JARGON BUSTER

Research terms explained.

Demystifying Research Series

Read more

Analysis

In simple terms, analysis means to break something down into its smaller parts to understand more about it. In research terms, analysis is the systematic process of examining and interpreting collected data to identify patterns, themes, relationships, and trends. It's about making sense of data to answer research questions and draw meaningful conclusions. There are different forms of analysis used in research and these depend on the research questions being asked and the type of data being collected to answer them.

  • Quantitative analysis is used for numerical data. It can involve describing trends and patterns in data or summarising data in tables and charts. These are known as descriptive statistics. Inferential statistics are used to explore relationships between data, make predictions or to generalise from the sample of data to a larger population of interest. This is known as generalisation.

  • Qualitative analysis is used where data is not numerical and where research questions are more concerned with understanding meaning and experiences. There are many forms of qualitative analysis including, thematic, discourse, narrative, content and grounded theory approaches.

  • Researchers interested in combining qualitative and quantitative data may choose to adopt what are known as a mixed methods approaches to analysis.

Bias

Bias in research is a systematic error or slant that causes findings to consistently lean in a particular direction, away from the true picture. It's like a faulty scale that always adds a little bit extra, or a camera lens that always makes things look a bit redder than they are. Bias can occur at any point in a study and, importantly, is often unintentional.

There are many common misconceptions about bias in research:
✗ Qualitative research is more biased than quantitative research
Both types of research are susceptible to different forms of bias. Quantitative research, while relying on numbers, can suffer from sampling bias, measurement errors, or statistical manipulation. Qualitative research, which explores experiences and perspectives, is more prone to observer bias or confirmation bias from the researcher's interpretation. Rigour in both approaches involves specific strategies to address their respective biases.
✗ A large sample size eliminates bias
A large sample size helps reduce random errors in quantitative research but it does not eliminate bias. A large, biased sample will still give you precisely biased results.

There are many ways to address the risk of bias in research including:
✓ In quantitative research: ensuring representative sampling, randomly assigning participants to groups then blinding people to the group they are in.
✓ Across all research types: using standard procedures to ensure data collection and measurement are done consistently, having more than one researcher collecting or analysing data and using multiple types of data to explore a specific topic (also known as triangulation).

Blinding, blind

Blinding is a technique used in randomised controlled trials where participants, researchers, or both are kept unaware of which treatment group participants have been assigned to. This prevents knowledge of the treatment allocation from influencing behaviour, expectations, or assessment of outcomes.

Blinding helps minimise several types of bias including performance bias (where knowledge of treatment affects how participants or researchers behave), detection bias (where knowledge influences how outcomes are measured or reported), and expectation bias (where beliefs about treatment effectiveness influence perceived or reported results). By preventing these influences, blinding helps ensure that any observed differences between groups are more likely due to the actual treatment effect rather than psychological or behavioural factors related to knowing which treatment was received.

There are different types of blinding used:
• Single-blind: Either participants or researchers (but not both) are unaware of treatment assignments
• Double-blind: Both participants and researchers are unaware of treatment assignments
• Triple-blind: Participants, researchers, and data analysts are all unaware of treatment assignments

Cochrane Reviews

Cochrane Reviews are systematic reviews of primary research in health care and health policy, produced by the Cochrane Collaboration which is an international network of researchers and health professionals. These reviews represent the gold standard for systematic reviews in healthcare.

Confounding variables, confounders 

Confounding variables (also called confounders) are factors that are associated with both the exposure/treatment being studied and the outcome of interest but are not part of a causal pathway between them. These variables can create a false impression of a relationship between the exposure and outcome or hide a true relationship. 

For example, age could be a confounder in a study examining the relationship between exercise and heart disease, as older people may exercise less and have higher rates of heart disease regardless of exercise. 

Researchers control for confounders through randomisation, matching participants, statistical adjustment, or stratification to increase confidence that the groups being compared are similar except for the exposure being studied. 

