Research design and paradigms: a glossary

Abductive reasoning

Also called: abduction, abductive approach, abductive inference, inference to the best explanation, IBE, abductive logic, abductive method, abductive research

Abductive reasoning is reasoning from what has been observed to the explanation that would best account for it, held as a working conclusion rather than a proven one. Philosophers use the word in two senses: the modern one, inference to the best explanation, which judges a hypothesis by how well it explains the evidence, and Peirce's older one, which is about thinking up hypotheses in the first place.

Action research

Also called: action-research, action inquiry

Action research is research in which people tackle a practical problem in their own setting and study the effects of what they change, in repeated cycles of planning, acting, observing and reflecting. It aims at improvement more than at testing theory, and it is often carried out by practitioners, such as teachers, nurses or social workers, in their own workplace.

Alternative hypothesis

Also called: H1, Ha, H₁, experimental hypothesis, alternate hypothesis, alternative hypotheses

The alternative hypothesis is the statement, set against the null hypothesis, that a difference or relationship does exist in the population, written H1 or Ha. It never contains an equals sign: it says a value is not equal to, greater than or less than another, and which of these it says decides whether the test is two-tailed or one-tailed. It is usually the claim the researcher hopes to support.

Applied research

Also called: applied studies

Applied research is original research directed at a specific practical problem or use, such as how using a mobile phone affects driving. It differs from basic research in its purpose rather than its methods, and the two feed each other: an applied study can reveal something fundamental about attention or perception, and basic findings often find a practical use later.

Assumptions

Also called: research assumptions, assumptions of the study, assumption

Assumptions are the things a study takes to be true without testing them, such as that participants answer honestly or that a questionnaire measures what it is meant to. A thesis or proposal often states them beside the scope, delimitations and limitations so readers can judge the conclusions. Every study also rests on deeper philosophical assumptions about what is real and how it can be known.

Attribution (evaluation)

Also called: attribution of impact, attributing impact

Attribution, in evaluation, is the claim that observed changes were caused by a particular intervention, together with the work of judging how much of a change can be credited to it. The OECD's definition does not require the intervention to be the only cause. It is usually contrasted with contribution, the more modest claim that an intervention played a part alongside other influences.

Axiology

Also called: axiological assumptions, axiological position, axiological assumption, axiological stance

Axiology is the branch of philosophy concerned with values, and in research it names a researcher's position on the part values play in inquiry. Positivists aim for value-free research, while interpretivist, critical and transformative researchers hold that values shape every study and should be stated openly. A paradigm is often described by its ontology, epistemology and axiology together.

Basic research

Also called: pure research, fundamental research, basic science, blue-skies research, curiosity-driven research

Basic research is research that seeks new knowledge about how things fundamentally work, pursued for understanding rather than for any particular use. The OECD's Frascati Manual, the international standard for measuring research and development, sets it beside applied research and experimental development. The line is not sharp, since a basic study of how people talk to each other can later inform counselling.

Boundary conditions

Also called: boundary condition, theoretical boundary conditions

Boundary conditions are the assumptions and limits of time, place, population or values that mark where a theory is expected to hold and where it is not. Whetten counts them, with constructs, propositions and logic, among the building blocks of a theory. A theory of workplace motivation built on studies of large firms, for example, may not hold for unpaid volunteers.

Bradford Hill criteria

Also called: Hill's criteria, Bradford Hill viewpoints, Hill criteria for causation, Bradford-Hill criteria, Hill's criteria of causation

The Bradford Hill criteria are nine considerations, set out by Austin Bradford Hill in 1965, for judging whether an observed association is likely to be causal: strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment and analogy. Hill himself called them viewpoints, and of the nine only temporality, the cause coming before the effect, is widely agreed to be essential.

Burrell and Morgan's four paradigms

Also called: Burrell and Morgan's paradigms, Burrell and Morgan framework, Burrell and Morgan 1979, sociological paradigms and organisational analysis

Burrell and Morgan's four paradigms are a 1979 framework that sorts social theories along two axes, subjective against objective assumptions about social science and regulation against radical change, giving the functionalist, interpretive, radical humanist and radical structuralist paradigms. It remains influential in business and organisation research, though critics argue it needs updating for newer ways of producing knowledge.

Case study research

Also called: case study, case study design, case study method, case study approach, case studies, case study methodology, case-study research

Case study research is the in-depth study of one or a few bounded cases, such as a person, a school, a policy or an event, in its real-life setting, drawing on several kinds of evidence. It suits how and why questions where the thing studied cannot be separated from its context. Stake distinguished intrinsic cases, studied for their own sake, from instrumental ones, studied to understand a wider issue.

Causal-comparative research

Also called: causal comparative research, causal-comparative design, ex post facto research, ex post facto design, ex-post-facto research, causal-comparative study

Causal-comparative research is non-experimental research that compares groups already differing on some characteristic or experience, looking for its possible effect on an outcome after both have occurred. It is also called ex post facto research, meaning after the fact. Because the researcher cannot assign people to groups or manipulate the cause, it can suggest but not establish cause and effect.

Causality

Also called: causation, cause and effect, causal relationship, cause-and-effect relationship, causal relationships

Causality is the relationship in which one event, behaviour or condition brings about another. To claim a cause in a nomothetic study, researchers usually look for four things: the two vary together, the link is plausible, the cause comes first, and no third variable explains the association. A correlation on its own meets only the first of these.

Citizen science

Also called: public participation in scientific research, PPSR, participatory science, citizen scientist, citizen scientists, citizen science project

Citizen science is research in which members of the public take part, usually as volunteers, in scientific projects. Their part may be as small as recording observations or lending computer time, or extend to shaping questions and analysing data. US government agencies also call it participatory science, and NOAA's projects alone drew more than half a million participants in its 2025 financial year.

Co-operative inquiry

Also called: cooperative inquiry, co-operative enquiry, cooperative enquiry, co-inquiry

Co-operative inquiry is a form of participatory research, developed by John Heron and Peter Reason, in which everyone involved is both co-researcher and co-subject. Together they decide the questions and methods, try them out in their own practice, and reflect on the experience in repeated cycles. It is the research method that the participatory paradigm calls for.

Community-based participatory research

Also called: CBPR, community based participatory research, community-based participatory research approach

Community-based participatory research is a collaborative approach in which researchers and members of a community affected by a problem share power and decisions across every stage of a study, from setting the question to using the findings. It is common in public health. Unlike research that is simply carried out in a community, it aims at building capacity and action as well as knowledge.

Comparative research

Also called: comparative method, comparative design, comparative study, cross-national research, cross-national comparative research, comparative research design

Comparative research is research that seeks explanations by systematically comparing two or more cases, such as countries, institutions or policies, on the factors thought to matter. It is the defining method of comparative politics and suits questions where cases are few and experiments impossible. Its two classic strategies are the most similar and the most different systems designs.

Conceptual framework

Also called: conceptual frameworks, conceptual frame

A conceptual framework is a network of linked concepts that together explain the phenomenon a study is about, showing how the researcher expects those concepts to relate. Usage varies widely: some guides treat it as another name for a theoretical framework, while others keep theoretical framework for a structure drawn from existing theory and conceptual framework for the researcher's own, study-specific argument.

Conceptual replication

Also called: inexact replication, constructive replication

A conceptual replication is a replication study that tests the same hypothesis or theory as an earlier study but deliberately changes important features, such as the measures, the population or the procedure. Success suggests the finding holds beyond the original method. It sits at the opposite end of a continuum from direct replication, and different authors name the distinction in different ways.

Confirmatory research

Also called: confirmatory study, confirmatory analysis, hypothesis-testing research, hypothesis testing research, confirmatory studies

Confirmatory research is research that tests hypotheses or predictions stated before the relevant data are seen, so that the data can genuinely show them to be wrong. It contrasts with exploratory, hypothesis-generating work. Nosek and colleagues argue that presenting explanations devised after seeing the results as if they had been predicted overstates confidence, and that preregistration keeps the two apart.

Constructivism

Also called: constructivist paradigm, constructivist approach, constructivist

Constructivism is the research paradigm holding that people build their own understanding of reality, so there are many realities rather than one, and that knowledge comes from interpreting how participants make sense of their experience. It usually leads to qualitative methods. Some writers treat it as the same paradigm as interpretivism, and some keep it for meaning made in individual minds, as against social constructionism.

Context of discovery and context of justification

Also called: context of discovery, context of justification, discovery and justification, context distinction

The context of discovery and the context of justification are a distinction in the philosophy of science between how a new idea or hypothesis is thought up and how it is then tested and defended. For much of the twentieth century philosophers held that only justification could be analysed logically. Nosek and colleagues treat it as one version of the line between generating hypotheses and testing them.

Contribution analysis

Also called: contribution analyses

Contribution analysis is a theory-based evaluation method that builds a reasoned case, step by step, that an intervention contributed to observed outcomes, by testing its theory of change against several kinds of evidence. It is used where an experimental design is not possible. The UK Magenta Book notes that it offers a defensible line of reasoning about contribution rather than proof of cause.

