Research – Definition, Types, Methods, Process and Examples

Research is a systematic and critical process used to answer questions, investigate problems, test ideas, interpret evidence, or create new knowledge. It involves selecting appropriate sources or methods, collecting or examining evidence, analysing it transparently, and drawing conclusions that are justified by the evidence and the limitations of the study.

Introduction

Research is central to academic study, scientific discovery, policy development, business decisions, healthcare, education, technology, and public understanding. Students use research to investigate questions and complete assignments. Scholars use it to develop and test explanations. Organisations use it to evaluate programmes, understand users, improve products, and make evidence-informed decisions.

However, research is more than searching online, collecting quotations, or gathering large amounts of information. Formal research connects a clear question to an appropriate method, a reasoned analysis, and a transparent conclusion.

This guide explains:

• What research means.
• What distinguishes formal research from ordinary information gathering.
• The main types of research.
• How methods, methodology, design, and approach differ.
• How the research process works.
• How to choose a suitable method.
• What makes research trustworthy and ethical.
• How open science, digital tools, and artificial intelligence affect modern research.

Key Takeaways

• Research is a systematic investigation of a defined question, problem, phenomenon, text, theory, or body of evidence.
• A study can belong to several research categories at the same time.
• The research question should guide the design and methods, not the other way around.
• Good research is transparent about its evidence, reasoning, uncertainty, and limitations.
• Ethics and research integrity apply throughout the project, not only during data collection.
• AI can assist selected research tasks, but human researchers remain responsible for accuracy, decisions, disclosure, and authorship.

What Is Research?

Research is a planned process of inquiry undertaken to develop understanding, answer a question, solve a problem, evaluate a claim, or contribute knowledge. It may involve experiments, surveys, interviews, observations, documents, datasets, texts, artefacts, simulations, proofs, or existing studies.

The exact form of research varies across disciplines. A laboratory scientist may manipulate variables. A historian may evaluate archival sources. A sociologist may conduct interviews. A computer scientist may develop and test an algorithm. A literary scholar may interpret texts. A mathematician may construct a proof.

Research in Simple Words

In simple terms, research means:

Asking a worthwhile question and using an organised, appropriate, and transparent process to develop a justified answer.

The answer may be provisional rather than final. Research often reduces uncertainty without eliminating it completely.

Everyday Information Searching Versus Formal Research

People commonly say they are “doing research” when comparing phones, reading news, or searching for travel information. These activities involve information gathering, but they are not always formal academic research.

Everyday information searching	Formal research
May begin with a broad personal interest	Begins with a defined problem, question, or objective
Often uses convenient sources	Uses a planned and justified source or data strategy
May stop when a useful answer is found	Continues until the evidence has been systematically examined
Rarely documents every decision	Records methods, criteria, and analytical decisions
Usually serves an immediate personal need	Seeks a defensible contribution, explanation, evaluation, or synthesis
May not assess uncertainty or bias	Examines limitations, bias, alternative explanations, and uncertainty
Usually does not require formal review	May require ethical, institutional, legal, or professional review

Everyday searching can become part of formal research when it is conducted systematically and linked to an explicit research design.

What Makes an Activity Research?

An activity is more likely to qualify as formal research when it includes the following elements:

1. A defined focus: The researcher identifies a question, problem, objective, proposition, or phenomenon.
2. Engagement with existing knowledge: The project considers relevant literature, evidence, theories, or prior work.
3. A systematic process: Sources, cases, participants, observations, or data are selected through stated procedures.
4. Appropriate evidence: The evidence is capable of addressing the research question.
5. Critical analysis: Information is not merely collected; it is evaluated, compared, interpreted, modelled, or tested.
6. Transparency: Important decisions and limitations are explained.
7. An intended contribution: The project adds evidence, interpretation, synthesis, application, method, theory, or understanding.
8. Communication: Findings are shared in a form that others can examine.

Not every project must produce a revolutionary discovery. Replication, verification, reinterpretation, evidence synthesis, and application to a new context can all make valuable contributions.

Academic, Policy, and Regulatory Definitions

The OECD defines research and experimental development as creative and systematic work intended to increase knowledge and develop new uses of existing knowledge. It identifies novelty, creativity, uncertainty, systematic planning, and transferability or reproducibility as important criteria for research and development activity (OECD, 2015).

In United States human-subject regulation, research has a narrower legal meaning: a systematic investigation designed to develop or contribute to generalisable knowledge. That definition helps institutions determine whether particular activities fall under human-research rules. It should not be treated as the only definition applicable to every discipline or country (U.S. Department of Health and Human Services [HHS], n.d.).

Why Is Research Important?

Research is important because it provides a disciplined way to investigate claims rather than relying only on assumption, tradition, authority, or personal experience.

Its major purposes include:

Expanding Knowledge

Basic or fundamental research develops theories, explanations, concepts, measurements, and observations without requiring an immediate practical application.

Solving Practical Problems

Applied research investigates defined needs, such as improving an educational programme, reducing manufacturing defects, evaluating a public-health intervention, or designing a more accessible digital service.

