitative data comparison
summarized further to form crisp information for inclusion as of two groups

10 MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017

 Labani, et al.: Data analysis and writing results and discussion

involvement of a bio-statistician as a team member would
reflect definite fulfillment of the felt need in the sections of
results and discussion of the publication. Key messages on
data analysis and data presentation are given in Table 7.
The results section should be organized based on the sequence
of tables and figures included. The tables and figures should
be prepared as soon as all the data is analyzed. It is a good
practice to note one or two key results from each table or
figure to be addressed in the text portion of the results.

WRITING OF RESULTS SECTION Numbers should be assigned to tables and figures in the
sequence to be referred in the text as Table 1, Table 2, etc. and
Suppose a testable hypothesis that can be answered
Figure 1, Figure 2, and so forth. Table is referred in the text as
experimentally, or a question that can be answered by
(e.g., Table 1), whereas figure is abbreviated as Fig. (e.g.,
collecting samples, we accumulate observations about
Fig. 1). The table and figure legends should include the brief
those organisms or phenomena. Those observations are
description of the results being presented and any other
then analyzed to yield an answer to the question called
necessary essential information. Results are a text-based
key result. In the results section, it is required to present
presentation of the key findings with reference to the
data that supports the results of research question. Data
tables and figures. The key results should be stressed
analysis and summarization becomes the first step and
clearly providing the answers to the questions investigated.
especially presentation of data in terms of tables, graphs,
The tables should be mentioned sequentially clearly
and computation of numerical values of central location and
indicating the key results that they convey.[31,32]
dispersion such as mean, median, and SD. The second step is
summarization using inferential statistics results in terms of The results section objectively explains the key results in an
significance and exact P values. orderly and logical sequence without interpretation, using
both text and illustrative materials (tables and figures). The
Generally, the order of presentation followed in methods
results section always begins with key results in the form of
section will be adopted for writing results section. For
text followed by reference to figures and tables. The
example, if the sequence in methods section is about −
summaries of the statistical analyses can be done either in
study subjects, measurements, and statistical analysis, then
text (usually parenthetically) or in relevant tables or figures
the results section should include − presentation of
(in the legend or as footnotes to the table or figure). The
characteristics in terms of one-way or two-way tabular
results section should highlight the evidence needed to
form, this follows and further analysis of the
answer the questions/hypotheses investigated, reporting the
summarizations both descriptive and inferential done on
negative results as well. The text of the results section should
data based on objectives. Detailed data is presented as
be concise and objectively written. Past tense should be used
tables and figures to keep written data to a minimum.
as the writing style. Repetitive paragraph structures should be
Some data descriptions are directly presented in the text.
avoided and data should not be interpreted in this section.
Generally, tables are used for large or complicated
Some points to remember while writing results are given in
data sets that would be difficult to fully explain in the
Table 8.
text. A textual presentation of key findings that emerge or
reveal from figures and tables must be included in the The results of the experiment should be presented in logical
results section. The tables or figures must always be sequence to support (or provide evidence against) the
referred in text. hypotheses, or answer to the question, stated in the

Table 7: Key messages on data analysis and data presentation

1. As most manuscripts require presentation of baseline demographic and clinical characteristics, tables such as one-analysis of categorical data and
means along with standard deviation or medians with quartile range/minimum or maximum values must be computed. This step also helps in data
exploration for better understanding.
2. No statistical test of significance is to be attempted unless the data is fully explored and understood with summaries in the step 1 above.
3. For want of one single table for inclusion in results section of a publication, many one way and two tables along with means and standard
deviation for various groups under comparison requires being prepared. Between standard deviation and standard error, inclusion of standard
deviation with mean is recommended.
4. Computer software packages are available for ease of computation. Use of such packages requires appropriate training on basic methods of
biostatistics for what situations a particular test is to be chosen.
5. A warning in use of statistical packages is that it provides lot of statistics by default and just copying whatever is available on the output is
disastrous. Journal’s review process can easily screen for manuscripts to give negative marking.
6. Tables and figures decided for inclusion in the publication must describe the study subjects or answer in part or full the research questions
targeted in the article under publication.
7. Figures and tables included as part of results section must be self explanatory including abbreviations and symbols.
8. The sample size presented in various tables must be consistent and if there is any loss of sample due to non-response, must be stated as footnote
in the table.
9. Appropriate precision needed for expression of measured values and P-values. Too much precision for measured values is not recommended. For
very low P-values containing most zeros must be terminated with three decimal places such as P < 0.001.

MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017 11

 Labani, et al.: Data analysis and writing results and discussion

introduction. It is important to report negative results. The
hypothesis needs to reformulated if the anticipated results are
not obtained or there is absence of an effect. Even though the
results do not support the hypothesis, they may be of
importance to others. Results obtained contrary to that
expected should not be considered as “bad data”, as many
important discoveries can be traced from it. Appropriate units
should be used while reporting data or summary
statistics. For an individual value, units should be written
as “the mean age was 30 years.” On the other hand, for a series
of numbers all having the same unit, the unit should be placed
after the last value (e.g., height of 156, 154, 157, and
172 cm).
The use of the word, “significant” in scientific studies implies
the statistical test indicated a larger difference in the particular
variable in the study than expected to get by chance alone. If
P-value in the statistical information indicates significance
(usually when P < 0.05), it is unnecessary to use the word
“significant” as the P-value is interpreted in the same way.
Similarly, when the difference in means is reported, it is
understood that it is tested and the difference is found to be
statistically significant, especially if P-value <0.05 is
reported.
DISCUSSING RESULTS OF RESEARCH STUDY
In manuscript writing, one of the faltering blocks is
preparation of a write-up on discussion on results.
Discussion writing is a concern and a scary one even for
many young faculty and students attempting to write their PG
thesis. Discussion section presents the findings in a research
context and explains the meaning of findings in its main
purpose. The section provides answers to the questions posed
in the introduction section and follows results section where
new information related to findings on research question or
questions presented. The sequence is that in the introduction,
we arrive at a research question from a background of the
research topic while in the discussion we begin with
summary of findings and arrive at discussing the results by
presenting interpretations, discrepancies, unexpected
findings, weaknesses, and limitations.[20-22]
Checklists of guidelines for writing various sections of an
original research article in IMRAD are available freely on
internet sites such as STROBE[33] for observational studies,
CONSORT[34] for clinical trials, and journal instructions on
writing various sections of a research article. Yet the
presentation of discussion section by authors in their
research articles has wide variations. Authors adopt
writing approach by reiterating the aim of the study or by
specifying the major findings or a few start comparing the
findings in the beginning. Even some authors view this
section as if it is a review of literature and describe
without relating to the findings of the study. Re-writing of
most results again in discussion section is also a deviation
from guidelines. The order of presentation by some authors is
quite haphazard and not as dictated by the importance of the
results presented in results section of the research article.
Since every research paper has its unique results and
findings, the order of discussion requires certain rules and
guidelines following which the section is written to improve
the quality of the manuscript. The narration and
demonstration of available rules and guidelines of writing
discussion in the light of published papers are helpful. Some
important guidelines of writing discussion section are given
in Table 9.
Citing of references in discussion section
The work of others is used as citation primarily in the introduction
section to develop back-group and rationale on the topic of
research question and in the discussion section of the article or a
thesis. No referencing of others work is cited in the results section
while in methods section if necessary have some citations of
others or standard methodology. Discussion of our results by
comparing and contrasting the findings of other studies is of
importance by citing and referencing of those studies. Whenever
other studies’ findings are used, we must refer and cite the
sources from which such information is drawn. Citing means
acknowledging within our own text the related document from
which information is obtained. References are the details of
document included in a separate section at the end of our text. As
against reference, the term bibliography is used in thesis/
dissertation document which includes the total publications
consulted. We generally look at the aspects in others studies
for inclusion in our discussion of our results such as their
objectives, findings and conclusions. We must take the
information of other studies and put it in our own words to
avoid plagiarism. This is related to use of others’ work or ideas or
words without proper citing of their reference. In writing or
reporting research, credit must be given for others’ ideas or
writing or research.