Control, controlled, control for, control conditions, control groups 

To control in an RCT means to manage, restrict, or account for variables that might affect the outcome being studied (often confounding variables). By controlling variables through design (randomisation, matching) or analysis (statistical adjustment), researchers try to reduce confounding bias and increase confidence that observed differences are due to the treatment rather than other factors. For example, "controlling for age and gender" means adjusting the analysis to account for differences in these factors between groups. 

Control conditions (also known as control groups) are a comparison group in an experiment that does not receive the treatment being tested or receives a standard or placebo treatment. This provides a baseline against which to measure the effect of the intervention. 

Data 

Data refers to the information collected during a research study. This can include numbers, measurements, observations, responses to questions, or any other facts gathered to answer the research question. 

Data forms the evidence base that researchers analyse to draw conclusions, test hypotheses, and answer research questions. The quality and reliability of data directly affect the validity of research findings.  

Effect, effective, effectiveness, efficacy 

An effect is the change or difference that occurs because of an intervention or treatment. This can be measured as the difference between treatment and control groups, or before and after an intervention. 

An intervention is considered effective when it produces the intended beneficial outcome under controlled research conditions (often in clinical trials). This demonstrates that the intervention can work when delivered under optimal circumstances. 

Effectiveness relates to how well an intervention works in real-world, everyday conditions outside of controlled research settings. This measures whether the intervention continues to produce benefits when used in routine practice with usual care conditions. This differs from efficacy, which describes whether an intervention works in a research setting. For example, a psychological intervention may show an effect in research trials (efficacy) but show reduced effectiveness when assessed in routine practice due to factors like poor adherence. 

Evidence based medicine 

Evidence-based medicine in health is the practice of making clinical decisions by integrating the best available research evidence with clinical expertise and patient values and preferences. Arguably, research evidence is given a greater emphasis than clinical expertise and patient values and preferences. 

Experiment, experimental research 

An experiment is a controlled study where researchers deliberately manipulate one or more variables to observe the effect on an outcome, while controlling other factors that might influence the results. 

 Experimental research uses experiments to establish cause-and-effect relationships by systematically manipulating variables under controlled conditions. Experimental designs include randomised controlled trials. 

Fidelity 

Fidelity refers to how closely an intervention is delivered according to its original design, protocol, or intended specifications. It measures whether the intervention was implemented as planned and consistently across different settings, providers, or participants. Researchers use checklists, audio/video recordings, manuals, supervision, or independent observers to assess and maintain fidelity. Ensuring fidelity increases confidence that observed effects (or lack thereof) are truly due to the intervention rather than variations in how it was delivered. 

Follow up  

Follow-up refers to the continued monitoring and assessment of research participants after an initial intervention or baseline measurement has occurred. This involves collecting additional data at specified time points to track changes, outcomes, or long-term effects. 

Generalisation, generalisability, generalise 

Generalisation (also called generalisability or external validity) refers to the extent to which research findings from a specific study can be applied to other people, settings, times, or circumstances beyond those directly studied. Factors affecting generalisation include a study setting and the people recruited. For example, an anxiety self-management intervention tested in university students may not generalise to older people. 

Hierarchy of evidence 

The hierarchy of evidence is an influential ranking system commonly applied in health research. Studies ranked higher in the hierarchy are considered to provide stronger, more trustworthy evidence, helping researchers, clinicians, and policymakers identify and prioritise the most reliable evidence when making decisions. The hierarchy generally ranks systematic reviews and meta-analyses at the top, followed by randomised controlled trials, then observational studies, with expert opinion and case studies at the bottom.  

Limitations of the hierarchy of evidence: 

  • Not all research questions can be answered with RCTs 

  • Study quality within each level can vary significantly 

Context and relevance also matter, not just methodology 

Hypothesis 

A hypothesis is a testable prediction or educated guess about the relationship between variables that a researcher makes before conducting a study. It states what the researcher expects to find and can be proven right or wrong through data collection and analysis. For example: children who eat breakfast will perform better in exams than students who do not.  Predefining hypotheses before doing any research is considered a core feature of the scientific method. It can reduce bias and cherry-picking and makes research falsifiable (a good hypothesis can be proven wrong), which helps build cumulative knowledge. 

