Data collection methods in research: a glossary
The methods and instruments researchers use to gather data, from questionnaires, rating scales and interviews to observation, diaries, digital traces and existing records, and the steps that turn recordings into usable data. Every definition was checked against the sources listed below.
- Acquiescence bias
- Actigraphy
- Administrative data
- Ambiguous question
- Anthropometry
- Application programming interface
- Archival data
- Archival research
- Artefact
- Arts-based research
- Automated transcription
- Back translation
- Behavioural measure
- Big data
- Biomarker
- Bipolar scale
- BRUSO
- Case report form
- Check-all-that-apply question
- Clinical outcome assessment
- Closed question
- Codebook (data documentation)
- Cognitive interviewing
- Computer-assisted interviewing
- Computer-assisted personal interviewing
- Computer-assisted self-interviewing
- Computer-assisted telephone interviewing
- Consensus methods
- Covert observation
- Critical incident technique
- Data archive
- Data cleaning
- Data collection method
- Data dictionary
- Data entry
- Data linkage
- Data management plan
- Delphi technique
- Demographic question
- Denaturalised transcription
- Diary method
- Dichotomous question
- Digital trace data
- Documentary sources
- Don't know option
- Double-barrelled question
- Double-negative question
- Drop-off survey
- Ecological momentary assessment
- Electronic health record
- Elite interview
- Event sampling
- Event-contingent sampling
- Experience sampling method
- Eye tracking
- Face-to-face survey
- FAIR principles
- Field notes
- Fieldwork
- Filter question
- Focus group
- Focus group moderator
- Forced-choice question
- Found data
- Galvanic skin response
- Go-along interview
- Group interview
- Group-administered questionnaire
- Guttman scale
- In-depth interview
- Intelligent verbatim
- Interview guide
- Interview schedule
- Interviewer effect
- Interviewer-administered questionnaire
- Key informant
- Leading question
- Life-history interview
- Likert item
- Likert scale
- Loaded question
- Matrix question
- Metadata
- Mixed-mode survey
- Mode effect
- Modified Delphi
- Multiple-choice question
- Narrative interview
- Naturalised transcription
- Naturalistic observation
- Neutral midpoint
- Nominal group technique
- Non-participant observation
- Numeric rating scale
- Observation
- Observation schedule
- Observer effect
- Observer roles
- Online focus group
- Online interview
- Online survey
- Open question
- Oral history
- Overt observation
- Paradata
- Participant observation
- Passive data collection
- Patient registry
- Patient-reported outcome measure
- Photo elicitation
- Physical trace measure
- Physiological measure
- Pilot test (questionnaire)
- Postal survey
- Pretesting
- Primary data
- Probe
- Projective technique
- Prompt
- Psychometric test
- Q sort
- Question order effect
- Questionnaire
- Questionnaire design
- Ranking question
- Rapport
- Rating scale
- Repertory grid
- Research data management
- Research instrument
- Research interview
- Response option
- Response order effect
- Response set
- Reverse-worded item
- Routinely collected data
- Satisficing
- Scale anchor
- Secondary data
- Secondary data analysis
- Selective transcription
- Self-administered questionnaire
- Self-report measure
- Semantic differential
- Semi-structured interview
- Sensitive question
- Social desirability bias
- Social media data
- Straightlining
- Structured interview
- Structured observation
- Survey
- Survey mode
- Telephone interview
- Test battery
- Think-aloud protocol
- Thurstone scale
- Time sampling
- Time-contingent sampling
- Time-use diary
- Transcription
- Trusted research environment
- Unipolar scale
- Unobtrusive measure
- Unstructured interview
- Unstructured observation
- Verbal probing
- Verbal rating scale
- Verbatim transcription
- Video interview
- Vignette
- Visual analogue scale
- Visual methods
- Walking interview
- Wearable sensor
- Web scraping
Acquiescence bias
Also called: acquiescence, acquiescent responding, acquiescent response set, acquiescence response style, yea-saying, agreement bias, yea saying
Acquiescence bias is the tendency of some respondents to agree with statements, or to answer yes, whatever the statements say. It is strongest with agree-disagree items and when an interviewer is present, and less educated and less informed respondents show more of it, so designers offer balanced pairs of statements or mix the direction of items. The opposite habit, answering no regardless of content, is called nay-saying.
Actigraphy
Also called: actigraph, wrist actigraphy
Actigraphy is the continuous recording of a person's movement, usually with a small accelerometer worn on the wrist or body, to estimate rest and activity over days or weeks. Sleep researchers use it to estimate sleep and wake patterns in everyday settings, as a cheaper and less intrusive alternative to overnight laboratory recording of sleep.
Administrative data
Also called: administrative records, admin data, administrative datasets
Administrative data are records created when people use public services, such as schools, hospitals, courts, and tax or benefits offices, and kept by government to run those services rather than for research. They cover everyone who uses a service, not only those who agree to a survey, but researchers must work with variables defined for someone else's purpose.
Ambiguous question
Also called: vague question, unclear question, ambiguous item, ambiguous wording
An ambiguous question is a question that different respondents can reasonably read in different ways, because of a vague word, jargon, an undefined term or an unclear time frame, as when a survey asks about 'the media' without saying which. Their answers cannot then be compared, since people were in effect answering different questions. Pretesting, especially cognitive interviewing, is the usual way to catch such items.
Anthropometry
Also called: anthropometric measurement, anthropometric measurements, anthropometrics, body measurements
Anthropometry is the measurement of the human body, such as height, weight, waist circumference and skinfold thickness, taken with standard instruments and procedures for research or comparison. The word also names the scientific study of how body size and proportions vary with age, sex and other characteristics.
Application programming interface
Also called: API, APIs
An application programming interface, or API, is a mechanism through which one piece of software passes data to another in a structured form, and researchers use the ones platforms publish to download posts, records or metadata. Official APIs are usually well documented and stable, but platforms decide what they release and have restricted access, which pushes some researchers towards web scraping.
Archival data
Also called: archival records, archival sources, archive material, archival materials
Archival data are records that already exist in storage, such as historical documents, organisational files, official statistics or old newspapers, used as evidence in a new study. Because they were produced without the researcher's involvement they avoid reactivity, but the researcher cannot choose what was recorded and must work out who created each record and why.
Archival research
Also called: archival method, archival study, archive research
Archival research is a method that answers a research question from records someone else has already made and stored, such as letters, minutes, court files, newspapers or earlier datasets. It suits questions about the past and about behaviour that cannot be observed directly, and it is unobtrusive, but the records were kept for other purposes and support weaker causal claims than experiments.
Artefact
Also called: artifact, artefacts, artifacts, material artefact
An artefact, in research, is a physical object made or used by people, such as a tool, a building, clothing, a gravestone or a flyer, studied as evidence of the beliefs, values and practices of those who made or used it. Artefacts are a form of unobtrusive data, and in field research they are collected or photographed alongside observation and interviews.
Arts-based research
Also called: arts-based methods, arts-based research methods, ABR, arts-based inquiry
Arts-based research is research that uses an art form, such as drawing, collage, photography, poetry, drama or dance, at some stage of the work, whether to generate data, to interpret it or to communicate findings. Participants may create art that is then discussed, which can bring out experiences hard to put into words. Whether it is a new paradigm or a set of added techniques is still debated.
Automated transcription
Also called: automatic transcription, speech-to-text transcription, speech recognition transcription, machine transcription, AI transcription
Automated transcription is the conversion of recorded speech to text by speech recognition software instead of a person. It is fast and cheap, but accuracy suffers with strong accents, specialist terms, similar-sounding words and poor audio, so every transcript must be checked against the recording, and researchers must make sure the service meets data protection requirements.
Back translation
Also called: back-translation, backward translation
Back translation is a check on a translated questionnaire in which an independent translator, who has not seen the original, translates the new version back into the source language so the two can be compared. Differences expose words or items whose meaning shifted. It normally follows forward translation by at least two translators and comes before review by an expert committee and pilot testing of the translated version.
Behavioural measure
Also called: behavioral measure, behavioural measures, behavioural measurement
A behavioural measure is a measure based on what participants actually do, observed and recorded by the researcher, rather than on what they report about themselves, such as counting aggressive acts during play or timing how long someone persists at a task. When self-report, behavioural and physiological measures of the same construct agree, this is called converging operations.
Big data
Also called: big data sources
Big data is a loose label for very large collections of digital records, such as search logs, social media posts, phone and payment records, sensor readings and government administrative files, usually created for purposes other than research. There is no agreed definition: a common one lists volume, variety and velocity, while social researchers stress that the data were produced by companies and governments for their own ends.
Biomarker
Also called: biological marker, biomarkers
A biomarker is a defined biological characteristic, measured objectively, that indicates a normal body process, a disease process or a response to an exposure or treatment, such as blood pressure, a blood test result or a scan finding. It does not tell you how a person feels, manages daily life or how long they live, which is why trials pair biomarkers with clinical outcome assessments.
Bipolar scale
Also called: bipolar rating scale, bipolar response scale
A bipolar scale is a rating scale whose two ends are opposites, such as strongly disagree to strongly agree, usually with a neutral point in the middle. It differs from a unipolar scale, which runs from none of a quality to a great deal of it. One widely used methods textbook suggests seven response options for a bipolar scale and five for a unipolar one.