Supporting Decisions

Research allows policymakers, organisations, practitioners, and individuals to assess alternatives using evidence. Evidence does not make decisions automatically, but it can clarify likely benefits, harms, costs, uncertainties, and trade-offs.

Testing Existing Claims

Researchers may examine whether earlier findings can be reproduced, replicated, extended, or applied in another setting.

Challenging Assumptions

Research can reveal that a widely accepted belief is incomplete, context-dependent, poorly measured, or unsupported.

Preserving and Interpreting Knowledge

Historical, archival, linguistic, archaeological, legal, and humanities research helps societies understand documents, events, cultures, ideas, and creative works.

Developing Methods and Tools

Some research produces new instruments, algorithms, datasets, analytical methods, conceptual frameworks, software, or research protocols.

Characteristics of Good Research

Good research is not defined by one method. Its quality depends on whether its question, evidence, design, analysis, and conclusions are appropriately aligned.

Characteristic	What it means	A useful quality-control question
Clear	The problem, question, and objectives are understandable	What exactly is being investigated?
Systematic	Important procedures follow a reasoned plan	How were sources, cases, or data selected?
Appropriate	The design fits the question and context	Can this method produce the required evidence?
Evidence-based	Conclusions are linked to observations, sources, data, or reasoned argument	What evidence supports each major claim?
Critical	Alternative explanations and weaknesses are considered	What else could explain the findings?
Ethical	People, communities, animals, environments, knowledge, and data are treated responsibly	What risks or unfair consequences could arise?
Transparent	Decisions, exclusions, assumptions, and limitations are reported	Could another reader understand what was done?
Coherent	Questions, theory, methods, analysis, and conclusions fit together	Does each stage support the next?
Reflexive	The researcher considers how assumptions and position may affect the work	How might the researcher influence interpretation?
Proportionate	Claims do not go beyond the evidence	Is the conclusion stronger than the design permits?
Reusable where possible	Data, code, materials, or procedures can be examined or reused when ethically and legally appropriate	What could responsibly be shared?
Open to correction	Findings can be questioned, updated, or tested	What evidence would change the conclusion?

What Are the Main Types of Research?

Research can be classified by its application, purpose, approach to evidence, data source, design, time frame, reasoning, and relationship to observation. These categories overlap. A single study may be applied, explanatory, mixed-methods, longitudinal, and quasi-experimental at the same time.

Research Types at a Glance

Classification dimension	Main types
Application	Basic, applied, translational
Purpose	Exploratory, descriptive, correlational, explanatory, evaluative, predictive
Evidence approach	Qualitative, quantitative, mixed methods
Data source	Primary, secondary
Design	Experimental, quasi-experimental, observational, case-based, review-based
Time	Cross-sectional, longitudinal, prospective, retrospective
Reasoning	Deductive, inductive, abductive
Evidence relationship	Empirical, non-empirical, theoretical, formal
Researcher–practice relationship	Action research, participatory research, evaluation research

Types of Research by Application

Basic Research

Basic research seeks to expand knowledge or theory without requiring an immediate practical use.

Example: Investigating how human working memory represents sequences.

Basic research can later support practical innovation, even when its original purpose was theoretical.

Applied Research

Applied research addresses a particular real-world problem.

Example: Testing whether a revised classroom feedback system improves assignment completion.

Applied research still requires theoretical and methodological justification. “Practical” does not mean methodologically informal.

Translational Research

Translational research moves findings, methods, or discoveries between stages of development and practical use.

Example: Adapting a laboratory diagnostic technique for clinical testing and eventual implementation.

The term is especially common in health and biomedical fields, but the broader idea also appears in engineering, education, and public policy.

Types of Research by Purpose

Exploratory Research

Exploratory research investigates a topic that is not yet sufficiently understood. It helps identify concepts, possible explanations, variables, stakeholder concerns, or future questions.

Example: Interviewing remote workers to explore emerging forms of digital workplace isolation.

Exploratory findings should not be presented as definitive population estimates unless the design supports such claims.

Descriptive Research

Descriptive research documents what exists, how often it occurs, or what characteristics it has.

Example: Estimating the proportion of students who use public transport to reach a university campus.

Descriptive research answers questions such as “what,” “who,” “where,” “when,” and “how much.”

Correlational Research

Correlational research examines whether variables are associated.

Example: Examining the relationship between weekly study time and examination scores.

An association does not by itself demonstrate that one variable causes the other.

Explanatory Research

Explanatory research investigates why or how an outcome occurs. It may test causal mechanisms, compare explanations, or develop theory.

Example: Examining whether feedback timing influences learning by changing how quickly students correct misunderstandings.

Strong causal claims require a design capable of addressing confounding, selection effects, reverse causation, and alternative explanations.

Evaluative Research

Evaluative research assesses the design, implementation, effectiveness, efficiency, relevance, or consequences of a policy, programme, service, or intervention.

Example: Evaluating whether a university mentoring programme reaches intended participants and improves retention.

An evaluation may be primarily local, while formal research often aims to produce broader knowledge. The boundary depends on purpose, design, governance, and applicable institutional rules.