Inferential statistics 

Inferential statistics are mathematical methods that allow researchers to draw conclusions about a larger population based on data from a smaller sample (i.e., generalisation). They aid hypothesis testing by determining whether findings are likely due to real effects or just random chance. They also help to quantify uncertainty with confidence intervals and p-values that tell researchers how confident they can be in their conclusions. 

Examples include: 

  • T-tests (comparing group averages) 

  • Correlation analysis (measuring relationships) 

  • Regression (predicting outcomes) 

  • Chi-square tests (analysing categorical data)

The choice of which tests to use, how to interpret results, and what significance levels to accept all involve subjective decisions that can influence conclusions. 

Intervention 

An intervention is a deliberate action or treatment that researchers introduce to study its effects on participants or outcomes. It is what the researcher actively does or gives to see if it causes a change. Interventions are typically compared against a control group, who don't receive the intervention or who receive an alternative. Researchers measure specific variables before and/or after the intervention to assess its impact. 

Journal publication 

Journal publication is the process by which research findings are formally shared with the scientific community through academic journals. It serves as the primary method for disseminating and validating new research knowledge. Published articles become part of the permanent scientific record, allowing others to cite, replicate, or build upon the work. Journals employ peer review, whereby experts in the field anonymously evaluate the research for quality, methodology, significance, and accuracy. They can recommend acceptance, revision, or rejection and, in combination with editorial oversight, publication procedures theoretically ensure only rigorous, valid research reaches the scientific community. However, journal publication is subject to a number of biases and quality issues.  

Mechanism (of change) 

When researchers refer to mechanisms of change or mechanisms of action they're asking: how does something actually work? What are the underlying processes that lead to the outcomes we observe? Think of it like understanding how a treatment like a medicine works, rather than just knowing that people who take it often get better. Understanding mechanisms matters for several reasons. First, it helps explain why interventions work (or don't work) in different contexts - if you know the active ingredients, you can better understand what needs to be present for success. Second, it informs better research design - studies can be designed to examine specific mechanisms rather than treating the intervention as a "black box." Third, it supports implementation - if we know what's actually creating change, we can focus on those elements when rolling out programmes. 

Methods, methodology 

Methods (or methodology) refers to the systematic procedures, techniques, and approaches researchers use to collect, analyse, and interpret data to answer their research questions.  

Examples include: 

  • Qualitative interview based studies 

  • Randomised controlled trials 

  • Ethnographic fieldwork with participant observation 

  • Surveys with validated scales 

Decisions about which methods to use, how to implement them, and what to prioritise reflect underlying assumptions and can introduce bias. 

Outcome 

Outcome refers to the specific results or effects that researchers measure to determine whether their intervention, treatment, or variable of interest has had an impact. It's what the study is designed to detect or change. There are many different types of potential outcome and identifying which outcomes to prioritise and measure can be highly contested in research.  

Participants  

Participants are the people who take part in a research study and provide the data that researchers collect and analyse. 

Qualitative 

Qualitative research explores and understands experiences, meanings, behaviours, and social phenomena through non-numerical data like words, observations, and narratives. Key characteristics include rich descriptions (a focus on depth rather than breadth, seeking detailed understanding of "how" and "why" rather than "how many") and flexible methods (interviews, focus groups, observations, case studies, and document analysis to gather data). Qualitative research uses subjective interpretation; researchers interpret meaning from participants' own words and perspectives rather than predetermined categories and aims to understand phenomena within their natural settings and cultural contexts. 

Quality  

Quality in research refers to the standards and criteria used to evaluate how well a study is designed, conducted, and reported. It encompasses the rigor, reliability, and validity of the research process and findings. 