BRUSO
Also called: BRUSO model, BRUSO principles
BRUSO is a checklist for writing questionnaire items, which holds that each item should be brief, relevant, unambiguous, specific and objective. In practice that means short, plain wording, no questions that serve no purpose in the study, only one possible reading, a clear referent and time frame, and nothing in the wording that hints at the answer the researcher expects.
Case report form
Also called: CRF, eCRF, electronic case report form
A case report form is the paper or electronic form on which a clinical trial records the information its protocol requires about each participant, for reporting to the sponsor. Current good clinical practice guidance treats it as one kind of data acquisition tool, alongside data that arrive directly from devices, wearables, laboratory systems or electronic health records.
Check-all-that-apply question
Also called: select all that apply, select-all-that-apply question, tick all that apply, multiple-response question, checklist question, CATA question
A check-all-that-apply question lists several options and lets respondents tick as many as apply, so each option has to be analysed as a separate yes or no variable. Pew Research Center found that asking a forced-choice yes or no question about each option instead gives more accurate answers, particularly on sensitive topics, and it generally avoids the check-all format for that reason.
Clinical outcome assessment
Also called: COA, clinical outcome assessments
A clinical outcome assessment is a measure of a patient's health based on a report by the patient, a clinician or another observer, or on the patient's performance of a standard task. The four kinds are patient-reported, clinician-reported and observer-reported outcomes, and performance outcomes such as a timed walk or a memory test. Biomarkers, by contrast, measure biological characteristics.
Closed question
Also called: closed-ended question, close-ended question, closed-ended item, fixed-response question, structured question
A closed question is a question that gives respondents a fixed set of answers to choose from, such as yes or no, a list of categories or a rating scale. Answers are quick to give and easy to count, but the options shape the results: in one Pew Research Center comparison far more people chose the economy from a list than volunteered it unprompted.
Codebook (data documentation)
Also called: data codebook, survey codebook, dataset codebook
A codebook, in quantitative research, is a document that describes every variable in a dataset: its name, the exact question wording, the codes used for each answer and the construct it is meant to measure. It lets other researchers, and the original team later, use the data correctly. In qualitative research the same word means the list of codes used in analysis.
Cognitive interviewing
Also called: cognitive interview, cognitive pretesting, cognitive testing
Cognitive interviewing is a way of pretesting survey questions in which a small number of volunteers answer draft items while the interviewer explores how they understood each question, recalled the information and chose an answer, through thinking aloud, verbal probing or both. It exposes wording problems before fieldwork. It is unrelated to the cognitive interview used with eyewitnesses in police investigations.
Computer-assisted interviewing
Also called: CAI, computer assisted interviewing, computer-assisted data collection
Computer-assisted interviewing is the family of survey methods in which a computer program presents the questions and records the answers, whether an interviewer is present, in person or by telephone, or respondents complete the survey themselves on a device or the internet. It includes computer-assisted personal, telephone and self-interviewing, which are defined separately in this glossary.
Computer-assisted personal interviewing
Also called: CAPI, computer assisted personal interviewing, computer-aided personal interviewing
Computer-assisted personal interviewing, or CAPI, is face-to-face survey interviewing in which the interviewer reads the questions from a laptop or tablet and enters the answers into it as they are given. Surveys such as the UK Household Longitudinal Study combine it with computer-assisted self-interviewing, in which respondents answer some sections on the same kind of device themselves, out of the interviewer's sight.
Computer-assisted self-interviewing
Also called: CASI, computer assisted self-interviewing, computer-assisted self-administered interviewing
Computer-assisted self-interviewing, or CASI, is survey data collection in which respondents read and answer questions on a computer or tablet themselves, often during an interviewer's visit, without the interviewer seeing their answers. Taking the interviewer out of the exchange reduces socially desirable answering, so it is used for sensitive sections. In the audio version, ACASI, the device also reads each question aloud.
Computer-assisted telephone interviewing
Also called: CATI, computer assisted telephone interviewing
Computer-assisted telephone interviewing, or CATI, is telephone survey interviewing in which the interviewer reads each question from a screen and types the answers straight into the computer. The system also logs every attempt to reach each sampled number, with its date and time, and these call records are among the most widely used kinds of paradata.
Consensus methods
Also called: consensus development methods, formal consensus methods, consensus techniques
Consensus methods are structured ways of drawing together the judgement of a panel of experts when published evidence is lacking or contradictory, so that a group can agree on guidance, priorities or definitions. The best known are the Delphi technique and the nominal group technique, both of which control how members interact so that no single voice dominates.
Covert observation
Also called: disguised observation, undisclosed observation, concealed observation, hidden observation
Covert observation is observation in which the people being studied do not know that they are being observed, or that the observer is a researcher. It avoids reactivity and can reach groups that would refuse access, but it involves deception and proceeds without informed consent, so it raises serious ethical questions and needs strong justification. Its opposite is overt observation.
Critical incident technique
Also called: CIT, critical-incident technique, critical incident method
The critical incident technique is a method of collecting accounts of specific events in which someone's behaviour or a feature of a system led to an unusually good or poor outcome, gathered from observers, interviews or written reports. The incidents are then classified to find the factors that make the difference. It is widely used in work settings, for example in accident prevention.
Data archive
Also called: research data archive, data repository, social science data archive
A data archive is an organisation or service that preserves research datasets with their documentation and makes them available for other researchers to reuse, often under conditions that protect participants. National archives hold major surveys and collections of qualitative interviews, which makes secondary analysis possible without collecting new data.
Data cleaning
Also called: data cleansing, cleaning data
Data cleaning is the process of finding, diagnosing and correcting errors and oddities in a dataset, such as impossible values, duplicates, inconsistent answers and unexpected gaps, before analysis. It runs in repeated cycles of screening, diagnosis and editing, every change should be recorded so the analysis can be traced back to the raw data, and it is best planned when the study is designed.
Data collection method
Also called: data collection methods, data collection technique, method of data collection, data gathering
A data collection method is the procedure a study uses to gather its evidence, such as a questionnaire, an interview, observation, a test, a diary, physiological recording or the use of existing records. The choice follows from the research question and design, and many studies combine several methods so that the weaknesses of one are offset by the strengths of another.
Data dictionary
Also called: data dictionaries
A data dictionary is a structured description of the variables in a dataset, giving each one's name, meaning and format. Data providers publish them so that researchers can see what a dataset contains before applying to use it. It covers much the same ground as a codebook, which usually adds the exact wording of each question.
Data entry
Also called: data input
Data entry is the transfer of answers from paper forms, questionnaires or other records into a computer file for analysis. To catch typing errors, careful studies use double entry, in which the data are keyed a second time and a program flags every place where the two versions disagree so the original can be checked.
Data linkage
Also called: record linkage, data matching, dataset linkage
Data linkage is the joining of records about the same people, places or organisations from two or more datasets, for example school records with health records, using a variable the datasets share. Linked data can answer questions no single source can, and in the UK linked administrative data are de-identified before researchers can use them in a trusted research environment.
Data management plan
Also called: DMP, data management and sharing plan, DMSP, research data management plan
A data management plan is a document, written before a study starts and updated as it goes, that sets out what data will be collected, how they will be organised, documented, stored and protected, who is responsible and what it will cost, and how and when the data will be shared or archived. Many research funders require one with a grant application.
Delphi technique
Also called: Delphi method, Delphi study, Delphi survey, Delphi process
The Delphi technique is a consensus method in which a panel of experts answers a series of questionnaires in rounds, receiving an anonymous summary of the group's responses after each round and a chance to revise their views, until agreement or stable answers are reached. Anonymity keeps dominant personalities from steering the group. It can also be run online, as an e-Delphi.
Demographic question
Also called: demographic questions, demographic item, demographics, background question
A demographic question asks about a respondent's background, such as age, gender, education, income or ethnicity, so that answers can be described and compared across groups. Questionnaires usually place them at the end, since they are easy to answer but dull, and should include only those the analysis needs. Age and income are often asked in bands because some people will not give exact figures.
Denaturalised transcription
Also called: denaturalized transcription, denaturalism in transcription
Denaturalised transcription is, like its partner term, used in opposite senses. Bucholtz used it for a transcript that keeps the features of speech, such as ums, repetitions and pauses, and McMullin equates it with full verbatim, while Oliver, Serovich and Mason used it for a tidied transcript in which grammar is corrected and interview noise removed. Readers should check whose definition a study follows.
Diary method
Also called: diary study, diary studies, participant diary, solicited diary, diary survey
The diary method is a way of collecting data in which participants record their own behaviour, experiences or feelings, usually as they happen or at the end of each day, over a set period. Diaries capture events that are easily forgotten by the time of an interview, and a follow-up interview about the entries is often added. Some diaries use set questions and others free text.
Dichotomous question
Also called: yes/no question, yes-no question, binary question, dichotomous item
A dichotomous question is a closed question with only two possible answers, such as yes or no, true or false, or agree or disagree. It is quick to answer and simple to analyse, but the agree-disagree form invites acquiescence, which is why Pew Research Center prefers to offer two balanced statements and ask which comes closer to the respondent's view.
