Terminology

…

The vocabulary of doctoral research, defined plainly and cross-linked so the connections between concepts are visible, not just the concepts themselves.

Also searchable as: a research glossary.

Core Concepts

Rigour: Rigour is the discipline of following a method carefully enough that someone else could check your work and get the same result. It's less about being clever and more about not cutting corners when nobody's watching.
Validity: Validity asks whether you're actually measuring what you think you're measuring. A survey can be perfectly reliable and still invalid if it measures the wrong thing entirely.
Reliability: Reliability is about consistency: would you get the same result if you repeated the measurement under the same conditions? A stopwatch that's always ten seconds fast is reliable but not accurate.
Reproducibility: Reproducibility means someone else, using your data and your exact steps, arrives at the same result. It's a check on the process, not necessarily on whether the underlying finding is true.
Positivism: Positivism holds that reality is objective and measurable, and that knowledge is built by testing hypotheses against observable, quantifiable data.
Realism: Realism holds that an objective reality exists independently of our perception of it, but accepts that our knowledge of it is always partial and filtered through theory.
Interpretivism: Interpretivism holds that reality is socially constructed, and that understanding people's meanings and experiences matters more than measuring an external, fixed truth.
Pragmatism: Pragmatism holds that the right method is whichever one answers your research question best, mixing quantitative and qualitative approaches rather than committing to a single worldview.
Ontology: Ontology is your assumption about the nature of reality itself: whether it's a single objective truth waiting to be discovered, or something constructed differently by different people.
Epistemology: Epistemology is your assumption about how knowledge can be known and justified: whether it comes from objective measurement, lived experience, or something in between.
Axiology: Axiology is the role your own values play in the research: whether you aim to stay neutral and detached, or acknowledge that your position shapes what you notice and how you interpret it.
Phenomenology: Phenomenology studies how people subjectively experience a phenomenon, aiming to describe the lived experience itself rather than explain its causes.
Deductive reasoning: Deductive reasoning starts from an existing theory, derives a testable hypothesis from it, and checks that hypothesis against data.
Inductive reasoning: Inductive reasoning starts from specific observations and builds toward a general pattern or theory, rather than testing one that already exists.
Abductive reasoning: Abductive reasoning moves back and forth between data and theory, forming the most plausible explanation for a surprising finding and checking it against further evidence.
Action research: Action research studies a real problem by intervening in it directly, cycling between taking action, observing the effect, and refining the approach, usually alongside the people affected.
Experimental research: Experimental research manipulates one variable under controlled conditions to isolate its effect on another, usually against a comparison group that didn't receive the manipulation.
Case study research: Case study research examines one instance, be it a person, organisation, or event, in enough depth to understand it in its real context, rather than sampling broadly across many.
Time horizon: A time horizon is whether your study captures a single point in time (cross-sectional) or tracks change over multiple points (longitudinal).
Reflexivity: Reflexivity is the practice of examining how your own background, assumptions, and position might be shaping your research, and being explicit about it rather than pretending it isn't there.
Research limitations: Research limitations are the specific constraints of your study, such as sample size, method, or scope, that bound how far its findings can reasonably be trusted or applied.
Empirical: Empirical means based on observation or data gathered directly, rather than on theory, opinion, or reasoning alone.
Generalisability: Generalisability is the extent to which findings from your specific sample can reasonably be expected to hold true for a wider population.
X variable (independent variable): The X variable is standard notation in quantitative research for the predictor variable, the one assumed to influence an outcome.
Y variable (dependent variable): The Y variable is standard notation for the outcome variable being predicted or explained, paired with the X variable in a regression relationship.
Constructivism: Constructivism is a paradigm holding that individuals actively construct their own understanding of reality through experience, closely related to interpretivism but emphasising the active building of meaning rather than just its interpretation.
Social constructionism: Social constructionism is the view that concepts and categories people treat as natural or fixed are actually produced through shared social processes, distinct from constructivism's focus on individual meaning-making.
Post-positivism: Post-positivism is a refinement of positivism acknowledging that objective truth exists but can only ever be imperfectly and probabilistically known, not directly and completely observed.
Objectivity: Objectivity is freedom from personal bias in observing, interpreting, or reporting research, an ideal positivist research aims for, and one interpretivist research treats as neither fully possible nor always desirable.
Subjectivity: Subjectivity is the influence of a researcher's own perspective, values, or experience on their research, treated as bias to minimise in positivist work and as a meaningful, disclosable factor in interpretivist work.
Bracketing: Bracketing is a phenomenological technique of deliberately setting aside one's own assumptions and prior knowledge to encounter participants' experience as freshly as possible.
Verstehen: Verstehen is an interpretivist concept meaning empathetic understanding, grasping social action from the meaning it holds for the people involved, rather than only from external, observable behaviour.
Falsifiability: Falsifiability is Popper's criterion that a genuinely scientific claim must be capable of being proven false by evidence, a claim that can never be disproven under any circumstance isn't testable in the scientific sense.
Paradigm shift: A paradigm shift is Kuhn's term for a fundamental change in a field's accepted assumptions and methods, not just an incremental addition to existing knowledge but a genuine change in the underlying framework itself.
Grand theory: A grand theory is a broad, abstract theoretical framework attempting to explain large-scale social phenomena in general, as opposed to a narrower, more specific theory.
Middle-range theory: A middle-range theory is a theory pitched between grand, abstract theorising and pure empirical description, specific enough to be tested directly against data, broad enough to apply beyond a single case.
Tacit knowledge: Tacit knowledge is knowledge that's difficult to articulate or write down, held through experience and skill rather than formal instruction, a foundational concept in Knowledge Management research.
Explicit knowledge: Explicit knowledge is knowledge that can be codified, written down, and transferred formally through documents or systems, distinct from tacit knowledge, which resists this kind of transfer.
Value-free research: Value-free research is the ideal, most closely associated with positivism, that a researcher's own values should not influence the conduct or interpretation of a study.
Value-laden research: Value-laden research is the view, associated with interpretivist and critical traditions, that a researcher's values inevitably and legitimately shape a study, and should be disclosed rather than pretended away.
Nomothetic approach: A nomothetic approach aims to establish general laws or patterns applicable across many cases, typical of positivist, quantitative research.
Idiographic approach: An idiographic approach focuses on understanding a single case or individual in its own depth and particularity, typical of interpretivist, qualitative research.
Emic perspective: An emic perspective is an insider's perspective, describing a culture or setting using the concepts and categories meaningful to the people within it.
Etic perspective: An etic perspective is an outsider's perspective, describing a culture or setting using categories and frameworks external to it, often for comparison across settings.
Bounded rationality: Bounded rationality is the idea that decision-making is limited by available information, cognitive capacity, and time, rather than being fully rational, a concept relevant to organisational and management research.
Systems thinking: Systems thinking is an approach to understanding a phenomenon by examining how its parts interrelate within a whole, rather than studying components in isolation.
Complexity theory: Complexity theory is a theoretical lens treating organisations and social systems as complex, adaptive systems, where outcomes emerge from interaction rather than being directly controlled or fully predictable.
Praxis: Praxis is the practical application of a theory or idea, particularly in practice-based and action research traditions where theory and practice are meant to inform each other continuously.
Theory-practice gap: The theory-practice gap is the recognised distance between what academic theory suggests and what actually happens in applied, real-world practice, a common motivating driver for practice-based doctoral research.
Parsimony: Parsimony is the principle that, between two explanations that fit the evidence equally well, the simpler one should generally be preferred.
Grounded theory: Grounded theory is a methodology building theory directly from systematically collected and analysed data, rather than testing a theory decided in advance.
Hermeneutics: Hermeneutics is the theory and practice of interpretation, especially of text and lived meaning, underpinning much interpretivist qualitative work.
Symbolic interactionism: Symbolic interactionism is a theoretical perspective holding that people act toward things based on the meanings those things hold for them, meanings built through social interaction.
Critical theory: Critical theory is a theoretical tradition examining how power, ideology, and social structures shape and constrain what's taken as normal or true, aiming toward emancipation, not just description.
Feminist research: Feminist research is a research tradition centering gender as an analytical lens, often explicitly rejecting claims of value-free neutrality in favour of disclosed positionality.
Postmodernism: Postmodernism is a stance sceptical of grand, universal explanations of reality, emphasising multiple, fragmented, and contested truths instead.
Structuralism: Structuralism is an approach explaining phenomena by reference to underlying structures and systems of relationships, rather than individual actors or events.
Poststructuralism: Poststructuralism is a critique and extension of structuralism, arguing that meaning is unstable and structures themselves are historically and socially constructed, not fixed.
Naturalistic inquiry: Naturalistic inquiry is research conducted in real-world settings as they naturally occur, without researcher manipulation of variables.
Verification vs. falsification: Verification seeks evidence confirming a theory. Falsification seeks evidence capable of disproving it, Popper argued only the latter is genuinely scientific.
Inference to the best explanation: Inference to the best explanation is a reasoning approach favouring whichever explanation, among plausible alternatives, best accounts for the available evidence.
Causal mechanism: A causal mechanism is the actual process or pathway through which a cause produces an effect, distinct from merely establishing that a correlation exists.
Determinism: Determinism is the view that events are fully caused by prior conditions, leaving no room for genuine chance or choice, a background assumption some research paradigms hold more strongly than others.
Reductionism: Reductionism is explaining a complex phenomenon entirely in terms of its smaller, simpler components, a strategy some research designs embrace and others explicitly resist.
Holism: Holism is studying a phenomenon as an integrated whole, arguing it cannot be fully understood by examining its parts in isolation.
Emergence: Emergence is a property arising from the interaction of a system's parts that isn't present in, or predictable from, any single part alone.
Relativism: Relativism is the view that truth or validity is relative to a particular framework, culture, or perspective, rather than universal.
Scientific method: The scientific method is the systematic cycle of observation, hypothesis formation, testing, and revision underpinning positivist research.
Replication crisis: The replication crisis is the widely documented finding that many published results, particularly in psychology and some other fields, fail to replicate when the same study is repeated, prompting methodological reform.
Open science: Open science is a movement promoting transparency in research through open data, open methods, and preregistration, partly a response to the replication crisis.
Preregistration: Preregistration is publicly registering a study's hypotheses and analysis plan before data collection begins, preventing later adjustment of hypotheses to fit whatever results emerge.
Registered report: A registered report is a publication format where a study's introduction and method are peer-reviewed and provisionally accepted before results are known, reducing publication bias toward positive findings.
Ideology (research context): Ideology, in a research context, is a structured set of beliefs shaping how a researcher or a field interprets evidence, worth surfacing explicitly rather than treating one's own framework as neutral.
Discourse analysis: Discourse analysis is a qualitative method examining how language use itself constructs meaning, identity, and power relations within text or talk.
Narrative inquiry: Narrative inquiry is a qualitative approach studying the stories people tell about their experiences, treating narrative structure itself as meaningful data.
Ethnography: Ethnography is a qualitative methodology studying a culture or group through sustained immersion and observation in its natural setting, aiming to understand it from the inside.
Mixed methods research (approach): Mixed methods research is a research approach deliberately combining qualitative and quantitative methods within one study, distinct from Mixed Methods Appraisal Tool, which evaluates such studies rather than describing the approach itself.
Sequential explanatory design: Sequential explanatory design is a mixed methods design collecting quantitative data first, then using qualitative data afterward to explain or deepen those results.
Sequential exploratory design: Sequential exploratory design is a mixed methods design collecting qualitative data first, then using quantitative data afterward to test or generalise what emerged.
Interpretive phenomenological analysis (IPA): Interpretive phenomenological analysis is a specific phenomenological approach focused on how individuals make sense of significant personal experiences, combining close textual analysis with interpretation.
Empiricism: Empiricism is the philosophical position that genuine knowledge comes primarily from sensory experience and observation, a foundational assumption underlying positivist research.
Interpretive repertoire: An interpretive repertoire is a recognisable, culturally shared way of talking about a topic that people draw on to construct meaning, relevant in discourse-focused qualitative work.
Situated knowledge: Situated knowledge is the view that all knowledge is produced from a particular position, time, and perspective, rather than from a neutral, view-from-nowhere standpoint.
Methodological pluralism: Methodological pluralism is the view that no single method or paradigm is universally superior, and that different research questions call for different methodological choices.
Ecological validity: Ecological validity is the extent to which a study's findings reflect real-world conditions, distinct from internal validity, which concerns whether the study design itself supports its causal claims.
Bounded case: In case study research, a bounded case is a case explicitly defined with clear limits, such as a specific time period or organisational unit, so it's clear what does and doesn't belong within it.
Interpretive validity: Interpretive validity is whether a qualitative researcher's interpretation genuinely reflects participants' own meanings, closely related to credibility but specifically focused on the act of interpretation itself.
Theoretical sampling: Theoretical sampling is a grounded theory sampling strategy where which data to collect next is decided based on what's needed to develop the emerging theory, rather than being fixed in advance.
Multiparadigm research: Multiparadigm research deliberately draws on more than one paradigm within a single study, a contested but increasingly discussed approach for tackling complex, multifaceted problems.
First-order vs. second-order constructs: First-order constructs are participants' own concepts and language. Second-order constructs are the researcher's own analytical categories built from interpreting that data.
Epoché: Epoché is a phenomenological practice of suspending prior judgments and assumptions to encounter a phenomenon as freshly as possible, closely related to bracketing.