Table 8: Points to remember in results section

1. Key results are an answer to the analysis made on the observations obtained while testing the hypotheses.
2. The results should provide as much information as possible about the nature of differences and relationships. A clear comment should be made on
the differences, directionality, and magnitude of differences between the different groups studied.
3. While commenting on the different groups in the study, we should answer the following questions. (a) How are they different? (b) How much are
they different? (c) We need to report on the direction of differences (greater, larger, smaller, etc.) and (d) the magnitude of differences (percent
difference, how many times, etc.).
4. Results relevant to the question(s) presented in the introduction irrespective of whether or not the results support the hypothesis(es).
5. The text should not be the repetition of the same information in figures/tables.
6. Data should be accurate and consistent throughout the manuscript.

12 MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017

 Labani, et al.: Data analysis and writing results and discussion

There are standardized systems for referring to materials used
in our writing with the help of a citation style or a reference
system. Most biomedical journals including Maulana Azad
Medical College Journal of Medical Sciences today use
Vancouver style. In this style, a number is allocated to a
source in the order in which it is cited in the text beginning
from introduction to discussion sections. In case if the source
is referred to again, the same number is used and the
references are listed in numerical order in the reference list
at the end of the paper. Academic institutions use, including
for the purpose of thesis and dissertations, the other famous
style of referencing called Harvard style. This is an author-
date referencing style which varies in some features between
institutions/journals such as punctuation, capitalization,
abbreviations, and the use of italics.
Check list for observational and clinical trials
According to the checklist STROBE statement,[33]
observational studies the number of individuals at each
stage of study should be reported (e.g., potentially eligible
participants, eligibility confirmed, those included in the
study, those who came for follow up, and finally those
who were analyzed). The reasons for non-participation at
each stage should be given. Flow diagram showing the
different stages of study should be considered. The
demographic, clinical, and social characteristics of the
participants should be given with information on exposures
and potential confounders. The number of participants should
be mentioned with mention about the missing data for each
variable of interest. For a cohort study, the average and total
follow-up time should be summarized. The number of
outcome events or summary measures over time for a
cohort study should be reported. In a case–control study,
the numbers in each exposure category, or summary measures
of exposure should be reported. For a cross-sectional study,
the numbers of outcome events or summary measures should
be reported. While reporting the main results, the unadjusted
estimates should be given along with the confounder-adjusted
estimates and their precision (e.g., 95% CI), if applicable.
Explanation for including the confounders and how they were
adjusted should be given. If the quantitative data is
categorized, the category boundaries should be reported.
For a time period, the relative risk and absolute risk
should be translated into meaningful data. If any other
analyses are done such as, analyses of subgroups and
interactions, and sensitivity analyses should also be
reported.
For a clinical trial, the CONSORT 2010 checklist[34] provides
the necessary information to be included while reporting. It is
strongly recommended that the participant flow should be
explained in the form of flow diagram. The numbers of
participants who were randomly assigned into each group
should be mentioned, along with those who received
treatment and who were analyzed for the primary outcome.
After randomization, those participants who were lost or
excluded for each group should also be mentioned. The
recruitment period and follow up date should be stated. The
reason for ending or stopping the trial should be declared. The
baseline demographic and clinical characteristics should be
illustrated in the form of table. The number of participants
(denominator) for each group included in each analysis should
be mentioned and whether the analysis was by original assigned
groups. The results report the primary and secondary outcomes
for each group with the estimated effect size and its precision
(e.g., 95% CI). The absolute and relative effect size should be
presented for binary outcomes. The subgroup analyses and
adjusted analyses for pre-specified and exploratory analyses
should be stated. As per the specific guidelines under ethics for
harms, it is necessary to report all important harms or
unintended effects in each group. The above description is
on checklists for observational and clinical trial studies and
there are checklists for other kinds of studies too in the
literature.For discussion section, the desired checklist is to
ensure whether the following are included − (i) key results
summarized with reference to objectives, (ii) limitations with
reference to source of bias along with magnitude and direction,
(iii) interpretation of results in the light of objectives,

