Research and Writing

This page covers how to actually carry out the research, structure the essay, manage sources, and meet the formatting rules.

The process at a glance

Broad topic --> Focused topic --> Preliminary investigation --> Research question
     |                                                                |
     v                                                                v
Review and edit  <--  Write conclusion  <--  Analyse and synthesise <-- Plan and apply
   final essay              findings                                     research methods

The arrows are not strictly one-way – you will loop back, especially between methodology and results – but the high-level shape holds.

Sources

Source quality directly affects your mark – it is not a presentation detail. Examiners read your bibliography as evidence of how rigorously you researched the topic.

Use these as primary sources

Original data you collected – algorithm benchmarks, simulation output, dataset analysis.
Public datasets – UCI, Kaggle, government data portals, ImageNet, MNIST and similar.
Outputs from your own implementations.

Use these as secondary sources

Peer-reviewed journal articles.
Conference proceedings (ACM, IEEE).
Scholarly books.
Official technical documentation.

Use these only for background

Technical blogs and websites (do not base analysis on them).
Wikipedia (a starting point, not a citation).
Journalistic articles about technology.

If your bibliography is dominated by blogs, YouTube videos, or Wikipedia, your knowledge mark will suffer. Google Scholar is the minimum starting point.

AI tools

May be used to: identify academic sources, generate test datasets, assist with testing infrastructure.
Must not be used to: generate or structure your written content.
Any AI use must be acknowledged.

Essay structure

A CS extended essay typically follows this structure. Section names can vary, but every element must be present.

#	Section	What goes in it
1	Title page	Student code, research question, subject (Computer Science). No name, school, or supervisor anywhere in the file.
2	Contents page	Sections and page numbers.
3	Introduction	Why this topic matters, the research question, a preview of the method, the line of argument.
4	Background / Literature Review	Relevant CS concepts, existing research, theoretical framework. Evaluate, do not just define.
5	Methodology	What you did, the metrics you used, why each choice was made and what the alternatives were. Must be reproducible: dataset version(s), library versions, hyperparameters, random seeds.
6	Results / Findings	Data tables, charts, graphs. Standardised format, clearly labelled, units stated.
7	Discussion / Analysis	Interpret results. Examine patterns, trade-offs, anomalies. Compare with existing literature.
8	Evaluation	Integrated into discussion (not a separate section). Strengths, limitations, alternative interpretations, methodological reflection.
9	Conclusion	Synthesise findings. Answer the research question. Note unresolved issues. Do not introduce new material.
10	Reference list / Bibliography	Every source cited (and, if you use a bibliography, every source consulted).

Where does code go?

In the appendix. The body of the essay should discuss the outcomes of your implementation, not its line-by-line structure. Examiners are not required to read appendices, so anything that supports your argument must also appear in the body.

Include only the code that is essential to the investigation – the algorithm under analysis, the experimental harness, the metric implementation. Do not paste the full codebase, configuration files, or boilerplate. Pasting hundreds of pages of unrelated source signals that you do not know what is essential.

Visuals

Data tables, charts, and graphs are encouraged in the body.
Number every figure and table sequentially: Figure 1, Figure 2, Table 1, Table 2.
Add a brief caption beneath each (one short sentence on what the visual shows).
Reference each visual in the running text – “as shown in Figure 3” – otherwise the visual is decoration, not evidence.
If a visual is from another source, attribute it (“adapted from Smith, 2024”).
They are not included in the word count.

Word count

Maximum: 4,000 words. Examiners stop reading at the 4,000th word – mid-sentence if necessary. If your conclusion sits past the limit, the examiner will not see it, and Criteria C and D suffer.

Counted	Not counted
Introduction, body, conclusion	Contents page
Quotations	Headers
Non-reference footnotes / endnotes	Maps, charts, diagrams, annotated illustrations
	Tables
	Equations, formulas, calculations
	Citations and references (parenthetical, footnote, endnote)
	Bibliography
	Reflective statement on the RPF

The total word count must be stated on the title page or first page of the essay.

Anti-circumvention note. Examiners may, at their discretion, count “tables” that are really paragraphs in cell form. Do not try to dodge the word limit by reformatting prose as a single-column table – it does not work and it costs marks for poor structure.

Formatting requirements

Font size 12, line spacing 1.5.
Page numbering, beginning after the contents page.
Anonymity: no student, supervisor, or school name anywhere in the file.
File size under 10 MB.

Citation and referencing

Use a consistent referencing style throughout. The IB does not prescribe which one. The realistic options for CS are APA, Harvard, or IEEE (the latter is most common in computer-science publications). Pick one early and stick with it.
“Cite as you write.” Trying to add references after the fact is how things get lost.
Minimum information per reference: author, publication date, title, page numbers if applicable, URL and access date for electronic sources.
Reference list = sources you actually cited.
Bibliography = every source you consulted (including uncited ones). Use one or the other, but be consistent.

Building a literature matrix

A literature matrix is a structured way to organise the sources you read so you can compare them later. Build it as you go – it will save hours during the writing phase.

For each source, record:

Field	What goes here
Topic / aspect	Which part of your research it relates to
Title	Full title
Author / Year	Author(s) and publication year
Peer-reviewed?	Yes / no
Full citation	In your chosen referencing style
Source type	Journal article, conference paper, book, dataset, etc.
Purpose	What the source set out to investigate or explain
Key findings	Main results or conclusions
Strengths	Methodological strengths
Limitations	Methodological limitations
Relevance	How it connects to your research question
Quotes	Direct quotes you may want to cite
Notes	Anything else worth remembering

A literature matrix is not assessed directly, but it powers the comparison-with-literature work that distinguishes a 6/8 from an 8/8 on Criterion D.

Academic integrity reminders

The essay must be authentically yours.
All sources must be acknowledged. Distinguish your words from others’ (quotation marks, indentation).
Do not fabricate data.
The same work cannot be submitted twice. The EE must not duplicate content from your IA or any other Diploma Programme component.
Your supervisor may comment on one full draft only – no direct editing.
The final version must be clean – no comments left in the file.
Breaching IB academic integrity puts your diploma at risk.