RESEARCH / BLOG

Research,
Explained.

Read the research question, follow the authors’ methods and explore the evidence behind the article.

Research Blog

Research methodology · Research insights

From Research Gap to Question, Objectives and Hypothesis

Align a research gap, research question, objectives and methods. Use PICO, SPIDER or an appropriate engineering framework to prepare a coherent PhD research proposal.

By NPT Research7 min read
The novelty toolkit · S204 / 09
Connect the question to the test.A credible proposal makes its evidence path visible.
NPT Research

The Novelty Toolkit · Series 2 · Part 4 of 9

A navigator may know that a region is poorly mapped. That does not yet tell them where to sail. Likewise, discovering a research gap does not automatically produce an excellent research question. The gap describes an uncertainty; the question selects a tractable part of it; the objectives organize the work needed to answer it.

This tutorial shows how to move from 'we do not know enough' to a coherent study plan. We will use one hypothetical field-monitoring example throughout.

Treat the research gap as a map, not a destination

Suppose your literature review suggests that existing studies have not adequately distinguished temperature-driven error from ageing-driven error over a specified outdoor deployment period. That is a candidate evidence gap. It tells you where uncertainty may exist, but you still need to decide whether your study will estimate effects, test an explanation, compare methods or explore experiences.

A weak question is 'How can we build the best smart water sensor?' It assumes the solution before clarifying the problem. A stronger quantitative question is: 'Among defined low-cost sensors, how does the error relative to a reference method vary with ambient temperature after accounting for sensor age and calibration schedule?'

A good question contains enough structure that a reader can imagine what evidence would count as an answer.

Select a question framework that fits the design

PICO/PICOT can help structure comparative intervention questions: population/problem, intervention, comparison, outcome and optionally time. It is not mandatory for all engineering or observational research. SPIDER can help frame qualitative evidence questions using sample, phenomenon of interest, design, evaluation and research type (Gutiérrez Ubeda’s research-question framework).

For a descriptive engineering study, a simpler question structure may be more appropriate: system + condition + variable + comparison/reference + outcome. Avoid forcing every project into a clinical-trial template.

The novelty toolkit · Series 2Build a study whose parts answer each other.Start with the uncertainty and work towards the evidence.
  1. 01
    GapIdentify a supported unresolved issue.
  2. 02
    QuestionState the specific inquiry.
  3. 03
    ObjectivesDescribe what will be estimated or compared.
  4. 04
    EvidenceChoose measurements or source materials.
  5. 05
    DesignPlan comparisons and analysis that answer the question.
A learning framework, not a measured result. Verify evidence before making a research claim.
Figure 1. A research gap must be translated into aligned decisions before a design becomes credible.

Draft one main question and supporting questions

Main question: 'What proportion of observed measurement error under defined field conditions is explained by ambient temperature and elapsed time since calibration?'

Supporting question A: 'Does the size of the temperature-associated error differ between sensor models?'

Supporting question B: 'Is error reduced after an independently defined recalibration procedure?'

Supporting questions should logically support the main question; they should not introduce unrelated projects. Define the specific sensor, comparison standard and deployment conditions in the protocol.

Convert questions into objectives with measurable verbs

A research objective should specify what will be estimated, compared, described or explained. Use verbs such as quantify, compare, test, estimate, characterize or evaluate. Avoid vague verbs such as 'study', 'understand better' or 'make advanced'.

Example objective 1: Estimate agreement between low-cost sensor readings and reference measurements at pre-specified time points.

Objective 2: Model the association of temperature and elapsed calibration time with measurement error, accounting for repeated measures.

Objective 3: Compare the performance of a pre-specified calibration strategy against a baseline using data not used to fit the correction.

The third objective adds an important safeguard: a method that performs well on the same data it was trained on may fail when evaluated on independent measurements.

Know when a hypothesis is appropriate

A hypothesis states an expectation that can be confronted with data. It is particularly useful in explanatory, comparative and experimental studies. Qualitative, exploratory and some design-science projects may instead use research questions, propositions or design requirements, depending on their epistemology.

For the example, a directional hypothesis might be: 'After accounting for elapsed time since calibration, increases in temperature are associated with systematic changes in measurement error.' A corresponding null hypothesis might state that the specified association is absent under the model. Whether a formal hypothesis test is suitable depends on the design, sample size and analysis plan.

Do not convert every objective into a forced 'significant difference' hypothesis. Statistical significance cannot compensate for invalid measurements or a poorly specified comparison.

The novelty toolkit · Series 2Choose a framework that fits the inquiry.No single template suits every research discipline.
  1. 01PICO or PICOTUseful for comparative intervention questions.
  2. 02SPIDERA structure for qualitative evidence questions.
  3. 03Engineering structureSystem, condition, variable, reference and outcome.
  4. 04FINERCheck feasibility, interest, novelty, ethics and relevance.
A learning framework, not a measured result. Verify evidence before making a research claim.
Figure 2. Match the question framework, worth-answering criteria, and design before collecting evidence.