Digital trace data
Also called: digital traces, trace data, digital footprint data, behavioural trace data
Digital trace data are the records that people's everyday activity leaves in digital systems, such as clicks, searches, purchases, calls, app use and social media posts, kept by companies and governments and later reused for research. They record behaviour without asking, so they avoid recall and self-presentation problems, but they were not designed for research and rarely cover a whole population.
Documentary sources
Also called: documents, documentary data, documentary evidence, personal documents
Documentary sources are written, printed or digital documents used as data, such as letters, diaries, minutes, policies, reports, newspapers and websites. Researchers study what they say, how and for whom, treating them as products of their context rather than neutral records. Personal documents such as private diaries and letters show how people saw their own lives.
Don't know option
Also called: don't know response, don't know category, no opinion option, DK option
A don't know option is a response choice that lets respondents say they have no answer or no opinion instead of picking a substantive one. Leaving it out pushes some people with no real view to choose an answer anyway, a behaviour one textbook calls floating, but offering it lets others avoid the effort of thinking, and frequent don't know answers are counted as a sign of satisficing.
Double-barrelled question
Also called: double-barreled question, double barrelled question, double barreled question, compound question
A double-barrelled question asks about two separate things but allows only one answer, as in asking how satisfied someone is with the pay and the hours of their job. Anyone who feels differently about the two cannot answer accurately, and the researcher cannot tell which part an answer refers to. Splitting it into two questions fixes it, and the word 'and' is often the giveaway.
Double-negative question
Also called: double negative, double negative question, double-negative item
A double-negative question contains two negatives, so respondents have to untangle the logic before they can answer, as in asking whether people disagree that the council should not raise taxes. People often misread such items and give the opposite of the answer they intend. Survey guides advise wording questions positively and keeping negative terms out of question wording wherever possible.
Drop-off survey
Also called: household drop-off survey, drop-off questionnaire, drop-off and pick-up survey
A drop-off survey is a self-administered questionnaire that a researcher delivers by hand to a home or business, to be completed in private and then posted back or collected later. It combines the personal contact of a visit, which lets people ask about the study and makes them likelier to take part, with the privacy and convenience of filling in the form alone.
Ecological momentary assessment
Also called: EMA, momentary assessment, ecological momentary assessments
Ecological momentary assessment is the repeated collection of data about people's current behaviour, feelings or surroundings, many times over days or weeks, as they go about their normal lives. Short questionnaires are triggered at random or set times or after particular events, usually by smartphone, and may be combined with sensor readings. Asking in the moment reduces recall error.
Electronic health record
Also called: EHR, electronic health records
An electronic health record is a digital record of a patient's health and care, built up over time and including diagnoses, test results, medicines and clinical notes, designed to be shared across the services involved in that person's care. An electronic medical record, in the stricter sense, is a single practice's digital chart. Both are major sources of routinely collected health data.
Elite interview
Also called: elite interviewing, elite interviews, interviewing elites
An elite interview is an interview with someone who has, or once had, a position of power or special access, such as a minister, senior official, judge, executive or party leader, chosen for what they know about decisions and events. Political scientists use such interviews to trace how decisions were made. Who counts as elite is not agreed and depends on the study's context.
Event sampling
Also called: event recording, event sampling observation
Event sampling, in observational research, is a recording strategy in which the observer notes every occurrence of a clearly defined behaviour whenever it happens during the observation period, such as each time a child hits another. It shows how often and in what sequence an event occurs, and it differs from time sampling, which checks for the behaviour only at set intervals.
Event-contingent sampling
Also called: event-contingent recording, event-based sampling, event-contingent assessment, event-based assessment
Event-contingent sampling is a schedule for diary and experience sampling studies in which participants make an entry whenever a defined event happens, such as a social interaction, a panic attack or a cigarette, instead of at set times. It suits events that are rare or of particular interest, but it relies on participants noticing each event and reporting it.
Experience sampling method
Also called: ESM, experience sampling, experience sampling methodology
The experience sampling method is a structured diary technique in which participants report their thoughts, feelings, activities or symptoms several times a day in daily life, usually when prompted by a phone at random or fixed times. It avoids the distortion of recalling experiences later. The same approach is also described as ecological momentary assessment, and the definitions of the two overlap heavily.
Eye tracking
Also called: eye-tracking, eyetracking, eye tracker, gaze tracking
Eye tracking is the recording, with a device called an eye tracker, of where a person looks and for how long, fixation by fixation, while they view a display such as a screen or a page. It is used across psychological research, and survey methodologists have used it to see how respondents read questions and answer options.
Face-to-face survey
Also called: in-person survey, face-to-face interview survey, personal interview survey, in-person interview
A face-to-face survey is a survey in which an interviewer asks the questions in person, usually at the respondent's home, and records the answers on paper or a device. It is harder to refuse than other modes and lets interviewers clarify and probe, but it is one of the costliest and slowest modes, and the interviewer's presence tends to increase socially desirable answers.
FAIR principles
Also called: FAIR data, FAIR, FAIR data principles, FAIR guiding principles
The FAIR principles are guidelines, published in 2016, for making research data findable, accessible, interoperable and reusable, by giving datasets persistent identifiers, rich metadata, standard formats and clear licences. They concern how well people and computers can discover and use data, and they do not require data to be open to everyone, since access can be controlled where necessary.
Field notes
Also called: fieldnotes, field-notes, ethnographic field notes
Field notes are the written record an observer or interviewer makes during fieldwork of what was seen, heard and felt, including descriptions of settings, people, conversations and events and the researcher's own reflections. Brief jottings made on the spot are expanded into full notes as soon as possible afterwards, recording the date, place and people present.
Fieldwork
Also called: field work, field research
Fieldwork is the stage of a study in which researchers go to the settings where people live or work to collect data first-hand, by observing, taking part, interviewing and writing notes, rather than working in a laboratory or from existing records. Anthropological fieldwork has traditionally lasted about a year, although shorter periods of weeks are now also used.
Filter question
Also called: contingency question, screening question, screener question, filter questions
A filter question is a question used to decide which later questions a respondent should answer, such as asking whether someone has ever smoked before asking how much they smoke. People who do not qualify skip the follow-up items, which spares them irrelevant questions. Some textbooks call the same device a contingency question, and guides advise keeping chains of filters short so respondents are not confused.
Focus group
Also called: focus groups, focus group discussion, FGD, focus group interview
A focus group is a planned discussion, led by a moderator, in which a small group of people who share a relevant experience talk about a defined topic, commonly six to ten people though guides differ. Its distinctive data come from the interaction between participants, who question, agree and disagree with one another, which is what separates it from interviewing several people at once.
Focus group moderator
Also called: focus group facilitator, group moderator, discussion moderator
A focus group moderator is the person who runs a focus group: they set ground rules, introduce topics from a guide, encourage everyone to speak, stop dominant members from taking over and steer the talk back when it stalls or drifts, while leaving participants free to respond to one another. Another team member usually takes notes, including on body language and who spoke.
Forced-choice question
Also called: forced choice question, forced-choice item, forced-choice format
A forced-choice question, in survey research, requires an explicit answer to every item, for example yes or no for each option in a list, instead of letting respondents tick only those that apply. Pew Research Center found this format gives more accurate answers than a check-all list, especially for sensitive questions, and Pew generally avoids check-all lists as a result.
Found data
Also called: repurposed data
Found data are data created for a purpose other than research, such as business transactions, platform activity or government records, which researchers take over and repurpose. Salganik cautions that such data are not simply found: someone designed them for their own aims, and understanding those aims is essential before drawing conclusions. The contrast is with data designed by researchers for a study.
Galvanic skin response
Also called: GSR, electrodermal activity, EDA, electrodermal response, skin conductance response
Galvanic skin response is a change in how readily the skin conducts electricity, caused by the activity of sweat glands in the palms and fingers, recorded through electrodes as a physiological sign of arousal. Researchers treat it as an indicator of emotional arousal or stress, although strictly it shows how aroused a person is, not which emotion they feel.
Go-along interview
Also called: go-along, go-along method, go-along interviews
A go-along interview is a form of walking interview in which the researcher accompanies a participant on an outing they would have made anyway, at the usual time and place, asking questions, listening and watching as they go. It mixes interviewing with participant observation and shows how people engage with their surroundings as part of their everyday routines.
Group interview
Also called: group interviewing, group interviews
A group interview is an interview in which one or more interviewers put questions to two or more people at the same time. The term is sometimes used loosely for any group discussion, but focus group methodologists stress that a focus group is designed to generate data from the interaction between participants, whereas a group interview can simply question several people together.
Group-administered questionnaire
Also called: group administered questionnaire, group-administered survey
A group-administered questionnaire is a self-completion questionnaire handed out to respondents gathered in one place, such as a class or a staff meeting, each of whom fills in their own copy. Response rates tend to be high and people can ask about unclear questions on the spot. It differs from a focus group, in which people respond together through discussion instead of individually.