Contribution to knowledge

Also called: original contribution to knowledge, contribution to the field, original contribution

A contribution to knowledge is the new understanding a piece of research adds to its field, the test at the heart of a doctorate. The UK Quality Assurance Agency expects every doctoral graduate to have made an original contribution, through original research or by applying existing knowledge in a new way, and notes that most examiners look for work of publishable quality.

Cost-benefit analysis

Also called: CBA, benefit-cost analysis, social cost-benefit analysis, cost benefit analysis

Cost-benefit analysis is an economic evaluation method that expresses an intervention's costs and benefits in money where possible, so that they can be compared directly, often as a net present value or a benefit-cost ratio. It asks whether an intervention is worth its cost to society. Benefits that cannot sensibly be priced should still be described rather than left out.

Cost-effectiveness analysis

Also called: CEA, cost effectiveness analysis, cost-effectiveness evaluation, cost-effectiveness ratio

Cost-effectiveness analysis is an economic evaluation method that compares options by their cost per unit of a single outcome, such as cost per life saved or per case prevented, without putting a money value on the outcome itself. It is used where benefits are hard to price, especially in health, and cost-utility analysis is one form of it.

Cost-utility analysis

Also called: CUA, cost utility analysis

Cost-utility analysis is a form of cost-effectiveness analysis that measures benefits in a single utility-based unit, most often the quality-adjusted life year, or QALY, which combines length of life with its quality. Options can then be compared by their cost per QALY gained. It is widely used in health to compare treatments that aim at similar outcomes.

Counterfactual

Also called: the counterfactual, counterfactual scenario, counterfactual condition, counterfactual analysis

The counterfactual is what would have happened to the same people, places or organisations if an intervention had not taken place. It can never be observed directly, so evaluations estimate it, usually with a control or comparison group, and the difference between actual outcomes and the counterfactual is the intervention's impact. Theory-based methods are used where no credible counterfactual can be built.

Critical rationalism

Also called: critical rationalist, Popper's critical rationalism

Critical rationalism is Karl Popper's philosophy of knowledge, which holds that theories can never be proved but can be improved by exposing them to the severest criticism and testing and discarding those that fail. It opposes both scepticism and relativism. Falsificationism, his account of how science tests its theories, is its best-known application.

Critical realism

Also called: critical realist paradigm, critical realist approach, critical realist

Critical realism is a research philosophy, developed by Roy Bhaskar, holding that a real world exists independently of our knowledge of it but can only be known partly and fallibly. It sees reality in layers: what is experienced, the events that actually happen, and the underlying mechanisms that cause them. Some textbooks class it as a form of post-positivism, others as a paradigm of its own.

Critical theory

Also called: critical paradigm, critical research, critical social theory

Critical theory, in research methods, is the paradigm that treats inquiry as a means of exposing and changing unequal power, so research is openly value-laden and aimed at emancipation. In its narrow sense the name belongs to the Frankfurt School of social philosophy, begun in Germany in the 1930s, and in its broad sense it covers feminist, critical race, queer and postcolonial approaches.

Cross-sectional design

Also called: cross-sectional study, cross-sectional research, cross sectional design, cross-sectional

A cross-sectional design collects data from a sample at a single point in time, giving a snapshot of how things stand. It is quick and relatively cheap and can show how characteristics go together, but because nothing is followed over time it cannot show which came first, so it is weak evidence of cause. It contrasts with a longitudinal design.

Decolonising methodologies

Also called: decolonizing methodologies, decolonising research, decolonizing research, decolonising methodology, decolonizing methodology

Decolonising methodologies are approaches to research that question the assumptions, motives and values behind research practices shaped by colonialism, and put the priorities and knowledge of Indigenous and colonised peoples at the centre. The phrase comes from Linda Tuhiwai Smith's 1999 book, which critiques research done through imperial eyes and sets out an agenda for research by Indigenous peoples themselves.

Deductive reasoning

Also called: deduction, deductive approach, deductive research, top-down approach, theory-testing approach, deductive logic, deductive inference, deductive method

Deductive reasoning in research moves from the general to the specific: the researcher starts from a theory, derives a hypothesis from it and collects data to test that hypothesis. In logic, a valid deduction from true premises guarantees its conclusion. It is the usual logic of quantitative and experimental work and runs the opposite way to inductive reasoning, which builds theory from data.

Deductive-nomological model

Also called: DN model, D-N model, deductive nomological model, Hempel-Oppenheim model, Hempel's model of explanation

The deductive-nomological model is an account of scientific explanation, set out most influentially by Carl Hempel, in which a phenomenon is explained by deducing it from at least one law of nature together with statements of the particular conditions. Reactions to it shaped later debate. Counterexamples, such as deriving a flagpole's height from its shadow, showed that a valid deduction is not always an explanation.

Delimitations

Also called: delimitation, delimitations of the study

Delimitations are the boundaries a researcher deliberately sets on a study, such as the population, place, period, variables or theories it will cover and those it will leave out. They are decided and justified at the design stage. They differ from limitations, which are weaknesses that restrict what the findings can show and are often only fully known once the study is done.

Demarcation problem

Also called: problem of demarcation, demarcation criterion, criterion of demarcation

The demarcation problem is the question of how to tell science apart from pseudoscience and other kinds of non-science. Karl Popper, who named it, proposed falsifiability as the dividing line. Most philosophers now doubt that any single criterion works and use lists of several signs instead, although there is wide agreement on which particular doctrines count as pseudoscience.

Descriptive research

Also called: descriptive study, descriptive design, descriptive studies

Descriptive research is research that sets out to describe a population, situation or phenomenon accurately, answering who, what, where, when and how many rather than why. A count of how many children are in foster care and for how long is a typical example. Description usually comes first, since something has to be described well before its causes can be studied.

Design-based research

Also called: DBR, design based research, design experiment, design experiments, educational design research

Design-based research is an approach, mainly in education, in which researchers design an intervention such as a teaching tool, try it in real classrooms and refine it over repeated cycles, producing both a better intervention and theory about how learning happens. Unlike a controlled experiment, it treats the messy real setting as part of what is studied, and it aims to yield design principles others can use.

Developmental evaluation

Also called: developmental evaluations

Developmental evaluation is an iterative evaluation approach in which the evaluator is embedded in a programme and gives its staff feedback in real time or close to it, so that they can learn and adjust while the work is under way. The OECD describes it as suited especially to complex or uncertain settings, where an intervention is still taking shape.

Direct replication

Also called: exact replication, close replication, literal replication

A direct replication is a replication study that repeats an earlier study's methods, materials and analysis as closely as possible with a new sample. It checks whether the original result was a matter of chance, an artefact of the procedure or an error. No two studies can be identical, so even a direct replication differs from the original in time, place and people.

Directional hypothesis

Also called: one-tailed hypothesis, one-sided hypothesis

A directional hypothesis predicts not only that a difference or relationship exists but which way it runs, for example that pupils given feedback will score higher than pupils without it. Its alternative hypothesis uses greater than or less than, which leads to a one-tailed test. Predicting that two variables rise together, or that one falls as the other rises, is also directional.

Ecological fallacy

Also called: ecological inference fallacy

The ecological fallacy is the error of drawing conclusions about individuals from data that describe only groups, such as deciding that richer people favour a party because richer districts vote for it more. It happens when a conclusion is pitched at a smaller unit of analysis than the data. The reverse error, claiming things about a group from data on a few individuals, is sometimes called reductionism.

Economic evaluation

Also called: value for money evaluation, value-for-money evaluation, VfM evaluation, economic evaluations

Economic evaluation is evaluation that weighs what an intervention costs against what it achieves, to judge whether it is a good use of resources. The UK Magenta Book calls it value for money evaluation and treats it as one of three main types, alongside process and impact evaluation. Its methods include cost-benefit, cost-effectiveness and cost-utility analysis.

Empirical question

Also called: empirical questions, empirically answerable question

An empirical question is a question about how the world actually is that can be answered by systematic observation, such as whether a reading scheme raises children's test scores. Science deals in questions of this kind. It contrasts with questions of value or ethics, such as whether something ought to be done, which evidence can inform but cannot settle on its own.

Empirical research

Also called: empirical study, empirical investigation, empirical, empirical studies

Empirical research is research that answers its question by systematic observation or measurement of the world, collecting data rather than relying on argument, authority or intuition alone. Experiments, surveys, interviews and analyses of records are all empirical. It contrasts with purely theoretical work and with normative questions about what ought to be, which data alone cannot settle.

Empiricism

Also called: empiricist, empiricist epistemology, empirical philosophy

Empiricism is the philosophical view that knowledge comes mainly or only from experience, through the senses, rather than from reason alone. Locke and Hume are its classic figures, and it underlies science's insistence on checking ideas against systematic observation. It contrasts with rationalism, which holds that some knowledge can be gained independently of experience.