Framing Your Question

SPIDER: SPIDER (Sample, Phenomenon of Interest, Design, Evaluation, Research type) is a framework for turning a qualitative research question into a structured, searchable form. It's the qualitative counterpart to PICO.
PICO(C): PICO(C) (Population, Intervention, Comparison, Outcome, and optionally Context) breaks a clinical or intervention question into parts you can actually search for. Each letter becomes a search term or filter.
Research Gap: A research gap is a specific, demonstrable absence in the existing literature, not just a topic nobody has written the exact paper you want to write yet. Naming it precisely is what justifies your study existing at all.
Research topic: A research topic is a broad area of interest, not yet a research problem. It becomes a problem only once it's paired with a specific context, a demonstrated gap, and a clear reason the gap matters.
Research problem: A research problem is a topic sharpened into three held-together elements: the context it arises in, evidence a gap exists, and why closing that gap matters. Missing any one of the three, a proposal reads as a survey of a field rather than an argument for a study.
Gap-spotting: Gap-spotting is the dominant, safer strategy for constructing a research question: identifying an empirical, theoretical, methodological, or contextual hole in existing literature and proposing a study to fill it (Sandberg and Alvesson, 2011).
Problematisation: Problematisation is a stronger, less common strategy than gap-spotting: challenging an assumption underlying an entire body of literature, rather than incrementally extending it (Sandberg and Alvesson, 2011). Strong doctoral proposals often combine gap-spotting with at least one problematised assumption.
Theoretical gap: A theoretical gap is a gap type where a phenomenon lacks adequate theoretical explanation in the existing literature.
Empirical gap: An empirical gap is a gap type where a claim exists in theory but hasn't actually been tested in a given setting.
Methodological gap: A methodological gap is a gap type where a topic has been studied, but never with the method being proposed.
Contextual gap: A contextual gap is a gap type where something is well studied elsewhere, but not in a specific sector, region, or organisation type.
Population gap: A population gap is a gap type where existing studies sample a different group than the one actually intended to be studied.
Research aim: A research aim is a single, overarching statement of intent for a study, the top of the aim-objectives-questions chain.
Research objective: A research objective is one of a small number (commonly three to five) of achievable, evidence-based steps that decompose a research aim. A common weakness is objectives that just restate the aim in different words rather than genuinely distinct steps.
Research question: A research question is the specific, answerable query or queries that, taken together, let the research objectives be met.
Hypothesis: A hypothesis is a testable proposition derived from theory, used mainly in positivist, deductive designs, not typically used in purely qualitative or interpretivist work.
Critical realism: Critical realism is a research paradigm holding that reality exists independently of the mind but is layered and only partly observable, making it well suited to applied research examining real but complex organisational or social mechanisms.
Paradigm: A paradigm is a coherent set of beliefs about reality and knowledge (ontology and epistemology together) that shapes every later methodological choice, approach, strategy, and technique.
Research onion: The research onion is Saunders, Lewis and Thornhill's layered model placing philosophy as the outermost layer, working inward through approach, strategy, time horizon, and technique, each layer constrained by the ones outside it.
PEO: PEO is a question-framing framework (Population, Exposure, Outcome) for observational questions with no intervention, distinct from PICO's comparative structure.
SPICE: SPICE is a question-framing framework (Setting, Perspective, Intervention, Comparison, Evaluation) suited to service or policy evaluation questions.
ECLIPSE: ECLIPSE is a question-framing framework (Expectation, Client group, Location, Impact, Professionals, Service) suited to policy, management, or organisational questions.
CIMO: CIMO is a framework (Context, Intervention, Mechanism, Outcome) used particularly in realist reviews, explaining not just whether something works but why, for whom, and under what conditions.
Conceptual framework: A conceptual framework is a researcher-built structure showing how the study's own key concepts relate to each other, distinct from a theoretical framework, which draws on an existing, named theory.
Theoretical framework: A theoretical framework is an existing, named theory used to underpin and interpret a study, distinct from a conceptual framework built specifically for the study itself.
Research design: Research design is the overall plan connecting philosophy, approach, and strategy to the research questions, decided only after the paradigm is fixed, not before.
Practitioner-researcher: A practitioner-researcher is someone researching a setting they already work within. Requires explicit handling of positionality and reflexivity, since prior insider knowledge will shape what's noticed and asked, whether acknowledged or not.
Research contribution: A research contribution is the specific, stated addition a study makes to existing knowledge, what an examiner or reviewer will ultimately test the whole thesis against.
PICOT: PICOT is a variant of PICO adding Time as a fifth component, common in nursing and clinical research questions.
PICOS: PICOS is a variant of PICO adding Study design as a fifth component, commonly used specifically for structuring systematic review search strategies.
FINER criteria: FINER criteria is a checklist for evaluating whether a research question is workable: Feasible, Interesting, Novel, Ethical, Relevant.
SMART objectives: SMART objectives is a checklist for well-formed research objectives: Specific, Measurable, Achievable, Relevant, Time-bound.
Research proposal: A research proposal is the formal document setting out a planned study's problem, questions, literature grounding, and intended method, submitted for approval before the research itself begins.
Working title: A working title is a provisional title for a study, expected to change as the research question and focus sharpen through the process.
Delimitations: Delimitations are the boundaries a researcher deliberately sets on a study's scope, distinct from limitations, which are constraints imposed on the study rather than chosen by it.
Scope of study: The scope of study is the defined extent of what a study will and will not cover, in terms of population, time period, geography, or theme.
Assumptions (of a study): Assumptions of a study are things taken as true for the purposes of a study without being directly tested within it, stated explicitly so a reader understands what the findings do and don't depend on.
Significance of the study: The significance of the study is the stated case for why a study's contribution matters, to theory, practice, or policy, distinct from its research questions, which state what it asks, not why it matters.
Background of the study: The background of the study is the contextual information a reader needs before understanding the research problem itself, setting, history, or circumstances leading up to the study.
Rationale: A rationale is the reasoned justification for why a particular study, question, or method was chosen, the argument connecting the gap to the proposed research.
Central research question: In qualitative research designs, the central research question is the single broad, open question guiding the whole study, from which more specific sub-questions are then derived.
Sub-questions: Sub-questions are more specific questions derived from a central research question, breaking a broad inquiry into researchable, answerable parts.
Moderating variable: A moderating variable changes the strength or direction of the relationship between two other variables, without itself being the cause of the outcome.
Mediating variable: A mediating variable explains the mechanism through which one variable influences another, sitting between a cause and its effect in the causal chain.
Operationalization: Operationalization is the process of defining an abstract concept in terms of concrete, measurable indicators, so it can actually be observed or tested in a study.
Construct (research): A construct is an abstract concept a study aims to measure or examine, e.g., job satisfaction or organisational trust, made measurable through operationalization.
Unit of analysis: The unit of analysis is the specific entity a study is actually examining and drawing conclusions about, an individual, a team, an organisation, a policy, or a document.
Level of analysis: The level of analysis is the scale at which a phenomenon is examined and conclusions are drawn, individual, group, organisational, or societal, distinct from the unit of analysis but closely related.
Research proposition: A research proposition is a statement, often derived from qualitative or theory-building research, proposing a relationship expected to hold, less formally testable than a quantitative hypothesis.
Working hypothesis: A working hypothesis is a provisional, tentative hypothesis held during early stages of research, expected to be refined or replaced as data collection proceeds, common in inductive and abductive designs.
Null hypothesis: The null hypothesis states there is no real effect or relationship, which quantitative statistical testing is actually designed to try to reject, not directly to prove the alternative true.
Alternative hypothesis: The alternative hypothesis states an effect or relationship does exist, tested against the null hypothesis, supported only when the null hypothesis can be rejected.
Directional vs. non-directional hypothesis: A directional hypothesis predicts the specific direction of an effect (e.g., higher X leads to higher Y). A non-directional hypothesis predicts a relationship exists without specifying its direction.
Research design vs. research method: Research design is the overall plan; research method is the specific technique used within that plan, often confused but operating at different levels.
Exploratory research: Exploratory research is conducted when little is known about a topic, aiming to generate initial insight rather than test a specific hypothesis.
Descriptive research: Descriptive research aims to accurately describe a phenomenon, its characteristics or frequency, without necessarily explaining why it occurs.
Explanatory research: Explanatory research aims to establish cause-and-effect relationships, explaining why or how a phenomenon occurs.
Cross-sectional design: A cross-sectional design collects data at a single point in time, showing a snapshot rather than change over time.
Longitudinal design: A longitudinal design collects data from the same subjects at multiple points over time, allowing change to be observed directly.
Single case vs. multiple case design: A single case study examines one instance in depth. A multiple case design examines several, allowing comparison and stronger analytic generalisation.
Comparative research design: A comparative research design deliberately compares two or more cases, settings, or groups to identify similarities and differences.
Retrospective vs. prospective design: A retrospective design looks backward at events that already happened. A prospective design follows subjects forward from the present.
Feasibility study: A feasibility study is a small preliminary study assessing whether a larger, planned study is practically achievable, distinct from a pilot study, which tests the study's actual procedures.
Pilot study: A pilot study is a small-scale trial run of a study's procedures and instruments before the full study, identifying practical problems in advance.
Research gap justification: Research gap justification is the explicit argument connecting an identified gap to why it specifically justifies the proposed study, not just that a gap exists somewhere.
Contribution to knowledge vs. contribution to practice: A theoretical/academic contribution advances understanding within a field. A practical contribution offers direct value to practitioners or organisations, doctoral work is often expected to state both.
Theoretical contribution: A theoretical contribution is a study's specific addition to existing theory, extending, challenging, or synthesising it in a new way.
Practical contribution: A practical contribution is a study's specific, usable value to practitioners, policymakers, or organisations, distinct from its theoretical contribution.
Methodological contribution: A methodological contribution is a study's contribution through a new or adapted research method itself, rather than through its findings.
Iterative question development: Iterative question development is the normal, non-linear process of refining a research question repeatedly as literature review and early data collection reveal what's actually researchable.
Funnel approach: The funnel approach narrows a broad area of interest down to a specific, researchable question in progressive stages, from general topic to precise focus.
Mind mapping (topic generation): Mind mapping, for topic generation, is a visual brainstorming technique for generating and organising possible research angles before committing to one.
Concept mapping (research planning): Concept mapping, in research planning, is diagramming the relationships between a study's key ideas early in the planning process, a precursor to a formal conceptual framework.
Brainstorming (research context): Brainstorming, in a research context, is open, unfiltered idea generation used early to surface possible questions or angles before critical evaluation narrows them down.
Literature-driven vs. practice-driven question: A literature-driven question originates from a gap noticed in existing research. A practice-driven question originates from a real problem encountered in professional practice.
Research question refinement: Research question refinement is the ongoing process of tightening a research question's wording and scope as understanding deepens, continuing well past the initial proposal stage.
Theoretical lens: A theoretical lens is the specific theory or framework a researcher chooses to view and interpret their data through, shaping what's noticed and how it's explained.
Unit of observation: The unit of observation is the specific entity data is actually collected from, which may differ from the unit of analysis a study ultimately draws conclusions about.
Boundary conditions (of a theory): Boundary conditions are the specific contexts, populations, or circumstances within which a theory is expected to hold, stating these explicitly clarifies what a study's contribution does and doesn't claim.
A priori vs. emergent themes: A priori themes are decided before analysis, often from theory. Emergent themes arise from the data itself during analysis, a key distinction in how a qualitative study's coding approach is justified.
Sensitising concepts: Sensitising concepts are broad, initial ideas a researcher brings into qualitative data collection, offering a starting point for looking without rigidly predetermining what will be found.
Stakeholder analysis: Stakeholder analysis is systematically identifying and understanding the people or groups affected by or interested in a research problem, often used early to frame a practice-based study's real-world relevance.
Working proposition: A working proposition is a tentative, statement-form claim guiding early-stage qualitative or exploratory research, more provisional than a formal hypothesis.
Problem-driven vs. theory-driven research: Problem-driven research starts from a real-world issue needing a solution. Theory-driven research starts from a gap or tension within existing theory itself.
Aims vs. objectives vs. outcomes: The aim states overall intent, objectives break it into achievable steps, and outcomes describe the actual, measurable results once the study concludes, three related but distinct levels.
Scoping the literature: Scoping the literature is an early, broad reading pass across a topic area, done before committing to a specific research question, to understand what's already known and where real gaps might sit.
Problem tree analysis: Problem tree analysis is a visual technique mapping a problem's root causes and effects, useful for clarifying a research problem's real structure before framing a question around it.
Researchable question: A researchable question is phrased precisely and narrowly enough to actually be investigated with available time, resources, and methods, distinct from an important question that may still be too broad to research directly.
Diagnostic research question: A diagnostic research question is aimed at identifying the cause or nature of a problem, often a necessary first step before an evaluative or interventional question can be meaningfully asked.
Evaluative research question: An evaluative research question assesses the effectiveness, value, or impact of something already in place, such as a policy or intervention.
Predictive research question: A predictive research question aims to forecast a future outcome based on current or historical data, common in quantitative and data-driven research.
Comparative research question: A comparative research question examines similarities or differences between two or more groups, settings, or time periods.
Question refinement funnel: The question refinement funnel is the practical narrowing sequence from a broad area of curiosity to problem to question to specific, answerable sub-questions, a concrete restatement of the funnel approach applied specifically to question wording itself.