Guttman scale
Also called: Guttman scaling, cumulative scale, cumulative scaling, scalogram, scalogram analysis
A Guttman scale is a set of statements ordered from mild to strong so that anyone who agrees with a stronger statement should also agree with all the milder ones, which means a person's total score shows exactly which items they endorsed. Real answers rarely fit perfectly, so scalogram analysis is used to keep the items that come closest to this cumulative pattern.
In-depth interview
Also called: in depth interview, depth interview, IDI, in-depth interviewing
An in-depth interview is a long, one-to-one qualitative interview designed to explore a person's experiences, views and feelings in detail, using open questions and follow-up probes that let the participant answer in their own words. It is usually semi-structured or unstructured, and the term describes the aim of depth rather than a fixed format.
Intelligent verbatim
Also called: intelligent verbatim transcription
Intelligent verbatim is a style of transcription that keeps every meaningful word a speaker says but drops fillers, false starts, repetitions and slips, adapting speech to the norms of written text so that it reads easily. It suits analyses of what was said more than how it was said. Some writers, following Bucholtz, call this naturalised transcription.
Interview guide
Also called: topic guide, interview protocol, interview topic guide, question guide
An interview guide is the list of topics and questions, with possible follow-up probes, that an interviewer plans to cover in a semi-structured or unstructured interview. It works as a flexible checklist rather than a script, so questions can be reworded, reordered or added as the conversation develops, and it is often revised between interviews as the analysis progresses.
Interview schedule
Also called: interview script, structured interview schedule
An interview schedule is the set of questions, in fixed wording and order, that an interviewer reads out in a structured interview or an interviewer-administered survey, so that every respondent hears the same thing and interviewer effects are kept small. It differs from an interview guide, which lists topics and questions for a flexible semi-structured interview.
Interviewer effect
Also called: interviewer effects, interviewer influence
An interviewer effect is any influence that an interviewer's characteristics or behaviour, such as age, sex, experience, manner or reactions, has on the answers a respondent gives. People may, for instance, talk more openly about sensitive topics to an interviewer of the same sex. Interviewer bias is the narrower case in which the interviewer's own expectations or prejudices shape the interview or its interpretation.
Interviewer-administered questionnaire
Also called: interviewer-administered survey, interviewer administered questionnaire, interview survey
An interviewer-administered questionnaire is one in which an interviewer reads the questions aloud, in person or by telephone, and records the respondent's answers, following an interview schedule. Interviewers can clarify questions and keep people engaged, but they add cost, and their presence tends to increase acquiescence and socially desirable answers compared with self-completion on paper or online.
Key informant
Also called: key informants, key informant interview, key informant interviews, KII
A key informant is a person chosen because their role or experience gives them first-hand knowledge of a community, organisation or issue, such as a community leader, a professional or a long-standing resident. Key informant interviews are in-depth interviews with a small number of such people. In ethnography a key informant is also someone who helps the researcher gain access to and understand the setting.
Leading question
Also called: leading questions, leading survey question
A leading question is a question worded so that it suggests the answer the researcher expects or prefers, for example by opening with 'Don't you agree that' or by asking only about the benefits of a policy. It pushes respondents towards one answer and so distorts the results. Neutral wording, balanced response options and asking about both sides of an issue avoid it.
Life-history interview
Also called: life history interview, life-history method, life history research
A life-history interview is an interview in which a person recounts their life as a whole, or a long stretch of it, so that the researcher can see how experiences, choices and wider events connect over time. Oral historians use it alongside interviews on particular events, and in therapy a similar account of a client's development from birth is also called a life history.
Likert item
Also called: Likert-type item, Likert-type question, Likert question, Likert scale item
A Likert item is a single statement to which respondents give their level of agreement on an ordered set of options, typically five running from strongly disagree to strongly agree, with or without a neutral middle point. Strictly, a Likert scale is the sum of many such items, so one item on its own is better described as a Likert-type item or a rating question.
Likert scale
Also called: Likert scaling, summated rating scale, Likert summated rating procedure, Lickert scale
A Likert scale is an attitude measure made of several statements, some favourable to the topic and some unfavourable, each answered on the same agreement scale, with the answers added together into a single attitude score. It is named after the psychologist Rensis Likert. The name is widely applied to any agree-disagree question or to single items, which methodologists regard as loose usage.
Loaded question
Also called: loaded questions, loaded wording
A loaded question is a question whose wording carries emotionally charged or value-laden language, or builds in an assumption, that tilts respondents towards one answer. In Pew Research Center surveys, asking about welfare instead of assistance to the poor produced different levels of support for the same policy. The term overlaps with leading question, and survey guides often treat the two together as biased wording.
Matrix question
Also called: matrix questions, grid question, question grid, matrix item, matrix table
A matrix question sets out several items that share the same answer options as rows of a grid, with the options as columns, so respondents can rate many things quickly and the layout saves space. The risk is that people tick the same column all the way down without reading each item, a pattern known as straightlining.
Metadata
Also called: meta-data, data about data
Metadata are information that describes a dataset or data item, such as who collected it, how, when and where, what it covers, its format and its licence, as distinct from the data themselves. Good metadata let others find, understand, cite and reuse data, and they are central to the FAIR principles and to depositing data in an archive.
Mixed-mode survey
Also called: mixed mode survey, multimode survey, multi-mode survey, mixed-mode design
A mixed-mode survey is a survey that collects responses in more than one mode, such as online for most people with telephone or paper for those who cannot or will not respond online, or that changes mode between waves. It can reduce costs and reach more people, but answers may differ by mode, so mode effects must be checked before results are combined or trends compared.
Mode effect
Also called: mode effects, survey mode effect, mode of interview effect, mode of administration effect
A mode effect is a difference in answers to the same survey question that is caused by the mode in which it was asked, such as by an interviewer over the telephone rather than on the web. In a 2015 Pew Research Center experiment, telephone respondents gave more favourable, socially desirable answers on some questions, while many other questions showed no difference at all.
Modified Delphi
Also called: modified Delphi technique, modified Delphi method, modified Delphi study, modified-Delphi
A modified Delphi is a consensus study that keeps the Delphi technique's anonymous rounds with feedback but changes parts of the classic procedure, for example by using two rounds instead of the four or more of early designs. There is no standard definition of what counts as modified, and many published studies do not explain their changes, so reports should state what was altered and how consensus was defined.
Multiple-choice question
Also called: multiple choice question, MCQ, multiple-choice item, single-answer question
A multiple-choice question is a closed question that offers three or more answer options and asks respondents to choose one, such as a list of occupations. The options should cover every likely answer without overlapping, with an 'other, please specify' choice to catch the rest. When respondents may choose more than one option it becomes a check-all-that-apply question and is analysed differently.
Narrative interview
Also called: narrative interviewing, narrative interviews
A narrative interview is an unstructured interview that invites the participant to tell the story of an experience, or of part of their life, in their own way, with the interviewer holding back questions until the story is complete. In the form set out by Schütze it moves from an opening invitation, through uninterrupted narration and questions about the story, to a closing conversation.
Naturalised transcription
Also called: naturalized transcription, naturalism in transcription
Naturalised transcription is a term with two opposite meanings in the methods literature. For Bucholtz and the writers who follow her, it is a transcript adapted to the conventions of written text, close to intelligent verbatim. For Oliver, Serovich and Mason it is a transcript that keeps every utterance, pause and stutter in full detail. A methods section should say which sense it uses.
Naturalistic observation
Also called: natural observation, field observation, naturalistic observational method
Naturalistic observation is the observation of behaviour in the setting where it normally happens, such as a playground, a ward or a street, without the researcher manipulating anything. It shows behaviour as it really occurs, but the observer cannot control events, and people who know they are watched may act differently. It contrasts with structured observation in a controlled setting.
Neutral midpoint
Also called: midpoint, middle option, neutral option, neither agree nor disagree, middle category, scale midpoint
A neutral midpoint is the middle option on a rating scale, such as 'neither agree nor disagree', for respondents whose view lies between the two sides. Including it gives genuinely neutral people an honest answer, but some who do hold a view choose it to avoid committing themselves, sometimes called fence-sitting, and heavy use of the midpoint is treated as a sign of satisficing.
Nominal group technique
Also called: NGT, nominal group process, nominal group method
The nominal group technique is a structured meeting for generating and ranking ideas: members first note their ideas privately, the ideas are then listed for the whole group to discuss and clarify, and members vote on them anonymously, often in more than one round. It gives each member an equal say and reduces pressure to conform, and it is common in health services research.
Non-participant observation
Also called: nonparticipant observation, non-participatory observation, detached observation
Non-participant observation is observation in which the researcher watches and records a setting without joining in its activities, whether openly or unnoticed, for example sitting at the back of meetings. It limits the observer's influence on events more than participant observation does, but gives less of an insider's understanding. Gold's complete observer role is its fullest form.
Numeric rating scale
Also called: numerical rating scale, NRS, 0 to 10 rating scale, 0-10 scale, 11-point numeric rating scale
A numeric rating scale is a rating scale on which respondents choose a number, typically from 0 to 10, where 0 means none of the quality and the top number the most imaginable, as with pain from no pain to the worst pain possible. Unlike a visual analogue scale it allows only whole-number answers, eleven possible scores on a 0 to 10 scale.
