How to Write a Research Paper, Step by Step

Pick a question you can actually answer

Most research papers fail in week one, before anyone opens a database, because the writer picked a topic instead of a question. 'Social media and mental health' is a topic — it has no edges, so no amount of writing will finish it. 'Did limiting Instagram to 30 minutes a day change loneliness scores in college students?' is a question: it names a population, an intervention and an outcome you can look up. The test takes ten seconds. If you can describe the evidence that would settle the question, it is answerable. If you cannot, it is still a topic, and it will quietly eat your semester.

Narrow by adding one constraint at a time: a population (first-generation students), a place (community colleges in Ohio), a window (2019-2024), or a live disagreement between researchers. Every constraint cuts your reading pile roughly in half, and the pile matters because a 10-page paper runs about 2,500-2,800 words and comfortably supports eight to twelve sources — not forty. Before committing, run three checks. First, search Google Scholar for ten minutes; fewer than five relevant hits means the question is phrased oddly or genuinely empty. Second, confirm that someone could disagree with at least one plausible answer. Third, log the due date in the Deadline Planner and set yourself a question deadline one week out.

• State the question in one sentence with no 'etc.' and no 'and related issues' — vagueness here becomes fourteen pages of padding later.
• Match the question to the page count: a 2,000-word paper supports one clean comparison; save three-variable designs for term papers of 5,000 words or more.
• Check that your library actually has access — JSTOR, PsycINFO and ERIC coverage varies by school, and a question you cannot source is a dead question.
• Keep the question visible at the top of your notes file; every source that does not help answer it gets cut, no matter how interesting.

What students get wrong about research papers

The most common failure is not bad prose — it is the annotated summary. The student gathers nine sources, hands each one a paragraph, and staples an introduction to the front. Nothing is argued; the paper is a book report with citations. Graders spot it in about ninety seconds, because each body paragraph opens with a source's name instead of the writer's claim. On a typical 100-point rubric, that paper caps out in the C range even when the writing is clean, since 'analysis' and 'argument' usually carry 40 or more points together and a summary paper earns none of them.

The second failure is time allocation. Students pour 60-70 percent of their hours into searching, treat drafting as typing, and leave revision for the final ninety minutes. That ratio produces exactly the papers graders complain about: current, well-stocked, and unshaped. Flip it. Reading deserves about a third of your total hours, drafting another third, and revising the last third — and revision means restructuring, not comma-hunting. A paper with eight sources and two full revision passes beats a paper with twenty sources and no passes, and the difference shows up on the rubric line that says 'thesis and organization.'

Search like a librarian

Your library pays for databases Google cannot touch, and searching two of them well beats skimming ten badly. Start in Google Scholar to map the terrain: put quotation marks around exact phrases, filter to the last five years for anything fast-moving, and note which authors keep appearing. Then move to the subject databases — PsycINFO for psychology, ERIC for education, PubMed for health topics, JSTOR for humanities, Business Source Complete for anything with a case study. The move most students miss is citation chaining. Find one recent literature review, mine its reference list for the classic studies, then click 'Cited by' on Scholar to see who answered it. Twenty minutes of chaining typically beats two hours of new keyword guessing.

• Search in quotes first ('zero-tolerance policy' with the quotation marks), then broaden only if the hits come back thin.
• Read abstracts in a batch: ten abstracts in fifteen minutes, keep three. Full-text reading is for keepers only.
• Mine one good review article's reference list — it is a pre-filtered bibliography chosen by an expert.
• Request anything your library lacks through interlibrary loan the same day; delivery runs 3-10 days, which is why late requests cost real time.
• Log every source into the Citation Generator the moment you open it. Ten minutes of filing now prevents the 2 a.m. reconstruction of where a statistic came from.
• Set a stopping rule: when two consecutive searches return nothing new, you have enough. Saturated searches end; compulsive ones consume weekends.

Build a reverse outline

A reverse outline is built from your notes instead of from a blank page. After reading, write one sentence per cluster of material — and make each sentence a claim, not a topic label. 'Studies show mixed results' is a label. 'Later school start times raised attendance in both district studies, but only when the bell moved past 8:30' is a claim. Stack the claims in the order that builds your argument, then hunt the gaps: a heading with no sources means more research is needed, and a source that fits no heading means your question has drifted. Ten rows is plenty for a 10-page paper, because ten rows is roughly one body paragraph each.

The Essay Structure Builder on this page does the same job interactively: it takes your thesis and scaffolds body sections with claim-evidence-analysis slots, so the outline arrives pre-tested against paragraph anatomy. However you build it, give every row three fields: the claim in your words, the two or three sources that carry it, and the evidence type (interview data, statistics, a court ruling, a close reading). If you cannot fill the source field, you have found your next search — which is the outline doing its real job, catching holes while they cost minutes instead of weekends.