Searching

Boolean search string: A Boolean search string combines your keywords with AND, OR, and NOT so a database returns exactly the intersection of concepts you mean, not everything that mentions any one of them.
Database: A database indexes published academic literature so you can search it by keyword, author, or subject, with Scopus and Web of Science as typical examples.
Registry: A registry (like PROSPERO) records a study's protocol, meaning its planned methods and outcomes, before results exist, so anyone can later check what was promised against what was actually reported.
Database vs. registry: Reach for a database when you need to find what's been published on a topic. Reach for a registry when you need to check what was planned against what was reported. If you're searching for results, use a database; if you're auditing reported outcomes against a pre-specified plan, use a registry.
PROSPERO: PROSPERO is an international registry where systematic review protocols, meaning their planned methods and outcomes, are recorded before the review begins.
Grey literature: Grey literature is material that hasn't gone through commercial or peer-reviewed publishing: theses, government reports, conference proceedings, preprints. It often contains findings that never made it into a journal at all.
Primary source: A primary source is original, uninterpreted data or material: experiments, interviews, organisational records, archival documents. Doctoral data collection draws mainly on primary material.
Secondary source: A secondary source is material that interprets or evaluates primary sources: literature reviews, textbooks, critiques. Doctoral literature reviews draw mainly on secondary material. Treating a textbook summary as the original finding is a common, avoidable error.
Institutional report: An institutional report is grey literature from governments, NGOs, or international organisations, credible for policy and sector context, but author and purpose should be evaluated carefully before relying on it.
Conference paper: A conference paper presents emerging or work-in-progress research, often not yet peer-reviewed. Useful for tracking the frontier of a field, worth checking against a later journal version if one exists.
Thesis vs. dissertation: A thesis and a dissertation are both original, extended research documents with full literature reviews, useful for methodology exemplars. Usage of the two terms varies by country and level, not a fixed universal distinction.
Dataset: A dataset is raw or structured data, essential for empirical and quantitative work, sourced from repositories or collected directly.
Archival material: Archival material is unpublished primary sources such as internal records or documents, valuable but often access-restricted.
Citation chaining: Citation chaining is following a source's own references backward (what it cites) or forward (what later cites it) to find related literature beyond a single keyword search.
Search saturation: Search saturation is the point in a literature search where new searches stop turning up genuinely new, relevant sources, a practical signal the search is approaching completeness.
Truncation: Truncation is a search technique using a wildcard symbol (commonly an asterisk) to match a word's variant endings in one search, e.g., a truncated root matching both singular and plural forms.
Phrase searching: Phrase searching is searching for an exact multi-word phrase, usually by enclosing it in quotation marks, rather than the words being matched separately and scattered across a result.
Subject heading vs. keyword searching: Keyword searching matches the words you type. Subject heading searching matches a database's own controlled vocabulary tags for a concept, often catching relevant sources that use different wording than your search terms.
Systematic search strategy: A systematic search strategy is a documented, justified combination of databases searched, search strings used, and date ranges applied, the auditable core of a rigorous literature search.
Indexing: Indexing is the process by which a database evaluates and tracks a journal's citation performance and impact, distinct from simply hosting a journal's full text.
Content database vs. indexing database: A content database (e.g., ScienceDirect, JSTOR) gives access to full-text material. An indexing database (e.g., Scopus, Web of Science) evaluates and tracks citation performance and journal quality, they serve different jobs, not interchangeable ones.
ScienceDirect: ScienceDirect is a major full-text content database, particularly strong in science, technology, and business literature, useful for exploratory as well as confirmatory searching.
JSTOR: JSTOR is a full-text archive strong in humanities and social sciences, including older and foundational material harder to find elsewhere.
ProQuest: ProQuest is the primary database for theses and dissertations, useful for finding methodology exemplars from completed doctoral work.
Google Scholar: Google Scholar is a free, broad academic search covering journals, theses, and grey literature together, useful for a first, wide pass, though it lacks the quality curation of Scopus or Web of Science.
PubMed: PubMed is the primary free database for biomedical and life sciences literature, maintained by the US National Library of Medicine.
IEEE Xplore: IEEE Xplore is the primary database for engineering, computer science, and electronics literature, essential for technical and computer science research.
Emerald: Emerald is a publisher and database particularly strong in management, business, and Knowledge Management literature.
VOSviewer: VOSviewer is a free bibliometric mapping tool that visualises a field's citation and co-citation structure before deep reading begins, useful for spotting a field's key clusters and gaps quickly.
Connected Papers: Connected Papers builds a visual citation graph from a single seed paper, a fast way to snowball related literature visually rather than manually.
Litmaps: Litmaps tracks and visualises citation networks over time, useful for seeing how a field has evolved and staying current on a specific paper's citing literature.
Wildcard search: A wildcard search uses a symbol to represent any character or set of characters within a word, broader than truncation, which typically only matches word endings.
Proximity search: A proximity search requires two terms to appear within a specified distance of each other in the text, tighter than a general keyword search but looser than an exact phrase search.
Field searching: Field searching restricts a search to a specific part of a record, such as title only, or abstract only, rather than searching the full text.
Nesting (Boolean): Nesting uses parentheses to group search terms and control the order Boolean operators are applied, the same logic as order of operations in mathematics.
Controlled vocabulary: A controlled vocabulary is a fixed, standardised set of terms a database uses to tag and organise its content, distinct from the free-text keywords a researcher might type.
MeSH terms (Medical Subject Headings): MeSH terms are PubMed's specific controlled vocabulary system, used to tag articles by standardised medical concepts rather than relying on keyword matching alone.
Thesaurus (database): A thesaurus, in this sense, is a database's own structured list of preferred terms and their relationships, used to guide effective subject heading searches.
Backward snowballing: Backward snowballing is following a source's own reference list to find earlier, related literature it cites.
Forward snowballing: Forward snowballing is finding later literature that has cited a given source, moving forward in time rather than backward through references.
Search string documentation: Search string documentation is the practice of recording the exact search string, database, and date used for every search run, the raw material a PRISMA-style audit trail depends on.
Search Record: A Search Record is the structured, ongoing record of every search performed during a review, database, exact string, date, and result count, together forming the PRISMA-style audit trail a systematic review's methodology depends on.
Date range restriction: Date range restriction is limiting search results to sources published within a specified period, common when a review deliberately focuses on recent literature or a specific historical window.
Language restriction: Language restriction is limiting search results to sources in a specific language, a decision that should be stated and justified explicitly, since it can introduce its own bias into a review.
Publication type filter: A publication type filter restricts search results to a specific kind of source, such as peer-reviewed articles only, or excluding conference papers, applied and documented as part of a search strategy.
Full-text vs. abstract-only access: Full-text vs. abstract-only access is a practical distinction in whether a database or institutional subscription provides the complete article or only its summary, directly affecting what can actually be screened in depth.
Interlibrary loan: Interlibrary loan is a service allowing a researcher to request a source their own institution doesn't have direct access to, borrowed from another library or institution.
Institutional repository: An institutional repository is a free, open repository many universities host of their own theses, dissertations, and research output, often the easiest way to access a specific institution's doctoral work directly.
Preprint server: A preprint server is a platform for sharing research publicly before peer review, useful for tracking a fast-moving field's frontier, though the work hasn't yet been formally vetted.
arXiv: arXiv is the dominant preprint server for physics, mathematics, computer science, and related quantitative fields.
SSRN (Social Science Research Network): SSRN is the dominant preprint server for social science, economics, and law, widely used for early-stage working papers.
ResearchGate: ResearchGate is a researcher networking platform where authors often share their own papers directly, sometimes the only free access point for a paywalled article.
Academia.edu: Academia.edu is another researcher-sharing platform, similar in function to ResearchGate, useful for finding author-uploaded copies of otherwise paywalled work.
Semantic Scholar: Semantic Scholar is a free, AI-assisted academic search engine that extracts and highlights influential citations and paper summaries, useful for quickly gauging a paper's actual impact within its field.
BASE (Bielefeld Academic Search Engine): BASE is one of the largest free search engines specifically indexing open access academic content, useful for finding freely available versions of research.
CORE: CORE is a large aggregator of open access research papers from repositories worldwide, another strong free option for finding accessible full text.
Web scraping (research context): Web scraping is extracting data directly from websites using automated tools, a legitimate data collection method in some research contexts, but one requiring careful attention to a site's terms of service and relevant data protection law before use.
Manual searching: Manual searching is deliberately browsing key journals or conference proceedings issue by issue, rather than relying solely on database search strings, useful for catching relevant work a search string might miss.
Systematic mapping review: A systematic mapping review is a review method charting the scope and distribution of existing research on a broad topic, identifying clusters and gaps, without synthesising findings in the depth a full systematic review would.
Rapid evidence assessment: A rapid evidence assessment is a streamlined literature review method balancing systematic rigor against a tighter timeline, similar in spirit to a rapid review but often applied specifically in policy contexts.
Search filters (methodological): Search filters, methodologically, are pre-built, validated search strings targeting a specific study design, such as a filter designed to retrieve only randomised controlled trials, saving a researcher from building this from scratch.
OpenAlex: OpenAlex is a free, open scholarly database providing citation and metadata coverage similar in scope to Scopus or Web of Science, without a subscription requirement.
Lens.org: Lens.org is a free platform combining scholarly and patent literature search, useful for research bridging academic and applied innovation contexts.
Snowball sampling (literature): Snowball sampling, for literature, is using an initial set of key sources to identify further relevant literature through their references and citations, the literature-search application of the same logic as citation chaining.
Berry-picking (search behavior): Berry-picking is a recognised, realistic model of search behavior where a researcher's search strategy evolves and shifts as they learn from what they find along the way, rather than following one fixed, linear plan.
Query expansion: Query expansion is automatically or manually adding related terms to a search query to broaden its results, often through synonym lists or a database's own suggested terms.
Dimensions.ai: Dimensions.ai is a free and paid research discovery platform linking publications, grants, datasets, and patents together, useful for tracing a topic's funding and policy footprint alongside its literature.
Unpaywall: Unpaywall is a free browser tool and database that locates legally available open-access versions of paywalled articles automatically.
Citation pearl growing: Citation pearl growing is an iterative search technique using promising early results to identify better search terms, refining the search strategy as understanding of the field's actual vocabulary improves.
Search string sensitivity testing: Search string sensitivity testing is deliberately checking whether a search string retrieves known, already-identified key papers, confirming the string is well constructed before relying on it for the full search.
Full-text search vs. metadata search: A full-text search checks the entire document's content. A metadata search checks only structured fields like title, author, or keywords, the two can return meaningfully different results for the same query.
TRIP Database: TRIP Database is a specialised search engine for evidence-based medicine, ranking results partly by evidence quality, useful for clinical and health-related research specifically.
WorldCat: WorldCat is a global catalogue aggregating library holdings worldwide, useful for locating which libraries hold a specific book or thesis.
Hand-searching: Hand-searching is a specific form of manual searching applied systematically as part of a documented review methodology, not informal browsing.
Expert consultation (search strategy): Expert consultation, for a search strategy, is having a subject expert or librarian review a planned search strategy before running it, catching missing terms or databases early.
PRESS checklist (Peer Review of Electronic Search Strategies): The PRESS checklist is a formal, structured checklist for peer-reviewing a systematic review's search strategy before it's run, checking for missed terms, incorrect syntax, or overly narrow scope.
Precision vs. recall: Precision measures how many retrieved results are actually relevant. Recall measures how many of all relevant results were actually retrieved, the two typically trade off against each other.
Saved search / search alert: A saved search or search alert is a stored search that automatically notifies a researcher when new matching results are published, keeping a literature base current without manually re-running searches.
Citation alert: A citation alert is a notification triggered when a specific paper is newly cited elsewhere, useful for tracking a field's ongoing response to a key source.
Author search vs. topic search: Searching by a known author's name finds their body of work, distinct from a topic or keyword search looking for a subject regardless of who wrote about it.
Institutional subscription: An institutional subscription is database or journal access provided through a researcher's university or organisation, often the actual access point behind what looks like a personal login.
Walk-in access: Walk-in access is physical, in-person access to subscription databases or archives at a library, sometimes the only access route for non-affiliated researchers.
Open access repository: An open access repository is a freely accessible archive of research outputs, distinct from a subject-specific preprint server, often run by an institution or a discipline-wide initiative.
Federated search: A federated search is a single search interface querying multiple databases simultaneously, convenient but sometimes less precise than searching each database individually with its own syntax.
Discovery layer / service: A discovery layer or service is a library-provided search tool aggregating results across many of an institution's subscribed resources into one interface.
Link resolver: A link resolver is a library tool that connects a citation to the actual full text a specific institution has access to, rather than a generic external link.
Metadata (research context): Metadata, in a research context, is structured descriptive information about a source, author, title, date, keywords, that enables it to be found, filtered, and correctly cited.
Persistent identifier: A persistent identifier is a permanent, stable reference to a digital object (a paper, dataset, or author) that won't break even if the object's location changes, DOI and ORCID are both specific types.
Handle system: The Handle system is an underlying technical infrastructure for assigning persistent identifiers to digital objects, DOI is one well-known implementation built on it.
Search strategy translation: Search strategy translation is adapting a search string's exact syntax to match a different database's specific search rules, since Boolean and field-code conventions vary across platforms.
Interoperability (metadata): Interoperability, for metadata, is the ability of different databases and systems to correctly exchange and understand each other's metadata, underpinning tools like federated search and citation tracking.
Subject repository: A subject repository is a repository focused on one discipline's outputs (like arXiv for physics), distinct from an institutional repository organised around one university's outputs regardless of subject.
Sensitivity vs. specificity (search context): A sensitive search strategy retrieves more results, risking irrelevant ones. A specific strategy retrieves fewer, more precisely targeted results, risking missed relevant sources.
Grey literature database: A grey literature database is a specialised database or repository specifically indexing grey literature, such as OpenGrey, distinct from general academic databases.
Data repository: A data repository is a platform for storing and sharing research datasets themselves, distinct from a literature repository storing papers.
Search log: A search log is the running documentation of every search performed during a review, database, string, date, and result count, the raw material a PRISMA-style report is built from.
Boolean operator precedence: Boolean operator precedence is the default order in which AND, OR, and NOT are applied within a search string when no parentheses are used, varies by database and is a common source of unexpected results.