Table 9: Important guidelines for writing discussion section

1. For the benefit of the reader, reiterate first the research problem being investigated and then move on to major findings beginning with most
important and then decreasing order of their impact. If the objective is otherwise clear from title of the paper, it is advised to begin the section
with major findings.
2. In the results section of the article, the revealing findings are just stated without any interpretation. The meaning of each finding should be
described, as the author is the best person to talk about their own study. There may be other possible explanations that fitted into our hypotheses
and they all need to be considered.
3. It is helpful to support the importance of our results by comparing and contrasting the findings of other studies. The aspects that we look at in
others studies are: their objectives, findings and conclusions. After ensuring common methodology, the agreement in findings if exists may be
stated. In case of major differences in findings between studies, the reason why findings differed or unanticipated should be mentioned.
4. It is important to acknowledge our study weaknesses, limitations and strengths. For example if the limitation is related to method adopted, an
explanation of influence of findings due to that reason be added. Unanswered questions that our study did not address, the extent of non-possible
generalization of results to other situations need mention.
5. Last paragraph of the section generally on conclusions. This should be a take home message from the study findings and a general statement
reiterating our answer to the research question along with scientific implications and advice if any on future studies required.
6. Avoid re-stating results or adding new results. The sequence of discussion of findings should be in the same sequence that we described them in
the results section with sub-headings if necessary.

MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017 13

 Labani, et al.: Data analysis and writing results and discussion

limitations and results of other similar studies, and
(iv) generalization of study results in terms of external
validity.
There are many statistical terms[27,28] used while writing the
results section. Some commonly used statistical terms are
given in Table 10.
SUMMARY
Writing a manuscript of a research article or a thesis on public
health or any scientific subject is a stupendous process for
learning writers. An original research article should follow the
format of IMRAD following the sections in the sequence of
Introduction, Methods, Results, and Discussion. Even though
the IMRAD format is followed, the results section is written
first that provides new information related to answering the
research question. Sometimes first in the order of writing
manuscript, methods section is preferred to results section as
the former is already written section at the protocol level and
only requires to be re-written in past tense. To make a beginning
of results section writing, it is essential to understand its
association with data summaries expected to give answers to
the research questions attempted to answer. Results section
attempts to present the results of the analysis obtained on data
collected on the investigation of interest after application of
appropriate statistical tools. This communication attempts to
present the link between data summarization and writing text
part of the results section. This attempt provides tips to young
researchers in writing paper and PG/UG medical students in the
thesis writing. The completion of results section along with
methods section constitutes 50% of the paper writing work and
the young researcher or student who is attempting first research
paper or thesis writing can gain confidence in the process of
research document.
Questionnaire, schedule or proformae are the commonly used
data collection tools. Data can be either numerical (or
quantitative) or categorical. Numerical or quantitative data
can be further classified as discrete or continuous. Categorical
data can be nominal or ordinal. Various methods of
summarization of data include tables, graphs or numbers.
Depending on the type of analysis the tables can be one-
dimension (or one-way) or two-dimension (two-way). Excel
sheet is used for data entry and data analyses are done using
statistical packages.
Data summaries are generally prepared and presented through
tables and graphs. Trends are displayed using graphs such as
bar, pie, line, and scatter diagram. Data summarization can be
descriptive or inferential. Descriptive summarization can be
performed in the form of graphs, as histograms,
quantile–quantile plot or scatter plot. The data needs to be
classified into different types and then depending on the type of
data, appropriate numerical and visual summaries are used. The
estimation and test of hypothesis are the components of
inferential summarization. This is dependent on whether the
data are qualitative or quantitative. Mean, SD, etc. are
quantitative measures whereas frequency, proportion, rate or
ratios are qualitative measures. Involvement of a qualified bio-
statistician is recommended as a team member for contribution
in study design, statistical analysis and publication.
The text part of results is presented in the past tense and all
the contents of what is presented in tables and figures should
not be repeated in the text results. As this section provides
new information in the related research question attempted
to answer, key consideration and rules of writing this section
help to produce a good text part of the results section.
Following data summarization based on the objectives,
the results section is written. The text part of all the
contents of what is presented in tables and graphs should
not be repeated in the text results. In this section, each
paragraph begins with a result. Most of the data is presented
in the form of tables and figures and its use is mentioned in
the text. Results section objectively explains the key results
in an orderly manner and logical sequence without
interpretation. The evidence needed to answer the