Predictive Research

Predictive research develops or tests models intended to forecast an outcome.

Example: Developing a model to estimate which equipment components are most likely to fail within 30 days.

Prediction and explanation are not identical. A model may predict accurately without identifying the true causal mechanism.

Types of Research by Approach to Evidence

Qualitative Research

Qualitative research examines meanings, experiences, interpretations, processes, language, practices, relationships, or context.

Common methods include:

• Interviews.
• Focus groups.
• Participant observation.
• Ethnography.
• Case studies.
• Document analysis.
• Discourse analysis.
• Narrative analysis.
• Thematic analysis.
• Grounded-theory approaches.

Example: Exploring how first-generation university students experience academic belonging.

Qualitative research is not defined merely by small samples or the absence of numbers. Its defining features concern the type of question, evidence, interpretation, and knowledge sought.

Quantitative Research

Quantitative research uses numerical measurement and mathematical or statistical analysis to describe patterns, estimate effects, test hypotheses, compare groups, or make predictions.

Common methods include:

• Structured surveys.
• Experiments.
• Quasi-experiments.
• Cohort studies.
• Analysis of administrative records.
• Measurement studies.
• Statistical modelling.
• Simulation.

Example: Estimating the average difference in test performance between students using two teaching approaches.

Quantitative research is not automatically objective or generalisable. Measurement quality, sampling, missing data, model assumptions, and design limitations still matter.

Mixed-Methods Research

Mixed-methods research intentionally integrates qualitative and quantitative evidence within one study or programme of inquiry.

Example: A researcher first surveys employees about job satisfaction and then interviews selected participants to explain unexpected patterns.

Simply including an open-ended question in a survey does not necessarily create a rigorous mixed-methods design. The two forms of evidence should be connected at the design, analysis, or interpretation stage.

Types of Research by Data Source

Primary Research

Primary research generates or collects evidence specifically for the current project.

Examples include:

• Conducting interviews.
• Administering a new survey.
• Running an experiment.
• Recording observations.
• Collecting specimens or measurements.
• Creating a new corpus or dataset.

Secondary Research

Secondary research analyses evidence that already exists.

Examples include:

• Analysing census data.
• Reusing an archived dataset.
• Conducting a systematic review.
• Examining historical documents.
• Synthesising published studies.
• Analysing organisational records.

Secondary-data research can be original. Originality may lie in the question, dataset combination, analytical method, theoretical interpretation, time period, or comparison rather than in collecting new data.

Types of Research by Design

Experimental Research

Experimental research deliberately changes an independent variable and observes its effect on an outcome.

Random assignment can help create comparable groups and strengthen causal inference. Experiments are not always feasible or ethical.

Quasi-Experimental Research

Quasi-experimental research evaluates an intervention or exposure without full random assignment.

Examples include:

• Interrupted time-series studies.
• Regression-discontinuity designs.
• Natural experiments.
• Difference-in-differences analyses.
• Matched comparison-group studies.

A quasi-experiment can support stronger causal reasoning than a simple association study when its assumptions are credible.

Observational Research

Observational research examines conditions without assigning the exposure or intervention.

Examples include:

• Cross-sectional surveys.
• Cohort studies.
• Case–control studies.
• Naturalistic observation.
• Archival analysis.

Observational studies are essential when experiments would be impossible, unethical, or too artificial.

Case Study Research

A case study investigates a bounded case in depth, such as an organisation, event, community, project, person, decision, or programme.

Case studies may use qualitative, quantitative, or mixed evidence. A case study is a research strategy, not simply a short description of an example.

Review Research

Review research systematically or critically examines existing studies or sources.

Forms include:

• Narrative reviews.
• Scoping reviews.
• Systematic reviews.
• Rapid reviews.
• Umbrella reviews.
• Meta-analyses.
• Qualitative evidence syntheses.

The review question and intended use should determine the review type.

Types of Research by Time Frame

Cross-Sectional Research

Cross-sectional research examines a population, sample, corpus, or phenomenon at one point or during a limited period.

It is useful for estimating current characteristics or relationships but usually offers limited evidence about change over time.

Longitudinal Research

Longitudinal research collects or examines evidence across multiple time points.

It can study:

• Development.
• Trends.
• Durability of effects.
• Sequences of events.
• Changes within individuals, organisations, or societies.

Prospective Research

Prospective research specifies a plan and follows events forward from a defined starting point.

Retrospective Research

Retrospective research examines records or evidence relating to events that have already occurred.

Types of Research by Reasoning

Deductive Research

Deductive research begins with a theory, model, or proposition and derives expectations that can be examined using evidence.

A simplified sequence is:

1. Start with a theory.
2. Form a hypothesis or expectation.
3. Collect or examine evidence.
4. Assess whether the evidence is consistent with the expectation.
5. Revise the explanation when necessary.

Inductive Research

Inductive research develops concepts, patterns, or theory from detailed observations or evidence.