Screening & Selecting

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses): PRISMA is a reporting checklist and flow diagram for systematic reviews, documenting exactly how many records were identified, screened, excluded, and why, at each stage. It exists so the selection process is auditable, not just asserted.
EQUATOR Network: EQUATOR Network is a hub of research reporting guidelines across disciplines and study types, including PRISMA and its extensions.
Systematic review: A systematic review follows a predefined protocol to answer a focused, narrow question as exhaustively as possible, with every inclusion and exclusion decision documented.
Narrative review: A narrative review synthesizes literature informally around a broad theme, without a predefined protocol dictating exactly which studies must be included.
Scoping review: A scoping review maps the size, shape, and key concepts of a field before a focused research question has even been settled.
Systematic review vs. narrative review vs. scoping review: All three synthesize existing literature, but answer different needs. Pick a systematic review when your question is narrow and you need an exhaustive, auditable answer. Pick a narrative review when you're building broad background on a theme. Pick a scoping review when you need to map a field before your question is even settled.
PRISMA flow diagram: The PRISMA flow diagram is the visual funnel documenting a systematic review's search and screening process, from initial records found through to the final included studies, the core evidence of rigour an examiner looks for.
Identification: Identification is the first PRISMA stage: records located through database searching and other sources (citation chaining, grey literature), with duplicates removed.
Screening (PRISMA stage): Screening is the second PRISMA stage: titles and abstracts screened against pre-defined inclusion and exclusion criteria.
Eligibility: Eligibility is the third PRISMA stage: full-text articles assessed for eligibility, with reasons for exclusion recorded and reported, not just discarded silently.
Included: Included is the fourth and final PRISMA stage: studies meeting all criteria, carried forward into the synthesis, with the full selection process documented.
Deduplication: Deduplication is removing duplicate records found across multiple databases or sources before screening begins, standard reference managers automate much of this step.
Title and abstract screening: Title and abstract screening is the first, faster pass of screening, filtering out clearly irrelevant records before committing to a full-text read.
Full-text screening: Full-text screening is the slower, deeper pass applied only to records that survived title and abstract screening, where eligibility is finally confirmed or rejected.
Inclusion criteria: Inclusion criteria are the explicit, pre-defined conditions a source must meet to be kept in a review, stated before screening begins, not adjusted afterward to fit a preferred outcome.
Exclusion criteria: Exclusion criteria are the explicit, pre-defined conditions that remove a source from a review, reported alongside the reasons, not left implicit.
Dual reviewer screening: Dual reviewer screening is having two people independently screen the same records and compare results, a standard rigour check against one person's individual bias or oversight.
Inter-rater reliability: Inter-rater reliability is the degree to which two independent reviewers agree on a screening or coding decision, a measurable signal of how consistent and defensible a review's selection process actually is.
Cohen's kappa: Cohen's kappa is a statistical measure of agreement between two independent reviewers, correcting for the agreement expected purely by chance, the specific number typically used to report inter-rater reliability.
Meta-analysis: A meta-analysis is a statistical technique combining numerical results across multiple quantitative studies into one pooled estimate, only appropriate when the underlying studies are similar enough in design and measurement to combine meaningfully.
Meta-synthesis: Meta-synthesis is the qualitative equivalent of meta-analysis, systematically combining findings across multiple qualitative studies into a new, higher-level interpretation, rather than pooling numbers.
Umbrella review: An umbrella review synthesises existing systematic reviews themselves, rather than individual primary studies, useful for mapping a very broad or mature field quickly.
Rapid review: A rapid review is a streamlined version of a systematic review, using a simplified or accelerated method to answer a question faster, at some cost to comprehensiveness, a legitimate approach when timelines are short and the trade-off is stated openly.
Risk of bias assessment: A risk of bias assessment is a structured evaluation of how much a study's own design or conduct might have distorted its results, applied to each included study, not just to the review as a whole.
Quality appraisal: Quality appraisal is the broader practice of judging how trustworthy an individual study is before relying on its findings, of which risk of bias assessment is one common method.
Data extraction form: A data extraction form is a standardised template used to pull the same fields (method, sample, findings, and so on) consistently from every included study, the direct precursor to a literature matrix.
CASP (Critical Appraisal Skills Programme): CASP (Critical Appraisal Skills Programme) provides free, widely used checklists for appraising different study types (randomised trials, qualitative studies, cohort studies), a practical starting point for quality appraisal.
JBI Critical Appraisal Tools: JBI Critical Appraisal Tools are a set of checklists from the Joanna Briggs Institute, covering a broad range of study designs, commonly used alongside or instead of CASP.
MMAT (Mixed Methods Appraisal Tool): MMAT (Mixed Methods Appraisal Tool) is the standard appraisal tool specifically for reviews that include mixed-methods studies, less commonly used when a review is purely quantitative or purely qualitative, where CASP or JBI's tools are the more typical choice instead.
Rayyan: Rayyan is a free, widely used tool for collaborative title and abstract screening, letting multiple reviewers screen independently and compare decisions.
Covidence: Covidence is a paid, structured platform covering the whole screening and data extraction workflow, often provided through university library subscriptions.
ASReview: ASReview is an open-source tool using AI-assisted prioritisation to help a reviewer work through large result sets faster, with the prioritisation process itself logged and transparent, keeping it auditable rather than opaque.
Living systematic review: A living systematic review is a systematic review continuously updated as new evidence emerges, rather than being a single, fixed snapshot, common in fast-moving fields where a static review would quickly go out of date.
Realist review: A realist review asks not just whether an intervention works, but why, for whom, and under what conditions, closely tied to the CIMO framework.
Critical interpretive synthesis: Critical interpretive synthesis is a qualitative review method that critiques and synthesises literature to build new theoretical understanding, rather than just summarising existing findings.
Configurative review vs. aggregative review: An aggregative review pools comparable findings together (as in meta-analysis). A configurative review arranges diverse findings to build a new interpretation, closer to how qualitative synthesis works.
Review protocol: A review protocol is the pre-registered plan for a systematic review, stating its questions, search strategy, and eligibility criteria before the review begins, registered through PROSPERO or similar.
Protocol amendment: A protocol amendment is a documented, justified change to a review's registered protocol after it has begun, reported transparently rather than made silently.
Reporting guideline: A reporting guideline is a structured standard (such as PRISMA) for what a review or study must report, distinct from a methodology itself, it governs how the work is described, not how it's conducted.
Two-stage screening: Two-stage screening is screening in two distinct passes, title and abstract first, full text second, the standard PRISMA approach.
Single-stage screening: Single-stage screening is screening full texts directly without a separate title and abstract pass first, faster but less commonly used in a rigorous systematic review.
Screening disagreement resolution: Screening disagreement resolution is the documented process for resolving cases where two independent reviewers disagree on whether to include a source, commonly a third reviewer or a discussion to consensus.
Pilot screening: Pilot screening is testing inclusion and exclusion criteria on a small sample of records before applying them to the full set, catching ambiguous criteria early.
Evidence synthesis: Evidence synthesis is the overall process of combining findings across multiple studies into a coherent conclusion, the umbrella term covering meta-analysis, meta-synthesis, and narrative synthesis as specific methods within it.
Narrative synthesis: Narrative synthesis is a synthesis method summarising findings across studies in structured text rather than through statistical pooling, used when studies are too varied to combine numerically.
Thematic synthesis: Thematic synthesis is a synthesis method identifying and developing themes across multiple qualitative studies, similar in logic to thematic analysis but applied across studies rather than within one dataset.
Effect size: Effect size is a quantitative measure of how large a difference or relationship actually is, distinct from statistical significance, which only indicates whether an effect is likely real, not how big it is.
Statistical significance: Statistical significance is an indication that an observed result is unlikely to have occurred by chance alone, commonly reported via a p-value, but not itself a measure of how important or large the result is.
P-value: A p-value is the probability of observing a result at least as extreme as the one found, assuming there is genuinely no effect, a small p-value is conventionally treated as evidence against that assumption.
Confidence interval: A confidence interval is a range of values likely to contain the true effect, given the data observed, generally considered more informative than a p-value alone since it also conveys precision.
Heterogeneity: Heterogeneity is the degree to which results vary across the studies included in a review, high heterogeneity is often a signal that combining results into a single pooled estimate may not be meaningful.
Forest plot: A forest plot is a standard visual summary in a meta-analysis, showing each included study's effect estimate alongside the pooled overall estimate.
Funnel plot: A funnel plot is a visual check for publication bias in a meta-analysis, an asymmetric, lopsided plot can suggest smaller studies with null results are missing from the published literature.
GRADE approach: The GRADE approach is a structured system for rating the overall certainty of evidence across a body of studies, from high to very low, widely used in health-related systematic reviews.
Sensitivity analysis: Sensitivity analysis is re-running a review's analysis while excluding certain studies (such as lower-quality ones) to check whether the overall conclusion still holds.
Subgroup analysis: Subgroup analysis is analysing a defined subset of included studies separately, to check whether an effect differs across specific populations or conditions.
Sample attrition: Sample attrition is the loss of participants or data points over the course of a study, relevant when appraising an included study's own reliability, not just a review's own screening process.
Reference deduplication software: Reference deduplication software describes tools within reference managers that automatically detect and remove duplicate records pulled in from multiple databases.
Screening software: Screening software describes dedicated platforms (such as Rayyan or Covidence) built specifically to manage title, abstract, and full-text screening at scale with multiple reviewers.
Machine learning-assisted screening: Machine learning-assisted screening uses AI-based prioritisation to rank likely-relevant records higher during screening, speeding up the process while keeping a human making every final decision.
Priority screening: Priority screening is reviewing records in order of predicted relevance rather than a random or chronological order, common in machine learning-assisted tools.
Stopping rule: A stopping rule is a predefined, justified criterion for when screening can reasonably stop, such as a set number of consecutive irrelevant records, rather than an arbitrary decision.
Quality threshold: A quality threshold is a minimum quality appraisal score or standard a study must meet to be included in a review's synthesis, decided and stated before appraisal begins.
Weighted quality assessment: Weighted quality assessment is giving higher-quality included studies more influence in a review's overall conclusions than lower-quality ones, rather than treating every included study equally.
Vote counting: Vote counting is a weak synthesis method simply tallying how many studies found a positive versus negative result, without accounting for study quality or size, generally discouraged in favour of proper meta-analysis.
Best-evidence synthesis: Best-evidence synthesis is a synthesis approach giving more weight to higher-quality studies without full statistical meta-analysis, a middle ground between vote counting and formal pooling.
Framework synthesis: Framework synthesis is a qualitative synthesis method organising findings from multiple studies against a pre-existing theoretical framework, rather than letting themes emerge purely from the data.
Qualitative comparative analysis (QCA): Qualitative comparative analysis is a method systematically comparing cases to identify which combinations of conditions are associated with a particular outcome, bridging case-based and variable-based approaches.
Fixed-effect model: A fixed-effect model is a meta-analysis model assuming all included studies are estimating one single true effect, appropriate only when studies are genuinely similar enough to share that assumption.
Random-effects model: A random-effects model is a meta-analysis model assuming the true effect varies across studies, generally the more realistic and commonly used choice when heterogeneity is present.
I² statistic: The I² statistic is a specific, commonly reported measure of the percentage of variation across studies attributable to real heterogeneity rather than chance alone.
Egger's test: Egger's test is a statistical test checking a funnel plot for asymmetry, providing a formal, numerical check for possible publication bias rather than relying on visual inspection alone.
Trim and fill method: The trim and fill method is a statistical technique estimating what a meta-analysis's pooled result might look like if missing studies, suspected due to publication bias, were accounted for.
Standardized mean difference: Standardized mean difference is a meta-analysis effect size measure expressing the difference between two groups in standard deviation units, allowing comparison across studies that used different measurement scales.
Odds ratio: An odds ratio is a measure comparing the odds of an outcome occurring in one group versus another, common in case-control studies.
Risk ratio: A risk ratio is a measure comparing the probability of an outcome occurring in one group versus another, common in cohort studies and randomised trials.
Hazard ratio: A hazard ratio is a measure comparing the rate at which an event occurs over time between two groups, common in survival and time-to-event analysis.
Number needed to treat: Number needed to treat is a clinically oriented measure indicating how many people would need to receive an intervention for one additional person to benefit, translating statistical effect into practical meaning.
Clinical significance vs. statistical significance: A result can be statistically significant without being large enough to matter practically, and vice versa, the two are genuinely different questions.
PICO-based data extraction: PICO-based data extraction structures a systematic review's data extraction form around the same Population, Intervention, Comparison, Outcome categories used to frame the original question, keeping extraction consistent with the review's own logic.
Overlapping reviews: Overlapping reviews occur when multiple systematic reviews on a related topic include many of the same primary studies, relevant to consider when conducting or citing an umbrella review.
Corrected covered area: Corrected covered area is a formal method for measuring the degree of primary-study overlap across multiple systematic reviews being combined in an umbrella review.
AMSTAR 2: AMSTAR 2 is a widely used critical appraisal tool specifically for assessing the methodological quality of systematic reviews themselves, not primary studies.
Newcastle-Ottawa Scale: The Newcastle-Ottawa Scale is a commonly used quality appraisal tool specifically for non-randomised studies, such as cohort and case-control designs.
Evidence gap map: An evidence gap map is a visual tool plotting what has and hasn't been studied across a field, often organised as a grid of interventions against outcomes, useful for identifying genuine research gaps systematically rather than anecdotally.
Harvest plot: A harvest plot is a visual method for synthesising evidence across outcomes when studies are too varied for formal meta-analysis, showing patterns of effect direction rather than pooled numeric results.
PICOS-based eligibility: PICOS-based eligibility structures a review's inclusion and exclusion criteria directly around the Population, Intervention, Comparison, Outcome, and Study design components used to frame the original search.
ROBIS: ROBIS is a tool specifically for assessing the risk of bias in the process of conducting a systematic review itself, complementing AMSTAR 2's focus on overall review quality.
SWiM (Synthesis Without Meta-analysis) guideline: The SWiM guideline is a reporting guideline for systematic reviews that synthesise results without statistical pooling, filling a documentation gap PRISMA alone doesn't fully cover.
Critical appraisal skills: Critical appraisal skills are the general competency of evaluating a study's methodological quality and trustworthiness, the umbrella skill that CASP, JBI, and MMAT are each specific tools for developing and applying.
Study selection flow: Study selection flow is the overall documented sequence a review follows from initial search results to final included studies, another common name for what the PRISMA flow diagram visually represents.
Data charting: Data charting is the process of systematically extracting and organising key information from included sources in a scoping review, the scoping-review-specific equivalent of data extraction in a systematic review.
STROBE guideline: The STROBE guideline is a reporting guideline specifically for observational studies (cohort, case-control, cross-sectional), distinct from PRISMA, which governs how reviews of studies are reported, not individual studies themselves.
CONSORT guideline: The CONSORT guideline is a reporting guideline specifically for randomised controlled trials, ensuring consistent, complete reporting of trial design and results.
COREQ guideline: The COREQ guideline is a reporting guideline specifically for qualitative research using interviews and focus groups, the qualitative equivalent of STROBE or CONSORT.
Reporting guideline selection: Reporting guideline selection is choosing the correct guideline (PRISMA, STROBE, CONSORT, COREQ) to follow based on a study's actual design, using the wrong guideline is a common, avoidable methodological error.
Living evidence synthesis: Living evidence synthesis is the broader practice underlying a living systematic review, continuously incorporating new evidence rather than treating a synthesis as a fixed, one-time product.

