About Us

Group Members / Profiles

Professor Christi Deaton

Dr Stephanie Tierney

Michael Connolly

Mary Dodd

Tarnya Hulme

Stephen Woods




Funding Opportunities






Glossary of Terms

Useful Links

Key Documents

MERG logo

Glossary of Terms

Glossary of useful terms relating to research and critical appraisal.

Jump To :   A  B  C  D  E  F  G  H  I  J  K  L  M  N  O  P  Q  R  S  T  U  V  W  X  Y  Z  +  


Action research - the study of a service with the intention to improve what is delivered based on findings from the research. Action research is usually cyclical, during which change and understanding is pursued, with action and critical reflection taking place in turn. Reflection is used to review the previous action and to plan the next one. Action research is often conducted by practitioners who recognise a problem or limitation in their workplace and, together, devise a plan to counteract this, carry out their plan, observe what happens, reflect on any outcomes, revise the plan, implement it, reflect, revise and so on.

Analytic generalisation - refers to the utility of concepts/constructs derived from qualitative data to explain the phenomenon being investigated. A study may be generalisable to theoretical propositions and not to populations, with a goal being to expand and generalise theories and not to compute frequencies. Qualitative research tends to focus on analytic generalisation (to theory) rather than empirical generalisation (to populations).


Bias - the deviation of results from the truth due to the manner in which a study has been conducted.

Blinding - refers to measures taken to disguise allocation to groups to avoid bias. In a single-blind study, participants are unaware of which group they have been put into, but the person delivering the intervention does know (or vice versa). A double-blind trial means that neither the participants nor the person delivering the intervention are aware of allocation to groups.


Case control study - a set of people with a particular condition or problem and another set without this condition or problem are examined to look for a potential cause of the condition or problem.

Case study - an intensive examination and description of a single client, or a group of clients with similar difficulties, or a specific setting, to advance understanding of an event or problem or condition.

Chi-square test - a statistical test used to assess the statistical significance of a finding, i.e. whether findings in a study occurred more than would be expected by chance (see also p-value and statistical significance below).

Confidence interval (CI) - indicates the uncertainty of a specific finding because it has been calculated from only a sample of the total population (Population used here refers to the wider set of individuals for whom you hope to draw conclusions from the results produced by the sample, e.g. all patients using a hospital). It tends to be reported as 95%, which means that one can be 95% certain that the true value for a total population rests within the confidence interval range that is based on data from the sample.

Content analysis - coding and organising data, often from documents or media sources. It generally involves a systematic approach and is carried out in an objective manner, with rules set in advance for how to code data.

Controls - a comparison group. In a randomised controlled trial (see below) controls are allocated to receive nothing, usual care, or an alternative intervention to the one being investigated.

Cost effectiveness - the cost of a service compared to the outcomes it produces.

Cross-sectional study - data are collected at one point in time and then the relationships between characteristics are considered. Hence, it provides a snapshot of the frequency or characteristics of a phenomenon at a particular point in time. But it cannot be used to confirm cause-effect relationships.


Data saturation - data collection in qualitative research is carried out until no new themes or ideas are emerging.


Ethnography - a qualitative research methodology that entails collecting and analysing data in a manner that considers the social and cultural settings of those involved. It has been closely associated with anthropology, for describing cultural patterns of groups, but can also be used to look at things like hospital cultures and a particular group of patients (e.g. those with an eating disorder).


Focus groups - a group of individuals brought together to discuss a specific subject, on which they have a common interest, experience or knowledge.


Grounded theory - an approach to qualitative research that rests on collecting and analysing information gathered with the intention of formulating and testing theory emerging from it.


Hawthorne effect - the phenomenon of those involved in research changing their behaviour simply because they know they are being studied.

Hypothesis - a statement to be tested through investigation/research.


Intention to treat analysis - whereby participants involved in a trial are followed up at the end, with individuals placed in the group they were randomised to, regardless of whether they actually received that intervention. Hence, participants are analysed in the group to which they were originally allocated, to avoid compromising randomisation.

Inter-rater reliability - using two or more researchers to see how far they come to the same conclusions using the same data.


Longitudinal study - the same data is collected at different time points from the same participants, over a certain period.


Mean - the average, created by adding all measures and then dividing them by the total number of measures.

Median - after placing all measures in numerical value (e.g. from smallest to the largest), the median is the value that comes in the middle.

Meta-analysis (relates to systematic reviews, see below) - refers to the assimilation of figures from studies in a systematic review. Single studies with a small sample size are prone to report false negative results, i.e., failing to show a statistically significant difference when one exists, which is why a meta-analysis can prove useful, combining data from different trials containing only small numbers of participants.

Mode - the value occurring most frequently in a series of numbers.


Narrative reviews - in contrast to systematic reviews, narrative reviews tend to be less methodically executed, often focusing on easily accessible research, such as that published in major journals. The problem with this approach is that authors may consciously or unconsciously refer to those studies that reflect their own biases.

Null hypothesis - proposes, for example, that no difference exists between two interventions apart from that occurring by chance.

Number needed to treat (NNT) - is a means of evaluating the effectiveness of an intervention. It examines how many people would need to receive a specific intervention (e.g. family therapy) to see one occurrence of a particular outcome (e.g. improvements in family functioning).


Observation - in terms of qualitative research it entails a researcher watching behaviour, conversations and other interactions, and recording them in words, on cassette/digital equipment, or as a video.

Odds - the probability of an event occurring.

Odds ratio (OR) - the odds for an experimental group divided by the odds for a control group. It is one means of assessing the effectiveness of an intervention.