A simplified sequence is:

1. Gather or examine evidence.
2. Identify patterns.
3. Develop concepts or categories.
4. Construct a broader interpretation or theory.

Induction does not mean that researchers begin without prior knowledge. Existing theory and assumptions still influence what is observed and how it is interpreted.

Abductive Research

Abductive reasoning moves between evidence and possible explanations to identify the most plausible account.

It is especially useful when researchers encounter surprising or incomplete findings.

Empirical and Non-Empirical Research

Empirical Research

Empirical research draws conclusions from observation, experience, measurement, documents, records, experiments, or other evidence about the world.

Non-Empirical Research

Non-empirical research develops knowledge primarily through concepts, logic, theory, interpretation, formal proof, or critical argument.

Examples include:

• Philosophical analysis.
• Mathematical proof.
• Theoretical modelling.
• Legal doctrinal analysis.
• Conceptual research.
• Some forms of literary criticism.

Empirical and non-empirical approaches can complement each other.

Research Approach, Design, Methodology and Methods

These terms are related but not interchangeable.

Term	Meaning	Example
Research paradigm or philosophy	Assumptions about knowledge, reality, values, and appropriate inquiry	Critical realism, constructivism, pragmatism
Research approach	Broad orientation to reasoning or evidence	Qualitative, quantitative, mixed methods, inductive, deductive
Methodology	The rationale and principles explaining why a research strategy and methods are appropriate	A justification for using interpretive phenomenology
Research design	The overall plan connecting the question to evidence and inference	Cross-sectional survey, randomised trial, longitudinal case study
Research method	A procedure for collecting, generating, or analysing evidence	Interviewing, observation, regression analysis
Research technique or instrument	The specific tool or operational procedure	Interview guide, questionnaire, sensor, coding framework

A coherent study aligns all these levels. For example, a researcher should not select a survey simply because it is familiar. The survey must fit the question, population, concepts, sampling strategy, measurement requirements, and intended claims.

The Research Process Step by Step

Research is often presented as a sequence, but it is usually iterative. Researchers may revise a question after reading the literature, modify an instrument after a pilot study, or reconsider an explanation after examining unexpected findings.

1. Identify a Topic or Problem

Begin with a broad area that is important, interesting, feasible, and relevant to a field or community.

A broad topic such as “social media and students” is not yet a research problem. The researcher must determine what remains uncertain and why it matters.

2. Conduct Preliminary Scoping

Read introductory and recent literature to learn:

• How the topic is defined.
• Which theories are used.
• What evidence already exists.
• Which populations and settings have been studied.
• Where experts disagree.
• What data or sources may be available.
• Which ethical or practical constraints apply.

Scoping prevents the researcher from designing a study around a question that has already been adequately answered or cannot be investigated with available resources.

3. Define the Research Problem

The research problem explains the specific uncertainty, contradiction, practical difficulty, evidence gap, or conceptual issue the project will address.

A useful problem statement identifies:

• What is known.
• What is not known.
• Why the missing knowledge matters.
• Who or what is affected.
• What kind of evidence is needed.

4. Formulate Questions, Objectives, or Hypotheses

A research question states what the study seeks to understand.

An objective states what the researcher will do.

A hypothesis states an expectation that can be examined.

Not all research requires a hypothesis. Exploratory, interpretive, historical, theoretical, and qualitative studies may be guided by open research questions rather than predicted relationships.

5. Review the Literature Systematically

The literature review should do more than list previous studies. It should:

• Define key concepts.
• Compare findings and methods.
• Identify patterns and disagreements.
• Assess the quality and relevance of existing evidence.
• Locate the current study within the field.
• Justify the new project.

The scope and search procedures should be proportionate to the project. A systematic review requires a more comprehensive and reproducible search than the background section of a small student project.

6. Develop a Conceptual or Theoretical Framework

A framework clarifies the ideas, variables, relationships, mechanisms, assumptions, or interpretive perspective guiding the study.

Not every project requires a formal named theory. However, every study benefits from explaining how its central concepts are understood.

7. Choose the Research Design and Methods

The researcher decides:

• What type of evidence is required.
• Where it will come from.
• Who or what will be included.
• How cases or sources will be selected.
• How concepts will be measured or interpreted.
• How evidence will be analysed.
• What comparisons will be made.
• What conclusions the design can and cannot support.

8. Plan Ethics, Governance and Data Management

Before collecting or accessing data, determine:

• Whether ethical or institutional approval is required.
• How consent will be obtained when applicable.
• How privacy and confidentiality will be protected.
• How risks will be minimised.
• Who may access the data.
• How files will be named, stored, secured, backed up, retained, and destroyed.
• Whether data, code, or materials can later be shared.
• How conflicts of interest will be managed.

Ethical approval cannot make a poorly designed study ethical. A project that exposes participants to inconvenience or risk but cannot answer its question may still be unethical.

9. Pilot the Procedures

A pilot or feasibility study can reveal:

• Confusing survey questions.
• Recruitment difficulties.
• Technical failures.
• Inadequate measurements.
• Excessive participant burden.
• Problems with coding or analysis.
• Unrealistic timelines.

