Survey Methodologies & Question Types
Survey methodologies and question types cover the strategies and specific question formats used to design reliable, inclusive, and actionable questionnaires.
This category explains terms such as cross‑sectional and longitudinal designs, sampling methods, Likert scales, open‑ended questions, response bias and routing logic — helping survey creators choose the right approach and wording for different audiences and goals.
Adaptive questioning
Adaptive questioning (also called branching or skip logic) displays or hides questions based on a respondent’s previous answers so each participant sees only relevant questions. It makes surveys shorter, more personal and easier to complete.
Cluster sampling
Cluster sampling is a method of selecting a study sample by grouping the population into clusters (like schools, neighborhoods or clinics) and then randomly choosing some whole clusters to survey. It’s often used when a complete list of individuals isn’t available or when surveying individuals directly would be costly or impractical.
Cognitive interviewing
Cognitive interviewing is a qualitative testing technique used to check how people understand, interpret and answer survey questions. It reveals hidden misunderstandings, ambiguous wording and cultural or language problems before a survey is launched.
Cohort study
A cohort study follows a group of people who share a defining characteristic over time to observe how different exposures or events relate to later outcomes. It's an observational way to track change and identify associations without assigning interventions.
Cross-sectional survey
A cross-sectional survey collects data from a group of people at a single point in time to measure the prevalence of opinions, behaviours or characteristics. It gives a snapshot of a population right now but does not track change or prove cause and effect.
Design effect
Design effect measures how much a survey's sampling design (clustering, weighting, stratification or unequal selection probabilities) increases the variance of estimates compared with a simple random sample of the same size. A design effect greater than 1 means your effective sample size and precision are reduced.
Experience sampling method (ESM)
The experience sampling method (ESM) is a research approach that prompts people to report their thoughts, feelings, behaviours or surroundings in real time or close to it. It collects short, repeated responses from participants while they go about their daily lives to reduce recall bias and capture context-rich data.
Interviewer-administered survey
An interviewer-administered survey is one where a trained person asks questions aloud and records the respondent’s answers. It is commonly used face-to-face or by phone to reach people who may not complete a written or online form on their own.
Item nonresponse
Item nonresponse is when a survey respondent skips or leaves blank a specific question (an item) while completing the rest of the survey. It reduces the amount of usable data for that question and can bias results if skipping is systematic.
Item response theory (IRT)
Item Response Theory (IRT) is a family of statistical models that describe how individual survey or test items relate to an underlying trait (like ability, attitude or satisfaction). IRT models estimate item properties (difficulty, discrimination, sometimes guessing) and place respondents on a common scale based on their pattern of answers.
Likert scale
A Likert scale is a common survey format that asks respondents to rate their level of agreement, satisfaction or frequency on an ordered set of response options (for example: Strongly disagree to Strongly agree). It measures the direction and intensity of attitudes or opinions.
Longitudinal survey
A longitudinal survey collects responses from the same population (or repeated samples from the same population) at multiple points in time to measure change, trends or the impact of interventions.
Margin of error
Margin of error is a number that summarises the sampling uncertainty in a survey result — it shows how much the reported percentage (or average) might differ from the true value in the whole population. It is usually reported alongside a confidence level (commonly 95%).
Matrix question
A matrix question (also called a grid question) presents a set of related items or statements in rows with a common response scale in columns, letting respondents answer the same question type for multiple items at once. It’s commonly used for Likert-style ratings (e.g., strongly agree → strongly disagree).
Missing data imputation
Missing data imputation is the process of filling in blank or incomplete answers in a dataset using estimated values so analyses can proceed without dropping incomplete records. It replaces missing responses with plausible substitutes based on rules or statistical models.
Mixed-mode survey
A mixed-mode survey collects the same questionnaire using two or more methods (for example: online, phone, paper or face-to-face) so people can respond in the way that suits them. It’s used to increase reach and representativeness when a single mode would miss important groups.
Mode effects
Mode effects are systematic differences in survey responses caused by the way a survey is delivered (web, phone, face-to-face, paper, mobile app, etc.). Different modes can change how people understand questions, what they choose to disclose, and how they use language.
Non-probability sampling
Non-probability sampling selects participants without giving every member of the target population a known chance of selection. It's common in quick, low-cost or targeted surveys but does not support statistical generalisation to a whole population.