Epistemic fallacy

Also called: the epistemic fallacy

The epistemic fallacy is Roy Bhaskar's name for the error of reducing questions about what exists to questions about what we can know or measure. Critical realists use it against research that treats an indicator, such as whether a country has ratified a treaty, as if it were the thing itself. Avoiding it means allowing that reality extends beyond what is observed.

Epistemology

Also called: epistemological assumptions, epistemological position, theory of knowledge, epistemological stance

Epistemology is the branch of philosophy concerned with knowledge: what counts as knowing something and how it can be known. In research it names the stance a study takes on how valid knowledge is produced, for example through objective measurement, as a positivist holds, or through interpreting people's own accounts, as an interpretivist holds. It pairs with ontology, which asks what exists to be known.

Evaluability assessment

Also called: evaluability, evaluability assessments, evaluation feasibility assessment

An evaluability assessment is a check of how far an intervention can be evaluated reliably and credibly, for example whether its objectives are clearly defined and its results can be verified, usually made early, before an evaluation is designed. The UK Magenta Book treats it as a formal feasibility stage. For complex or unstable interventions it may point instead towards an adaptive approach such as developmental evaluation.

Evaluation research

Also called: evaluative research, programme evaluation, program evaluation, policy evaluation, evaluation study

Evaluation research is research that judges whether a programme, policy or intervention is working, how and why, and at what cost, so that people can decide whether to keep, change or stop it. It borrows methods from every tradition. Its main types are process, impact and value for money evaluation, and a good one is expected to be useful and proportionate as well as robust.

Ex-ante evaluation

Also called: ex ante evaluation, ex-ante assessment, ex ante assessment

An ex-ante evaluation is an evaluation carried out before an intervention is put into practice, to judge in advance matters such as its relevance, design and likely results. The OECD links it with appraisal, the assessment used to decide whether a proposal should go ahead, and with evaluability. It contrasts with ex-post evaluation, which looks back once the intervention has finished.

Ex-post evaluation

Also called: ex post evaluation, ex-post assessment, ex post assessment

An ex-post evaluation is an evaluation of an intervention after it has been completed, either straight away or some time later. It aims to identify what made the intervention succeed or fail, to judge whether its results and impacts have lasted, and to draw lessons for other work. It contrasts with ex-ante evaluation, which is done before the intervention begins.

Experimental development

Also called: experimental development work

Experimental development is systematic work that draws on research and practical experience to produce new products or processes, or to improve existing ones. The OECD's Frascati Manual counts it as the third type of research and development, after basic and applied research. Designing and testing a prototype device that builds on earlier research findings would be an example.

Explanatory research

Also called: explanatory study, explanatory design, causal research

Explanatory research is research that tries to explain why something happens by identifying causes and how they work, for example whether harsh discipline in childhood leads to aggression later. It usually builds on exploratory and descriptive work. Showing a cause needs evidence that the two things are associated, that the cause came first and that no third factor explains the link.

Exploratory research

Also called: exploratory study, exploratory design, exploratory studies, hypothesis-generating research

Exploratory research is research on a question that has been little studied, carried out to become familiar with the topic, clarify its concepts and generate ideas or hypotheses for later work. It is flexible and often uses small samples, so its findings are tentative rather than conclusive. It can also test whether a larger study would be feasible.

Fallibilism

Also called: fallibilist, epistemological fallibilism

Fallibilism is the view that no belief, however well supported, can be justified so conclusively that it could not turn out to be false, so all knowledge, including science's best-established findings, stays open to revision. The term comes from the American philosopher Charles Sanders Peirce. Most philosophers accept it without concluding, as sceptics do, that nothing can be known.

Falsifiability

Also called: falsifiable, falsification principle, refutability

Falsifiability is the property of a claim or theory that some possible observation could show it to be false. Karl Popper proposed it as the line between science and non-science: no number of confirming cases can prove a universal law, but a single real exception can overturn it. A theory that fits every conceivable result therefore makes no testable claim.

Falsificationism

Also called: Popperian falsificationism, falsificationist, falsificationist methodology

Falsificationism is Karl Popper's account of scientific method, in which scientists propose bold conjectures and then try hard to refute them, keeping a theory only while it survives testing. A passed test corroborates a theory without proving it. Popper also judged that a theory rescued from refutation by ad hoc changes, made only to escape it, had stopped being scientific.

Feasibility study

Also called: feasibility studies, feasibility research

A feasibility study is a study that asks whether a planned piece of research can be done, whether it should go ahead and, if so, how, for example whether enough participants can be recruited. In the widely used 2016 framework by Eldridge and colleagues, feasibility is the umbrella term: every pilot study is a feasibility study, but not every feasibility study is a pilot.

Feminist research

Also called: feminist methodology, feminist research methodology, feminist methods, feminist research methods, feminist science

Feminist research is research shaped by feminist aims: showing how gender and power affect social life and knowledge itself, and producing knowledge that people in subordinate positions, especially women, can use. Writers disagree on method. Some hold that only qualitative methods true to women's own accounts are feminist, while others argue that feminists rightly use many methods, including quantitative ones.

FINER criteria

Also called: FINER, FINER framework, feasible, interesting, novel, ethical and relevant

The FINER criteria are a checklist for judging a proposed research question: is it feasible, interesting, novel, ethical and relevant. Suggested by Hulley and colleagues for clinical research, they test the question in general terms and are often paired with PICO or PICOT, which help to frame its specific parts, such as the population and the outcome.

Fixed design

Also called: fixed research design, fixed designs

A fixed design is a research design whose purpose, questions, methods and sampling are all settled before data collection begins and then followed as planned. Colin Robson's textbook Real World Research uses the term, setting it against flexible design. Experiments and questionnaire surveys are typical, but a fixed design can use any data that can be specified in advance, whether quantitative or qualitative.

Flexible design

Also called: flexible research design, flexible designs

A flexible design is a research design that is allowed to evolve during a study, with its questions, methods and sampling refined as the data come in. Robson contrasts it with a fixed design. Ethnography, action research and exploratory case studies are typical, though a design can be partly flexible, and either kind can use quantitative or qualitative data.

Formative evaluation

Formative evaluation is evaluation carried out while a programme is being developed, adapted or first delivered, to improve it rather than to pass a final verdict. It asks, before full roll-out, whether the plan can be delivered and whether the people it is meant for will find it suitable and acceptable. It contrasts with summative evaluation, which judges afterwards whether the programme met its goals.

Fourth-generation evaluation

Also called: fourth generation evaluation, constructivist evaluation, responsive constructivist evaluation

Fourth-generation evaluation is Guba and Lincoln's constructivist approach to evaluation, in which the claims, concerns and issues of all stakeholders are drawn out, then compared and negotiated, to reach judgements of a programme's merit and worth. It rests on a relativist ontology. Its authors warn against mixing its assumptions with positivist ones, though either kind of method may be used.

Goal-free evaluation

Also called: goal free evaluation, GFE, goal-free evaluator

Goal-free evaluation is evaluation carried out, at least in part, by evaluators who are deliberately not told a programme's stated goals, so that they judge what it actually does and look for all its effects, intended or not. Michael Scriven's Key Evaluation Checklist counts it among the main ways of finding unexpected side effects and treats a mix of goal-based and goal-free work as ideal.

Grand theory

Also called: grand theories, grand theorising, grand theorizing

Grand theory is highly abstract theorising that sets out to explain whole societies or social systems, often at the cost of contact with observable social life. C. Wright Mills coined the term in 1959 as a criticism, with Talcott Parsons's systems theory his main target. Merton's middle-range theory was proposed as a bridge between such theory and empirical research.

Historical research

Also called: historical research design, historical design, historical method, historical approach, historical study

Historical research is research that collects, verifies and interprets evidence from the past, such as diaries, official records, archives, maps and recordings, to establish what happened and to test explanations of it. It is unobtrusive, since the research cannot change the evidence, but it depends entirely on what documentation survives and on whether the sources are authentic.

Hypothesis

Also called: research hypothesis, hypotheses, scientific hypothesis, hypothesis statement

A hypothesis is a specific, testable prediction about what a study will find, usually stating an expected relationship between variables and often derived from a theory. A good one could be shown wrong by the data. Quantitative studies usually state hypotheses before collecting data, while much qualitative research starts from open questions instead and may produce hypotheses as a result.

Hypothetico-deductive method

Also called: hypothetico-deductive model, hypothetico-deductive approach, H-D method, hypothetico deductive method, hypothetico-deductive reasoning

The hypothetico-deductive method is the cycle of proposing a theory, deducing a testable hypothesis from it, testing that hypothesis against new observations and revising the theory in the light of the result. It is the textbook account of how quantitative science proceeds. On Popper's view, a passed test corroborates a theory without proving it, while a failed one counts against it.

Ideal type

Also called: ideal types, Idealtypus, Weberian ideal type

An ideal type is an analytical construct, introduced by Max Weber, that deliberately exaggerates certain features of a phenomenon, such as bureaucracy, to form a pure model against which real cases can be compared. It is not meant to describe any actual case, nor to be an ideal in the moral sense. Its worth lies in how useful it proves for analysis.