Pilot data should not automatically be combined with the final study unless this was planned and methodologically justified.

10. Collect or Generate Evidence

Follow the approved protocol while documenting:

• Recruitment and exclusions.
• Deviations from the plan.
• Missing data.
• Instrument changes.
• Fieldwork decisions.
• Software and equipment versions.
• Contextual events that may affect interpretation.

Consistency matters, but researchers should also respond appropriately to ethical or practical problems.

11. Prepare and Analyse the Evidence

Preparation may include:

• Data cleaning.
• Transcription.
• Anonymisation.
• Coding.
• Creating variables.
• Checking assumptions.
• Organising documents.
• Verifying source metadata.
• Recording exclusions.

The analysis should match the research question and design. Researchers should distinguish analyses planned in advance from exploratory analyses developed after seeing the data.

12. Interpret the Findings

Interpretation asks:

• What do the findings mean?
• How do they answer the question?
• How do they compare with previous work?
• Which alternative explanations remain?
• How does context affect the findings?
• What is uncertain?
• What claims are justified?
• What should not be concluded?

Statistical significance, thematic frequency, or a visually striking pattern is not sufficient by itself. Importance must be judged in relation to effect size, context, quality, theory, uncertainty, and practical consequences.

13. Report the Research Transparently

A research report should explain:

• Why the study was needed.
• What question was asked.
• What methods were used.
• What evidence was obtained.
• How it was analysed.
• What was found.
• What the findings mean.
• What limitations remain.
• What materials are available.

Design-specific reporting guidelines can help researchers report essential information. Examples include PRISMA 2020 for systematic reviews and CONSORT 2025 for randomised trials (Page et al., 2021; Hopewell et al., 2025).

14. Share, Evaluate and Extend the Work

Research may be communicated through:

• Journal articles.
• Theses and dissertations.
• Books.
• Conferences.
• Reports.
• Datasets.
• Code repositories.
• Policy briefs.
• Exhibitions.
• Public-engagement materials.
• Educational resources.

Peer review and publication are quality-control mechanisms, not guarantees that a study is correct. Research remains open to criticism, correction, replication, reinterpretation, and updating.

Worked Research Example

Consider the broad topic:

Sleep and academic performance among university students.

Step 1: Narrow the Question

A quantitative question might be:

Is average nightly sleep duration associated with first-semester grade-point average among first-year university students?

A qualitative question might be:

How do first-year students describe the ways their sleep routines affect studying and classroom participation?

A mixed-methods question might be:

What relationship exists between sleep duration and academic performance, and how do students explain the routines and pressures associated with that relationship?

Step 2: Select a Design

For the quantitative question, a researcher might use a longitudinal observational design:

• Recruit first-year students.
• Measure sleep repeatedly during the semester.
• Obtain academic outcomes with permission.
• Measure possible confounders such as employment hours, health, course load, and prior attainment.
• Analyse the association while reporting uncertainty.

For the qualitative question, the researcher might use semi-structured interviews and thematic analysis.

For the mixed-methods question, the researcher might analyse the quantitative pattern first and then interview selected students to explain it.

Step 3: Avoid Overclaiming

Even if sleep duration is associated with grades, an observational study may not prove that increasing sleep will cause grades to improve. Stress, health, employment, socioeconomic conditions, and academic workload could influence both sleep and performance.

A causal question might require an ethically acceptable intervention or a stronger quasi-experimental design.

This example demonstrates why the research topic alone does not determine the method. The precise question and intended claim do.

How to Choose the Right Research Method

The best method is the one that can produce credible evidence for the research question within ethical, practical, and disciplinary constraints.

Research goal	Potentially suitable designs or methods	Important caution
Estimate prevalence or describe a population	Representative survey, census, administrative records	Sampling and non-response affect estimates
Understand experiences or meanings	Interviews, focus groups, ethnography, diaries	Researcher interpretation and context must be addressed
Test an intervention’s causal effect	Randomised experiment	Feasibility, ethics, adherence, and generalisability matter
Estimate a causal effect without randomisation	Natural experiment, interrupted time series, regression discontinuity, matched comparison	Each design depends on assumptions that must be tested and defended
Examine associations	Correlational study, cohort study, regression analysis	Association is not automatically causation
Investigate change	Longitudinal panel, repeated measures, time series	Attrition and historical events may affect findings
Study a bounded case deeply	Case study using multiple evidence sources	Explain case selection and limits of transfer
Develop theory from evidence	Grounded-theory or iterative qualitative approaches	Procedures and researcher decisions should be transparent
Evaluate a programme	Process evaluation, outcome evaluation, mixed methods	Define success criteria and stakeholder perspectives
Synthesise existing studies	Systematic review, scoping review, meta-analysis, qualitative synthesis	Search, selection, appraisal, and synthesis methods must fit the question
Examine texts or discourse	Content, discourse, rhetorical, narrative, or textual analysis	Explain the corpus, interpretive framework, and context
Test a computational method	Benchmarking, simulation, ablation studies, external validation	Dataset leakage and inappropriate benchmarks can mislead

Before finalising a method, ask:

1. What claim do I want to make?
2. What evidence would justify that claim?
3. Can the evidence be obtained ethically?
4. Can the relevant population, cases, texts, or data be accessed?
5. What sources of bias are most likely?
6. What skills, time, software, equipment, and funding are required?
7. What will the design still be unable to establish?