Table 10: Some statistical terms commonly used in results section

Statistical term Meaning
Mean The average − the arithmetic mean
Standard deviation Most common and generally most appropriate measure of dispersion obtained as positive square root of variance.
(SD)
Ratio Strength or magnitude or number of one quantity relative to the other, such as sex ratio and albumin-globulin ratio.
Sample size (n) Total number of subjects involved in the study.
Percentage It is the number or ratio expressed as a fraction of 100.
Relative risk (RR) Risk of occurrence of an outcome in the presence of one factor (exposure) relative to the risk in the presence of another
(generally control, which means absence of a factor).
Confidence interval The interval within which a population parameter value is expected to lie in a certain interval with a confidence level of 95 or
(CI) 99% as could be revealed by repeated samples.
Range From minimum value of a measurement to the maximum value.
P-value It is a function of observed sample results (a test statistic) which measures how extreme the observation is, relative to a
statistical model.
Median The most middle value obtained after arranging values in increasing or decreasing order. Median seeks to divide the group in
two halves, each with n/2 individuals.

14 MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017

 Labani, et al.: Data analysis and writing results and discussion
questions/hypotheses investigated should be highlighted in
this section. It is important to highlight the negative results
as well. As compared to other sections of the manuscript,
discussion is considered hard to define. It describes
virtues; interpret results, state limitations and future
recommendations. It is focused to answer the research
question raised in the introduction section of the
manuscript. It includes statements on important results
and compares study results reported previously. The
approach of working on data summarization and begin
the writing of text part of the results and discussion of
results together is expected to help the young researchers or
UG and PG medical students to make a good beginning in
writing research paper or thesis writing.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
REFERENCES
1. Alexandrov AV. How to write a research paper. Cerebrovasc Dis
2004;18:135-8.
2. Ahmad HR. How to write a Doctoral Thesis. Pak J Med Sci
2016;32:270-3.
3. Cunningham SJ. How to write a thesis. J Orthod 2004;31:144-8.
4. Roberts P. Practical issues in ‘writing up’ a research thesis. Nurse Res
2000;7:14-23.
5. Kallestinova ED. How to write your first research paper. Yale J Biol
Med 2011;84:181-90.
6. Bajwa SJ, Sawhney C. Preparing manuscript: Scientific writing for
publication. Indian J Anaesth 2016;60:674-8.
7. Peh WC, Ng KH. Preparing a manuscript for submission. Singapore
Med J 2009;50:759-61.
8. Hong ST. Ten tips for authors of scientific articles. J Korean Med Sci
2014;29:1035-7.
9. Nahata MC. Tips for writing and publishing an article. Ann
Pharmacother 2008;42:273-7.
10. Pakes GE. Writing manuscripts describing clinical trials: A guide for
pharmacotherapeutic researchers. Ann Pharmacother 2001;35:770-9.
11. Veness M. Strategies to successfully publish your first manuscript. J
Med Imaging Radiat Oncol 2010;54:395-400.
12. Setiati S, Harimurti K. Writing for scientific medical manuscript: A guide