Idealism

Also called: idealist ontology, ontological idealism, philosophical idealism, idealist

Idealism is the philosophical view that reality is fundamentally mental, or that what we can know of the world is so shaped by the mind that it cannot be separated from our ideas of it. Philosophers distinguish ontological idealism, such as Berkeley's, from the epistemological version associated with Kant. In research philosophy it is usually set against realism and materialism.

Idiographic approach

Also called: idiographic, idiographic explanation, idiographic research, ideographic, ideographic approach

An idiographic approach seeks a full, detailed explanation of a particular person, group, event or setting at a particular time, rather than laws that hold across many cases. Its explanations tend to be rich, complex and specific to their context, and it is typical of qualitative and case study research. It is the counterpart of the nomothetic approach.

Impact evaluation

Also called: effectiveness evaluation

Impact evaluation is evaluation that asks what difference an intervention made: which outcomes, intended or not, followed from it and how much of the change can be put down to it rather than to other causes. It uses experimental, quasi-experimental or theory-based designs. Some guides separate outcome evaluation, which checks whether objectives were met, from impact evaluation of longer-term goals.

Implementation research

Also called: implementation science, dissemination and implementation research, D&I research, implementation studies

Implementation research is the study of methods for getting practices and interventions that have been shown to work adopted and built into routine health care and public health. It asks why proven practices are or are not taken up and tests ways of improving uptake, often in partnership with patients, practitioners and organisations. The field as a whole is called implementation science.

Incommensurability

Also called: incommensurable, incommensurability of paradigms, paradigm incommensurability, Kuhnian incommensurability

Incommensurability is Thomas Kuhn's claim that theories from different paradigms lack a common measure, because their standards, their observations and even the meanings of key terms differ, so they cannot be compared point by point. Kuhn stressed that this did not make them wholly incomparable. The idea was later applied to research paradigms, in the argument that they cannot be combined in one study.

Indigenous research methodologies

Also called: Indigenous methodologies, Indigenous research methods, Indigenous research paradigm, Indigenous methodology, Indigenous research methodology

Indigenous research methodologies are approaches to research that grow from Indigenous knowledge systems, values and ways of knowing rather than being adapted from Western paradigms. Shawn Wilson describes an Indigenous research paradigm with its own ontology, epistemology, axiology and methodology. Margaret Kovach notes that they share ground with some qualitative approaches while remaining distinct, for example in using story as method.

Inductive reasoning

Also called: induction, inductive approach, inductive research, bottom-up approach, theory-building approach, inductive logic, inductive inference, inductive method

Inductive reasoning in research moves from the specific to the general: the researcher collects data first, looks for patterns and builds a theory or explanation from them. Its conclusions go beyond the evidence, so they are probable rather than certain. It is most often linked with qualitative research, though quantitative studies use it too, and it runs the opposite way to deductive reasoning.

Instrumental case study

Also called: instrumental case, instrumental case studies

An instrumental case study is case study research in which a case is chosen to gain a broader understanding of an issue or phenomenon, rather than for its own sake. Stake set it beside the intrinsic and the collective case study, and the types can overlap, as when a study begun for a unique case turns out to say something more general.

Interdisciplinary research

Also called: IDR, interdisciplinarity, interdisciplinary study, interdisciplinary approach

Interdisciplinary research is research that integrates data, methods, tools, concepts or theories from two or more disciplines to understand a problem, or to solve one that lies beyond any single discipline. The US National Academies stress integration: placing disciplines side by side, each doing its own part, is multidisciplinary research, and merely borrowing another discipline's technique is not interdisciplinary either.

Interpretivism

Also called: interpretivist paradigm, interpretive paradigm, interpretivist approach, interpretivist, anti-positivism, antipositivism, interpretive approach

Interpretivism is the research paradigm holding that social reality is made up of the meanings people give to their experience, so the researcher's task is to understand those meanings from the inside rather than to measure objective facts. It favours qualitative methods such as interviews, values context and accepts that the researcher is part of what is studied. Many textbooks group it with constructivism.

Intrinsic case study

Also called: intrinsic case, intrinsic case studies

An intrinsic case study is case study research carried out because the case itself is of interest, usually for its uniqueness, rather than because it represents other cases. The case is chosen on its own merits and the aim is to understand it well, not to generalise from it. In Stake's typology it contrasts with the instrumental and the collective case study.

Kaupapa Māori research

Also called: Kaupapa Maori research, Kaupapa Māori, Kaupapa Maori, Kaupapa Māori methodology, Kaupapa Maori methodology

Kaupapa Māori research is an approach to research, developed in New Zealand, that privileges Māori values, knowledge and ways of knowing in order to build a research framework that is culturally safe. Linda Tuhiwai Smith traces the move from Māori as the researched to Māori as researchers, and argues that being Māori is essential to carrying out such research, though non-Māori may still take part.

Limitations

Also called: limitations of the study, study limitations, research limitations, limitation

Limitations are the weaknesses and constraints in a study that restrict how far its findings can be trusted or generalised, such as a small sample, missing data, self-reported answers or refused access. Guides differ on whether they include the consequences of the researcher's own design choices or only things outside the researcher's control. They are usually discussed in the discussion section, with what they mean for future research.

Logic model

Also called: programme logic model, program logic model

A logic model is a diagram of how a programme is meant to work, linking its inputs, such as staff and money, through its activities and outputs to the outcomes and longer-term impact it is expected to produce. Evaluators use it to decide what to measure at each stage. The UK Treasury's Magenta Book treats it as one way of drawing a programme's theory of change.

Logical framework

Also called: logframe, log frame, logical framework approach, LFA, logframe matrix, logical framework matrix

A logical framework, or logframe, is a management tool for designing an intervention that sets out its inputs, activities, outputs, outcomes and impacts, how each is expected to lead to the next, the indicators for each and the assumptions or risks that may affect success. It is used most often for single projects and supports planning, monitoring and evaluation.

Logical positivism

Also called: logical empiricism, neopositivism, logical positivist, logical positivists

Logical positivism is the philosophy of science developed by the Vienna Circle in the 1920s and 1930s, which held that a factual statement is meaningful only if experience could test it, so metaphysics is not false but meaningless. It is also called logical empiricism or neopositivism. It no longer survives as a movement, partly because no satisfactory version of its criterion of meaning could be found.

Longitudinal design

Also called: longitudinal study, longitudinal research, longitudinal

A longitudinal design collects data from the same people, groups or organisations repeatedly over a period of time, so that change and the order of events can be studied. It can show which came first, which a cross-sectional design cannot, but it is slow, often costly and tends to lose participants along the way. It contrasts with a cross-sectional design.

Meta-evaluation

Also called: metaevaluation, meta evaluation, evaluation of evaluations, evaluation of an evaluation

Meta-evaluation is the evaluation of an evaluation, or of a body of evaluations, to judge its quality or to scrutinise how well the evaluators performed. It helps decide how far an evaluation's findings can be trusted before they are acted on. Patton's utilisation-focused approach ends with a meta-evaluation of how far the findings were actually used.

Meta-research

Also called: metascience, meta-science, metaresearch, research on research

Meta-research is the scientific study of research itself, aimed at evaluating and improving how research is done, reported, checked, corrected and rewarded. It grew quickly in the 2010s alongside concern about the replication crisis, and its projects include large collaborative replications, checks on whether published analyses can be reproduced, and studies of the incentives researchers face.

Methodological choice

Also called: methodological choices, choice of methodology

Methodological choice is the layer of Saunders's research onion at which a researcher decides whether a study will be quantitative, qualitative or mixed methods, and whether it will use one method or several. It sits between the approach to theory development and the research strategy, and together with strategy and time horizon it makes up what Saunders and colleagues call research design.

Methodological holism

Also called: holism, holist explanation, holistic explanation, methodological holist

Methodological holism is the view that social phenomena such as institutions, structures and cultures must feature in social science explanations and cannot always be replaced by accounts of individuals. It contrasts with methodological individualism. An explanation that credits a school's culture with its pupils' results is holist, while one built from each pupil's choices and motives is individualist.

Methodological individualism

Also called: methodological individualist, individualist explanation

Methodological individualism is the principle that social phenomena should be explained by showing how they arise from the actions of individuals and the beliefs and intentions behind those actions. Max Weber introduced it as a rule for the social sciences, and it is sometimes put as the demand that large-scale explanations rest on individual-level foundations. It contrasts with methodological holism.

Middle-range theory

Also called: middle range theory, theory of the middle range, theories of the middle range, mid-range theory, middle-range theories

Middle-range theory is theory pitched between the small working hypotheses of everyday research and all-embracing grand theory, specific enough to be tested against evidence yet general enough to apply across many cases. Robert Merton advocated it in 1957 as the way to link sociological theory with empirical research, in reaction to the abstract systems theory of Talcott Parsons.