Evidence, Data, Sampling and Measurement

Evidence Is Broader Than Numerical Data

Research evidence can include:

• Measurements.
• Survey responses.
• Interview transcripts.
• Observations.
• Images.
• Audio and video.
• Documents.
• Artefacts.
• Administrative records.
• Biological samples.
• Digital traces.
• Computer logs.
• Source code.
• Simulations.
• Published studies.
• Formal proofs.
• Historical sources.

The quality of evidence depends on its relationship to the question, not merely on its format.

Population and Sample

A population is the wider group, set, corpus, or collection about which the researcher wants to draw conclusions.

A sample is the subset actually examined.

Probability sampling is useful when researchers need population estimates and can construct an appropriate sampling frame. Purposive sampling is suitable when cases are selected for their relevance, experience, variation, or theoretical significance.

Sample size should be justified by the design and analytical goal. There is no universal rule that qualitative studies are small and quantitative studies are large.

Measurement and Operationalisation

Operationalisation converts an abstract concept into observable indicators or analytical procedures.

For example, “academic engagement” might be represented through:

• Attendance.
• Time spent on course activities.
• Participation records.
• A validated questionnaire.
• Interviews about learning behaviour.

Each option captures only part of the concept. Researchers should explain why their chosen indicators are appropriate.

Bias

Bias is a systematic influence that distorts selection, measurement, analysis, interpretation, or reporting.

Examples include:

• Selection bias.
• Non-response bias.
• Recall bias.
• Measurement bias.
• Observer bias.
• Confirmation bias.
• Confounding.
• Publication bias.
• Selective outcome reporting.
• Algorithmic bias.

Bias cannot always be eliminated, but it should be anticipated, reduced, examined, and reported.

What Makes Research Trustworthy?

Trustworthy research uses methods and reasoning appropriate to its claims and reports enough information for readers to evaluate the work.

Quantitative Quality Criteria

Common quantitative criteria include:

Construct Validity

Does the measure adequately represent the intended concept?

Internal Validity

Is the observed relationship plausibly attributable to the proposed explanation rather than bias, confounding, or alternative causes?

External Validity

To which people, settings, periods, or conditions may the findings reasonably apply?

Reliability

Does a measurement procedure produce sufficiently consistent results under relevant conditions?

Reliability does not guarantee validity. An instrument can measure the wrong concept consistently.

Precision

How much uncertainty surrounds the estimate?

Precision may be expressed through confidence intervals, credible intervals, standard errors, or other design-appropriate measures.

Robustness

Do conclusions remain reasonably stable when plausible analytical choices or assumptions change?

Qualitative Quality Criteria

Qualitative traditions use different terminology, but frequently discussed criteria include:

Credibility

Are interpretations well supported by the evidence and context?

Dependability

Is the research process documented and logically consistent?

Confirmability

Can readers distinguish evidence from the researcher’s assumptions and preferences?

Transferability

Does the report provide enough contextual detail for readers to judge relevance to another setting?

Reflexivity

Does the researcher examine how identity, assumptions, relationships, access, and position may shape the research?

Techniques may include detailed field notes, analytic memos, triangulation, negative-case analysis, peer discussion, participant engagement where appropriate, and clear documentation of interpretive decisions.

Reproducibility and Replicability

Definitions vary among disciplines. A widely used distinction is:

• Reproducibility: obtaining consistent computational results using the same data, code, methods, and conditions.
• Replicability: obtaining findings that are consistent with a previous study by collecting or generating new evidence.

Both depend on transparent reporting. In 2026, NIH expanded its central resources supporting replication, reproducibility, rigor, and transparency across the research lifecycle (National Institutes of Health [NIH], 2026).

A failed replication does not automatically prove misconduct or make the original study worthless. Differences may result from context, sampling, implementation, measurement, chance, analytical decisions, or an effect that is less stable than initially believed.

Research Ethics and Integrity

Research ethics concerns what researchers ought to do. Research integrity concerns honest, responsible, and professionally accountable conduct.

Human Participants

Research involving people may require attention to:

• Voluntary informed consent.
• Capacity to consent.
• Privacy and confidentiality.
• Fair participant selection.
• Minimisation of harm and burden.
• Safeguards for vulnerable populations.
• Appropriate compensation.
• Withdrawal procedures.
• Community or stakeholder involvement.
• Institutional or research-ethics review.

Requirements differ by country, institution, funder, field, and study type. Researchers should follow the rules applicable to their project rather than relying on a general online guide.

Data Ethics

Researchers should determine:

• Whether data were obtained lawfully and fairly.
• Whether consent permits the intended reuse.
• Whether individuals can be reidentified.
• Whether combining datasets creates new risks.
• Whether communities could be harmed even when individuals are anonymised.
• Whether access restrictions are required.
• Whether data should be retained, archived, or destroyed.