Quantitative 

Quantitative research uses numerical data and statistical methods to measure, compare, and analyse variables to test hypotheses and identify patterns or relationships. Quantitative methods aim to minimise subjective interpretation through structured data collection and statistical analysis. They tend to use consistent, replicable methods to ensure reliability and reduce researcher bias. 

Random assignment/allocation, randomisation  

Random assignment/allocation (randomisation) is the process of using chance (like flipping a coin or computer-generated numbers) to assign participants to different groups or conditions in a quantitative study, ensuring each participant has an equal probability of being placed in any group. Randomisation reduces bias and is intended to create comparable groups in a study. It allows researchers to conclude that differences between groups are likely due to the intervention rather than pre-existing differences. 

Randomised Controlled Trials (RCTs) 

Randomised Controlled Trials (RCTs) are experimental studies where participants are randomly assigned to receive either an intervention (treatment group) or a control condition (control group), allowing researchers to more confidently test whether the intervention causes specific outcomes. 

Relationship 

In research terms, a relationship refers to a connection, association, or pattern between two or more variables that can be observed, measured, and analysed. Quantitative researchers use various statistical methods to identify, measure, and test the significance of relationships, helping to understand patterns in data and inform theories about how and if different factors connect to each other. 

Reliability 

In quantitative research, reliability refers to the consistency, stability, and repeatability of a measurement instrument or research procedure. It indicates whether a measurement tool produces the same or very similar results when used repeatedly under the same conditions. 

Replication crisis  

The replication crisis refers to the widespread failure to reproduce the findings of many published research studies, particularly prominent in psychology but affecting other sciences as well. This crisis has fundamentally challenged confidence in psychological research findings. 

Sample/ Sampling 

In research, a sample refers to a subset of individuals, cases, or units selected from a larger population to participate in a study. The sample represents the group that researchers actually collect data from. Sampling is the process of selecting participants from the target population. There are various approaches to sampling including random sampling and convenience sampling. The more representative the sample, the stronger the basis for generalisation. 

Scientific method  

The scientific method is a systematic approach to investigating phenomena, acquiring knowledge, and testing ideas through empirical observation and experimentation. Key characteristics of this way of knowing include: 

  • Structured approaches and standard procedures, including rigorous documentation of methods to support replication. 

  • Taking a theory driven approach beginning with existing theoretical frameworks or models and explaining how findings support, refute, or modify existing theory. 

  • Publication of findings with peer review, adding to cumulative knowledge. 

Statistical significance 

Statistical significance is a statistical concept that indicates whether an observed result in a study is likely due to a real effect rather than random chance or sampling error. Reported with a P-value, which is the probability of obtaining the observed results (or more extreme) if the null hypothesis were true. The null hypothesis is the assumption that there is no real effect or relationship between variables - any observed difference is due to chance. It does not provide an indication of the size of any effect, which is calculated separately. 

Systematic reviews 

Systematic reviews are comprehensive, methodical studies that collect, critically evaluate, and synthesise available research evidence on a specific research question using transparent, systematic methods. There are numerous approaches to systematic review, often determined by the type of studies to be included. 

Treatment, Treatment as usual (TAU) 

In health and social research, treatment refers to any intervention, therapy, or procedure administered to participants to address a health condition, symptom, or to promote wellbeing. Treatments are the active components being tested in clinical studies, often RCTs. 

Umbrella review 

An umbrella review (also called an "overview of reviews") is a systematic review that synthesises evidence from multiple systematic reviews rather than individual primary studies. It represents the highest level of evidence synthesis. 

Validity, validation 

Validity in research refers to the extent to which a study or measurement tool accurately measures what it claims to measure and produces trustworthy, meaningful results. It's a fundamental concept for ensuring research quality and credibility. Validation refers to the process of establishing and demonstrating validity through systematic testing and evidence collection. 

Variable 

A variable in research is any characteristic, attribute, or factor that can vary or change across individuals, situations, or time. Variables are the building blocks of research studies and represent the phenomena that researchers measure, manipulate, or control.