Observation
Also called: observation method, observational method, observational methods, direct observation
Observation, as a data collection method, is the systematic watching and recording of behaviour, events and settings as they happen, rather than asking people about them. It ranges from structured counting of predefined behaviours to open-ended participant observation, and it can be overt or covert. Its strength is that it records what people do, not what they say they do.
Observation schedule
Also called: observation checklist, observation protocol, observation sheet, code sheet, behavioural coding scheme
An observation schedule is the instrument used in structured observation, listing the behaviours or events to be recorded, with a precise definition of each, and providing a grid for noting when and how often they occur. Precise definitions let different observers record the same scene in the same way, which is checked by comparing their records.
Observer effect
Also called: observer effects, observation effect, observer reactivity
An observer effect is a change in how people behave because they know an observer is present and watching them, as when pupils behave better while a visitor sits in on a lesson. It is the form of reactivity that matters most in observational research, and the Hawthorne effect is a well-known name for it. It differs from observer bias, which lies in the observer's own expectations.
Observer roles
Also called: Gold's typology, Gold's observer roles, observer stances, participant observation roles
Observer roles are the positions a field researcher can take, set out by Gold in 1958 as four points on a scale: the complete participant, who joins a group and hides the research, the participant as observer, a known member who studies the group, the observer as participant, a known outsider who joins in to observe better, and the complete observer, who watches without the group's awareness.
Online focus group
Also called: virtual focus group, online focus groups, internet focus group, e-focus group
An online focus group is a focus group held over the internet. In the synchronous form participants meet at the same time by video or text chat, while in the asynchronous form they post to a moderated discussion board over days or weeks, answering when it suits them. It reaches people kept away by distance, disability or caring duties, though some studies report weaker interaction.
Online interview
Also called: online interviewing, e-interview, internet interview
An online interview is a research interview conducted over the internet, either in real time by video call or text chat, or spread over days by email or messaging, which gives participants time to reflect before replying. It saves travel and reaches people who are far away, but it depends on participants having the right equipment and being comfortable using it.
Online survey
Also called: web survey, internet survey, online questionnaire, web-based survey, web-based questionnaire, e-survey
An online survey is a self-administered questionnaire that respondents complete on a website or app, often reached through a link sent by email or text message. It is cheap, quick and private, and removing the interviewer reduces socially desirable answering, but it misses people without internet access or the skills to use it, and nobody is present to explain a confusing question.
Open question
Also called: open-ended question, open-ended item, free-text question, open response question, open-ended questions
An open question is a question that lets respondents answer in their own words instead of choosing from a list, such as asking what the most important problem facing the country is. It captures answers the researcher did not anticipate and is useful early in a project for building later answer lists, but answers take longer to give and must be coded before they can be counted.
Oral history
Also called: oral history interview, oral histories, oral history method
Oral history is both a method of gathering people's spoken memories of the past through recorded interviews and the body of recordings it produces, which may be archived for others to use with the permission of narrator and interviewer. Interviews may cover a narrator's whole life or a particular event, and the narrator shares control of what is told with the interviewer.
Overt observation
Also called: open observation, disclosed observation, undisguised observation
Overt observation is observation in which the people being studied know that a researcher is present and what the research is about. It allows informed consent and lets the researcher ask questions openly, but people may change their behaviour because they know they are being watched. It is the opposite of covert observation and is generally the more ethically straightforward choice.
Paradata
Also called: survey paradata
Paradata are data about the process of collecting survey data rather than the answers themselves, such as the number and timing of call attempts, interview length, keystrokes and the time spent on each question. They are often recorded automatically as a by-product of computer-assisted data collection and are used to monitor fieldwork, study measurement error, evaluate interviewers and adjust for non-response.
Participant observation
Also called: participant-observation, participant observer, participative observation
Participant observation is a method in which the researcher joins the life of a group, taking part in its activities over a long period while observing, talking with members and keeping field notes, to understand the group from the inside. The researcher's role may be open or hidden. It is the core method of ethnography, and its risk is that involvement colours what the researcher sees.
Passive data collection
Also called: passive sensing, passive measurement, mobile sensing, passive data
Passive data collection is the automatic recording of data about people's behaviour or surroundings by phones, wearables, websites or other devices, without participants having to answer questions each time. Examples include location tracking, step counts and records of calls and messages. It reduces the burden on participants but raises particular questions of consent and privacy.
Patient registry
Also called: disease registry, clinical registry, registry data, outcomes registry, clinical data registry
A patient registry is a standing collection of standardised information about people who share a disease, condition or exposure, gathered without intervening in their care, so that their outcomes can be tracked and compared for scientific, clinical or policy purposes. Disease, clinical and outcomes registries are all kinds, and their records are a leading source of routinely collected health data.
Patient-reported outcome measure
Also called: PROM, PROMs, patient-reported outcome, PRO measure
A patient-reported outcome measure is a questionnaire or interview that records a patient's own report of their health, symptoms, functioning or quality of life, without interpretation by a clinician or anyone else. Symptoms only the patient can feel, such as pain or fatigue, can be measured in no other way. It is one of the four kinds of clinical outcome assessment.
Photo elicitation
Also called: photo-elicitation, photo elicitation interview, photo-elicitation interview, PEI
Photo elicitation is an interviewing technique in which photographs, often taken by the participants themselves, are the starting point for conversation, so that the images draw out memories, meanings and details that questions alone might not reach. When participants photograph their own lives and then discuss the pictures in groups with the aim of reaching decision-makers, the approach is called photovoice.
Physical trace measure
Also called: physical traces, trace measures, physical trace evidence
A physical trace measure is evidence of behaviour inferred from what people leave behind or wear away, such as paths worn across a lawn, rubbish in bins, or how quickly floor tiles in front of each museum exhibit need replacing. It is an unobtrusive measure, since nobody is asked or watched, but it can be hard to know who left the trace and why.
Physiological measure
Also called: physiological measurement, physiological measures, psychophysiological measure, physiological data
A physiological measure is a record of a bodily process, such as heart rate, blood pressure, skin conductance, hormone levels or brain activity, used as data in a study. It does not depend on what participants choose to report, which makes it useful alongside self-report and behavioural measures, but what a given reading means psychologically still has to be established.
Pilot test (questionnaire)
Also called: pilot testing, piloting a questionnaire, questionnaire piloting, survey pilot
A questionnaire pilot test is a trial run of a draft questionnaire with a small group like the intended respondents, to find confusing items, check timing and routing, and see how answers spread before the main data collection. Usage varies: some guides treat piloting and pretesting as the same thing, while others keep pilot test for a larger trial of the final version that also checks reliability.
Postal survey
Also called: mail survey, postal questionnaire, mailed questionnaire, mail questionnaire, mail-out survey
A postal survey is a self-completion questionnaire sent to respondents by post, to be filled in and returned, usually in a stamped addressed envelope. It is relatively cheap and lets people answer privately in their own time, but response rates are often low, so advance letters, reminders and incentives are commonly used, and nobody is on hand to explain unclear questions.
Pretesting
Also called: pre-testing, questionnaire pretesting, survey pretesting, question testing
Pretesting is the testing of draft survey questions before the survey goes into the field, to find items that are misunderstood, hard to answer or badly routed. It can involve cognitive interviews, focus groups and small trial runs of the questionnaire, and problems found lead to rewording and further testing. Its purpose is to check that respondents understand each question as the researcher intends.
Primary data
Also called: primary data collection, first-hand data, original data
Primary data are data that researchers collect themselves for the purposes of their own study, through methods such as surveys, interviews, observation, experiments or tests. They fit the research question closely and the researcher controls how they are gathered, but collecting them costs more time and money than reusing data that already exist.
Probe
Also called: probing, probing question, interview probe, probes
A probe is a follow-up question or cue an interviewer uses to draw out a fuller answer, such as 'Could you tell me more about that?', repeating the participant's words, summarising what they said, or simply staying silent and waiting. Some probes are written into the interview guide and others are improvised. In cognitive interviewing, verbal probes ask how a survey question was understood.
Projective technique
Also called: projective method, projective techniques, projective test
A projective technique is a way of collecting data that presents deliberately ambiguous material, such as inkblots, pictures to describe, sentences to complete or words to associate, on the assumption that people's responses reveal personality, motives or attitudes they might not state directly. Psychologists disagree about its value, from those who think assessment incomplete without it to those who doubt its reliability and validity.
Prompt
Also called: interview prompt, prompts, prompting
A prompt, in interviewing, is a cue that gets a participant talking or keeps them talking, such as an opening invitation to 'tell me about' an experience. Methods texts draw the line between prompts and probes differently, and some use the two words interchangeably, so a report should make clear which follow-up questions were planned in advance and which were improvised.
Psychometric test
Also called: psychological test, psychometric assessment, psychometric instrument
A psychometric test is a standardised instrument, such as a scale, inventory or task, used to measure psychological attributes like ability, aptitude, personality, attitudes or emotional functioning. Its value depends on evidence that it is reliable and valid for the group being tested, and scores are usually interpreted against norms.
Q sort
Also called: Q-sort, Q-sort technique, Q sorting
A Q sort is a data collection task in which a participant arranges a set of statements, usually printed on cards, into ranked piles from most to least characteristic, often with a fixed number allowed in each pile so that the result follows a roughly normal distribution. It is used in personality research and is the data collection step of Q methodology.