Mill's methods

Also called: Mill's methods of induction, Mill's canons, Mill's methods of experimental inquiry, Mills methods

Mill's methods are John Stuart Mill's rules for identifying causes by comparing instances: the methods of agreement, difference, residues and concomitant variations, with a joint method of agreement and difference. The method of difference, for example, looks for the one factor present when an effect occurs and absent when it does not. Their conclusions are provisional, since unnoticed background conditions can mislead.

Model

Also called: research model, theoretical model, conceptual model

A model, in research, is a simplified representation of all or part of a system, built to study how that system works or what drives it. Where a theory tries to explain a phenomenon, a model tries to represent it, and a model can be useful for decisions without explaining anything. Models may be mathematical, network or path models, and descriptive, predictive or normative.

Monitoring and evaluation

Also called: M&E, monitoring & evaluation, M and E

Monitoring and evaluation, often shortened to M&E, is the pairing of monitoring, the continuous collection of data on indicators to track an intervention's progress, with evaluation, the periodic and systematic assessment of its design, implementation and results. Monitoring shows whether things are on track and whether funds are being used as planned, while evaluation asks why, and with what effect.

Most different systems design

Also called: MDSD, most-different systems design, most different cases design

A most different systems design is a comparative research strategy that compares cases differing in most respects but sharing the same outcome. The researcher looks for the few features the cases have in common, since one of them may explain the shared outcome. It is the reverse of the most similar systems design and, like it, suits studies with only a few cases.

Most significant change

Also called: MSC, most significant change technique, most significant change method, MSC technique

Most significant change is a participatory monitoring and evaluation method in which stories of change are collected from people involved in a programme, and panels of stakeholders then discuss them and select those they judge most significant. It suits complex interventions whose outcomes are hard to predict or vary widely between people. Rick Davies developed it in the mid-1990s.

Most similar systems design

Also called: MSSD, most-similar systems design, most similar cases design

A most similar systems design is a comparative research strategy that compares cases alike in most respects but different in the outcome of interest. Because their shared features are in effect held constant, the few differences between the cases become the candidate explanations. It is the reverse of the most different systems design.

Multidisciplinary research

Also called: multidisciplinarity, multidisciplinary approach, multi-disciplinary research, multidisciplinary study

Multidisciplinary research is research that brings together more than one discipline, each making its own separate contribution to a shared problem, without their ideas being integrated. Researchers may share facilities and approaches while working on distinct parts. The US National Academies distinguish it from interdisciplinary research, which synthesises the disciplines' methods and ideas into something new.

Multiple-case study

Also called: multiple case study, collective case study, multiple-case design, multi-case study

A multiple-case study is case study research that examines two or more cases, together or one after another, so that findings can be compared across them. Stake calls it a collective case study. Comparing cases shows whether a pattern found in one holds in others, a line of reasoning often called replication logic, which a single-case study cannot offer.

Needs assessment

Also called: needs analysis, needs assessments, community needs assessment

A needs assessment is a systematic process for identifying needs, meaning gaps between the current situation and the desired one, and setting priorities among them before a programme is designed or resources are allocated. It is an early step in developing a programme and can strengthen a bid for funding. Witkin and Altschuld divide it into pre-assessment, assessment and post-assessment.

Nominalism

Also called: nominalist, nominalist ontology

Nominalism is the philosophical view that general categories and universals, such as redness or justice, do not exist in themselves but are only names for groups of similar particular things. In social research, Burrell and Morgan used it for the ontological position that social reality is little more than the concepts and shared meanings people use to describe it, the opposite of realism.

Nomothetic approach

Also called: nomothetic, nomothetic explanation, nomothetic research

A nomothetic approach seeks general explanations that hold across many cases, usually expressed as relationships between variables that allow predictions beyond the people studied. It is typical of quantitative, explanatory research that samples from a population in order to generalise. It gives up the detail of any single case for breadth, and it is the counterpart of the idiographic approach.

Non-directional hypothesis

Also called: nondirectional hypothesis, two-tailed hypothesis, two-sided hypothesis

A non-directional hypothesis predicts that a difference or relationship exists without saying which way it runs, for example that two teaching methods will produce different scores. Its alternative hypothesis uses not equal to, which leads to a two-tailed test that can pick up an effect in either direction. It contrasts with a directional hypothesis.

Normal science

Also called: Kuhn's normal science, Kuhnian normal science

Normal science is Thomas Kuhn's term for the ordinary, cumulative work of a scientific community within an accepted paradigm, which he likened to puzzle-solving: the problems and methods are familiar and success depends mainly on the researcher's skill. It alternates with rare revolutionary periods, in which a crisis over unresolved anomalies leads to a paradigm shift.

Null hypothesis

Also called: H0, H₀, H nought, null hypotheses

The null hypothesis is the statement that there is no difference or no relationship in the population, written H0 and always containing an equality. A significance test asks whether the data would be unlikely enough if it were true to justify rejecting it. Failing to reject it does not prove it true, only that the sample did not give enough evidence against it.

Objectivism

Also called: objectivist, objectivist ontology, objectivist epistemology, epistemological objectivism

Objectivism is the position that things, and the facts and meanings attached to them, exist independently of anyone's awareness of them, so knowledge is there to be discovered rather than constructed. The word is used for both an ontological and an epistemological stance, and it opposes subjectivism and the view that knowledge is a social product. Research taking this position usually favours quantitative methods.

Objectivity

Also called: scientific objectivity, research objectivity, objective research

Objectivity, in research, is the ideal that findings should reflect the facts rather than the personal views, interests or values of whoever produced them. Philosophers understand it in several ways: as faithfulness to facts, as freedom from values, or as a property of communities that criticise one another's work. A completely perspective-free view from nowhere is now widely thought unattainable.

OECD DAC evaluation criteria

Also called: DAC criteria, OECD DAC criteria, DAC evaluation criteria, OECD evaluation criteria

The OECD DAC evaluation criteria are six standard headings for judging an intervention, set by the OECD's Development Assistance Committee: relevance, coherence, effectiveness, efficiency, impact and sustainability. Their definitions were revised and approved in 2019. First written for development co-operation, they are now used much more widely, including in the evaluation of domestic public policy.

Ontology

Also called: ontological assumptions, ontological position, ontological stance

Ontology is the branch of philosophy concerned with the nature of being: what exists and what reality is like. In research it names a study's assumptions about whether one reality exists independently of people, a realist position, or whether reality is constructed and multiple, a relativist one. These assumptions shape the epistemology and methods that follow.

Outcome evaluation

Also called: outcome evaluations, outcomes evaluation

Outcome evaluation is evaluation that measures a programme's effects on its target population by tracking progress towards the outcomes it set out to achieve, especially the shorter-term changes it is expected to affect most directly. The CDC pairs it with effectiveness evaluation and separates it from impact evaluation, which looks at a programme's longer-term, ultimate goals.

Outcome harvesting

Also called: outcomes harvesting

Outcome harvesting is a monitoring and evaluation method that first collects evidence of changes that have happened and then works backwards to establish whether and how an intervention contributed to them. It can run throughout a programme to give real-time information, and is designed to involve stakeholders. It was partly inspired by outcome mapping, and the two are often used together.

Paradigm shift

Also called: scientific revolution, paradigm change, paradigm shifts, Kuhnian revolution

A paradigm shift is the common name for what Thomas Kuhn called a scientific revolution, in which a field abandons its accepted framework of theories, methods and assumptions for a new one after anomalies it cannot explain produce a crisis. Kuhn argued that such change is not simply cumulative progress, since some achievements of the earlier paradigm are lost.

Parsimony

Also called: principle of parsimony, Occam's razor, Ockham's razor, law of parsimony, Occams razor

Parsimony is the principle that, other things being equal, a simpler explanation, one that assumes fewer entities or causes, should be preferred to a more complex one. It is best known as Occam's razor. Researchers invoke it when choosing between theories or statistical models that fit the evidence equally well, though philosophers still debate why simplicity should be a sign of truth.

Participatory action research

Also called: PAR, participative action research

Participatory action research is action research in which the people affected by a problem take part as co-researchers rather than subjects, helping to set the questions, collect and analyse the data and act on the findings. It aims at empowerment and social change as well as knowledge, often with marginalised communities. Community-based participatory research, common in health, is a closely related approach.

Participatory evaluation

Also called: participatory monitoring and evaluation, PM&E, participatory M&E

Participatory evaluation is evaluation in which the people with a stake in an intervention, including those it is meant to serve, take an active part in planning the evaluation, collecting and interpreting data, and drawing conclusions and recommendations. Michael Scriven notes that participation can improve how an evaluation is carried out but may cost it some credibility, a trade-off evaluators must manage.

Participatory paradigm

Also called: participatory worldview, participative worldview, participatory inquiry paradigm, participative paradigm

The participatory paradigm is a research worldview set out by John Heron and Peter Reason in 1997, which holds that reality is subjective-objective: people shape the world they know while also meeting a reality beyond themselves. Its extended epistemology recognises four ways of knowing (experiential, presentational, propositional and practical), and its method is co-operative inquiry.