Judging Evidence Quality

Journal quartile: Journal quartile (Q1 through Q4) ranks a journal's citation impact against others in its subject category, with Q1 being the top 25%. It's a rough proxy for influence, not a guarantee that any individual paper in it is good.
Predatory journal: A predatory journal charges publication fees while skipping or faking the peer review it claims to provide, existing mainly to collect fees rather than to advance a field. Citing one, even accidentally, can undermine your work's credibility.
DOI (Digital Object Identifier): A DOI is a permanent link to a specific piece of published work, staying valid even if the journal's website is redesigned or the paper moves. It's what a citation manager actually resolves against.
ISSN: An ISSN identifies a periodical, like a journal, as a whole, distinct from any single article within it.
ISBN: An ISBN identifies a specific book or edition, the way a DOI identifies a specific article.
ORCID (Open Researcher and Contributor ID): A free, permanent identifier that distinguishes one researcher from every other researcher, even ones who share the same name, and stays with you across institutions, publications, and your entire career.
APC: An APC is the fee a journal charges to publish your article, most common in open-access journals where the article is free for readers but not for authors.
Peer review: Peer review is the process where independent experts assess a manuscript's methods and conclusions before publication, checking for errors, gaps, and validity issues the authors may have missed. It's a filter, not a guarantee.
DOAJ (Directory of Open Access Journals): DOAJ is a curated, community-maintained directory listing open-access journals that meet baseline transparency and quality standards, like clear peer review policies and editorial information.
Scopus: Scopus is a large citation and abstract database covering peer-reviewed literature across disciplines, used to check whether a journal is genuinely indexed and to calculate citation metrics like CiteScore.
Web of Science: Web of Science is the other major citation database, indexing peer-reviewed literature and producing the official Journal Impact Factor through its Journal Citation Reports.
SCImago Journal Rank (SJR): SCImago Journal Rank is a free, public ranking that sorts journals into quartiles (Q1 to Q4) based on citation data drawn from Scopus.
Crossref: Crossref is the registration agency that assigns and maintains most academic DOIs, and also maintains records of retractions and corrections.
Think. Check. Submit.: Think. Check. Submit. is a practical checklist that walks you through assessing whether a journal or publisher is trustworthy before you submit or cite it.
Cabells Predatory Reports: Cabells Predatory Reports is a subscription-based, actively maintained database that specifically flags predatory journals against a documented set of criteria.
Beall's List: Beall's List was a list of journals and publishers considered predatory, created by librarian Jeffrey Beall. It was taken offline in 2017 under legal pressure from publishers it named, and it is no longer officially maintained. What exists today are unofficial, anonymously maintained archived continuations.
Journal vs. platform vs. publisher: A journal is a specific titled publication (e.g., Journal of Cleaner Production). A platform hosts many journals (e.g., ScienceDirect). A publisher owns and produces journals (e.g., Elsevier). Conflating the three makes it hard to correctly evaluate a source's actual quality signals.
CiteScore: CiteScore is a citation-based journal metric, based on Scopus data, averaging citations received over recent years against documents published in the same window.
Journal Impact Factor (JIF): Journal Impact Factor is the official citation metric from Clarivate's Journal Citation Reports, based on Web of Science data, distinct from CiteScore, which uses Scopus data instead, the two can rank the same journal differently.
h-index: The h-index is a metric measuring both the productivity and citation impact of an individual researcher or a journal, not the same measurement as a journal-level impact factor.
Open access: Open access is a publishing model making an article freely available to readers, often funded by an article processing charge rather than reader subscriptions.
Subscription journal: A subscription journal is a journal where readers or their institutions pay for access, the traditional model open access was created as an alternative to.
Hybrid journal: A hybrid journal is a subscription journal that also offers an open access option for individual articles, usually for an additional fee, a middle ground between the two models.
Single-blind peer review: Single-blind peer review is a peer review format where reviewers know the author's identity, but the author doesn't know who reviewed their work.
Double-blind peer review: Double-blind peer review is a peer review format where neither the author nor the reviewers know each other's identity, intended to reduce bias in either direction.
Open peer review: Open peer review is a peer review format where reviewer identities, reports, or both are published alongside the article, trading some reviewer anonymity for greater transparency.
Desk rejection: A desk rejection is when an editor rejects a manuscript before it's even sent out for peer review, usually for being out of scope or clearly below the journal's standard.
Revise and resubmit: Revise and resubmit is a peer review outcome asking the author to substantially address reviewer feedback and resubmit, often more than once, before a final accept or reject decision.
Retraction: A retraction is the formal withdrawal of a published article, usually due to serious error, misconduct, or fraud discovered after publication, checkable against the Crossref API using the retraction filter.
Study design hierarchy: A study design hierarchy is a ranked ordering of research designs by how much confidence can be placed in their findings, systematic reviews and meta-analyses at the top, expert opinion at the bottom, most relevant to quantitative and clinical research.
Randomised controlled trial (RCT): A randomised controlled trial randomly assigns participants to different conditions to isolate a causal effect, generally considered the strongest single-study design for establishing causation.
Cohort study: A cohort study is an observational study design following a group over time to observe outcomes, weaker for establishing causation than an RCT but often more practical or ethical.
Case-control study: A case-control study is an observational study design comparing a group with an outcome against a group without it, looking backward for associated factors.
Source hierarchy: A source hierarchy is a practical four-tier ranking for evaluating any source: strongest (peer-reviewed, indexed, recent), useful but secondary (textbooks, established grey literature), use with caution (non-indexed journals, older sources without corroboration), and avoid (unverifiable, no authorship, predatory signals present).
Z-score: A Z-score is a standardised value showing how many standard deviations a data point sits from the mean, the basis of a Z-test used to judge whether a result is statistically unusual.
Publication ethics: Publication ethics is the set of standards governing honest conduct in research publishing, covering authorship, plagiarism, data fabrication, conflicts of interest, and the proper handling of errors once discovered.
Impact factor vs. quartile: A common point of confusion: impact factor is a single numeric score, quartile (Q1 to Q4) is a ranked position relative to other journals in the same subject category, a journal's raw impact factor number means little without knowing its field's typical range.
Editorial board: An editorial board is the named group of scholars responsible for a journal's editorial decisions, a fake, non-contactable, or irrelevant editorial board is one of the clearest predatory journal warning signs.
Article processing charge waiver: An article processing charge waiver is a reduced or waived APC sometimes offered to researchers from lower-income countries or institutions, worth checking before assuming open access is unaffordable.
COPE (Committee on Publication Ethics): COPE (Committee on Publication Ethics) membership signals a publisher has committed to recognised ethical standards, one of the practical signals used to distinguish a legitimate publisher from a predatory one.
OASPA (Open Access Scholarly Publishers Association): OASPA (Open Access Scholarly Publishers Association) membership is another practical, checkable signal of a legitimate open access publisher, used alongside COPE and DOAJ membership when triangulating a journal's credibility.
Impact factor manipulation: Impact factor manipulation describes deliberate practices (such as encouraging excessive self-citation) intended to artificially inflate a journal's citation-based metrics.
Coercive citation: Coercive citation is when an editor pressures an author to add citations to that editor's own journal as a condition of publication, a recognised form of metric manipulation.
Editorial self-citation: Editorial self-citation is a journal's own editorial team citing that same journal excessively across published articles, inflating its apparent citation impact.
Special issue (predatory risk): A special issue is a themed journal issue, generally legitimate, but sometimes exploited by predatory operations to rapidly publish large volumes of poorly reviewed content under a credible journal's name.
Guest editor: A guest editor is an external academic invited to manage a special issue's editorial process, worth checking has genuine standing and oversight, particularly for special issues.
Book chapter vs. journal article: A book chapter appears within an edited volume with its own editorial process, distinct from and generally less rigorously vetted than a peer-reviewed journal article.
Monograph: A monograph is a single-authored, book-length work on one specific subject, distinct from an edited volume with contributions from multiple authors.
White paper: A white paper is an authoritative report or guide, often produced by an organisation or company, addressing a specific issue or proposing a solution, a common form of grey literature.
PMID (PubMed ID): A PMID is a unique identifier assigned to every record in PubMed, useful for precisely citing or retrieving a specific biomedical article.
PMCID: A PMCID is a unique identifier for articles archived in PubMed Central, PubMed's free full-text repository, distinct from a PMID, which just identifies the record itself.
ISNI (International Standard Name Identifier): ISNI is an identifier distinguishing individuals and organisations involved in creative and academic work, broader in scope than ORCID, which is specific to researchers.
ROR (Research Organization Registry): ROR is a persistent identifier system for research institutions themselves, increasingly used alongside ORCID to unambiguously link authors to their affiliated organisations.
Funder ID / grant number: A funder ID or grant number is an identifier linking a published study to the specific funding body and grant that supported it, increasingly required for transparency and research assessment.
CRediT taxonomy: The CRediT taxonomy is a standardised set of roles (such as conceptualization, data curation, writing) used to specify exactly what each listed author contributed to a paper.
Data availability statement: A data availability statement is a formal statement in a publication specifying whether and how the underlying research data can be accessed, increasingly required by journals and funders.
Code availability statement: A code availability statement specifies whether and where analysis code used in a study can be accessed, relevant particularly for reproducibility in quantitative and computational research.
Reproducibility badge: A reproducibility badge is a visible marker some journals award to articles that meet specific open-data or open-materials standards, a practical, checkable signal of transparency.
Post-publication peer review: Post-publication peer review is ongoing evaluation and commentary on a paper after it's already published, supplementing rather than replacing the pre-publication peer review process.
PubPeer: PubPeer is a platform for public, post-publication discussion and critique of published papers, sometimes surfacing errors or concerns missed during formal peer review.
Journal transparency checklist: A journal transparency checklist is a structured set of disclosure criteria (data availability, funding, conflicts of interest, ethics approval) some journals now require authors to complete before submission.
Preprint server credibility: Preprints have not passed peer review, so their claims should be treated as provisional evidence, useful for currency but requiring more caution than a peer-reviewed source.
Handle vs. DOI: A Handle is the broader underlying identifier infrastructure; a DOI is one specific, widely used implementation of it for scholarly content.
Institutional biosafety committee: An institutional biosafety committee is an institutional body reviewing research involving biological materials or agents for safety, distinct from a human-subjects ethics committee.
Author contribution statement: An author contribution statement is a section in a publication detailing each author's specific role, often built directly from the CRediT taxonomy categories.
Conflict of interest statement: A conflict of interest statement is a formal, published disclosure of any financial or personal interests that could bias a study, now a standard requirement across reputable journals.
Vanity press: A vanity press is a publisher producing work primarily for a fee with minimal or no editorial vetting, the book-publishing equivalent of a predatory journal.
Eigenfactor score: The Eigenfactor score is a journal-level metric weighting citations by the influence of the citing journal itself, intended to better reflect genuine field-wide importance than a raw citation count.
DORA (San Francisco Declaration on Research Assessment): DORA is a widely signed initiative arguing against the misuse of journal-level metrics, like Impact Factor, to judge individual researchers or articles.
Leiden Manifesto: The Leiden Manifesto is a widely cited set of ten principles for the responsible use of research metrics in evaluation, cautioning against over-reliance on any single quantitative indicator.
Responsible metrics: Responsible metrics is the broader movement advocating for thoughtful, contextualised use of research evaluation metrics, of which DORA and the Leiden Manifesto are two prominent expressions.
Predatory publisher vs. predatory journal: A predatory publisher may operate many journals under one umbrella using deceptive practices across all of them, distinct from a single predatory journal, which may exist independently.
Impact factor inflation: Impact factor inflation is a general term for any practice artificially raising a journal's reported impact metrics, of which coercive citation and editorial self-citation are two specific methods.
Journal scope mismatch: Journal scope mismatch is when a submitted paper's topic doesn't genuinely fit a journal's stated aims and scope, a common and avoidable reason for desk rejection.
Author-pays vs. reader-pays model: Author-pays (common in open access, via APC) shifts cost to the author. Reader-pays (traditional subscription) shifts cost to the reader or their institution, the two fundamentally different economic models underlying most journal publishing.
Diamond open access: Diamond open access is an open access model with no fee charged to either the author or the reader, typically funded by an institution, society, or grant instead, distinct from standard open access, which usually relies on an APC.
Green open access: Green open access is self-archiving a copy of a paper (often the author accepted manuscript) in a repository, distinct from gold open access, where the official published version itself is made freely available directly by the journal.
Gold open access: Gold open access is the official, publisher-hosted version of an article made freely available immediately upon publication, typically funded through an APC.
Bronze open access: Bronze open access is an article made freely readable on a publisher's site without a clear open license or archiving guarantee, considered a weaker, less durable form of open access than green or gold.
Sherpa Romeo: Sherpa Romeo is a free tool for checking a specific journal or publisher's self-archiving policy, clarifying exactly what version of a paper an author is permitted to share and where.
Retraction notice: A retraction notice is the formal published statement accompanying a retraction, explaining why a paper was withdrawn, worth reading directly rather than assuming a retraction's reason from the paper's topic alone.
Bibliometrics: Bibliometrics is the quantitative study of publications and their citation patterns, the broader field CiteScore, Journal Impact Factor, and h-index all belong to.
Altmetrics: Altmetrics are alternative metrics tracking a paper's attention outside formal citations, social media mentions, news coverage, policy references, a supplementary signal, not a replacement for citation-based metrics.
Self-citation: Self-citation is an author citing their own previous work, legitimate in moderation, but excessive self-citation can artificially inflate a researcher's apparent citation impact.
Citation cartel: A citation cartel is an arrangement, sometimes coordinated between journals or authors, to excessively cite each other's work specifically to inflate citation metrics, a recognised form of metric manipulation.
Salami slicing: Salami slicing is splitting what is really one study's worth of findings into multiple separate publications, inflating a researcher's publication count without adding genuinely new knowledge.
Duplicate publication: Duplicate publication is publishing substantially the same findings in more than one venue without proper disclosure, a recognised publication ethics violation.
Ghost authorship: Ghost authorship is when someone who made a substantial contribution to a paper is not listed as an author, a violation of proper authorship attribution.
Guest authorship: Guest authorship is when someone is listed as an author despite not having made a substantial contribution, often for status or influence rather than actual involvement.
Corresponding author: The corresponding author is the author designated as the main point of contact for a publication, responsible for correspondence with the journal and, often, with readers after publication.
Version of record: The version of record is the final, definitive published version of an article, as opposed to a preprint or an earlier accepted-but-unformatted version.
Erratum, corrigendum, and retraction: An erratum corrects a publisher's error. A corrigendum corrects an author's error. A retraction withdraws the paper entirely, three different levels of correction, not interchangeable terms.
Article-level metrics: Article-level metrics measure the impact of one specific article, such as its own citation count, as opposed to journal-level metrics measuring the publication venue as a whole.
Journal-level metrics: Journal-level metrics measure a journal's overall performance, such as Journal Impact Factor or CiteScore, which say nothing about the quality of any one specific article published within it.
Field-weighted citation impact: Field-weighted citation impact is a metric adjusting raw citation counts for the fact that citation norms differ hugely by field, allowing fairer comparison across disciplines.
PsycINFO: PsycINFO is the primary indexing database for psychology literature, maintained by the American Psychological Association.
ERIC (Education Resources Information Center): ERIC is the primary free database for education research literature, maintained by the US Department of Education.
CINAHL: CINAHL is the primary indexing database for nursing and allied health literature.
Predatory metrics: Predatory metrics are fabricated or meaningless-sounding journal metrics (such as "Global Impact Factor" or "Universal Impact Factor") designed to look legitimate but issued by unaccountable, non-standard bodies, distinct from genuine metrics like the real Journal Impact Factor or CiteScore.
Journal hijacking: Journal hijacking is when a fraudulent website impersonates a real, legitimate journal to collect submission fees, a distinct threat from predatory publishing, since the underlying journal itself is genuine, only the website is fake.
Clone journal: A clone journal uses a name deceptively similar to a real, established journal, intended to confuse authors into submitting or citing the wrong one.
Sting operation (publishing): A sting operation is a deliberate test submitting a flawed or fabricated paper to check whether a journal's peer review process actually catches it, the most well-known example (Bohannon, 2013) exposed widespread acceptance of a deliberately flawed paper across many open access journals.
Discipline-specific indexing: Discipline-specific indexing is the principle that indexing status should be checked against databases relevant to a paper's actual field, a journal absent from Scopus might still be properly indexed in a smaller, discipline-specific database like PsycINFO or CINAHL.
Citation context: Citation context is whether a citing paper actually supports, contradicts, or merely mentions a cited source in passing, a more precise signal of influence than a raw citation count alone.
Author identifier vs. journal identifier: An author-level identifier (ORCID) verifies a person; a journal-level identifier (ISSN) verifies a publication, they answer different credibility questions and shouldn't be confused.
Preservation and archiving (publishing): Preservation and archiving is the practice of ensuring a journal's content remains accessible even if the journal itself ceases operating, services like CLOCKSS and Portico exist specifically to guarantee this, worth checking for a journal you're relying on long-term.