Question order effect
Also called: question-order effect, question order effects, item-order effect, item order effect, question context effect
A question order effect is a change in answers caused by the questions asked beforehand rather than by the question itself. An earlier question can set up a comparison that pushes later answers apart, a contrast effect, or pull them into line with it, an assimilation effect. In one classic study, asking about dating before life satisfaction greatly strengthened the link between the two answers.
Questionnaire
Also called: questionnaires, survey questionnaire, survey instrument, survey form
A questionnaire is a written set of questions, with instructions and answer formats, that respondents complete themselves on paper or on screen, or that an interviewer reads out. It is the main instrument of survey research, and its wording, order, layout and response options all affect the answers it produces. A survey is the whole exercise of sampling and asking, the questionnaire the tool used.
Questionnaire design
Also called: questionnaire construction, questionnaire development, survey question design, question design
Questionnaire design is the work of deciding what to ask, how to word each question, which answer options to offer and in what order and layout to present them, so that respondents understand the questions as intended and can answer accurately. It draws on how people interpret a question, recall information, form a judgement and fit it to the options, and it ends in pretesting.
Ranking question
Also called: rank-order question, rank order question, ranking item, rank-ordering question
A ranking question asks respondents to put a set of items in order of preference or importance, for example numbering four candidates from first to fourth choice. It forces people to separate items they might otherwise rate equally, but it shows only order, not how far apart the items are, and some respondents misread the instruction and simply tick a favourite.
Rapport
Also called: building rapport, establishing rapport
Rapport is the sense of connection and trust between a researcher and a participant that makes the participant willing to speak openly. Interviewers build it through respect, active listening and a setting where the participant feels at ease, and field researchers through time spent with a group. It differs from a therapeutic relationship, and researchers need to keep to the limits of the research role.
Rating scale
Also called: rating scales, rating-scale question, rating question
A rating scale is a set of ordered response options on which respondents place a person, object or statement along a dimension such as agreement, frequency, satisfaction or quality, commonly with five or seven points. Scales differ in their number of points, whether every point carries a verbal label, and whether they run between two opposites or from none of a quality to a lot.
Repertory grid
Also called: repertory grid technique, rep grid, Kelly's repertory grid
A repertory grid is a technique, introduced by George Kelly, for uncovering the personal constructs a person uses to make sense of the world. The participant rates a set of important elements, such as people or objects, on a series of dimensions using a numerical scale, and the resulting grid of ratings can be analysed statistically to show how the constructs relate.
Research data management
Also called: RDM, data management
Research data management is the care of research data through a project and after it: organising, documenting, storing and securing data, handling personal information lawfully, and preparing data for sharing or archiving. It is usually set out in a data management plan at the start and reviewed as the study progresses.
Research instrument
Also called: data collection instrument, data collection tool, research tool
A research instrument is the tool used to collect data in a study, such as a questionnaire, interview schedule, observation schedule, test, rating scale or diary form. Using an established instrument allows results to be compared across studies, while a new one has to be developed and tested first. In qualitative research the researcher is often described as the main instrument.
Research interview
Also called: interview, interviews, interviewing
A research interview is a method of collecting data through a conversation in which a researcher asks questions and a participant answers, face to face, by telephone or online. Interviews range from fully structured, with fixed questions and answer options, to unstructured, and they suit topics that need detailed explanation, follow-up questions or an account of a process such as a decision.
Response option
Also called: response options, response category, response categories, answer option, answer choices, response alternative
A response option is one of the answers offered for a closed question. The set of options should be mutually exclusive, so that no answer fits two options, and exhaustive, so that every possible answer has a place, if necessary through an 'other' or 'don't know' choice. The number, order and labelling of the options all influence which answers people give.
Response order effect
Also called: response-order effect, response order effects, primacy and recency effects
A response order effect is a change in which answer people pick caused by the order in which the options are presented. When options are read aloud, as on the telephone, people tend to choose those heard last, a recency effect, and when they read the list themselves they favour those near the top, a primacy effect. Randomising the order counters it, except on ordered scales.
Response set
Also called: response sets, response tendency
A response set is a tendency to answer questions in a systematic way that has nothing to do with their content, such as agreeing with everything, choosing the middle option, or giving socially desirable answers. When the tendency is a lasting trait of the person, seen across situations and over time, psychologists call it a response style. Both distort measures built from self-report.
Reverse-worded item
Also called: reverse-scored item, reverse-coded item, reversed item, negatively worded item, reverse-keyed item, item reversal
A reverse-worded item is a questionnaire item phrased in the opposite direction to the others on the same scale, so that agreeing with it indicates less of the quality being measured. Mixing directions discourages respondents from agreeing with everything, and before scores are added up the answers to these items are reversed so that a high score means the same thing throughout.
Routinely collected data
Also called: routine data, routinely collected health data, routinely-collected health data
Routinely collected data are records gathered in the ordinary running of services, without a research question in mind, and later used for research, such as health administrative databases, electronic health records, disease registries, primary care databases and surveillance returns. Studies based on them are reported using the RECORD guideline, an extension of STROBE, which asks authors to name their data sources and any linkage.
Satisficing
Also called: survey satisficing, satisficing behaviour, satisficing behavior
Satisficing is answering survey questions with less effort than an accurate answer needs, giving a response that seems good enough instead of the best one. Common signs are choosing the middle option, answering don't know and giving the same answer to every item in a grid. In Krosnick's theory of survey satisficing it depends on how able and how motivated respondents are to answer well.
Scale anchor
Also called: anchor, anchors, verbal anchor, anchor label, scale anchors, end anchor
A scale anchor is a word or phrase attached to a point on a rating scale, most often the two ends, such as 'not at all satisfied' and 'completely satisfied', which tells respondents what the points mean. One widely used textbook advises showing respondents verbal labels and converting them to numbers only at analysis, and the labels on a bipolar scale should be balanced around its centre.
Secondary data
Also called: existing data, second-hand data, pre-existing data
Secondary data are data that were collected by someone else, such as another research team, a government agency or an organisation, and are reused in a new study. Census and survey datasets, administrative records and archived interview collections are common examples. Whether data count as primary or secondary depends on who collected them and for what purpose, not on their type.
Secondary data analysis
Also called: secondary analysis, secondary data analyses, SDA, reanalysis of existing data
Secondary data analysis is the analysis of data someone else collected, often for a different purpose, to answer a new question, such as reanalysing a national survey held in a data archive. It saves time and money, gives access to larger samples than one researcher could gather and allows replication, but the analyst cannot change how variables were measured.
Selective transcription
Also called: summary transcription, gist transcription
Selective transcription is transcribing only the parts of a recording that matter to the research question, or noting the key points with a few quotations word for word, instead of typing out everything said. It saves time and can help decide what to transcribe in full later, but it is too thin for detailed analysis, and what was left out should be reported.
Self-administered questionnaire
Also called: self-completion questionnaire, self-administered survey, self-report questionnaire, SAQ, self-completed questionnaire
A self-administered questionnaire is one that respondents read and complete on their own, on paper or online, without an interviewer asking the questions. It is cheaper than interviewing and tends to draw more candid answers on sensitive topics, but nobody is present to clarify questions, and people who find written questions hard to follow may answer less accurately.
Self-report measure
Also called: self-report, self-reported data, self-report data, self-report measures
A self-report measure is any method in which participants describe their own thoughts, feelings, behaviour or circumstances, through questionnaires, interviews, diaries or rating scales. It is the only direct route to inner states that others cannot see, but it depends on people's honesty, self-awareness and memory, so self-reports are often checked against behavioural or physiological measures.
Semantic differential
Also called: semantic differential scale, semantic differential technique, bipolar adjective scale
A semantic differential is a rating scale on which respondents place a concept, such as a product, a person or an idea, between pairs of opposite adjectives, such as bad and good or weak and strong, often on seven points. The pairs tend to reflect three dimensions, evaluation, potency and activity, and ratings on the evaluative pairs can be combined into an attitude score.
Semi-structured interview
Also called: semistructured interview, semi structured interview, semi-structured interviewing, patterned interview
A semi-structured interview is an interview that follows a prepared guide of topics and open questions but lets the interviewer change their order and wording, follow up interesting answers and explore issues the participant raises. It is the most common source of qualitative data in health services research, balancing comparability across participants with room for each person's own account.
Sensitive question
Also called: sensitive questions, sensitive item, threatening question, sensitive survey question
A sensitive question is a survey or interview question about something private, stigmatised, illegal or socially judged, such as income, drug use, sexual behaviour or voting, which people may be reluctant to answer honestly. Placing such questions after easier ones, offering answer bands instead of exact figures and letting people answer privately instead of to an interviewer all reduce refusals and social desirability bias.
Social desirability bias
Also called: social desirability, social desirability effect, socially desirable responding, social desirability response set, social desirability response bias
Social desirability bias is the tendency of respondents to give answers that make them look good, over-reporting approved behaviour such as voting or giving to charity and under-reporting disapproved behaviour such as drug use or prejudice. It is stronger when an interviewer is present than when people answer alone on paper or online, and careful wording and private answering reduce it.