Philosophical research

Also called: philosophical research design, philosophical design, philosophical analysis, philosophical inquiry

Philosophical research is an approach that uses argument and conceptual analysis, rather than new data, to examine and challenge the assumptions underlying a field, such as its ideas about what is real, how it can be known and what is valued. It is valued for clarifying concepts and guiding ethical practice, but it can be abstract and hard to apply to practical problems.

Pilot study

Also called: pilot, pilot trial, pilot project

A pilot study is a smaller-scale run of a planned study, or of part of it, carried out to see whether its procedures work before the full study begins. It may include random allocation, as in a randomised pilot trial, or not. In the 2016 framework by Eldridge and colleagues it is one kind of feasibility study, whereas older UK guidance treated pilot and feasibility studies as separate kinds.

Positivism

Also called: positivist paradigm, positivist approach, positivist, positivist research

Positivism is the research paradigm holding that a single objective reality exists and can be known through systematic observation and measurement, with the researcher standing apart from what is studied. It seeks general, law-like explanations, tests hypotheses deductively and favours quantitative methods such as experiments and surveys. It aims for research free of the researcher's own values.

Post-positivism

Also called: postpositivism, post-positivist paradigm, postpositivist, post-positivist, post positivism

Post-positivism is the research paradigm holding that an objective reality exists but can only be known approximately and never with certainty, because observation is shaped by researchers' theories, values and biases. It keeps the aim of objectivity while treating every finding as provisional and open to correction, and it accepts qualitative as well as quantitative methods. Some textbooks count critical realism as one form of it.

Postmodernism

Also called: postmodern paradigm, postmodernist approach, post-modernism, postmodernist, postmodern research

Postmodernism, as a research paradigm, is the view that what counts as truth depends on history and culture, so no single explanation of social life can hold everywhere. Researchers working in it question the claims of established knowledge and treat every account of reality, including their own, as one version among many rather than the final word.

Poststructuralism

Also called: post-structuralism, poststructuralist, post-structuralist, poststructuralist research

Poststructuralism is a movement in philosophy and social theory, associated with French thinkers such as Derrida and Foucault, that grew out of and against structuralism. It holds that meaning is never fixed but is produced through language and discourse, so that seemingly stable categories and truths are open to challenge. It overlaps heavily with postmodernism, and the labels are often used together.

Pragmatism

Also called: pragmatist paradigm, pragmatic paradigm, pragmatic approach, pragmatist, pragmatic worldview

Pragmatism, as a research paradigm, sets aside debates about the ultimate nature of reality and judges an approach by whether it answers the research question and has practical use. It lets researchers choose and combine methods to suit the problem, which is why it is often linked to mixed methods research. It draws on the American philosophers Charles Peirce, William James and John Dewey.

Predictive research

Also called: predictive study, predictive design, prediction research, predictive studies

Predictive research is research that uses known, systematic relationships between variables to forecast whether an event or behaviour will occur, such as which students are likely to leave a course early. Prediction is one of the three classic goals of science, with description and explanation. It needs only a reliable relationship, while working out why that relationship holds is the separate task of explanation.

Primary research

Also called: original research

Primary research is research that gathers fresh data for the question at hand, through surveys, interviews, experiments, observation or other methods the researcher sets up. The data fit the question closely, but gathering them takes more time and resources than using what already exists. It contrasts with secondary research, which analyses information others have already collected.

Problem of induction

Also called: Hume's problem, Hume's problem of induction, induction problem

The problem of induction is the difficulty, set out by David Hume in the eighteenth century, of justifying the step from observed cases to unobserved ones, such as from past sunrises to tomorrow's. Any argument for that step seems to assume that the future will resemble the past, which is exactly what needs showing. Popper's falsificationism and Bayesian approaches are among the responses.

Process evaluation

Also called: implementation evaluation

Process evaluation is evaluation of how an intervention was delivered: whether it was carried out as intended, what worked well or badly and why, and how the setting affected delivery. It uses qualitative and quantitative methods and is often run alongside an outcome or impact evaluation, since a programme that shows no effect may simply not have been delivered as planned.

Process tracing

Also called: process-tracing, process tracing method

Process tracing is a method for testing whether a hypothesised causal mechanism explains how a particular outcome came about in a single case. The researcher works out what evidence should appear if the explanation is true and what if it is false, then looks for it, giving most weight to evidence that could exist under only one explanation. It is used in case study research and theory-based evaluation.

Programme theory

Also called: program theory, intervention logic, programme theories, program theories

Programme theory is an explicit account of how and why an intervention is expected to produce its results, including the assumptions behind each causal step. The UK Magenta Book notes that it is often mapped as a theory of change, a logic model or a systems map, and the OECD's term intervention logic means much the same. Evaluators use it to decide what to test.

Proposition

Also called: theoretical proposition, propositions, theoretical propositions

A proposition is a statement of the relationship between two or more constructs at the level of theory, such as that social support reduces stress. Because constructs are abstract, a proposition cannot be tested directly. It is tested through hypotheses, which restate it as relationships between measurable variables, such as scores on a support questionnaire and a stress scale, and one proposition can yield many hypotheses.

Pseudoscience

Also called: pseudo-science, pseudoscientific, pseudo-scientific

Pseudoscience is a belief or practice presented as scientific that does not meet the standards of science, for example because it ignores contrary evidence, cannot be tested or keeps its evidence away from public scrutiny. Astrology and homeopathy are standard examples. What separates it from ordinary scientific error is the sustained effort to claim scientific authority for it.

Quantitative research

Also called: quantitative methods, quantitative approach, quantitative study, quantitative methodology, quantitative research methods, quantitative inquiry

Quantitative research is research that collects numerical data, or turns observations into numbers, and analyses them statistically to describe, compare, predict or explain, often generalising from a sample to a population. It relies on structured instruments and objective measurement and usually tests hypotheses. It contrasts with qualitative research, which works mainly with words and meanings, and the two are combined in mixed methods research.

Rationale

Also called: research rationale, rationale for the study, justification for the study, significance of the study

The rationale of a study is the argument for why it is worth doing: what gap in knowledge or practical problem it addresses, who will benefit and what it adds to earlier work. It usually appears in the introduction of a proposal or thesis and draws on the literature review. A clear rationale is what persuades a supervisor or funder that the study deserves time and resources.

Rationalism

Also called: rationalist, rationalist epistemology

Rationalism is the philosophical view that some significant knowledge and concepts are gained through reason, independently of sense experience, for example by intuition and deduction or because they are innate. Descartes, Spinoza and Leibniz are its classic figures. It contrasts with empiricism, and the dispute between the two concerns how far knowledge of the world depends on experience.

Realist evaluation

Also called: realistic evaluation

Realist evaluation is a theory-based approach to evaluation that asks what works, for whom, in what respects, how and under what circumstances, rather than simply whether a programme works. It tests theories about the mechanisms, often people's reasoning and choices, through which an intervention produces its results, and about how context affects them. The UK Treasury's Magenta Book lists it among the main theory-based methods.

Realist ontology

Also called: ontological realism, realism

A realist ontology is the position that reality exists independently of anyone observing or thinking about it, so there are facts to be found whether or not anyone knows them. Positivism, post-positivism and critical realism all rest on it, though they differ over how well that reality can be known. It contrasts with a relativist ontology.

Reductionism

Also called: reductionist, reductionist explanation, reductionist approach

Reductionism is the approach of explaining something complex by breaking it into simpler parts or lower-level units, such as explaining an organisation through its individual members. In research design the word also names an error: drawing conclusions about a group or larger unit from data about individuals or smaller units, the reverse of the ecological fallacy.

Relativist ontology

Also called: ontological relativism, relativism, anti-realism

A relativist ontology is the position that reality, or at least social reality, is constructed by people and so exists in many versions that depend on the observer and the context. Interpretivism and constructivism rest on it. It contrasts with a realist ontology, and the choice between the two shapes whether a study sets out to measure or to interpret.

Replicability

Also called: replicable, replicability of findings

Replicability is the extent to which a finding holds up when a new study asks the same question and collects its own data. The US National Academies define it this way and keep reproducibility for rerunning the original analysis on the original data. Other fields use the two words the other way round or interchangeably, so it is worth checking what an author means.

Replication crisis

Also called: reproducibility crisis, replicability crisis, crisis of replication

The replication crisis is the name given, from the 2010s, to the discovery that many published findings, especially in psychology, medicine and the life sciences, did not hold up when other researchers repeated the studies. The best-known test, the Reproducibility Project: Psychology, published in 2015, repeated 100 studies and reproduced between 36 and 47 per cent of results, depending on the measure used.

Replication study

Also called: replication, replication research, replication studies

A replication study is a study that repeats an earlier one to see whether its findings hold. It may follow the original methods as closely as possible, a direct replication, or test the same hypothesis in a deliberately different way, a conceptual replication, and in practice most sit somewhere between. Replication is how science checks that a result was not a fluke or an error.

Reproducibility

Also called: reproducible research, computational reproducibility, reproducible

Reproducibility, as the US National Academies use the word, means that someone rerunning the original analysis, on the same data with the same code and settings, gets consistent results. It checks that an analysis was reported fully and done correctly, not that the finding is true. Some fields use the word for what others call replicability, which means repeating a study with new data.

