Types of research are the recognised categories researchers use to classify a study according to its purpose, its approach to data, its level of application, its data source, its mode of reasoning and its design. In short, the main types of research are: by purpose (exploratory, descriptive, explanatory and correlational), by approach (qualitative, quantitative and mixed methods), by application (basic and applied), by data source (primary and secondary), by reasoning (inductive and deductive), and by design (experimental and non-experimental). You choose a type — usually a combination of these — by working backwards from your research question and aim.
These categories are not mutually exclusive labels but overlapping layers. A single dissertation might be applied, explanatory, quantitative, deductive and experimental all at once. The skill lies in describing your study accurately along each dimension so that your methodology is coherent and defensible.
What do we mean by “types of research”?
When methodologists talk about types of research, they are describing the choices that sit beneath any empirical study. Saunders, Lewis and Thornhill capture this layering in their well-known “research onion“, which presents methodology as a series of nested decisions — from philosophy and approach on the outer layers, through strategies and choices, down to data collection and analysis at the core. Each layer you peel back forces a decision, and the type of research you end up conducting is simply the sum of those decisions described in plain language.
It helps to think in dimensions rather than a single label. A study has a purpose (what it is trying to achieve), an approach (the kind of data and logic it relies on), a level of application (whether it advances theory or solves a practical problem), a data source (whether you gather new data or reuse existing data), a mode of reasoning (whether theory drives the study or emerges from it), and a design (whether you manipulate variables or observe them as they are). Naming your study correctly along all six dimensions is what makes a methodology chapter convincing.
The full taxonomy of research types (master table)
The table below summarises the major types of research grouped by classification dimension, with a one-line definition and a recognisable dissertation example for each. Use it as a quick reference, then read the detailed sections that follow.
| Classification | Type | One-line definition | Dissertation example |
|---|---|---|---|
| By purpose | Exploratory | Investigates a problem that is poorly understood to generate insight, not firm answers. | Scoping how UK first-year students experience “academic belonging” before any scale exists to measure it. |
| Descriptive | Documents the characteristics, frequency or distribution of a phenomenon without explaining causes. | Profiling the social-media habits of 16–18 year-olds across three colleges. | |
| Explanatory | Tests why something happens — the causal or mechanistic relationship between variables. | Why flexible working hours reduce employee turnover in SMEs. | |
| Correlational | Measures whether and how strongly two or more variables move together, without manipulating them. | The relationship between sleep duration and exam performance in undergraduates. | |
| By approach | Qualitative | Explores meaning, experience and context using non-numerical data such as words and images. | Interviewing nurses about moral distress during understaffed shifts. |
| Quantitative | Measures and tests relationships using numerical data and statistical analysis. | Surveying 400 consumers to test whether price perception predicts purchase intention. | |
| Mixed methods | Integrates qualitative and quantitative data within one study to draw on the strengths of both. | A survey of patient satisfaction followed by interviews explaining the low-scoring items. | |
| By application | Basic (pure) | Builds or extends theory and knowledge for its own sake, without a specific application in mind. | Developing a theoretical model of how working memory limits second-language acquisition. |
| Applied | Solves a concrete, real-world problem in a specific setting. | Evaluating whether a new onboarding programme cuts staff attrition at a named retailer. | |
| By data source | Primary | Collects new, first-hand data directly from participants or observations. | Running an original questionnaire on remote-work productivity. |
| Secondary | Analyses existing data gathered by others, such as datasets, reports or prior studies. | Re-analysing the UK Labour Force Survey to study gig-economy growth. | |
| By reasoning | Inductive | Builds theory bottom-up from observed patterns in the data. | Generating a grounded model of “founder burnout” from start-up interviews. |
| Deductive | Tests a pre-stated hypothesis derived from existing theory, top-down. | Testing whether transformational leadership raises team performance, per existing theory. | |
| By design | Experimental | Manipulates an independent variable under controlled conditions to establish cause. | Randomly assigning students to gamified vs standard revision to test recall. |
| Non-experimental | Observes variables as they naturally occur, without manipulation (e.g. surveys, case studies). | A cross-sectional survey of job satisfaction across departments. |
Types of research by purpose
The purpose dimension answers: what is this study trying to achieve? It is usually the first thing to settle, because it shapes everything downstream.