Social media data
Also called: social media research data, social networking site data, platform data
Social media data are the posts, images, profiles, reactions and networks of connections that people create on social media platforms, collected for research through official APIs, web scraping or tools that participants install to share their own data. They offer large volumes of naturally occurring talk and behaviour, but users are not a representative sample, and access, terms of service and ethics all constrain their use.
Straightlining
Also called: straight-lining, straight lining, nondifferentiation, non-differentiation, non-differentiated responding
Straightlining is giving the same answer to every item in a set of questions that share a response scale, such as ticking 'agree' all the way down a grid, whatever the items say. It is treated as a sign of satisficing, because respondents who read and weighed each item would usually vary their answers, and it is usually measured across the items of a grid.
Structured interview
Also called: standardised interview, standardized interview, structured interviewing
A structured interview is an interview in which every respondent is asked the same questions, in the same words and order, usually with fixed answer options, so that answers can be compared and combined across people. Survey interviews and many job selection interviews work this way. It contrasts with semi-structured and unstructured interviews, which leave the interviewer freedom to follow the conversation.
Structured observation
Also called: systematic observation, structured observational method
Structured observation is observation in which the researcher decides in advance which behaviours to record, defines each one precisely and notes how often they occur, often in a controlled setting and using an observation schedule. It produces quantitative data and allows checks of agreement between observers, but it cannot capture anything outside the categories chosen beforehand.
Survey
Also called: surveys, sample survey, questionnaire survey
A survey is the collection of information from a sample of people by asking them the same questions, through self-completed questionnaires or interviews, usually to describe a larger population. The word covers the whole exercise, from sampling and questionnaire design to fieldwork and analysis, whereas a questionnaire is only the instrument used to ask the questions.
Survey mode
Also called: mode of data collection, data collection mode, mode of administration, survey administration mode
A survey mode is the channel through which questions are put and answered, such as face-to-face, telephone, post, web or text message, and whether a person or a computer administers them. The mode affects cost, who can be reached and the answers themselves, since people respond differently to an interviewer than to a screen or a printed form.
Telephone interview
Also called: telephone survey, phone interview, phone survey, telephone interviewing
A telephone interview is an interview conducted by phone, used both for structured surveys and for qualitative research. It is quicker and cheaper than meeting in person, reaches people over a wide area and can suit participants who value privacy, but the interviewer loses visual cues, and one comparison found qualitative telephone interviews shorter than face-to-face ones.
Test battery
Also called: battery of tests, assessment battery, test batteries
A test battery is a set of tests given together as one assessment, so that several aspects of a person or condition can be measured in a single sitting. The tests may cover the same or different areas and may be scored separately or combined into a single total.
Think-aloud protocol
Also called: think aloud, think-aloud, thinking aloud, think-aloud method, think-aloud interview
A think-aloud protocol is a method in which participants say out loud everything that goes through their mind while doing a task, such as answering survey questions or solving a problem, and the recording is analysed afterwards. In cognitive interviewing it shows how people interpret and answer questions, though participants need training, find it tiring and often drift away from the task.
Thurstone scale
Also called: Thurstone scaling, method of equal-appearing intervals, equal-appearing interval scale, Thurstone attitude scale
A Thurstone scale is an attitude scale built by asking a panel of judges to rate many statements for how favourable they are to the topic, commonly on an 11-point scale, and keeping statements the judges agree on that spread evenly across the range. Respondents then simply agree or disagree with each statement, and their score is the average scale value of those they agree with.
Time sampling
Also called: time-sampling, interval recording, time-sample recording
Time sampling is an observation strategy in which the observer records whether a target behaviour occurs during short, set intervals, at fixed or random times, instead of watching continuously, for instance checking what each pupil is doing for a few seconds every five minutes. Scores are based on the number of intervals in which the behaviour appears or its rate, and it differs from event sampling.
Time-contingent sampling
Also called: time-based sampling, time-contingent assessment, time-based assessment
Time-contingent sampling is a schedule for experience sampling and diary studies in which participants report at set times instead of after particular events, either at fixed intervals, such as every two hours or each evening, or when signalled at random moments within blocks of the day so that time is evenly covered. It is contrasted with event-contingent sampling.
Time-use diary
Also called: time diary, time-use diaries, time use diary, time-budget diary
A time-use diary is a diary in which respondents record in sequence what they were doing through the day, so that researchers can measure how people divide their time between paid work, care, leisure and other activities. Time-budget studies of this kind date back to the 1930s and underpin large national and cross-national surveys of how people spend their time.
Transcription
Also called: transcribing, interview transcription, audio transcription
Transcription is the conversion of recorded speech, from interviews, focus groups or meetings, into written text that can be analysed. It is never a neutral copy: the transcriber decides how much detail to keep, from every pause and overlap to a tidied version, and those choices should fit the planned analysis and be reported. An hour of audio can take several hours to transcribe.
Trusted research environment
Also called: TRE, trusted research environments, secure research environment
A trusted research environment is a highly secure computing platform in which sensitive data, such as linked administrative or health records, are held in de-identified form and analysed without leaving the platform. Only accredited researchers working on approved projects can gain access, which lets valuable personal data be used for research while protecting the people in them.
Unipolar scale
Also called: unipolar rating scale, unipolar response scale
A unipolar scale is a rating scale that measures how much of a single quality is present, running from none of it to a great deal, such as from not at all satisfied to completely satisfied, with no opposite quality at the other end. It contrasts with a bipolar scale, whose ends are opposites, and one widely used textbook recommends five response options for it.
Unobtrusive measure
Also called: unobtrusive measures, unobtrusive methods, unobtrusive research, non-reactive measure, nonreactive measure
An unobtrusive measure is a way of collecting data that does not involve or alert the people being studied, so the research cannot change their behaviour, such as analysing records, physical traces or public behaviour watched from a distance. It avoids reactivity but gives the researcher little control over what data exist, and for some questions no unobtrusive measure is available.
Unstructured interview
Also called: non-directive interview, nondirective interview, unstructured interviewing
An unstructured interview is an interview with no fixed list of questions, in which the interviewer starts from a broad topic or opening question and lets the conversation follow the participant's lead, asking questions as they arise. It can produce rich, personal accounts, but answers are hard to compare across participants and much depends on the interviewer's skill.
Unstructured observation
Also called: unsystematic observation, open-ended observation, qualitative observation
Unstructured observation is observation without a predetermined list of behaviours to record, in which the observer notes whatever seems relevant to the research question, usually as detailed field notes, and lets the focus sharpen as the study goes on. Participant observation is usually of this kind, whereas structured observation fixes in advance who, what, when and how to record.
Verbal probing
Also called: verbal probe, verbal probes, cognitive probing
Verbal probing is the main technique of cognitive interviewing, in which the interviewer asks follow-up questions about a survey item once it has been answered, such as what a term meant to the respondent or how they arrived at their answer. Probes can come straight after each question, concurrent probing, or be saved until the end, retrospective probing, which suits self-completion questionnaires.
Verbal rating scale
Also called: VRS, verbal descriptor scale, adjectival rating scale
A verbal rating scale is a rating scale made of a short ordered list of words, such as none, mild, moderate and severe, from which respondents choose the one that best describes their state. Scales of four to six words are common in clinical trials. It is easy to understand and well accepted by respondents, but it offers fewer distinctions than numeric or visual analogue scales.
Verbatim transcription
Also called: verbatim transcript, full verbatim, full verbatim transcription, word-for-word transcription, exact transcription
Verbatim transcription is transcription that records everything said, word for word, including fillers such as um and er, repetitions, false starts and grammatical slips, and often laughter, pauses and interruptions. It suits analyses of how people speak as well as what they say. Because the word verbatim is used loosely, a methods section should state which features were kept.
Video interview
Also called: video-call interview, video call interview, videoconference interview, video interviewing, virtual interview
A video interview is a research interview held over a video call, so interviewer and participant see and hear each other at the same time from different places. It comes closest of the remote modes to meeting in person, but it can suffer from time lags, dropped calls and a head-and-shoulders view that hides body language. One comparison found in-person interviewees said somewhat more, on a similar range of topics.
Vignette
Also called: vignettes, vignette technique, vignette method, vignette question
A vignette is a short story about hypothetical people in a specified situation, presented to participants who are asked how they, or the characters, would respond. Vignettes help researchers explore context, moral judgements and sensitive experiences at one remove, used alone or alongside interviews. The stories need to be plausible and simple, since plots with many twists confuse participants.
Visual analogue scale
Also called: visual analog scale, VAS, unnumbered graphic rating scale
A visual analogue scale is a straight line, often 10 centimetres long, with a label at each end, such as no pain and worst pain imaginable, on which respondents mark a point to show the level of a sensation or feeling. The score is the distance from one end to the mark. Adding numbers or descriptive words along the line turns it into a graphic rating scale.
Visual methods
Also called: visual research methods, visual methodologies, visual methodology, image-based research
Visual methods are research methods that use images, such as photographs, drawings, maps, film or video, as data, as prompts in interviews or as a way of presenting findings. Images may be made by participants, by the researcher or drawn from existing media, and they can surface experiences that are otherwise hard to reach. Photo elicitation and photovoice are common examples.