Participant observation - involves a researcher actually partaking in a setting as well as simply observing.

Phenomenology - aims to discover how individuals interpret their own situation and life. Research from this standpoint attempts to provide an insiderís perspective on a topic. Key influences include Husserl - a Husserlian approach to phenomenology involves bracketing preconceptions when investigating a phenomenon, so the researcher can look at it in its essential form (i.e. study it without preconceptions). Another key influence is Heidegger - a Heideggerian approach to phenomenology would argue that it is impossible to investigate a phenomenon without pre-understanding; therefore, the emphasis is not on bracketing out preconceptions but bringing these to the interpretation. The researcher will reflect on how their own understanding and influence is part of their interpretation of data collected.

Power - the likelihood that a sample is large enough to detect a statistically significant difference (see below) between, for example, a control and an intervention group, if such a difference actually exists.

Publication bias - refers to a tendency for research to be published if it produces positive results.

Purposeful (purposive) sampling - choosing specific participants, with particular characteristics, rather than being based on random selection.

P-value - a statistical measure examining how far a difference shown in findings from research could have occurred by chance. Convention states that a p-value greater than 0.05 (5%) suggests that a difference is likely to have occurred by chance, and cannot be related directly to an intervention.


Quasi-experimental study - study designs in which the participants are not randomised to conditions, although experimental procedures are still employed. Researchers do not have complete control of an independent variable because the intervention is already in process, or because it is impossible or ethical to manipulate the variable. For example, when measuring effects of smoking on peopleís health, it would be unethical to randomise people to a smoking or non-smoking group. Therefore, researchers have to rely on existing populations, such as people already smoking versus those who do not smoke.


Randomisation - when researchers allocate participants to one of two or more groups. Assignment to a group is determined by chance, which should mean that variables not controlled for (especially those that are difficult to measure) are evenly distributed between groups involved in the research.

Randomised controlled trial (RCT) - are designed to minimise bias, as far as possible, eradicating elements that would favour an experimental or control arm. The main features of RCTs are randomisation and the presence of a control group. Randomisation helps produce a representative sample, typical of the population under investigation (e.g. consumers of a specific service). The use of a control group, receiving nothing, or an alternative intervention (often standard care), means that any differences between the two groups can be more firmly related to the effect of the independent variable (e.g. the intervention under consideration), as long as the groups are of a similar make up from the outset on key variables.

Reliability - is about ensuring that results are not simply a quirk of the particular sample involved or manner in which the study has been conducted. For example, steps should be taken to ensure that coding and data analysis are consistent when carried out at different times by the same researcher or by two different researchers.


Semi-structured interview - an interview in which the researcher has a set of themes and areas they wish to discuss with an interviewee. They will have thought about questions to ask in advance. However, they are not bound by these, and can investigate emerging issues arising from the interview.

Standard deviation (SD) - measures how far results scatter from the mean (i.e. the spread of scores that make up the mean around this average figure). Approximately two thirds of values will fall within one standard deviation of the mean and 95% within two standard deviations of the mean.

Statistical significance - a research result or finding that is unlikely to have occurred by chance. Conventionally, a p-value equal to or less than 0.05 is held to be statistically significant, although some studies are stricter and use 0.01 as a cut off point to determine whether a finding is likely or not to have been a random occurrence or representative of a true difference.

Structured interview - an interview (e.g. face to face or via the telephone) in which the same predetermined questions are asked to each participant in the same way.

Survey - information gathered via questionnaire or structured interview at one time point to obtain responses from more than one person, which can then be quantified and subjected to statistical analysis.

Systematic reviews - are referred to as secondary forms of research because they gather, analyse and collate findings from a number of primary studies. They are a useful means of combining vast quantities of research into coherent reports. If a number of studies on a similar theme all have similar conclusions, findings are strengthened.


Theme - an issue from qualitative data that represents something important to the research topic and reflects, at some level, patterns within the data.

Theoretical sampling - used in studies in which developing theory is key (e.g. a piece of grounded theory). It involves seeking out participants who allow the researcher to test the robustness of a theory they are developing, under different conditions. Hence, individuals are selected to test an identified theoretical issue.

Triangulation - a comparison of two or more theories, methods, data sources, investigators, or analytical methods.

T-test - a statistical test for comparing the means of a particular variable between two groups (e.g. the control and intervention groups in a study).

Type I error - exists when a null hypothesis is rejected when really it is true.

Type II error - exists when a null hypothesis is accepted when really it is false.


Unstructured interview - an interview in which a researcher asks participants very general questions, enabling them to shape the interview in whichever way they see fit, without a predetermined plan to the flow of the conversation. They will not usually have developed specific questions to ask, but will wait for the interviewee to talk, which will then shape the flow of data collection.


Validity - refers to the rigour of a study. If a study is valid, it has been carried out in a manner that ensures that results are unbiased, i.e. giving a true picture of the effectiveness of an intervention. Internal Validity relates to how far the study has established whether a variable or condition under scrutiny has had an effect. If you can control against extraneous variables, internal validity can said to have been achieved. External Validity relates to whether findings from a specific sample involved in a study can be generalised to a larger, target population.

Variables - a principle factor of experimental studies is that one element is manipulated on purpose by the researcher to see whether it has any impact upon another. The element or factor that is being manipulated by researchers is known as the independent variable, whereas the change (or outcome) resulting from the implementation of the independent variable is the dependent variable.


Other glossaries that you might like to look at include:

Project GOLD (a distance learning project run by the Royal College of Nursing).

MORI (the market research company).

An e-learning course in data analysis at the University of the West of England.

Behind the news: A guide to the scientific terms in the media.