Draft the middle first

Introductions promise; only a finished body knows what it can deliver. So draft the section you have the most notes for — usually the second theme or the methods — and let the opening paragraph wait until the argument exists. Budget by words, not by mood: at 250 words per double-spaced page, a 10-page paper is roughly 2,500 words, which splits into a 250-word introduction, a 250-word conclusion, and about 2,000 words of body. At a realistic 300-500 words per focused hour, that is five to seven drafting sessions, not one heroic night. Type '[FIX]' wherever a fact is missing and keep moving; momentum is worth more than precision in a first draft.

Write ugly on purpose. First drafts with perfect sentences are usually being edited while drafted, which halves your speed and hides structural problems under polish. A better deal with yourself: 45-minute sprints, 300 words minimum each, phone in another room. After each sprint, write one margin note naming what the section just proved — that note becomes the paragraph's topic sentence during revision, and it is the cheapest fix for sections that wander. The Readability Checker will catch the tangled sentences later. Nothing in a first draft needs to be elegant; it needs to exist.

Cite as you write, not after

Citation is the cheapest grade insurance available, and the price drops when you pay mid-draft. The student who cites while writing produces accurate page numbers and reference entries that match. The student who cites afterward produces guesses, [CITE] brackets and a 1 a.m. hunt through fourteen browser tabs. Lock the style before drafting, because the choice changes real things: APA 7 runs author-date citations with its own title-page layout, MLA 9 uses author and page number, and Chicago 18 offers notes-bibliography footnotes. If the syllabus is silent, email the instructor this week — a style switch on page six of a finished draft costs an entire evening.

• Type the in-text citation with the sentence; brackets you promise to fill later multiply overnight.
• Quotation marks go around borrowed words even in rough notes. Patchwriting starts as lazy note-taking, and detectors flag it the same way graders do.
• Keep page numbers for every quote: APA 7 wants (Smith, 2022, p. 14), MLA 9 wants (Smith 14).
• File each source into the Citation Generator the day you open it, then spot-check two fields (year, journal title) against the database record — generators misread messy metadata more often than students expect.
• Before submission, reconcile the two lists: every in-text citation appears in the references, and every reference is cited somewhere. Mismatches are the most common five-point loss on an otherwise clean paper.

Revise in three passes

Revision fails when it tries to fix everything at once, so split it into three passes with different jobs. Pass one is argument: read only the first sentence of every paragraph, in order, and check that the sequence proves the thesis. If it does not, reorder paragraphs before touching a single sentence — moving furniture is cheaper than repainting it. Pass two is sentences: cut 10 percent of your word count, break the 60-word monsters, and vary the rhythm. Pass three is proofreading, done on paper or in a different font, because familiar screens hide familiar errors.

• Pass 1 — argument: the first sentences alone should read like a slide deck of your case; where they do not, the paragraph is misordered or redundant.
• Pass 2 — sentences: run the draft through the Readability Checker, then rewrite anything scoring below 20 on reading ease — those passages are overwritten, not advanced.
• Pass 3 — proofread: print it, or switch to a 14-point serif font; the new format exposes typos your brain has stopped seeing.
• Sleep between passes. Two passes on consecutive days catch more than four passes in one night, because distance does the work fatigue cannot.

Rubric weight explains why the order matters. Argument and organization typically carry about half the grade; sentence-level style, a quarter; mechanics and citation accuracy, the rest. Students with only two hours should revise backwards from that weight — fix paragraph order, then the reference list, then the commas. The commas were almost never the problem. Three passes work because they spend scarce attention where the rubric actually pays, and they leave the mechanical sweep for a brain too tired to argue. If you doubt it, grade your own draft with the rubric open; the missing points announce themselves.

A realistic hour budget for a 10-page paper

The honest total for a course research paper runs 15-40 hours across two to six weeks, and the spread has a logic. For a typical 10-pager: three hours to pick and narrow the question, eight to search and screen (roughly thirty abstracts read, twelve kept), two to outline, eight to ten for drafting, four for citations, and four for the three-pass revision. That is about 30 hours, which is three focused weeks at ten hours each — or one miserable, source-starved weekend. Same hours, opposite outcomes. The variable that changes everything is start date, not speed.

Put the schedule somewhere you will obey it. The Deadline Planner on this page takes your due date and assigns the phases — question, sources, outline, draft, revise — as dated checkpoints, which converts a scary 30-hour blob into five evenings a week. Protect the last 20 percent of the calendar for revision no matter what. Papers die in the gap between 'finished drafting' and 'due tomorrow,' and that gap is exactly what the planner exists to defend. Treat a slipped checkpoint as information, not guilt; it tells you which phase was under-budgeted.