Walking interview
Also called: walking interviews, walking interview method
A walking interview is a qualitative interview conducted while the researcher and participant walk together through a place, with the route chosen by the researcher, by the participant or left open. Places passed on the way prompt memories and talk, and walking side by side can ease the power imbalance of a sit-down interview, which suits studies of people's relationship with places.
Wearable sensor
Also called: wearable device, wearables, wearable technology
A wearable sensor is a small device worn on the body, such as a wristband, watch or chest strap, that continuously records data such as movement, heart rate or sleep as people go about their lives. Studies pair such passive readings with self-reports, for example combining experience sampling with measures of physical activity and heart rate variability, and current clinical trial guidance names wearables as a recognised source of data.
Web scraping
Also called: web-scraping, scraping, web harvesting, web data extraction
Web scraping is automated data collection from websites or apps, using software that downloads pages and extracts information from them, such as prices, posts, reviews or listings, turning loosely structured content into a dataset. Researchers turn to it when no API provides the data, but scrapers break when sites change, and terms of service, the law and ethics all need careful thought.
Where these definitions were checked
Pew Research Center, Writing Survey Questions
Pew Research Center, From Telephone to the Web: The Challenge of Mode of Interview Effects in Public Opinion Polls (2015)
American Psychological Association, APA Dictionary of Psychology (entries for the survey, interview, observation, measurement and testing terms defined here)
Rajiv Jhangiani, I-Chant Chiang, Carrie Cuttler and Dana Leighton (open textbook), Research Methods in Psychology, 7.3: Constructing Surveys
Rajiv Jhangiani, I-Chant Chiang, Carrie Cuttler and Dana Leighton (open textbook), Research Methods in Psychology, 6.6: Observational Research
Rajiv Jhangiani, I-Chant Chiang, Carrie Cuttler and Dana Leighton (open textbook), Research Methods in Psychology, 4.2: Understanding Psychological Measurement
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 11.3: Types of surveys
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 11.4: Designing effective questions and questionnaires
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 13.1: Interview research, what is it and when should it be used?
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 13.2: Qualitative interview techniques
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 13.3: Issues to consider for all interview types
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 13.4: Focus groups
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 14.1: Unobtrusive research, what is it and when should it be used?
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 14.3: Unobtrusive data collected by you
Matthew DeCarlo (open textbook), Scientific Inquiry in Social Work, 14.4: Secondary data analysis
William M. K. Trochim, Research Methods Knowledge Base: Types of Surveys
William M. K. Trochim, Research Methods Knowledge Base: Types of Survey Questions
William M. K. Trochim, Research Methods Knowledge Base: Response Format
William M. K. Trochim, Research Methods Knowledge Base: Question Wording
William M. K. Trochim, Research Methods Knowledge Base: Question Content
William M. K. Trochim, Research Methods Knowledge Base: Question Placement
William M. K. Trochim, Research Methods Knowledge Base: Likert Scaling
William M. K. Trochim, Research Methods Knowledge Base: Guttman Scaling
William M. K. Trochim, Research Methods Knowledge Base: Thurstone Scaling
William M. K. Trochim, Research Methods Knowledge Base: Unobtrusive Measures
William M. K. Trochim, Research Methods Knowledge Base: Data Preparation
William M. K. Trochim, Research Methods Knowledge Base: Conducting Interviews
Gordon B. Willis, with Rachel A. Caspar and Judith T. Lessler, Research Triangle Institute, Cognitive Interviewing: A "How To" Guide (1999)
Patrick Sturgis and Ian Brunton-Smith, Personality and Survey Satisficing (Public Opinion Quarterly, 2023)
Brady T. West, Paradata in Survey Research (Survey Practice, 2011)
National Research Council, National Academies Press, The 2000 Census: Counting Under Adversity (2004), Glossary
Siny Tsang, Colin F. Royse and Abdullah Sulieman Terkawi, Guidelines for developing, translating, and validating a questionnaire in perioperative and pain medicine (Saudi Journal of Anaesthesia, 2017)
Mathias Haefeli and Achim Elfering, Pain assessment (European Spine Journal, 2006)
Melissa DeJonckheere and Lisa M. Vaughn, Semistructured interviewing in primary care research: a balance of relationship and rigour (Family Medicine and Community Health, 2019)
Sandra Jovchelovitch and Martin W. Bauer, LSE Research Online, Narrative interviewing (2000)
Oral History Association, Oral History: Defined
Penelope Kinney, University of Surrey, Walking Interviews (Social Research Update 67, 2017)
Louise Corti, University of Surrey, Using diaries in social research (Social Research Update 2, 1993)
Christine Barter and Emma Renold, University of Surrey, The Use of Vignettes in Qualitative Research (Social Research Update 25, 1999)
UCLA Center for Health Policy Research, Key Informant Interviews (Health DATA Program training materials)
Ozlem Tuncel, Elite Interviewing in Political Science: A Meta-Analysis of Reporting Practices (Perspectives on Politics, 2026)
Matthew Krouwel, Kate Jolly and Sheila Greenfield, Comparing Skype (video calling) and in-person qualitative interview modes in a study of people with irritable bowel syndrome (BMC Medical Research Methodology, 2019)
Shelagh K. Genuis, Westerly Luth, Garnette Weber, Tania Bubela and Wendy S. Johnston, Asynchronous online focus groups for research with people living with amyotrophic lateral sclerosis and family caregivers (BMC Medical Research Methodology, 2023)
Jeremy Jones and Duncan Hunter, Consensus methods for medical and health services research (BMJ, 1995), abstract
Prashant Nasa, Ravi Jain and Deven Juneja, Delphi methodology in healthcare research: How to decide its appropriateness (World Journal of Methodology, 2021)
Barbara B. Kawulich, Participant Observation as a Data Collection Method (Forum: Qualitative Social Research, 2005)
Jim McCambridge, John Witton and Diana R. Elbourne, Systematic review of the Hawthorne effect: new concepts are needed to study research participation effects (Journal of Clinical Epidemiology, 2014)
Inez Myin-Germeys and colleagues, Experience sampling methodology in mental health research: new insights and technical developments (World Psychiatry, 2018)
Matthew J. Salganik (open edition), Bit by Bit: Social Research in the Digital Age, 2.1: Introduction to observing behaviour
Matthew J. Salganik (open edition), Bit by Bit: Social Research in the Digital Age, 2.2: Big data
Matthew J. Salganik (open edition), Bit by Bit: Social Research in the Digital Age, 3.5: New ways of asking
Matthew J. Salganik (open edition), Bit by Bit: Social Research in the Digital Age, 3.5.1: Ecological momentary assessments
Megan A. Brown, Andrew Gruen, Gabe Maldoff, Solomon Messing, Zeve Sanderson and Michael Zimmer, Web Scraping for Research: Legal, Ethical, Institutional, and Scientific Considerations
Katherine Boydell, Brenda M. Gladstone, Tiziana Volpe, Brooke Allemang and Elaine Stasiulis, The Production and Dissemination of Knowledge: A Scoping Review of Arts-Based Health Research (Forum: Qualitative Social Research, 2012)
Caroline Wang and Mary Ann Burris, Photovoice: concept, methodology, and use for participatory needs assessment (Health Education and Behavior, 1997), abstract
Katharine A. R. Price and colleagues, A Mixed-Method Approach to Explore Successful Recruitment and Treatment of Minority Patients on Therapeutic Cancer Clinical Trials Using Photo-Elicitation Interviews (Health Equity, 2024)
Caitlin McMullin, Transcription and Qualitative Methods: Implications for Third Sector Research (Voluntas, 2023)
Leandro da Silva Nascimento and Fernanda Kalil Steinbruch, "The interviews were transcribed", but how? Reflections on management research (RAUSP Management Journal, 2019)
Irit Mero-Jaffe, 'Is that what I said?' Interview Transcript Approval by Participants (International Journal of Qualitative Methods, 2011)
University of Bath Library, Working with data: Transcription
UK Data Service, Data management planning
UK Data Service, Administrative data
Administrative Data Research UK, What is administrative data?
Administrative Data Research UK, Glossary
GO FAIR, FAIR Principles
Framework for Open and Reproducible Research Training, Codebook (FORRT Glossary)
Jan Van den Broeck, Solveig Argeseanu Cunningham, Roger Eeckels and Kobus Herbst, Data Cleaning: Detecting, Diagnosing, and Editing Data Abnormalities (PLOS Medicine, 2005)
Eric I. Benchimol, Liam Smeeth, Astrid Guttmann, Katie Harron, David Moher and colleagues, The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement (PLOS Medicine, 2015)
Office of the National Coordinator for Health Information Technology, US Department of Health and Human Services, EMR vs EHR: What is the Difference?
Agency for Healthcare Research and Quality, Registries for Evaluating Patient Outcomes: A User's Guide, Patient Registries
FDA-NIH Biomarker Working Group, BEST (Biomarkers, EndpointS, and other Tools) Resource, Glossary
International Council for Harmonisation, published by the European Medicines Agency, ICH E6(R3) Guideline for good clinical practice, Step 5, Glossary