for preparing manuscript submitted to biomedical journals. Acta Med
Indones 2007;39:50-5.
13. Ohwovoriole AE. Writing biomedical manuscripts part I:
Fundamentals and general rules. West Afr J Med 2011;30:151-7.
MAMC Journal of Medical Sciences ¦ Volume 3 ¦ Issue 1 ¦ January-April 2017
View publication stats
14. Huber VC, Vogt HB. So you want to be an author: A primer on writing
for publication in the medical literature part I: Manuscript preparation.
S D Med 2016;69:172-5.
15. Nte AR, Awi DD. Writing a scientific paper: Getting to the basics.
Niger J Med 2007;16:212-8.
16. Van Way CW 3rd. Writing a scientific paper. Nutr Clin Pract
2007;22:636-40.
17. Branson RD. Anatomy of a research paper. Respir Care 2004;49:1222-8.
18. Mauffrey C, Scarlat MM, Pećina M. Setting standards for medical
writing in orthopaedics. Int Orthop 2014;38:1-5.
19. Auvinen A. How do I write a scientific article − Advice to a young
researcher. Duodecim 2015;131:1460-6.
20. The Discussion − Organizing Your Social Sciences Research Paper.
Available from: libguides.usc.edu/writing guide/discussion. [Last
accessed on2016 Dec 06].
21. Bavdekar SB. Writing the discussion section: Describing the significance
of the study findings. J Assoc Physicians India 2015;63:40-2.
22. Doblecki W. The scientific article. Discussion section. J Endod
1993;19:480.
23. Bresson L. How to write the “ Discussion” part of your article, master’s
essay or thesis. Gynecol Obstet Fertil 2009;37:372-3.
24. Writing the Results Section for a Scientific Manuscript. Available
from: http://www.biosciencewriters.com/Writing-the-Results-Section-
for-a-Scientific-Manuscript.aspx. [Last accessed on 2016 Dec 06].
25. How to Write the Results Section of a Research Paper. Available from:
http://classroom.synonym.com/write-results-section-research-paper-
4246.html. [Last accessed on2016 Dec 06].
26. Rita N. How to Write a Strong Discussion in Scientific Manuscripts.
Available from: http://www.biosciencewriters.com/How-to-Write-a-
Strong-Discussion-in-Scientific-Manuscripts.aspx. [Last accessed
on 2016 Dec 06].
27. Indrayan A, Satyanarayana L. Biostatistics for Medical, Nursing and
Pharmacy Students. New Delhi, India: PHI Publishers; 2006. ISBN:
81-203-3054-4.
28. Indrayan A, Satyanarayana L. Simple Biostatistics for MBBS, PG
Entrance and USMLE. 4th ed. Delhi: Academa Publishers; 2013.
29. Labani S, Asthana S. Basics of biostatistics for understanding research
findings. MAMC J Med Sci 2015;1:136-41.
30. Satyanarayana L, Asthana S. Relevance of statistical significance in
medical research. G R J 2014;3: 107-213.
31. Results Writing Section. Available from: http://abacus.bates.edu/
∼ganderso/biology/resources/writing/HTWsections.html. [Last accessed
on 2016 Dec 06].
32. Twelve Steps to Writing Effective Results Section. Available from:
www.sfedit.net/results.pdf. [Last accessed on 2016 Dec 06].
33. CONSORT 2010 Checklist of Information to Include When Reporting
a Randomized Trial. Available form: http://www.consort-statement.
org/. [Last accessed on 2016 Dec 06].
34. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC,
Vandenbroucke JP; STROBE Initiative. The Strengthening the
Reporting of Observational Studies in Epidemiology (STROBE)
statement: Guidelines for reporting observational studies. J Clin
Epidemiol 2008;61:344-9. PMID: 18313558. Available from: http://
strobe-statement.org/index-php?id=available-checklists. [Last accessed
on 2016 Dec 06].
15