Work backward from the evidence needed

Ask what observation would answer the question. For a sensor-comparison study, that may require a traceable reference measurement, repeated readings, temperature records, calibration history, sensor identifiers and planned handling of missing observations.

Next ask what could bias the result: changing water chemistry, instrument ageing, operator procedures and dependence between repeated measurements. If you cannot separate the suspected mechanism from major alternatives, revise either the question or design.

Finally apply FINER—feasible, interesting, novel, ethical and relevant (Stanford’s research-question guidance). A perfectly formed sentence can still fail the feasibility test if no appropriate reference measurements are available.

The one-page alignment table

Use five rows in your planning document: (1) evidence-based gap, (2) principal question, (3) objectives, (4) expected measurements or source materials, and (5) planned design and analysis. Read across each row. If an objective has no measurement, it is currently aspirational. If a planned analysis does not answer an objective, it may be unnecessary.

This alignment should appear in the research proposal before expensive equipment or data collection decisions are finalized.

A ten-minute writing challenge

Start with: 'It is unclear whether X varies with Y under Z conditions.' Draft one research question, two specific objectives, and either a testable hypothesis or an explicit reason why an exploratory question is more appropriate. Ask a colleague to underline every term that could be measured, observed or independently checked. Rewrite the rest until the evidence path is visible.

How misalignment produces weak dissertations

Consider this pair: 'Objective: improve the accuracy of low-cost sensors' and 'Method: survey users about their confidence.' The survey might measure perceived reliability, but it cannot directly establish analytical accuracy. The objective and evidence do not match. Either change the objective to users' perceptions or plan a measurement study with an appropriate reference method.

Another mismatch: 'Hypothesis: temperature causes drift' and 'Design: cross-sectional correlation between temperature and error.' Association alone may not establish causation. A stronger causal claim may require experimental manipulation or careful identification of alternative explanations. Choose words proportional to what the design can support.

Different disciplines, different legitimate outputs

A humanities or qualitative study may produce an interpretive argument rather than a numerical hypothesis. A theoretical study may establish a new derivation or proof. Design-science research may evaluate an artefact against explicit requirements. A systematic review may synthesize evidence rather than conduct new field measurements. Each can contribute knowledge when its question, method and evaluation standard are aligned.

The practical check is not 'Does every study have H0 and H1?' It is 'Can the reader see what the study will contribute, how it will be evaluated, and what would count as a credible result?'

A supervisor-review exercise

Write a one-page proposal synopsis with five headings: (1) established knowledge, (2) unresolved uncertainty, (3) main question, (4) two objectives, (5) evidence and methods. Ask a supervisor or peer to draw a line from every objective to its associated measurement or source material. Any objective with no line needs revision; any measurement with no purpose needs justification. Finally, ask which assumption is most likely to fail under real study conditions.

What to remember

A research question is a promise about what evidence your study can deliver. The gap justifies the journey, the question sets the destination, the objectives define the route, and the methods determine whether you can get there. This completes the four human-led tutorials. The next five introduce AI strictly as an assistant to this reasoning—not as its owner.

Methodological context and limits

Cooke and colleagues’ SPIDER paper introduces a search-question structure for qualitative and mixed-methods evidence synthesis and calls for wider testing. Farrugia and colleagues’ question–hypothesis–objective guide discusses alignment in clinical studies. The engineering example adapts the reasoning; it is not an empirical validation of either framework.

Continue the research novelty series

Series 1: originality, evidence and scientific judgment

Begin with what makes research novel, then find the closest prior research and build a novelty matrix. Before using an assistant, examine the limits of AI novelty judgments and why the researcher owns the question. Develop the conversation with scientific prompting and six-round brainstorming, then challenge the idea against evidence. The complete worked research dialogue connects those decisions.

Series 2: from a problem to a defensible proposal

First separate a research gap from a research problem and identify a genuine problem worth investigating. Next search published literature for the gap and align the question, objectives and hypothesis. For AI-assisted work, clarify the problem without outsourcing judgment, verify the literature search and its references and compare papers in an evidence matrix. Then stress-test the research hypothesis before following the complete observation-to-proposal workflow.

Sources and further reading

  1. Stanford University. Selecting Your Research Question: FINER and study planning. Institutional guidance.
  2. Gutiérrez Ubeda. How to Build and Assess the Quality of Healthcare-Related Research Questions. Global Journal on Quality and Safety in Healthcare 5, 39–43 (2022). Published 2 Jun 2022
  3. Cooke, Smith & Booth. Beyond PICO: The SPIDER Tool for Qualitative Evidence Synthesis. Qualitative Health Research 22, 1435–1443 (2012). Published 24 Jul 2012
  4. Farrugia et al. Research questions, hypotheses and objectives. Canadian Journal of Surgery 53, 278–281 (2010).