- Exploratory research is appropriate when little is known about a topic. It is flexible and open-ended, often qualitative, and aims to clarify concepts, surface variables and generate hypotheses rather than confirm them. A scoping interview study or an open survey are typical formats. Findings are tentative and pave the way for later, more structured work.
- Descriptive research answers what, who, where and when. It paints an accurate picture of a population or phenomenon — its frequency, characteristics or distribution — but stops short of explaining causes. Surveys and observational studies dominate here.
- Explanatory (causal) research answers why and how. It tests cause-and-effect relationships between variables and is the natural home of experiments and strong inferential statistics. Establishing causation demands careful design to rule out confounding influences.
- Correlational research sits between description and explanation. It establishes whether variables are associated and how strongly (and in what direction), but — crucially — correlation alone cannot prove causation because variables are measured, not manipulated. It is invaluable when experiments are unethical or impractical.
If your aim involves cause and effect you will lean explanatory or correlational; learn the distinction in our guides to experimental research and correlational research, which spell out when each is appropriate and how to avoid the classic “correlation equals causation” error.
Types of research by approach
The approach dimension concerns the kind of data you work with and the logic you apply to it.
- Qualitative research works with words, images and meanings. It is interpretive, context-rich and well suited to exploring how people understand and experience the world. Methods include interviews, focus groups, ethnography and document analysis, with thematic analysis a common analytic strategy.
- Quantitative research works with numbers. It measures variables, tests hypotheses and seeks generalisable, statistically supported conclusions through surveys, experiments and structured observation.
- Mixed-methods research deliberately combines the two within a single study. Creswell describes core designs such as convergent (collect both strands in parallel and compare), explanatory sequential (quantitative first, then qualitative to explain the numbers) and exploratory sequential (qualitative first to build an instrument, then quantitative to test it). Mixing is justified when neither strand alone answers the question fully.
Choosing between them is one of the most consequential decisions in a dissertation; our dedicated comparison of quantitative vs qualitative research walks through the trade-offs in depth, and the practicalities of gathering either kind of data are covered in our guide to methods of data collection.
“Mixed methods research is an approach to inquiry involving collecting both quantitative and qualitative data, integrating the two forms of data, and using distinct designs that may involve philosophical assumptions and theoretical frameworks.” (Source: Creswell & Creswell, 2018)
Types of research by application
The application dimension distinguishes research conducted to advance knowledge from research conducted to solve a problem.
- Basic (pure or fundamental) research seeks to expand understanding and develop theory without an immediate practical use in view. It asks foundational questions and its value may only become apparent years later.
- Applied research targets a specific, practical problem in a defined context — evaluating an intervention, improving a process or informing a policy decision. Most professional and commissioned dissertations are applied.
The two are complementary: applied work frequently draws on theories that began as basic research, and findings from applied studies can feed back to refine theory.
Types of research by data source
This dimension concerns where your data come from.
- Primary research generates new, first-hand data through surveys, interviews, experiments or observation. It gives you control over what is measured and how, but is time-consuming and resource-intensive.
- Secondary research analyses data already collected by someone else — published datasets, government statistics, organisational records or prior studies. It is efficient and enables large-scale or longitudinal analysis, but you are limited to what was originally recorded and must scrutinise data quality and provenance.
Types of research by reasoning
The reasoning dimension describes the relationship between theory and data.
- Deductive research moves top-down: you begin with existing theory, derive a hypothesis, then collect data to test it. It is associated with quantitative, hypothesis-driven studies.
- Inductive research moves bottom-up: you begin with observations, identify patterns and build theory from them. It is associated with qualitative, exploratory studies. In practice many studies are abductive, moving back and forth between data and theory.
Types of research by design
Design refers to the structure that lets your data actually answer your question.
- Experimental research manipulates an independent variable, controls extraneous variables and (ideally) randomly assigns participants, allowing strong causal claims. True experiments, quasi-experiments and field experiments vary in how much control they achieve.
- Non-experimental research observes variables as they naturally occur — surveys, case studies, correlational and longitudinal designs. It offers realism and is often the only ethical option, but causal inference is weaker.
Whichever design you adopt, your conclusions are only as trustworthy as the people you study and how you select them; our guide to sampling methods of research explains probability and non-probability sampling and how each affects the validity of your claims.