Order effects
Order effects are biases in survey responses caused by the position of questions or answer options — items shown earlier or later can get systematically different answers. They change how respondents interpret and choose answers, reducing the survey's accuracy.
Panel survey
A panel survey gathers data from the same group of people (a panel) at multiple points in time. It’s used to track changes in attitudes, behaviour or outcomes within that defined group.
Pilot testing
Pilot testing is a small-scale trial of a survey or form to check questions, translations, flow and technical behaviour before the full launch. It helps you find and fix problems that would reduce response quality or participation.
Post-stratification
Post-stratification is a weighting technique applied after data collection to make survey results better reflect a known population. It adjusts responses so groups that were under- or over-represented in your sample count proportionally in the final estimates.
Probability sampling
Probability sampling is a way of selecting survey participants so every member of the target population has a known, non-zero chance of being chosen. Methods such as simple random, stratified, cluster and systematic sampling allow you to make statistically valid estimates about the whole population.
Question wording effects
Question wording effects are changes in survey answers caused by the specific words, phrases or structure of a question. Small wording differences — or imperfect translations — can produce systematically different responses.
Quota sampling
Quota sampling is a non‑random sampling method that sets targets (quotas) for subgroups you want in your survey so the final sample matches those proportions. It helps ensure voices from key groups — for example language, age or neighbourhood — are included even when you can't run a full probability sample.
Raking (iterative proportional fitting)
Raking (iterative proportional fitting) is a survey-weighting method that adjusts sample weights so the distribution of key characteristics (margins) in your respondents matches known population totals. It does this by repeatedly reweighting the sample across each characteristic until the margins align.
Respondent-driven sampling
Respondent-driven sampling (RDS) is a chain-referral survey method used to reach hard-to-reach or networked populations by having participants recruit peers. It combines peer recruitment with statistical weighting to allow limited population estimates from non-random samples.
Response rate
Response rate is the proportion of people who returned usable answers to a survey out of the group invited or exposed to it. It shows how many invited participants engaged with the survey and is usually expressed as a percentage.
Sampling frame
A sampling frame is the list or mechanism that defines who could be invited to take a survey — the set of people you can realistically reach. It determines the population your sample is drawn from and affects how representative your results are.
Self-administered survey
A self-administered survey is one that respondents complete on their own—online, on paper, or via a kiosk—without an interviewer present. It relies on clear questions and user-friendly design so people can read and respond independently.
Semantic differential scale
A semantic differential scale measures people's attitudes by asking them to place a concept between two opposite adjectives (e.g., friendly — unfriendly) on a numeric scale. It captures the connotative meaning or feeling people associate with an object, service or idea.
Simple random sampling
Simple random sampling is a basic method of selecting a subset of people from a larger population where every individual has an equal chance of being chosen. It’s used to produce unbiased, generalisable estimates when you have a complete list of the population.
Skip logic
Skip logic (also called conditional branching) directs respondents through different questions or pages based on earlier answers so each person only sees what’s relevant. It tailors the survey path to reduce burden and improve data quality.
Snowball sampling
Snowball sampling is a chain‑referral recruitment method where existing study participants invite people they know, so the sample grows like a rolling snowball. It’s useful for finding hard‑to‑reach or networked groups but is not a statistically representative sample.
Social desirability bias
Social desirability bias is the tendency for respondents to answer questions in a way they think will be viewed favorably by others, rather than reporting their true thoughts or behaviour. It often affects sensitive topics or situations where people feel judged.
Stratified sampling
Stratified sampling is a survey method that divides a population into distinct subgroups (strata) and draws random samples from each so all key groups are represented. It improves precision and ensures minority or important subgroups aren't missed.
Survey weighting
Survey weighting adjusts responses so a survey sample better matches the known characteristics of the population you care about. It rebalance results when some groups are over- or under-represented in your collected responses.
Systematic sampling
Systematic sampling is a probability sampling method that selects every k-th item from an ordered list after a random start. It’s a simple way to get a evenly spread sample when you have a complete list of the population.
Unit nonresponse
Unit nonresponse is when a selected person or household does not take part in a survey at all. It is measured as the share of invited sample units who provide zero data (no completed or partial responses).
Visual analogue scale
A visual analogue scale (VAS) is a continuous measurement tool that asks respondents to mark a position along a line between two labelled endpoints to indicate intensity, agreement or feeling. Online VAS are often implemented as sliders returning a numeric value (for example 0–100).
No glossary terms yet