Research Ethics

Ethical clearance: Ethical clearance is formal approval from an institutional review board or ethics committee confirming your study's methods meet the standards for treating participants, data, and risk responsibly. Most institutions require it before data collection can legally begin, not after.
Exemption/waiver: An exemption or waiver is a formal determination that a study doesn't need full ethics review, usually because it uses existing public data or poses minimal risk. It still has to be requested and granted, not assumed.
Data access agreement: A data access agreement is a formal contract specifying what you're allowed to do with a dataset you didn't collect yourself: how it can be stored, shared, and eventually deleted. Breaching it can jeopardize the arrangement for everyone after you.
Belmont Report: The Belmont Report is the 1979 foundational document establishing three core principles for ethical research involving human subjects, still the basis for most institutional ethics review today.
Respect for persons: Respect for persons is a Belmont Report principle: individuals should be treated as autonomous agents, and those with reduced autonomy are entitled to additional protection.
Beneficence: Beneficence is a Belmont Report principle: research should maximise possible benefits and minimise possible harms to participants.
Justice: Justice is a Belmont Report principle: the benefits and burdens of research should be distributed fairly, no group should bear disproportionate risk while another reaps the benefit.
Informed consent: Informed consent is a participant's voluntary agreement to take part in research, given with a genuine understanding of what participation involves, its risks, and their right to withdraw.
Right to withdraw: The right to withdraw is a participant's ongoing right to leave a study at any point, including after data has been collected, without needing to justify the decision or face any penalty.
Vulnerable population: A vulnerable population is a participant group requiring additional ethical safeguards due to reduced capacity to give full, free consent, e.g., children, people in custody, or those in a dependent relationship with the researcher.
Deception (in research): Deception in research is deliberately withholding or misrepresenting the true purpose of a study from participants, permitted only in limited circumstances, and typically requiring a debriefing afterward.
Debriefing: Debriefing is explaining a study's true purpose to participants after their involvement ends, particularly required where deception was used during the study itself.
Anonymisation: Anonymisation is removing or altering identifying information so a participant cannot be linked back to their data, distinct from confidentiality, which protects who can access identifiable data rather than removing the identity itself.
Confidentiality: Confidentiality is an agreement that identifiable participant data will not be disclosed to others, the data may still technically identify someone, but access to it is restricted and protected.
Data classification: Data classification is formal categorisation of data by sensitivity level, common in government and international organisations, restricting what data can leave an institution's own systems at all, beyond standard anonymisation.
Aggregated reporting: Aggregated reporting is presenting findings at a group level rather than an individual level, a technique for protecting identity in small or specialist units where even anonymised individual data could still be identifiable.
Role-based reporting: Role-based reporting is describing a participant by their role or function rather than any identifying detail, used alongside aggregated reporting in small, specialist research settings.
Gatekeeper approval: Gatekeeper approval is permission from someone controlling access to a research site or population, often required before formal university ethics approval can even be sought, and treated as a distinct, earlier step.
Institutional access: Institutional access is the broader arrangement, often formal and negotiated, allowing a researcher to conduct work within an organisation, particularly significant when the organisation is confidentiality-bound.
Insider research: Insider research is research conducted within a setting the researcher already works in or belongs to, requiring explicit handling of positionality since prior knowledge inevitably shapes what's noticed and interpreted.
Positionality: Positionality is a researcher's own position, background, and relationship to the research setting, and how that shapes what they notice, ask, and interpret, treated as a methodological variable to disclose, not a flaw to hide.
Reflexive journal: A reflexive journal is an ongoing written record, kept throughout data collection and analysis rather than reconstructed afterward, documenting how a researcher's own position may be shaping the research.
Dual role conflict: Dual role conflict is the tension a practitioner-researcher faces holding two roles at once, colleague or employee, and researcher, particularly relevant to consent, access, and whether participants feel free to decline.
Conflict of interest: A conflict of interest is a situation where a researcher's personal, financial, or professional interests could improperly influence their research or its reporting, required to be disclosed, not resolved by staying silent about it.
Research misconduct: Research misconduct is serious violation of accepted research standards, commonly defined as fabrication, falsification, or plagiarism, distinct from honest error or differences in interpretation.
Data fabrication vs. falsification: Fabrication is inventing data or results that were never collected. Falsification is manipulating real data or results to misrepresent what was actually found. Both are forms of research misconduct, not interchangeable terms.
Plagiarism: Plagiarism is presenting someone else's words, ideas, or data as one's own without proper attribution, one of the three core categories of research misconduct.
Ethics committee (Institutional Review Board): An ethics committee, or institutional review board, is the formal institutional body reviewing and approving research involving human subjects before it begins, the source of ethical clearance or exemption.
Predatory conference: A predatory conference charges registration fees while offering little genuine peer scrutiny or academic value, existing mainly to generate revenue from presenters. Listing one on a CV can raise the same credibility concerns as citing a predatory journal.
GDPR (General Data Protection Regulation): GDPR is the EU's data protection law, directly relevant to any research involving personal data of EU residents, regardless of where the researcher themselves is based.
Data protection impact assessment: A data protection impact assessment is a formal assessment of how a research project's data handling affects participant privacy, required under many data protection frameworks before certain kinds of data collection begin.
Data minimisation: Data minimisation is the principle of collecting only the data genuinely necessary for a study's stated purpose, not gathering extra data "just in case."
Pseudonymisation: Pseudonymisation is replacing identifying details with a code or pseudonym, where a separate key could still, in principle, re-identify the person, distinct from anonymisation, which removes that possibility entirely.
Data controller vs. data processor: The data controller decides why and how data is used. The data processor handles the data on the controller's behalf, an important distinction under most data protection law, including GDPR.
Data sharing agreement: A data sharing agreement is a formal agreement governing how a researcher's own collected data may be shared with others afterward, distinct from a data access agreement, which governs accessing someone else's existing data.
Assent: Assent is agreement to participate given by someone (commonly a minor) who cannot legally provide full informed consent themselves, typically given alongside a parent or guardian's actual consent.
Proxy consent: Proxy consent is consent given on behalf of someone unable to consent for themselves, by a legally authorised representative such as a parent or guardian.
Community consent: Community consent is agreement obtained from a community's own recognised leadership or structure before research proceeds, relevant particularly in indigenous and community-based research settings, in addition to, not instead of, individual consent.
Full board review: Full board review is the most rigorous level of ethics committee review, required for research carrying more than minimal risk to participants.
Expedited review: Expedited review is a faster ethics review track for research posing minimal risk, reviewed by one or a few committee members rather than the full board.
Continuing review: Continuing review is ongoing ethics oversight for a study already underway, particularly for longer projects, rather than a single one-time approval at the start.
Retrospective ethics approval: Retrospective ethics approval is ethics approval sought after data collection has already begun or finished, generally discouraged and, in most institutions, only accepted in unusual, justified circumstances.
Risk-benefit analysis: A risk-benefit analysis is a formal weighing of a study's potential harms against its potential benefits, central to an ethics committee's decision on whether a study should proceed.
Adverse event reporting: Adverse event reporting is the requirement to report any unexpected harm to a participant during a study promptly to the relevant ethics body, not just at the study's conclusion.
Safeguarding plan: A safeguarding plan is a specific, documented plan for protecting participants, particularly vulnerable ones, from harm during a study, going beyond general informed consent.
Equipoise: Equipoise is a state of genuine uncertainty about whether an intervention is beneficial, considered a necessary ethical condition for running a study (particularly a trial) comparing it against an alternative.
Research integrity office: A research integrity office is an institutional body responsible for investigating allegations of research misconduct, distinct from an ethics committee, which reviews proposed studies before they begin rather than investigating completed or ongoing ones.
Whistleblowing (research context): Whistleblowing is reporting suspected research misconduct or ethical violations, typically protected under institutional policy from retaliation against the person reporting it.
Ethics training certification: Ethics training certification is formal, often mandatory training (such as CITI Program certification) required before a researcher can conduct human subjects research at many institutions.
Indigenous research ethics: Indigenous research ethics is a distinct ethical framework recognising community ownership of knowledge and requiring community-level consent and benefit-sharing, alongside standard individual consent.
Cultural sensitivity (research ethics): Cultural sensitivity is awareness of and respect for cultural norms and values that may affect how consent, data collection, or dissemination should appropriately be handled in a given setting.
Dual-use research: Dual-use research is research with legitimate scientific value that could also be misused for harm, requiring additional ethical consideration around publication and dissemination, particularly relevant in security and technology-adjacent fields.
AI ethics disclosure: AI ethics disclosure is disclosing not just that AI tools were used in research, but any ethical considerations specific to that use, bias in AI-generated outputs, data privacy in tools processing participant data, distinct from the general AI use declaration covered under Writing Up.
Algorithmic bias (in research tools): Algorithmic bias is systematic distortion in an AI tool's outputs, favouring certain groups, languages, or perspectives over others, a real risk when using AI for literature synthesis, since underrepresented scholarship can be under-surfaced.
Confidentiality limits: Confidentiality limits are the circumstances, stated to participants in advance, under which a researcher may be legally or ethically required to break confidentiality, such as disclosed harm to self or others.
Mandatory reporting: Mandatory reporting is a legal or institutional obligation to report certain disclosures (such as child harm) regardless of a research confidentiality agreement.
Duty of care: Duty of care is a researcher's ethical obligation to avoid causing foreseeable harm to participants, extending beyond the minimum of formal consent procedures.
Undue inducement: Undue inducement is when compensation offered to participants is large enough to pressure them into accepting risks they otherwise wouldn't, undermining genuinely voluntary consent.
Voluntary participation: Voluntary participation is the principle that participation must be freely chosen, without coercion, undue pressure, or inducement.
Coercion (research ethics): Coercion, in research ethics, is any pressure, explicit or implicit, that compromises a participant's genuinely free choice to take part or withdraw.
Power imbalance (researcher-participant): A power imbalance between researcher and participant (such as employer-employee) is an unequal relationship that can compromise how freely consent is actually given.
Chain of custody (data): Chain of custody, for data, is a documented, traceable record of who has accessed or handled research data, particularly important for sensitive or legally significant material.
Secure data storage: Secure data storage refers to technical and procedural safeguards protecting research data from unauthorised access, loss, or breach, part of most ethics committees' formal requirements.
Data destruction / retention policy: A data destruction or retention policy is a stated plan for how long research data will be kept and how it will be securely disposed of once no longer needed.
Cross-border data transfer: Cross-border data transfer is moving research data between countries, which may trigger additional legal obligations depending on the data protection regimes involved, GDPR being one prominent example.
Data sovereignty: Data sovereignty is the principle that data is subject to the laws of the country in which it's collected or stored, a growing consideration in international and cross-institutional research.
Third-party data sharing restriction: A third-party data sharing restriction is a limitation, often stated in an ethics approval or data agreement, on whether and how collected data can be shared beyond the original research team.
Research with children: Research with children involves minors, requiring additional safeguards including parental or guardian consent alongside the child's own assent.
Research with prisoners: Research with prisoners involves incarcerated individuals, treated as a vulnerable population requiring heightened ethical scrutiny due to constrained autonomy in that setting.
Research with cognitively impaired participants: Research with cognitively impaired participants involves participants whose capacity to give informed consent may be limited, typically requiring proxy consent alongside the participant's own assent where possible.
Online research ethics: Online research ethics covers ethical considerations specific to research conducted via internet platforms, including consent mechanisms, data security, and the blurred line between public and private online spaces.
Social media research ethics: Social media research ethics addresses specific ethical questions around using publicly posted social media content as research data, including whether public posting implies consent to be studied.
Public vs. private data (ethics distinction): Whether data was posted with a reasonable expectation of privacy shapes what ethical obligations apply, publicly posted content is not automatically exempt from ethical consideration.
Deceased persons' data: Deceased persons' data is research using data belonging to people who have since died, ethics requirements vary by jurisdiction and institution, and shouldn't be assumed to simply not apply.
Historical research ethics: Historical research ethics covers ethical considerations specific to research using archival or historical personal data, balancing historical value against privacy considerations for people who cannot themselves consent.
Ethical review exemption categories: Ethical review exemption categories are the specific, defined categories of low-risk research (such as anonymous surveys or publicly available data) that many institutions allow to bypass full ethics committee review.
Debrief document: A debrief document is a written summary given to participants after a study concludes, explaining the research's purpose and findings, particularly important where deception was used.
Research ethics training: Research ethics training is institutional or externally provided training (formal certification programs are one example) required before conducting human subjects research, distinct from the specific approval process for one particular study.
Anticipated vs. unanticipated risk: Anticipated risks are identified and disclosed in advance during ethics review. Unanticipated risks emerge only once a study is underway, requiring prompt reporting and, potentially, a protocol amendment.
Declaration of Helsinki: The Declaration of Helsinki is a foundational international statement of ethical principles for medical research involving human subjects, predating and complementing the Belmont Report.
Broad consent: Broad consent is consent obtained for a general category of future research use, rather than for one specific, narrowly defined study, common in biobanking and long-term data repositories.
Dynamic consent: Dynamic consent is an ongoing consent model allowing participants to review and adjust their permissions over time, rather than a single fixed decision made once at the start.
Nuremberg Code: The Nuremberg Code is an early, foundational set of ethical principles for research involving human subjects, developed in response to historical atrocities, predating the Belmont Report and Declaration of Helsinki.
Research misconduct investigation process: The research misconduct investigation process is the formal, structured procedure an institution follows once an allegation of research misconduct is raised, distinct from the research integrity office, which is the body that carries this process out.
Third-party consent: Third-party consent is consent or approval obtained from someone other than the participant themselves, relevant when research involves data about people who aren't direct participants, such as family members mentioned in an interview.
Menlo Report: The Menlo Report is a framework extending Belmont-style ethical principles specifically to information and communications technology research, relevant to research involving digital systems, data, or cybersecurity.
Data subject rights: Data subject rights are the specific rights an individual holds over their own personal data under data protection law, including access, correction, and deletion, relevant whenever research involves identifiable participant data.
Ethics dumping: Ethics dumping is the practice of conducting research with weaker ethical oversight in one country or context that wouldn't be permitted under a researcher's home institution's standards, a recognised concern in international and cross-border research.