How to choose the right type of research
The single most reliable rule is this: let your research question and aim drive the type, never the other way round. Students often pick a method first (“I’ll do a survey“) and then bend the question to fit — a recipe for an incoherent methodology. Work through these steps instead:
- State your aim and question precisely. Is it about what (descriptive), why (explanatory), how strongly two things relate (correlational), or what is going on here at all (exploratory)? The wording usually reveals the purpose.
- Decide what counts as evidence. If your question is about meaning, experience or process, lean qualitative; if it is about measurement, prediction or testing a hypothesis, lean quantitative; if it genuinely needs both, consider mixed methods.
- Check the theory relationship. Are you testing an existing theory (deductive) or building one (inductive)? This aligns with your approach choice.
- Choose primary or secondary data based on whether the data you need already exist and are accessible, and on your time and budget.
- Select a design that can actually deliver the inference you need. Causal claims demand experimental control; if manipulation is impossible or unethical, a correlational or case-study design is the honest choice.
- Sense-check coherence. Read your six choices as a sentence (“This is an applied, explanatory, quantitative, deductive, experimental, primary-data study”). If any element jars, revisit it. This is exactly the alignment the research onion is designed to enforce.
- By purpose → Explanatory. The word “reduces” states a cause-and-effect claim: the programme is expected to cause a change in anxiety. The study tests why/whether X affects Y, not merely what anxiety looks like (which would be descriptive) or how two measured variables co-vary (correlational).
- By approach → Quantitative. The outcome is “anxiety symptoms,” which can be measured on a validated scale (e.g. GAD-7) and compared statistically before and after. Numbers and significance testing, not lived experience, answer the question — so the logic is quantitative.
- By application → Applied. It evaluates a concrete intervention in a named, real-world setting (one university’s student body) to inform a practical decision about whether to keep funding the programme — not to build abstract theory.
- By data source → Primary. No existing dataset records these specific students’ anxiety scores before and after this specific programme, so the student must collect first-hand questionnaire data directly from participants.
- By reasoning → Deductive. The student starts from established psychological theory (peer support buffers stress), derives a testable hypothesis (“programme participants will show a larger fall in GAD-7 scores than non-participants”), then gathers data to test it — theory drives data, top-down.
- By design → Experimental (quasi-experimental). The student manipulates the independent variable (programme vs no programme) and measures the dependent variable (anxiety score). Because students self-select rather than being randomly assigned, it is a quasi-experiment — strong on causal intent but with weaker control than a true randomised experiment, which the limitations section must acknowledge.
Read back as a single sentence: “an applied, explanatory, quantitative, deductive, quasi-experimental study using primary survey data.” Every layer reinforces the next — the aim demanded causation, causation demanded a comparison and a hypothesis, and the absence of an existing dataset demanded primary data. That internal consistency is exactly what an examiner means by a “coherent methodology.”
Strengths, limitations and common mistakes
Each type carries trade-offs. Quantitative designs offer generalisability and statistical rigour but can miss context; qualitative designs offer depth and nuance but limited generalisability; experiments give causal power but often at the cost of real-world realism. The art is matching the type to the question, not chasing the “most rigorous” method in the abstract.
The most common mistakes students make are:
- Choosing the method before the question. This forces an artificial fit and weakens the whole dissertation.
- Confusing correlation with causation. A correlational design cannot establish cause, however strong the relationship.
- Mislabelling the study. Calling a descriptive survey “explanatory”, or a deductive test “exploratory”, signals a shaky grasp of methodology to examiners.
- “Mixing” without integration. Bolting a few interviews onto a survey is not mixed methods unless the two strands are genuinely integrated and justified.
- Ignoring feasibility. An ambitious experimental design that your timeframe, ethics approval or access cannot support will fail in execution.
Avoiding these comes down to coherence: state the question, classify the study honestly along all six dimensions, and make sure each choice supports the others.
Not sure which type of research fits your dissertation?
Our academics can help you choose the right approach, design a coherent methodology and execute it end to end.
Bringing it together
There is no single “best” type of research — only the type, or combination of types, that best answers your question within your constraints. Treat the six dimensions as a checklist: purpose, approach, application, data source, reasoning and design. Describe your study accurately along each, keep the choices mutually consistent, and you will have a methodology that examiners recognise as rigorous and well reasoned. The research onion is simply a reminder that good methodology is layered, deliberate and aligned from the outside in.
Related methodology guides
- The Research Onion (Saunders)
- Cross-Sectional vs Longitudinal Studies
- Research Philosophy