Research aim

Also called: aim, research aims, aim of the study, purpose of the study, purpose statement, research goal, study aim

A research aim is a broad statement of the overall purpose of a study, the main thing it intends to find out or achieve. A study usually has one aim, or a few, which the research objectives then break into concrete steps. The aim says what is to be accomplished while the objectives say how, as in an aim to understand how first-year students choose sources for essays.

Research and development

Also called: R&D, research and experimental development, R and D

Research and development, as the OECD's Frascati Manual defines it, is creative and systematic work undertaken to increase knowledge, including knowledge of people, culture and society, and to find new uses for what is known. To count, an activity must be novel, creative, uncertain in outcome, systematic, and transferable or reproducible. It covers basic research, applied research and experimental development.

Research approach

Also called: approach to theory development, research approaches, approach to research

A research approach is the broad route a study takes between theory and evidence. In Saunders's research onion it means the approach to theory development: deductive (testing an existing theory), inductive (building one from data) or abductive (moving between the two). Other guides use the phrase for the broader choice between quantitative, qualitative and mixed methods research.

Research design

Also called: study design, research designs, design of the study

Research design is the overall plan of a study that links its question to the evidence needed to answer it: what data will be collected, from whom, when and how, and how they will be analysed. A good design is chosen to fit the question, rather than a favourite method being made to fit. Broad families include experimental, observational, survey, case study, action and mixed methods designs.

Research gap

Also called: gap in the literature, literature gap, knowledge gap, gap in knowledge

A research gap is a question, population, setting or method that existing research has not yet covered, or has covered poorly, and that a new study can address. It is usually found through a literature review and forms part of a study's rationale. A gap is worth filling only if the answer would matter, not merely because nobody has asked the question before.

How to find sources for a literature review fast

Research method

Also called: research methods, method, methods

A research method is a specific technique for collecting or analysing data, such as a questionnaire, an interview, an experiment, a focus group or a regression analysis. It is narrower than methodology, which is the reasoning that explains why those methods suit the question. A methods section describes what was done in enough detail for another researcher to repeat it.

Research methodology

Also called: methodology, methodological approach, research methodologies

Research methodology is the reasoning behind how a study is done: the strategy that links its philosophical assumptions and its question to the methods chosen, and the argument for why those methods are appropriate. In a thesis the methodology chapter justifies choices as well as describing them. In everyday use the word is often used loosely to mean the methods themselves.

Research objectives

Also called: objectives, research objective, objectives of the study, specific objectives, study objectives

Research objectives are the specific tasks a study must complete to fulfil its aim, usually written as a short numbered list starting with verbs such as identify, compare or assess. They are often expected to be SMART: specific, measurable, achievable, realistic and time-bound. Where the aim states what the study will accomplish, the objectives state how.

Research onion

Also called: Saunders' research onion, Saunders's research onion, Saunders research onion, research onion model

The research onion is a model by Saunders, Lewis and Thornhill that pictures the decisions behind a study as layers, worked through from the outside in: research philosophy, approach to theory development, methodological choice, research strategy, time horizon, and finally the techniques and procedures for collecting and analysing data. Each outer choice shapes the ones inside it.

Research paradigm

Also called: paradigm, research paradigms, worldview, philosophical worldview, research philosophy, philosophical paradigm

A research paradigm is a shared set of philosophical assumptions, about what is real, how it can be known and what part values play, that shapes the questions researchers ask and the methods they trust. Common examples are positivism, post-positivism, interpretivism or constructivism, critical theory and pragmatism. Textbooks list and group them differently, and some call the same idea a worldview or a research philosophy.

Research problem

Also called: problem statement, statement of the problem, research problem statement

A research problem is the specific issue, difficulty or gap in understanding that a study sets out to address, stated clearly enough to be investigated. It is narrower than a topic and comes before the research questions, which break it into answerable parts. The problem statement, usually a short paragraph, says what the problem is, why it matters and whom it affects.

Research programme (Lakatos)

Also called: research program (Lakatos), methodology of scientific research programmes, Lakatosian research programme, Lakatos research programme, scientific research programme

A research programme, in Imre Lakatos's philosophy of science, is a series of theories sharing a hard core of assumptions that researchers agree not to question, protected by a belt of auxiliary hypotheses that are adjusted when predictions fail. A programme is progressive while it keeps predicting new facts that are then confirmed, and degenerating when it only patches itself to fit results.

Research proposal

Also called: dissertation proposal, thesis proposal, prospectus, project proposal

A research proposal is a document that makes the case for a planned study, setting out its question, its importance and the way it will be carried out, written to win approval, a place or funding before the work begins. It usually covers the background and rationale, a short literature review, the questions, aims and objectives, the design and methods, often ethics and a timetable, and references.

Research proposal references

Research protocol

Also called: study protocol, protocol, trial protocol, clinical trial protocol, research study protocol

A research protocol is the document that sets out in full how a study will be carried out, including its background and rationale, objectives, design, methods, statistical considerations and organisation. Written before the study starts, it serves the team as a manual so that everyone follows the same methods, and it is used to monitor progress. Later changes are made through formal amendments.

Research question

Also called: RQ, research questions, study question, primary research question, main research question

A research question is the question a study is designed to answer, framed clearly and narrowly enough that evidence can answer it. A good one names its concepts and population, cannot be settled with a simple yes or no, and has more than one plausible answer. It is usually refined several times as the reading and planning go on.

Research strategy

Also called: research strategies, strategy of inquiry, strategies of inquiry

A research strategy is the overall plan of action by which a study will answer its question, such as an experiment, survey, case study, ethnography, action research, grounded theory or archival research. In Saunders's research onion it is the layer between methodological choice and time horizon. Methods are the specific techniques used within the chosen strategy, such as interviews or questionnaires.

Research topic

Also called: dissertation topic, thesis topic, research topics, topic of research

A research topic is the broad subject area a study is about, such as social media and teenagers' sleep, before it has been narrowed into a specific research problem and questions. Choosing one usually starts from personal interests, earlier modules and existing research. Students are advised to agree it with a supervisor early, since a topic is not yet a researchable question.

Retroduction

Also called: retroductive reasoning, retroductive inference, retroductive analysis, retroductive theorising

Retroduction is reasoning back from observed patterns to the underlying mechanisms and conditions that could have produced them, asking what must be true for these events to occur. It is central to critical realist and realist research, where it moves in repeated cycles between data and theory. Some writers, following Hanson, use it as another word for abduction, while critical realists treat the two as distinct.

Rigour

Also called: rigor, research rigour, research rigor, scientific rigour, scientific rigor, methodological rigour, methodological rigor

Rigour is the careful, systematic and transparent application of appropriate methods throughout a study, so that its design, analysis, interpretation and reporting are robust and as free from bias as possible. The US National Institutes of Health define scientific rigour in these terms and add full reporting so that others can reproduce the work. Qualitative researchers often judge rigour through trustworthiness criteria instead.

Rival hypothesis

Also called: rival explanation, alternative explanation, rival hypotheses, plausible rival hypothesis, competing hypothesis, competing explanation

A rival hypothesis is an alternative explanation that could account for a study's results as well as, or better than, the explanation the researcher proposes, such as a confounding variable or a coincidental event. Strong designs are built to rule plausible rivals out, and theory-based evaluation tests them explicitly. A finding is more convincing the more rival explanations it has eliminated.

Robustness (of findings)

Also called: robust results, robust findings, robustness of results, robust result

Robustness, in the vocabulary of reproducible research, is the quality of a result that holds when the same data are analysed in different reasonable ways, for example with another analysis pipeline or software. The Turing Way sets it beside reproducibility (the same analysis on the same data) and replicability (the same analysis on new data). Robust and replicable results together support generalisation.

Scientific method

Also called: the scientific method, scientific methods, scientific approach

The scientific method is the set of activities by which science builds reliable knowledge, usually said to include systematic observation and experiment, inductive and deductive reasoning, building models, and forming and testing hypotheses and theories. How these are carried out varies greatly between fields, and philosophers dispute whether any fixed set of methods is common to all the sciences and to them alone.

Scope

Also called: scope of the study, scope of research, research scope

The scope of a study is the extent of what it covers: the topic, population, setting, period and questions it takes on. Setting a realistic scope, given the time, skills and resources available, is one of the first design decisions. Delimitations are the stated choices that fix the scope, and limitations are the weaknesses found within it.

Secondary research

Also called: desk research, documentary research

Secondary research is research built on data or findings that already exist, such as census data, earlier studies, official records or archives, rather than on new data the researcher gathers. Literature reviews and meta-analyses are forms of it. It saves the work of new data collection, but the researcher must use data gathered for someone else's purpose.

Find sources

Secondary research question

Also called: sub-question, sub-questions, subsidiary research question, subsidiary question, sub-research question, secondary question

A secondary research question is a subsidiary question that a study also sets out to answer, supporting or extending its primary research question. Like the primary question, it should be set at the planning stage rather than added after the data are seen. It must never compromise the primary question, which drives the study's hypothesis and main objectives.