Open science does not mean that every dataset must be made public. Ethical, legal, cultural, commercial, security, and intellectual-property restrictions may require controlled access or non-disclosure.

Research Misconduct

Serious misconduct commonly includes:

• Fabrication: inventing data or results.
• Falsification: manipulating materials, procedures, data, or records deceptively.
• Plagiarism: presenting another person’s words or ideas without appropriate acknowledgement.

Other harmful practices include:

• Selective reporting.
• Inappropriate image manipulation.
• Undisclosed conflicts of interest.
• Misleading authorship.
• Citation manipulation.
• Duplicate publication.
• Inadequate supervision.
• Concealing important methodological changes.
• Misrepresenting exploratory analyses as preplanned.

Authorship

Authorship should reflect substantial intellectual contribution, participation in drafting or revising the work, approval of the final version, and accountability for the publication, subject to the relevant disciplinary policy.

People who contribute but do not meet authorship criteria should be acknowledged appropriately. AI systems cannot accept responsibility and should not be listed as authors (International Committee of Medical Journal Editors [ICMJE], 2026).

Research in the Digital and AI Era

Digital tools have changed how researchers find literature, collect data, collaborate, analyse evidence, document workflows, and communicate findings. These tools can improve efficiency, but they do not replace methodological judgement.

Common Categories of Digital Research Tools

Research task	Tool categories
Literature discovery	Academic databases, citation indexes, library catalogues, scholarly search engines
Reference management	Citation managers and collaborative reference libraries
Data collection	Survey platforms, electronic data-capture systems, sensors, laboratory systems
Quantitative analysis	Statistical software, programming languages, notebooks
Qualitative analysis	Transcription, coding, annotation, and qualitative-analysis software
Evidence synthesis	Screening, extraction, risk-of-bias, and meta-analysis tools
Collaboration	Version control, shared documents, project-management platforms
Open research	Repositories, preregistration platforms, data archives, protocol registries
Reproducibility	Code repositories, containers, computational notebooks, environment records
Communication	Visualisation, publishing, presentation, and public-engagement tools

A tool should be selected for its suitability, security, accessibility, interoperability, and long-term availability—not because it is fashionable.

Open Science

Open science seeks to make appropriate parts of research more accessible, transparent, collaborative, and reusable. Practices may include:

• Open-access publication.
• Preregistration.
• Registered reports.
• Open protocols.
• Open data.
• Open code.
• Open materials.
• Preprints.
• Persistent identifiers.
• Open peer review.
• Citizen and community participation.

The UNESCO Recommendation on Open Science emphasises transparency, collaboration, reproducibility, public accountability, and reducing inequalities in access to scientific knowledge (UNESCO, 2021).

Openness must remain compatible with consent, privacy, community rights, security, intellectual property, and legitimate confidentiality.

How AI Can Assist Research

Depending on institutional and disciplinary rules, AI tools may assist with:

• Generating preliminary search terms.
• Reformatting researcher-written text.
• Translating non-sensitive material that is later checked by a qualified person.
• Explaining unfamiliar code.
• Suggesting test cases for software.
• Classifying records during an evidence review, with validation.
• Drafting data dictionaries.
• Checking consistency in researcher-created tables.
• Producing alternative visualisation ideas.
• Transcribing approved recordings in a secure environment.
• Summarising the researcher’s own notes for verification.
• Identifying questions that require further investigation.

AI output should be treated as unverified material, not as authoritative evidence.

Major Risks of AI in Research

AI tools may:

• Generate false references.
• Misstate sources.
• Produce plausible but incorrect explanations.
• Reproduce social or linguistic biases.
• expose confidential or identifiable data;
• hide uncertainty.
• Produce non-reproducible results.
• introduce insecure or incorrect code;
• obscure who made an intellectual decision;
• encourage superficial reading;
• violate journal, university, participant-consent, or funder requirements.

Responsible AI Checklist

Before using AI in research:

1. Check institutional, funder, journal, and disciplinary rules.
2. Do not upload confidential, identifiable, restricted, proprietary, or unpublished material without authorisation.
3. Verify every factual statement, quotation, calculation, reference, and code output.
4. Read and cite the original sources rather than citing an AI answer.
5. Record the tool, version, date, purpose, and important prompts when required for transparency.
6. Disclose relevant use in the appropriate section.
7. Ensure a human researcher makes and can defend all substantive decisions.
8. Never list an AI system as an author.
9. Check outputs for bias, fabrication, plagiarism, and inappropriate similarity.
10. Retain a non-AI method for verifying important conclusions.

AI can support parts of a research workflow, but it cannot assume ethical, legal, or scholarly accountability.

Advantages and Limitations of Research

Advantages

Research can:

• Produce new knowledge.
• Test existing beliefs.
• Improve theory.
• Support evidence-informed decisions.
• Evaluate interventions and programmes.
• Reveal unintended effects.
• Develop practical solutions.
• Preserve cultural and historical knowledge.
• Improve measurement and analytical methods.
• Identify new questions.
• Reduce some forms of uncertainty.

