Research for Beginners: Finding, Reading and Organising the Literature
Research starts with finding the conversation you are joining, not with having an idea. The idea usually arrives later, out of the gap between what people have already shown and what they have not. That means the first real skill is searching and reading, and both are more structured than they look from outside.
This is how to do that part without drowning: how to search, how to read at speed without being careless, what to write down, and how to tell which kind of review you have been asked for.
Search more than one way
A single search phrased a single way finds a single corner of a field. Subfields name the same idea differently, and older papers use older vocabulary, so a query that returns twelve results is usually telling you about your phrasing rather than about the literature.
Search a scholarly database rather than relying on a general search engine, then follow citations in both directions: read the references of the best paper you find, and look up the newer work that cites it. That pair of moves finds more of the relevant literature than any amount of rephrasing, because it follows how the field actually connects.
Write down the terms you used and the date you used them. It feels like bureaucracy right up until you need to report a search strategy, and reconstructing one from memory is not possible.
Read in passes, not start to finish
Reading every paper you open from the first word to the last is the most common way to lose a month. Experienced readers make several passes and abandon most papers early.
The first pass is the title, the abstract, the section headings and the conclusion, and its only job is to decide whether there is a second pass. The second is the figures, the tables and the method, because that is where the claim actually lives and where a weak paper usually shows itself. Only work you have decided matters earns a close reading of the whole thing.
Most papers should not survive the first pass. That is not carelessness, it is the filter working.
Take the note while you are still reading
For every paper that survives, record five things: the exact citation, the question it asked, what it did, what it found, and one line on why it matters to you. That last line is the one people skip and the one that is impossible to reconstruct later, because it is about your argument rather than about their paper.
Keep the BibTeX entry next to the note. Citing then becomes a copy rather than a search, and the version you cite is the one you actually read. If you have never set up a bibliography, the citations and bibliography guide covers the mechanics, and citep against citet covers the one that trips everybody up first.
Which review are you actually writing
"Literature review" is used for four different documents with different methods and different reporting standards. A supervisor who asks for a systematic review is asking for a protocol, a documented search and a flow diagram, not a longer essay, and discovering that after writing the essay is a term of work lost.
Which review are you writing?
Four different documents share the name. Two questions usually settle which one your supervisor means.
Narrative or literature review
Use it when: You are mapping what is known in an area, usually to frame your own work or a thesis chapter. The question is broad and you are allowed to be selective, as long as you are honest that you were.
What it involves: Reading widely, organising by theme rather than by paper, and arguing a line through the field. No formal search protocol is required, which is exactly why it must not be called systematic.
Scoping review
Use it when: You want to know how much evidence exists and what shape it is in, before committing to a systematic review. Often the honest first step when nobody knows whether the literature is large or thin.
What it involves: A protocol written in advance, a documented search, and a description of what is out there rather than a verdict on it. It reports coverage, not effect.
Systematic review
Use it when: You have one specific, answerable question and you intend to find every study that bears on it, not a convenient sample.
What it involves: A pre-registered protocol, an explicit search strategy you publish so others can repeat it, stated inclusion and exclusion criteria, screening by more than one person, and a flow diagram accounting for every record. PRISMA is the reporting standard most fields expect.
Meta-analysis
Use it when: A systematic review whose included studies measure the same thing in a comparable enough way that their results can be combined into one estimate.
What it involves: Everything a systematic review involves, plus extracting effect sizes, choosing a model, and reporting heterogeneity. If the studies are not comparable, pooling them produces a confident number that means nothing.
If two still fit, that is a real answer: ask your supervisor which reporting standard they expect, because that is the thing the difference actually decides.
Organise by idea, not by paper
The commonest failure in a first review is a chain of paragraphs each summarising one paper in the order they were read. It is a reading list with sentences, and it tells a reader nothing about the field.
Organise by claim instead. What do these papers collectively establish, where do they disagree, and what has nobody checked? A paragraph that opens with the finding and cites three papers for it is doing work; a paragraph that opens with an author's name usually is not.
Write it somewhere your supervisor can reach
Feedback is the whole point of having a supervisor, and it arrives faster when they can read the current draft rather than an emailed PDF from nine days ago. Every template on LetX opens and compiles in the browser with no install, and more than one person can work in the same document, which matters most for the documents that take months. Your unpublished work is never used to train AI.
When the reading turns into a submission, how to publish your first paper covers what happens next, and the thesis templates cover 87 institutions if this is a chapter rather than a paper.
Frequently asked questions
What is the difference between a literature review and a systematic review?
A literature review maps what is known in an area and is allowed to be selective. A systematic review answers one pre-specified question by finding every study that bears on it, using a protocol registered in advance, a published search strategy, stated inclusion criteria and screening by more than one person. Calling a selective review systematic is a claim about method, not a description of length.
How do I search the literature properly?
Use a scholarly database rather than a general search engine as your main tool, search the same question more than one way because terminology differs between subfields, and follow citations in both directions: the references of a good paper, and the newer papers that cite it. Record the exact search terms and dates as you go, because a review may need to report them.
How should I read a research paper?
In passes rather than start to finish. First the title, abstract, section headings and conclusion, to decide whether to continue at all. Then the figures, tables and method, which is where the actual claim lives. Only read the whole thing closely for papers you have decided matter. Most papers you open should not survive the first pass, and that is the point.
What notes should I keep while reading?
For each paper: the exact citation, the question it asked, what it did, what it found, and one line on why it matters to you. Record it as you read, because reconstructing it later from a PDF you half remember is far slower than writing it once. Keep the BibTeX entry with the note so citing is a copy rather than a search.
Do I need a reference manager?
You need something that stores the citation data, and a reference manager is the usual answer. What matters is that your citation keys stay stable and your bibliography file is version-controlled alongside the document, so a citation cannot silently break between drafts.
Ready to try LetX?
Join researchers and students focusing on their work, not the tools.
Get Started Free
Written by Shihab Shahriar Antor , AI Engineer & Founder of Shahriar Labs. Builder of LetX (collaborative LaTeX) and QuantumSketch (AI STEM video).