Reasoning & Critique

Research gap types: Not all gaps are the same: a gap can be evidence based (nobody tested this), methods based (nobody tested it well), knowledge based (findings conflict), or population based (nobody tested it on this group). Naming the type sharpens your justification.
Contradiction: A contradiction is when two credible sources report incompatible findings on the same question. Ignoring it quietly is a shortcut; naming it and proposing why it might exist is the actual analysis.
Trustworthiness: Trustworthiness is Lincoln and Guba's umbrella term for rigour in qualitative research, made up of four criteria: credibility, transferability, dependability, confirmability, replacing the quantitative language of validity and reliability where it doesn't fit.
Credibility: Credibility is the qualitative equivalent of internal validity, whether the findings accurately reflect participants' actual experience, commonly checked through triangulation or member checking.
Transferability: Transferability is the qualitative equivalent of external validity or generalisability, whether findings could plausibly apply beyond the specific setting studied, established through rich, detailed description rather than statistical sampling.
Dependability: Dependability is the qualitative equivalent of reliability, whether the research process itself was consistent and well documented enough that another researcher could follow the same steps.
Confirmability: Confirmability is the qualitative equivalent of objectivity, whether findings are genuinely shaped by participants and data rather than by the researcher's own bias, often demonstrated through an audit trail of decisions made during analysis.
Internal validity: Internal validity is whether a study's design actually supports its causal claims, or whether some other explanation could account for the results just as well.
External validity: External validity is whether a study's findings can reasonably be generalised beyond the specific sample and setting studied.
Construct validity: Construct validity is whether a study is actually measuring the concept it claims to measure, rather than something related but distinct.
Confounding variable: A confounding variable is an unaccounted-for factor that influences both the variable being studied and the outcome, creating a false appearance of a relationship between them.
Triangulation: Triangulation is using multiple sources, methods, or researchers to examine the same question, checking whether independent lines of evidence converge on the same conclusion.
Member checking: Member checking is returning findings or interpretations to participants to confirm they accurately reflect what was meant, a direct check on credibility.
Peer debriefing: Peer debriefing is having a colleague, not directly involved in the study, review and challenge a researcher's interpretations during analysis, a check against unnoticed bias distinct from formal peer review of a finished manuscript.
CARS model (Create a Research Space): The CARS model is Swales' three-move model for how academic writing establishes a research space: describing the existing territory, identifying a niche or gap within it, then occupying that niche by stating the study's own contribution.
Toulmin's model of argument: Toulmin's model of argument is a structure for analysing any argument's components: a claim, the grounds supporting it, the underlying warrant connecting grounds to claim, backing for the warrant, a qualifier on how strongly it's held, and any rebuttal or exception.
Argument mapping: Argument mapping is visually laying out the structure of an argument, claims, supporting evidence, and objections, to make its logic and weak points easier to inspect.
Counterargument: A counterargument is a position that directly opposes a claim being made, addressing it strengthens an argument, ignoring it weakens the argument's credibility even if the original claim is correct.
Devil's advocate: Devil's advocate is deliberately arguing against a position, including one's own, to test whether it holds up under real challenge, a formal check against motivated reasoning.
Confirmation bias: Confirmation bias is the tendency to notice, favour, and remember evidence that supports what's already believed, while overlooking or discounting evidence that contradicts it.
Selection bias: Selection bias is a systematic error arising from how participants, sources, or data points were chosen, producing a sample that doesn't actually represent what it's meant to.
Publication bias: Publication bias is the tendency for studies with positive or striking results to get published more often than studies with null or unremarkable results, distorting what the overall published literature appears to show.
Circular reasoning: Circular reasoning is an argument that uses its own conclusion as one of its supporting premises, appearing to prove something while actually just restating it.
Straw man argument: A straw man argument misrepresents an opposing position as weaker or more extreme than it actually is, then argues against that weaker version instead of the real one.
Logical fallacy: A logical fallacy is a flaw in reasoning that undermines an argument's validity, regardless of whether its conclusion happens to be true.
Critical thinking: Critical thinking is the disciplined practice of evaluating evidence and arguments, including one's own, before accepting a conclusion, rather than accepting the most available or most comfortable answer.
Open coding: Open coding is the first stage of qualitative coding, breaking data into initial, provisional categories without yet imposing a predetermined structure.
Axial coding: Axial coding is the second stage of coding, identifying relationships and connections between the categories developed during open coding.
Selective coding: Selective coding is the final stage of coding, identifying a core category that the study's findings can be organised and integrated around.
Constant comparative method: The constant comparative method is a grounded theory technique of continuously comparing new data against previously coded data as analysis proceeds, rather than coding everything first and comparing afterward.
Data saturation: Data saturation is the point in data collection where new interviews or sources stop producing genuinely new themes or insights, a common (though debated) signal that enough data has been gathered.
Negative case analysis: Negative case analysis is deliberately searching for and examining data that contradicts an emerging pattern or theme, strengthening a finding by testing whether it holds even against its own exceptions.
Rich, thick description: Rich, thick description is detailed, contextualised qualitative description that lets a reader judge for themselves whether findings might transfer to another setting, the primary mechanism for establishing transferability.
Audit trail (qualitative research): An audit trail is a documented record of decisions made throughout data collection and analysis, the primary mechanism for establishing confirmability and dependability.
Type I error: A Type I error is incorrectly concluding an effect exists when it actually doesn't, a false positive.
Type II error: A Type II error is incorrectly concluding an effect doesn't exist when it actually does, a false negative.
Statistical power: Statistical power is a study's ability to detect a real effect if one genuinely exists, low power increases the risk of a Type II error, missing a real finding.
Sample size justification: Sample size justification is a stated, reasoned basis for why a study used the number of participants or data points it did, rather than an arbitrary or convenience-driven number.
Correlation vs. causation: Two variables moving together doesn't establish that one causes the other, one of the most common reasoning errors in interpreting research findings.
Spurious correlation: A spurious correlation is an apparent relationship between two variables that is actually coincidental or explained by a third, unaccounted-for factor, rather than a genuine causal connection.
Regression to the mean: Regression to the mean is the tendency for an extreme measurement to be followed by a more average one on remeasurement, sometimes mistaken for a real effect of an intervention when it's actually a statistical artefact.
Steelmanning: Steelmanning is deliberately presenting the strongest, most persuasive possible version of an opposing argument before responding to it, the opposite of a straw man argument.
Burden of proof: Burden of proof is the principle that the person making a claim is responsible for providing evidence for it, not the person questioning the claim for disproving it.
Appeal to authority (fallacy): Appeal to authority is treating a claim as true simply because an authority figure or credentialed source stated it, without independently evaluating the actual evidence behind the claim.
Hasty generalisation: A hasty generalisation is drawing a broad conclusion from an inadequate or unrepresentative sample of evidence, a common flaw in both everyday reasoning and under-powered research.
False dichotomy: A false dichotomy presents only two options as if they were the only possibilities, when other reasonable alternatives actually exist.
Source triangulation: Source triangulation is checking a claim against multiple independent sources rather than relying on just one, a specific application of triangulation focused on verifying facts rather than combining methods.
Source accuracy verification: Source accuracy verification is confirming a source actually says what it's being cited as saying, not just that it exists, a distinct check from confirming a DOI resolves.
Contradiction resolution: Contradiction resolution is the documented process of deciding how to handle two pieces of evidence that genuinely conflict, whether by favouring the stronger source, seeking a third source, or reporting the contradiction openly rather than silently picking a side.
Weight of evidence: Weight of evidence is an overall judgment of how strongly the total body of available evidence supports a claim, considering quantity, quality, and consistency together, not any single study in isolation.
Intellectual honesty: Intellectual honesty is representing evidence, arguments, and uncertainty accurately, including acknowledging what a study does not show, rather than overstating findings to appear more conclusive than the evidence supports.
Deductive vs. inductive coding: Deductive coding applies predetermined codes derived from theory. Inductive coding lets codes emerge directly from the data itself.
Framework analysis: Framework analysis is a structured qualitative analysis method organising data into a matrix of cases and themes, commonly used in applied and policy research.
Content analysis: Content analysis is a method systematically categorising and counting the occurrence of specific words, themes, or concepts within text or media.
Narrative analysis: Narrative analysis analyses the structure, sequence, and meaning of stories participants tell, distinct from narrative inquiry, which refers to the broader research approach itself.
Case study analysis: Case study analysis refers to the specific analytic techniques (such as pattern matching or explanation building) applied within a case study, distinct from the case study research design that frames the whole study.
Cross-case analysis: Cross-case analysis compares findings across multiple cases in a multiple case study design to identify shared patterns or meaningful differences.
Within-case analysis: Within-case analysis analyses one case in depth on its own terms before comparing it to others, a necessary step before cross-case analysis in multiple case study designs.
Pattern matching (analysis technique): Pattern matching compares an empirically observed pattern in case data against a pattern predicted by theory, a specific technique for strengthening internal validity in case study research.
Explanation building (case study technique): Explanation building is iteratively building and refining a causal explanation for a case's outcome across multiple rounds of analysis.
Analytic generalisation: Analytic generalisation is generalising a case study's findings to theory, rather than to a wider population, a different, non-statistical form of generalisation appropriate to qualitative case research.
Statistical generalisation: Statistical generalisation is generalising findings from a representative sample to a wider population, the form of generalisation quantitative research typically aims for.
Theoretical saturation: Theoretical saturation is the specific point in grounded theory analysis where new data stops adding new properties to the developing theoretical categories, distinct from general data saturation.
Memoing: Memoing is writing analytic notes throughout data collection and coding, capturing emerging ideas and decisions as they happen, a core grounded theory practice.
Researcher bias: Researcher bias is the general risk that a researcher's own expectations or preferences unintentionally shape data collection or interpretation, the broader category confirmation bias and other specific biases fall under.
Social desirability bias: Social desirability bias is a participant's tendency to answer in a way they believe will be viewed favourably, rather than accurately, particularly relevant in surveys and interviews on sensitive topics.
Recall bias: Recall bias is systematic error arising from participants inaccurately remembering past events or experiences, particularly relevant in retrospective study designs.
Response bias: Response bias is a general tendency for participants to answer questions in a way that doesn't reflect their true position, of which social desirability bias is one specific type.
Non-response bias: Non-response bias is systematic differences between people who respond to a study and those who don't, potentially skewing findings if non-responders differ meaningfully from responders.
Halo effect: The halo effect is the tendency for an overall positive or negative impression of a subject to unduly influence judgment of that subject's specific, unrelated characteristics.
Groupthink (research context): Groupthink is a tendency for members of a group (such as a focus group) to converge on a shared view, potentially suppressing genuine individual perspectives, worth accounting for when designing or analysing group-based data collection.
Anchoring bias: Anchoring bias is the tendency to rely too heavily on the first piece of information encountered when making subsequent judgments.
Availability heuristic: The availability heuristic is judging the likelihood or importance of something based on how easily examples come to mind, rather than on its actual frequency or significance.
Survivorship bias: Survivorship bias is drawing conclusions from a group that has already survived some selection process, while overlooking those that didn't, that were never included in the observed sample at all.
Base rate neglect: Base rate neglect is the tendency to ignore genuinely relevant background statistical information in favour of specific, memorable case details.
Overfitting: Overfitting is a statistical or analytical model that fits the specific data it was built on too closely, capturing noise rather than a genuine, generalisable pattern.
Rival explanations: Rival explanations are alternative causal accounts for a finding that must be considered and ruled out before the preferred explanation can be confidently accepted.
Plausibility (of an explanation): Plausibility is whether a proposed explanation is reasonably consistent with established knowledge and mechanisms, a basic check applied before accepting any causal claim, especially in qualitative or theory-building work.
Epistemic humility: Epistemic humility is acknowledging the genuine limits of what a study, or a researcher, can actually know or claim, a disposition closely tied to appropriate hedging in academic writing.
Sampling: Sampling is the process of selecting who or what to study from a larger population, the specific sampling approach chosen shapes what a study can and cannot claim to generalise.
Convenience sample: A convenience sample is a sample chosen based on ease of access rather than deliberate representativeness, quick to gather but limiting what a study can claim about a wider population.
Thematic analysis: Thematic analysis is a qualitative method for identifying, analysing, and reporting patterns (themes) within a single dataset, distinct from thematic synthesis, which applies the same logic across multiple studies rather than within one.