Limitations

Research is always constrained by factors such as:

• Available evidence.
• Measurement quality.
• Sampling.
• Researcher assumptions.
• Time and funding.
• Ethical restrictions.
• Missing data.
• Access to participants or sources.
• Model assumptions.
• Historical and cultural context.
• Publication and reporting practices.

A method can reduce particular limitations while introducing others. For example, a tightly controlled experiment may support causal inference but represent real-world settings poorly. An ethnographic study may provide deep contextual understanding but may not support numerical population estimates.

The goal is not to claim that a study has no limitations. The goal is to select a defensible design, reduce avoidable problems, and describe remaining limitations honestly.

Common Research Mistakes

Starting With a Preferred Method

“I want to conduct a survey” is not a research problem. Begin with the question and determine whether a survey can answer it.

Choosing a Topic That Is Too Broad

“Artificial intelligence in education” could support thousands of projects. Narrow it by population, setting, concept, outcome, evidence type, and time period.

Treating a Literature Summary as Analysis

A literature review should compare, evaluate, organise, and synthesise sources rather than describe them one by one.

Confusing Correlation With Causation

An association may result from confounding, reverse causation, selection, measurement problems, or chance.

Using an Unjustified Sample

Convenience may be unavoidable, but researchers should explain how it affects the claims that can be made.

Collecting Data Before Planning the Analysis

Researchers should decide in advance what evidence is needed and how it will address the question.

Ignoring Negative or Unexpected Findings

Unexpected findings may reveal measurement problems, contextual differences, incorrect assumptions, or new research questions.

Reporting Only Favourable Results

Selective reporting distorts the evidence base. Prespecified outcomes, transparent deviations, and complete reporting reduce this risk.

Overstating Generalisability

Findings from one institution, language, platform, country, period, or self-selected sample may not apply everywhere.

Hiding Limitations

Transparent limitations increase credibility. They help readers understand what the evidence can and cannot establish.

Treating AI Output as a Source

AI-generated references, quotations, summaries, and claims must be verified against original materials.

How to Recognise a Research Article

An original empirical research article commonly includes:

1. A clearly stated problem or question.
2. A review of relevant literature.
3. A methods or methodology section.
4. A description of participants, cases, sources, materials, or data.
5. An explanation of analysis.
6. Results or findings.
7. Interpretation and limitations.
8. References.
9. Information about ethics, funding, conflicts, data, or contributions where relevant.

Many scientific and social-science articles follow the IMRaD structure:

• Introduction.
• Methods.
• Results.
• Discussion.

However, not all research articles use this structure. Humanities, legal, mathematical, theoretical, and practice-based research may organise their argument differently.

A literature review, editorial, commentary, protocol, theoretical article, and original empirical study are all scholarly outputs, but they are not the same type of publication.

Peer review is useful, but readers should still evaluate:

• Journal credibility.
• Research design.
• Sample or source selection.
• Measurement.
• Analysis.
• Transparency.
• Conflicts of interest.
• Limitations.
• Whether the conclusion matches the evidence.

Research Proposal Template

A research proposal can use the following structure.

1. Working Title

State the central topic, population or material, and focus.

2. Background

Explain the context and what is already known.

3. Research Problem

Identify the specific gap, uncertainty, contradiction, or practical need.

4. Aim

State the overall purpose of the project.

5. Research Questions

List the questions the project will answer.

6. Objectives

Specify the activities required to achieve the aim.

7. Literature Review

Synthesise relevant theories, methods, and findings.

8. Conceptual or Theoretical Framework

Define the key concepts and proposed relationships or interpretive perspective.

9. Research Design

Explain the overall strategy and why it fits the questions.

10. Population, Cases, Sources, or Materials

Define what will be studied and how it will be selected.

11. Data or Evidence Collection

Describe instruments, procedures, sources, settings, and quality controls.

12. Analysis Plan

Explain how the evidence will be prepared, analysed, and interpreted.

13. Ethics and Governance

Address consent, risk, privacy, approval, permissions, data security, and conflicts.

14. Limitations and Delimitations

Distinguish unavoidable weaknesses from deliberate boundaries.

15. Data-Management and Sharing Plan

Explain storage, access, documentation, retention, archiving, and sharing.

16. Timeline

Provide realistic stages and milestones.

17. Resources or Budget

Identify required software, equipment, access, travel, personnel, and funding.

18. Expected Contribution

Explain what the study may contribute without promising results that have not yet been obtained.

19. Dissemination Plan

State how findings will be communicated to academic and non-academic audiences.

20. References

List every cited source using the required citation style.

Conclusion

Research is a systematic, critical, and transparent form of inquiry used to develop knowledge, interpret evidence, evaluate claims, and address problems. Its methods differ across disciplines, but strong research aligns its question, evidence, design, analysis, ethics, and conclusions.

No research type is universally best. The appropriate approach depends on what the researcher wants to know, what evidence is needed, what can be done ethically, and what conclusions the design can reasonably support.