Scope It: The One-Paragraph Plan

You’ve mapped the conversation and found your gap. The fastest way to find out whether the study it points to holds together is to write its abstract before you run it. Not the introduction, not a proposal — the ~250-word structured abstract, with the sections filled in as plans. If you cannot state the design, the sample, the analysis, and the contribution in a paragraph, the gaps you hit are exactly the gaps a reviewer (or your future self, six months in) will hit too. Writing it early is cheap; discovering it late is not.

This is the logic behind Registered Reports, where a study is reviewed and accepted on its question and design before any data exist (Chambers & Tzavella, 2022), and behind pre-registration, where you timestamp the hypotheses and analysis plan up front so confirmatory and exploratory work stay separate (Nosek et al., 2018). You do not need to submit a Registered Report to steal its best move: judge the study on the plan.

The structured abstract, re-slotted for planning

Keep the front of a normal structured abstract exactly as-is. Then, because the results do not exist yet, replace the Results section with what you expect the study to be worth — and force that into two branches.

Section	What you write (as a plan)
Background / objective	The gap in one or two sentences, then the specific question or hypothesis.
Methods — design	Observational, experimental, design-based; qualitative, quantitative, or mixed.
Methods — sample	Who, how many, and how recruited. (Sample size is a planning decision — see below.)
Methods — measures / data	What you collect and how you will analyze it — the analysis plan, written before data.
Results → Expected contribution	What the field gets if the study works — a guideline, a measure, a theory, a design.
Impact if results are POSITIVE	The takeaway if the effect shows up / the hypothesis holds.
Impact if results are NEUTRAL or OPPOSITE	The takeaway if it does not — a null, a small effect, a surprise.

That last split is the whole point.

Both branches have to carry a takeaway

A study is fragile when it is only interesting if it “works.” If the honest answer in the Impact if NEUTRAL or OPPOSITE box is “then we learned nothing” or “then we don’t publish,” you have designed a coin flip, not a study. The best experiments and studies have a useful takeaway regardless of the direction of the outcome — a null that is interpretable is a finding, not a failure.

▸See it filled in — the CS1 debugging-prompt study

Objective. Does a reflective “what have you already tried?” prompt change how intro-CS students seek help while debugging?

Design. Convergent mixed methods — a quasi-experiment across two CS1 sections plus follow-up interviews.

Sample. ~120 students across two sections; 12 students interviewed.

Measures / analysis. Help-request rate (logistic regression) + reflexive thematic analysis of the interviews, joined in a single display.

Expected contribution. A concrete, reusable CS1 lab-design guideline about reflective prompts.

Impact if POSITIVE. The prompt raises productive help-seeking → ship the guideline and the prompt wording.

Impact if NEUTRAL or OPPOSITE. No change in the rate — but the interviews explain why students route around the prompt (it reads as a gate, not an offer), which is itself a CS-education contribution and a redesign brief.

Robust to outcome? Yes — the qualitative strand guarantees a takeaway even if the number is null.

To draft your own, work top-down and stop the moment a box is hard to fill — that difficulty is the finding:

For very early, still-fuzzy ideas, a lighter one-pager can come first: the Heilmeier Catechism — eight plain-language questions (what are you trying to do, in no jargon? how is it done today, and what are the limits? what’s new, and why will it work? who cares? what are the risks and the cost? how long? what are the mid-term and final “exams” for success?). See the DARPA Heilmeier Catechism. Once the answers stabilize, they collapse neatly into the abstract above.

Plan your analysis: qualitative, quantitative, or both

Your analysis method is a planning decision, made when you write the abstract — not something you reach for once the data are sitting there. Let the question pick the method — open the one that matches yours:

▸My question measures, compares, or generalizes — how often, how much, does X change Y

Lean quantitative. You need measures, a defensible sample size, and a pre-specified test — decide the analysis before you collect data. (Exploratory quantitative methods like factor analysis and clustering exist too, so “quantitative” is not always confirmatory.)

▸My question is about meaning, process, or mechanism — how or why something happens, what it is like

Lean qualitative. Rigor rests on trustworthiness and thick description, not p-values, so a well-run qualitative study stands on its own at small n. Decide the coding approach up front — see Choosing Your Path.

▸I need a number AND the reason behind it

Mixed methods — but only if you decide the integration up front: which question each strand answers, which data feed each, the sequence and priority, and how you’ll join them (a joint display). Otherwise you have two studies stapled together, not one.

The reason a qualitative element so often makes a study robust to outcome is methodological: qualitative findings do not depend on statistical significance, so pairing a qualitative strand with a quantitative one decouples the paper’s value from the p-value. When you integrate them in a joint display, you get one of three takeaways no matter how the numbers land — the strands confirm each other, expand on each other, or disagree (discordance) — and disagreement is frequently the most interesting result of all (Fetters et al., 2013). In an explanatory-sequential design, a surprising or null quantitative effect becomes the reason for the qualitative phase (“the number was flat — the interviews tell us why”), turning a dead end into a contribution.

This lesson owns deciding the approach, not doing it. For the how-to, cross-link out and do not re-derive it here:

• How to choose and run a qualitative approach (codebook vs. reflexive thematic analysis, and who does what) → Choosing Your Path.
• How much data is enough — saturation and information power, which is how you justify qualitative sample size on the timeline → Trustworthiness, Rigor & Saturation.
• Which statistical test, and writing the analysis plan before data → the Quantitative Analysis module. Integrating the strands → the Mixed Methods module.

Where this lives

Everything above maps one-to-one onto Tab 1 · Structured Abstract of the companion sheet: one row per section, the two impact branches side by side, and the Robust to outcome? flag that goes red when the neutral-or-opposite branch is empty. Fill it in before you touch the timeline — the plan is the input to the schedule, not the other way around.