Single-case study

Also called: single case study, single-case case study

A single-case study is case study research that examines one case in depth, such as one organisation, programme or community. It suits a case that is unique or of special interest in itself, and it allows great detail, but it offers no comparison between cases. It differs from a single-case experimental design, which repeatedly measures one participant under changing conditions.

Social constructionism

Also called: constructionism, social constructionist paradigm, social constructionist

Social constructionism is the view that what people take to be real and true is produced through social interaction, language and shared agreement, and so varies between groups and changes over time. Research in this tradition studies how those shared meanings are made. Some textbooks treat it as another name for interpretivism, while others separate it from constructivism, which centres on individual minds.

Social return on investment

Also called: SROI, social return on investment analysis, SROI analysis

Social return on investment is a framework for measuring the social value an intervention creates by putting money values on both its costs and its outcomes and summarising the result as a ratio, such as pounds of social value per pound invested. It is used especially by voluntary and community organisations and social enterprises, and is often regarded as a form of cost-benefit analysis.

Standpoint theory

Also called: standpoint epistemology, feminist standpoint theory, feminist standpoint epistemology, standpoint theories

Standpoint theory is the view in epistemology that people's social position shapes what they can know, and that subordinated groups can have an advantage in understanding the relations of power that affect them. Feminist standpoint theory, developed by writers such as Nancy Hartsock and Sandra Harding, claims such an advantage for women's standpoint on matters where gender is involved.

Statistical hypothesis

Also called: statistical hypotheses

A statistical hypothesis is a claim about a population, such as the value of a mean or a difference between proportions, stated precisely enough to be tested with sample data. Hypothesis testing sets two against each other, the null and the alternative. It is narrower than the substantive research hypothesis it stands for, and confusing the two is an old criticism of significance testing.

Stratified ontology

Also called: domains of reality, the real, the actual and the empirical, layered ontology, stratified reality

A stratified ontology is critical realism's picture of reality in three layers: the empirical (what is observed or experienced), the actual (the events that happen whether observed or not) and the real (the structures and mechanisms whose powers generate those events). Because mechanisms such as discrimination are often invisible, research must reason from observed effects back to their causes.

Subjectivism

Also called: subjectivist, subjectivist epistemology, epistemological subjectivism

Subjectivism is the epistemological position that knowledge depends on the knower's own perspective, feelings and interpretations, so there is no external or objective test of its truth. In research it underlies approaches that study what people think and feel about a situation, usually through qualitative methods such as interviews. It stands opposite objectivism.

Summative evaluation

Summative evaluation is evaluation that reaches a verdict on whether a programme achieved what it set out to do, usually once it has run for long enough or has ended. It includes outcome and impact evaluation and informs decisions about continuing, expanding or stopping the programme. It contrasts with formative evaluation, which is done earlier to improve the programme.

Systematic empiricism

Also called: systematic observation (science)

Systematic empiricism is learning about the world through carefully planned, recorded and analysed observation. On one widely taught account it is the first of three features that mark out science, alongside asking empirical questions and producing public knowledge. It is what separates science from casual observation, which is haphazard and rarely checked against anything.

Theoretical framework

Also called: theoretical lens, theoretical frameworks, theoretical frame

A theoretical framework is the structure of one or more existing theories, with their concepts and definitions, that a study uses to frame its question, guide its design and interpret its findings. The researcher assembles it for the particular study rather than finding it ready-made in the literature. Whether it differs from a conceptual framework is disputed, and the two are often used interchangeably.

Theory

Also called: scientific theory, theories

A theory is an organised explanation of how or why a set of phenomena occur, built from related concepts and statements, that goes beyond what has been directly observed and can be used to derive testable predictions. It is broader than a hypothesis, which is one specific prediction. Theories range from broad ones that explain whole social systems to narrow ones about a single process.

Theory of change

Also called: ToC, theories of change

A theory of change is an explicit account of how and why an intervention is expected to bring about its intended results, setting out the chain from inputs and activities to outputs, outcomes and impact, with the assumptions and evidence behind each step and the context it depends on. Evaluators usually build one with the people running the intervention and use it to decide what to test.

Theory-based evaluation

Also called: theory-based impact evaluation, theory-driven evaluation, theory driven evaluation, theory based evaluation

Theory-based evaluation is evaluation that judges an intervention's impact by testing, against several sources of evidence, whether the causal chain set out in its theory of change actually happened and whether other explanations can be ruled out. It asks why change occurred as well as how much. Realist evaluation, contribution analysis and process tracing are among its methods.

Theory-ladenness of observation

Also called: theory-ladenness, theory-laden observation, theory ladenness, theory-laden

Theory-ladenness of observation is the idea that what scientists observe, and how they interpret data, is shaped by the theories and assumptions they already hold, so there are no purely neutral observations. Hanson, Kuhn and Feyerabend used it to question the objectivity of observation, though many philosophers now argue that it does not make fair tests of theories impossible.

Time horizon

Also called: time horizons, time dimension

The time horizon of a study is the period over which it collects data: a single point in time, making it cross-sectional, or a series of points, making it longitudinal. In Saunders's research onion it is the layer that follows research strategy. The choice depends on the question, since only data gathered over time can show change or the order in which things happen.

Transdisciplinary research

Also called: transdisciplinarity, transdisciplinary approach, transdisciplinary study

Transdisciplinary research is research that integrates knowledge across academic disciplines and with people outside academia, such as practitioners, policymakers and communities, who help frame the problem and produce the knowledge. It is used for complex societal problems such as sustainability. Definitions are contested: another sense means a synthesis that goes beyond disciplines altogether, without necessarily involving outsiders.

Transformative paradigm

Also called: transformative worldview, transformative research paradigm, transformative approach

The transformative paradigm is a research philosophy, associated especially with Donna Mertens, that puts social justice and human rights at the centre of research. It holds that whose view of reality counts is shaped by power, and it works with marginalised communities, often through mixed methods, so that research can support social change.

Translational research

Also called: bench to bedside, bench-to-bedside research, translational medicine

Translational research is research that moves discoveries along the path from basic science to benefits for people, through preclinical research, clinical research, clinical implementation and public health. The US National Center for Advancing Translational Sciences stresses that these stages do not run in a straight line, since each informs the others. Translational science, in its usage, is the study of that process itself.

Unit of analysis

Also called: units of analysis, level of analysis, analytic unit

The unit of analysis is the kind of entity a study's conclusions are about, such as individuals, households, schools, organisations, countries or texts. It is set by the research question and decides what the data must describe and what conclusions may be drawn. It is not always the same as the unit of observation, the entity from which data are actually collected.

Unit of observation

Also called: units of observation, observation unit, observational unit

The unit of observation is the entity a study actually gathers data from or measures, which may differ from the unit the study draws conclusions about. For example, a study of how well schools support new pupils, with schools as the unit of analysis, might gather its data by interviewing individual teachers, who are then the unit of observation.

Utilisation-focused evaluation

Also called: utilization-focused evaluation, utilisation focused evaluation, utilization focused evaluation, U-FE, UFE

Utilisation-focused evaluation is Michael Quinn Patton's approach to evaluation, which holds that an evaluation should be judged by its usefulness and actual use. The evaluator works with the primary intended users from the start, designing every step around how they will use the findings, and follows up afterwards to support that use. Patton sets it out as a seventeen-step checklist.

Value neutrality

Also called: value-free research, value freedom, value-freedom, Wertfreiheit, value-free ideal, value-neutral research, value-free science

Value neutrality is the ideal that researchers should keep their own moral and political values out of how they gather evidence and assess theories, even though values inevitably guide which problems they choose to study. Max Weber called it value freedom. Positivism embraces it, while interpretivist, critical and feminist researchers argue that research is always value-laden and that values should be declared.

Verificationism

Also called: verification principle, verifiability principle, verifiability criterion of meaning, verification criterion, principle of verification

Verificationism is the logical positivists' principle that a statement about the world is meaningful only if experience could in principle confirm or disconfirm it, so untestable metaphysical claims are neither true nor false but meaningless. Attempts to state the principle precisely kept failing. Popper replaced verification with falsification, not as a test of meaning but as the mark of science.

Verstehen

Also called: interpretive understanding, Verstehen sociology, empathic understanding

Verstehen, German for understanding, is the approach, associated above all with Max Weber, of explaining human action by grasping the meanings and intentions it has for the people involved, rather than only observing it from outside. It underpins interpretivism. Weber held that such understanding should be linked to causal explanation and not rest on empathy or intuition alone.

Working hypothesis

Also called: working hypotheses, provisional hypothesis, tentative hypothesis

A working hypothesis is a provisional hypothesis adopted to guide an investigation while it is under way, held loosely and expected to be refined, replaced or dropped as evidence accumulates. Robert Merton put the working hypotheses researchers form in their daily work at one end of a scale, with middle-range theory above them and all-inclusive grand theory at the other end.