Writing Up

Citation style: A citation style (APA, Chicago, Vancouver, and others) is a formatting convention for crediting sources, chosen by your field or institution rather than by preference. Getting it consistent matters more than which one you pick.
Preprint: A preprint is a manuscript shared publicly before it has gone through peer review.
Published version: A published version is a manuscript that has completed peer review and appeared in a journal, sometimes differing from earlier drafts in real ways.
Preprint vs. published: The practical difference is review status: a preprint hasn't been checked yet and its findings could still change, while a published version has passed review and is the citable, final form. When in doubt, check which one you actually have before citing it.
IMRaD structure: IMRaD structure is the standard journal article structure: Introduction, Methods, Results, and Discussion, the skeleton a doctoral thesis expands into a fuller chapter sequence.
Thesis structure: Thesis structure is the typical doctoral chapter sequence: introduction, literature review, conceptual framework, methodology, findings, discussion, conclusion and contribution, often with the conceptual framework as its own distinct chapter in practice-oriented theses.
Abstract: An abstract is a short, standalone summary of a study's problem, method, and key findings, written to let a reader decide whether the full text is relevant before committing time to it.
Limitations section: A limitations section is the part of a thesis or article stating what the study could not establish or account for, an honest limitations section strengthens credibility, it doesn't weaken the contribution.
Recommendations for future research: Recommendations for future research is the closing statement of what a study's own findings suggest should be investigated next, distinct from the study's own limitations, one names what wasn't done, the other suggests what should be done next.
Findings vs. discussion: Findings present what the data actually showed, without interpretation. Discussion interprets what those findings mean against the existing literature and theory. Conflating the two is a common structural weakness examiners flag directly.
Manuscript: A manuscript is the written version of a paper before or during the formal publication process, distinct from a preprint, which is specifically shared publicly before peer review.
Manuscript submission: Manuscript submission is the formal act of sending a manuscript to a journal, the first step of the publication pipeline, followed by editorial screening and peer review.
In press: In press is the status of an accepted manuscript that has completed peer review but not yet been formally published in an issue, citable but not yet final.
Harvard referencing: Harvard referencing is an author-date citation style common in UK business and social science programmes, in-text citations show author and year, full details appear in an alphabetical reference list.
APA referencing: APA referencing is an author-date citation style similar in structure to Harvard but with its own specific formatting rules, commonly required in psychology and US-based programmes, always confirm which specific style a programme or journal requires rather than assuming.
OSCOLA: OSCOLA is a footnote-based citation style used specifically for legal writing, citing cases and legislation, structurally different from author-date styles like Harvard or APA.
Reference manager: A reference manager is software (Zotero, Mendeley, EndNote) that stores sources, removes most manual citation formatting errors, and builds a structured, searchable personal library as a researcher's reading volume grows.
In-text citation vs. reference list: The in-text citation is the short marker appearing where a source is used in the writing itself. The reference list is the full bibliographic detail, appearing once, at the end. Every in-text citation must correspond to an entry in the list, and vice versa.
Et al.: Et al. is a Latin abbreviation meaning "and others," used in a citation when a source has more authors than a style's own threshold for naming them all individually.
Ibid.: Ibid. is a Latin abbreviation meaning "in the same place," used in footnote-based styles like OSCOLA to refer back to the immediately preceding citation without repeating it in full.
Hedging (academic writing): Hedging is using cautious, qualified language ("suggests," "may indicate") rather than absolute claims, reflecting appropriate uncertainty about findings rather than weak writing.
Word count and word limit: Word count and word limit are the stated length constraint for an assignment, chapter, or journal submission, treated as a firm requirement to design writing around, not a target to reach after writing freely.
Viva voce: Viva voce is the oral examination of a completed doctoral thesis, requiring a concise statement of the problem, a transparent account of method including its limitations, and a clear, evidenced statement of contribution.
AI use declaration: An AI use declaration is a formal statement disclosing which AI tools were used during research, for what purpose, and how outputs were verified, an explicit requirement in many current doctoral programmes, not an optional courtesy note.
Critical generative AI literacy: Critical generative AI literacy is the competency of using AI tools productively in research while remaining fully accountable for accuracy, originality, and disclosure, treating AI as a drafting and triage accelerator, never a substitute for the researcher's own verification and judgment.
Hallucinated citation: A hallucinated citation is a reference an AI tool generates that looks plausible but doesn't correspond to a real source, every AI-suggested citation requires independent verification before it's trusted or used.
Citation verification: Citation verification is confirming that every reference in a bibliography actually exists and resolves correctly, typically by checking its DOI, especially important for any AI-assisted drafting step.
Literature matrix: A literature matrix is a structured comparison table, one row per source, columns for method, sample, findings, and quality tier, turning a pile of sources into something that can actually be compared and argued from.
Conceptual framework chapter: In many DBA and practice-oriented theses, the conceptual framework appears as its own distinct chapter rather than folded into the literature review, making its links back to theory and identified gaps explicit and separately assessable.
Executive summary: An executive summary is a concise overview of a study's purpose, method, and key findings, written for a reader who may never read the full document, common in DBA and practice-oriented theses aimed partly at a practitioner audience.
Table of contents: The table of contents is the structured listing of a document's sections and page numbers, its own internal consistency (headings matching actual section titles) is a basic but commonly overlooked quality check.
Appendix: An appendix is supplementary material placed after the main text, supporting documents, full interview transcripts, additional data tables, kept separate so it doesn't interrupt the main argument's flow.
Chapter summary: A chapter summary is a short closing section within a chapter restating its key points and explicitly linking forward to the next chapter, helping a reader track the thesis's overall argument across many pages.
Direct quotation vs. paraphrase: A direct quotation reproduces a source's exact words, requiring quote marks and a precise page reference. A paraphrase restates the idea in the writer's own words, still requiring a citation, but not quote marks.
Secondary citation: A secondary citation is citing a source you found quoted or described within another source, rather than the original itself, generally discouraged in doctoral work in favour of locating and citing the original directly.
Page-specific citation: A page-specific citation includes the exact page number a quotation or specific claim was drawn from, required for direct quotations in most referencing styles, not just the source's year and author.
Reference list vs. bibliography: A reference list contains only sources actually cited in the text. A bibliography may also include sources consulted but not directly cited, the two terms are not interchangeable despite common informal use.
Signposting (academic writing): Signposting is explicit language telling the reader what a section will do before it does it ("This section examines..."), helping a reader navigate a long, complex document without getting lost.
Academic writing tone: Academic writing tone is the register expected in doctoral writing, formal, precise, and hedged where appropriate, avoiding both overly casual language and unsupported absolute claims.
Cohesion (writing): Cohesion is the quality of a piece of writing holding together as a connected whole, through consistent terminology, clear transitions, and a visible argumentative thread from paragraph to paragraph.
Voice (active vs. passive): Active voice states who performed an action directly ("The researcher collected data"). Passive voice omits or de-emphasises the actor ("Data was collected"), academic convention varies by field and should match the expected style of the target programme or journal.
Plain language summary: A plain language summary is a short, non-technical summary of a study written for a general audience, increasingly required by funders and some journals alongside the technical abstract.
Conference presentation: A conference presentation is sharing research findings orally at an academic conference, often before formal journal publication, a common way to get early feedback and establish priority on emerging findings.
Poster presentation: A poster presentation is a visual, self-contained summary of a study displayed at a conference, typically discussed informally with attendees rather than presented as a formal talk.
Research impact: Research impact is the demonstrable effect a piece of research has beyond academia itself, on policy, practice, or public understanding, increasingly a formal consideration in research assessment.
Knowledge translation: Knowledge translation is the active process of converting research findings into a form usable by practitioners or policymakers, distinct from simply publishing findings and hoping they're found and applied.
Lay summary: A lay summary is a summary of research written specifically for a non-academic public audience, similar in spirit to a plain language summary but aimed more broadly at public communication than at funders or ethics bodies.
Supervisor: A supervisor is the academic assigned to guide a doctoral candidate's research, providing feedback and oversight throughout the study, distinct from an external examiner, who evaluates the completed thesis independently.
External examiner: An external examiner is an academic, independent of the candidate's own institution and supervision, who evaluates a completed doctoral thesis, often alongside an internal examiner.
Upgrade or transfer viva: An upgrade or transfer viva is an interim examination, common in many doctoral programmes, assessing whether a candidate's progress justifies continuing from a provisional to a full doctoral registration.
Minor corrections vs. major corrections: Common outcomes of a viva, minor corrections require small, quick fixes before final submission, major corrections require substantial rework and often a further review before the degree is confirmed.
Embargo (thesis): An embargo is a period during which a completed, submitted thesis is kept restricted from public access, sometimes requested to protect sensitive data or allow time for journal publication before the full thesis becomes freely available.
Final submission vs. provisional submission: A provisional submission is made for examination purposes and may still require corrections. Final submission is the corrected, approved version formally lodged with the institution.
Deposit copy: A deposit copy is the official archived copy of a completed thesis lodged with an institution's library or repository, the permanent record of the completed work.
Introduction chapter structure: Introduction chapter structure refers to the typical components of a thesis introduction, background, problem statement, research questions, significance, and a roadmap of the chapters to follow.
Literature review chapter structure: A literature review chapter is commonly organised thematically or chronologically around a study's key concepts, building toward the research gap the study addresses, rather than simply summarising sources one by one.
Discussion chapter structure: A discussion chapter is typically organised around each research question or objective in turn, interpreting findings against existing literature and theory, distinct from the findings chapter, which reports results without this interpretation.
Conclusion chapter structure: A conclusion chapter commonly restates the study's contribution, summarises key findings briefly, and addresses limitations and future research, without introducing new findings or citations at this stage.
Implications for practice: Implications for practice is a section explicitly stating what a study's findings mean for practitioners or organisations, distinct from implications for theory.
Implications for theory: Implications for theory is a section explicitly stating how a study's findings extend, challenge, or refine existing theoretical understanding.
Implications for policy: Implications for policy is a section explicitly stating what a study's findings suggest for policymakers, particularly relevant in applied and public sector-oriented doctoral research.
Thesis by publication: A thesis by publication is a doctoral thesis format consisting of a set of published or publishable journal articles, linked by an overarching introduction and conclusion, an alternative to the traditional monograph format.
Traditional monograph thesis: A traditional monograph thesis is written as one continuous, extended document, the more conventional format compared to a thesis by publication.
Cumulative dissertation: A cumulative dissertation is another common term for a thesis by publication format, terminology varies by country and institution.
Track changes: Track changes is a word-processing feature recording edits visibly, commonly used between a candidate and supervisor during iterative drafting and review.
Version control (writing): Version control, for writing, is systematically managing successive drafts of a document so changes can be tracked and previous versions recovered if needed, increasingly important across a multi-year doctoral writing process.
Peer feedback (writing process): Peer feedback, in the writing process, is feedback from fellow researchers or colleagues on a draft's argument and clarity, distinct from formal peer review of a submitted, finished manuscript.
Proofreading vs. editing: Proofreading checks for surface errors, spelling, grammar, formatting. Editing addresses deeper issues of argument, structure, and clarity, both distinct passes, not the same task.
Copyediting: Copyediting is a specific, detailed editing pass focused on grammar, consistency, and style compliance, typically done close to final submission.
Similarity report: A similarity report is an automated report from plagiarism-detection software showing what percentage of a document's text matches existing sources, a tool for checking, not a final judgment on whether plagiarism occurred.
Self-plagiarism: Self-plagiarism is reusing substantial portions of one's own previously published or submitted work without proper citation or disclosure, a recognised ethical issue despite the material being one's own.
Paraphrasing tools (AI-assisted): Paraphrasing tools are software that reworks a text's wording, requiring the same disclosure discipline as any other AI tool use, and the same requirement that the underlying idea's citation is preserved.
Style guide: A style guide is an institution's or journal's specific formatting and writing conventions, covering everything from citation style to heading formats, always confirmed directly rather than assumed.
Publication manual: A publication manual is a comprehensive, published reference to a specific citation and formatting system, the APA Publication Manual being the best-known example.
Final defense: A final defense is, in some doctoral systems, a distinct formal examination event separate from or synonymous with the viva voce, terminology and process vary meaningfully by country and institution.
Research contribution statement (chapter-level): A research contribution statement is a specific, often required standalone statement, sometimes its own short section, explicitly naming the thesis's contribution in a few precise sentences.
Grammar-checking tools: Grammar-checking tools are software assisting with surface-level language correctness, a lower-stakes and generally uncontroversial use of AI or automated tools compared to content generation.
Plagiarism checking software: Plagiarism checking software refers to automated tools comparing a submitted document against a large database of existing text to flag potential unoriginal content, producing a similarity report as output.
Compilation thesis: A compilation thesis is another regional term sometimes used interchangeably with thesis by publication or cumulative dissertation, worth confirming which exact term a specific institution uses.
Author accepted manuscript: An author accepted manuscript is the final version of a paper accepted after peer review, but before a publisher applies its own typesetting and formatting, often the version authors are permitted to self-archive even when the final published version is paywalled.
Structured abstract: A structured abstract is an abstract format with explicit, labeled subsections (Background, Methods, Results, Conclusion), common in health and science journals, distinct from an unstructured, free-flowing abstract paragraph more common in humanities and some social sciences.
Rolling submission: Rolling submission is a journal or conference practice of accepting and reviewing submissions continuously throughout the year, rather than only during fixed, periodic submission windows.

Research Terminology

Core Concepts

Framing Your Question

Searching

Screening & Selecting

Judging Evidence Quality

Research Ethics

Reasoning & Critique

Writing Up