How to Plan a Research Paper: Question, Hypothesis, Outline

Most of the time wasted writing a research paper is wasted right here, before we even open up a source. A blank document with 'Working Title' typed at the top is not a plan. Knowing how to plan a research paper means turning that general area into three specific things, in order: a question, an answer you can defend or test, and a structure that carries a reader from one to the other. We may want to study social media and student attention. But a general area isn't something you can test, argue, or organize a paper around. It's a waste of time to do all that without having those three things ready.

Every guide to this stage tells you to pick a topic, do some research, then make an outline, which is true and almost useless on its own. It's whether each step really helps to narrow the previous one that makes the difference between a plan that works and one that stalls in week three. A topic isn't a question. A question isn't automatically a hypothesis. A hypothesis and a thesis statement do different jobs, and a single paper only ever needs one of them.

The following article will take you through one topic, social media use and student attention, in detail from its beginnings as just an idea to an outline that is complete. We'll show you exactly where this splits into an empirical version with a hypothesis and an argumentative version with a thesis statement. We won't address narrower questions, such as how to write a title or where to find sources. What follows stays at the level of the whole sequence, decision by decision.

Turn a broad topic into a research question

The first thing you should do is write down your topic as a complete sentence that has a hole in it, like this: not a noun phrase. Social media and attention is a label. Does daily social media use affect sustained attention span in university students names a population, a variable you can measure, and an outcome, which is what makes it answerable rather than just interesting to think about.

Not every answerable question is worth a semester of work. Researchers use a five-part check, known by the acronym FINER, to sort a workable question from one that only sounds workable. It was proposed by Hulley and colleagues for clinical research, but it holds up for any empirical topic.

Criterion	What it asks	Applied to our example
Feasible	Can you actually get the participants, time, and access this needs	University students and attention tasks are both realistic to recruit and run
Interesting	Would the answer matter to you, your supervisor, or your field	Attention and screen habits are live concerns in education research right now
Novel	Does it confirm, extend, or challenge something already published, rather than repeat it	Most existing work looks at total screen time, not social media specifically
Ethical	Would an ethics board approve how you would study it	Self-reported use and a standard attention task raise no major concerns
Relevant	Does it connect to a real theoretical or practical question	Connects directly to how universities think about digital wellbeing policy

Running a draft question through a research question generator before you commit to it does the same check faster, flagging when a question is too broad to finish or too narrow to matter. Either failure mode costs you weeks you won't get back once you're three sources deep.

What type of question are you actually asking?

Most research questions fall into one of three shapes: descriptive, comparative, or relationship-based, and knowing which one you are asking determines what kind of evidence would even count as an answer. The same starting topic can be phrased as any of the three, and each version sends you toward a different study design.

• Descriptive: how much time do university students spend on social media daily, which needs only measurement, no comparison group
• Comparative: do first-year and final-year students differ in daily social media use, which needs two groups measured the same way
• Relationship-based: does daily social media use affect sustained attention span in university students, which needs a measured outcome linked back to the variable you are studying

Health and nursing research has its own version of this instinct, based on a five-part formula known as PICOT (population, intervention, comparison, outcome, and time). Our example translated into PICOT reads: in first-year students, does high daily social media use, compared with low use, affect sustained attention scores over one semester. The approach provides researchers with a built-in system for identifying which pieces to include.

How to Plan a Research Paper: Research Question or Hypothesis?

A research question stays open. A hypothesis closes it down into one specific, testable prediction, stated before you collect a single data point. This is the actual difference: 'does social media use affect attention' is a question anyone could investigate in several directions, while 'students who use social media more than three hours a day show shorter attention spans than students who use it less than one hour' is a claim your data can only support or fail to support.

Every testable hypothesis comes in pairs. The first, usually written as Ha, is the prediction you just made. In this case it says heavier use leads to shorter attention spans. The second, usually called the null hypothesis and written as H0, is the boring default your prediction has to beat. It says there's no difference in attention span between heavier and lighter users. A key point: your statistical test isn't designed to prove your prediction directly, but rather to reject or fail to reject the boring default it's to beat.

Writing both halves of the pair, not just the one you expect to find, is what a hypothesis generator forces you to do by asking for the outcome variable and the comparison explicitly. Skipping the null half is a common gap examiners flag in an early-stage proposal.

What if your paper does not have a hypothesis?

Plenty of strong research papers never test a hypothesis at all. A literature review, a policy analysis, or an argumentative paper is organized around a thesis statement instead, a claim you defend with evidence rather than a prediction you test with data. The two tools solve different problems, and reaching for the wrong one is why some papers feel like they are trying to prove something a hypothesis was never built to prove.

But let's say you took the same starting topic in this argumentative direction: Although social media use is routinely blamed for shortening students' attention spans, sleep disruption and overall screen time better explain the pattern, which matters for how universities design digital wellbeing policy. This is a position, not a prediction. It can be wrong, it can be argued against, and every paragraph that follows exists to defend it with sources rather than with your own data.

A thesis statement generator does the same job here that a hypothesis check does on the empirical side: it will not let you submit a topic sentence, a fact, or a question dressed up as a claim, all three of which read as a thesis until a reader tries to argue against them and finds nothing to push back on.

Finding sources before you outline

An outline built before you know what the literature actually says tends to collapse the moment real sources arrive. Work backward from two or three papers you already trust, ones cited everywhere in your area, rather than starting a search from a blank query box.

• Pull the reference list of each anchor paper and note which sources appear in more than one of them, since repetition across papers is a rough proxy for a source's real weight in the field
• Use your database's cited-by function to chain forward from each anchor paper to newer work that built on it
• Stop adding sources once new searches mostly return papers you have already found, which is a genuine signal rather than a productivity trick
• Keep a one-line note on what each source actually claims as you go, since re-reading twenty abstracts the night before your outline is due is where most citation errors start

How do you actually outline a research paper?

An outline for an empirical paper follows the shape of the argument, not the order you thought of things in, which is exactly what an essay outline generator can enforce automatically once you feed it your research question and hypothesis. In the hypothesis version of our example, that means four blocks, each answering a different question a reader will actually have.

Section	What it has to do
Introduction	State the gap in existing research, then the research question and hypothesis, in that order
Methods	Describe participants, how attention span was measured, and the procedure in enough detail to repeat it
Results	Report what the data showed, including findings that complicate the hypothesis, not only the ones that support it
Discussion	Interpret what the result means, name its limits, and say what it does not show

If you search for an APA outline format, most of what you find online is actually describing APA's heading rules for a finished paper, five levels, each with its own combination of bold, italics, and centering, not the numbered outline you sketch in advance. APA's own style guide is explicit that headings in the final paper should not be labeled with numbers or letters at all. Use whichever numbered or lettered system helps you plan; just do not carry the numbering into the manuscript itself.

Writing the title, last

The title is often written before the outline, but should be written last. It's also likely to promise a paper you haven't written yet. Make sure it's shorter than you think it needs to be. A widely cited analysis of the most-cited papers published between 2007 and 2013 found that journals publishing shorter titles received more citations per paper, a relationship that held across the whole period even though it weakened at the level of individual papers in later years.

A title generator can produce a dozen candidates from your finding in the time it takes to write one yourself, which is useful less for the final wording than for seeing your own paper's contribution stated a dozen different ways before you pick the sharpest one.

Planning ends with an outline; drafting starts there. Drafting also has a way of going wrong: you produce a first draft that's stiffer than you intended. This post introduces the kind of AI humanizer you should know about if you ever need one. An AI humanizer built for academic writing is a tool for that later stage, worth knowing about now so you don't reach for it before you've anything worth revising.