Resources
Survey glossary
This glossary defines 18 terms used in survey design, survey logic, measurement and data protection, in plain British English. Each entry gives a tight definition first, then when to use the thing, the mistakes people usually make with it, and, in most entries, a worked example. Where a term maps onto something in NumoForms, the entry says exactly how — and where it does not, it says that too.
Written for people who have to run a survey next week, not for people studying survey methodology.
All terms
- Anonymous survey
- Branching
- Closed-ended question
- Completion rate
- Conditional logic
- Data Protection Impact Assessment (DPIA)
- Data residency
- Hidden field
- Likert scale
- Matrix question
- Net Promoter Score (NPS)
- Open-ended question
- Piping
- Ranking question
- Response rate
- Sampling bias
- Screener question
- WCAG
Question types
The shapes a question can take, and what each one does to your data.
Closed-ended question
A closed-ended question offers a fixed set of answer options and asks the respondent to choose among them.
Matrix question
A matrix question is a grid in which several statements — the rows — are all answered using the same set of response options, shown as columns.
Open-ended question
An open-ended question lets a respondent answer in their own words instead of choosing from a fixed list.
Ranking question
A ranking question asks a respondent to arrange a list of options into an order of preference rather than choose between them.
Logic & flow
How a survey decides what to show, and in what order.
Branching
Branching is the practice of sending respondents down different routes through a survey depending on how they answer.
Conditional logic
Conditional logic is a rule that shows or hides a survey question based on an answer given earlier in the same survey.
Hidden field
A hidden field is a value stored alongside a response that the respondent never sees and never types.
Piping
Piping inserts an answer a respondent has already given into text shown later in the same survey.
Screener question
A screener question is a qualifying question near the start of a survey whose only job is to establish whether the respondent belongs to the population you are studying.
Measurement
The numbers people quote about surveys, and what they actually mean.
Completion rate
Completion rate is the proportion of people who started a survey and reached the end of it, expressed as a percentage of starts rather than of invitations.
Likert scale
A Likert scale is an ordered, symmetric agreement scale — most often five or seven points running from strong disagreement to strong agreement through a neutral midpoint — used to measure an attitude rather than a fact.
Net Promoter Score (NPS)
Net Promoter Score (NPS) is a loyalty metric taken from a single question — how likely a respondent is to recommend an organisation, on a scale from 0 to 10.
Response rate
Response rate is the number of usable responses divided by the number of people invited to take part, expressed as a percentage.
Methodology
Survey design and sampling ideas that decide whether the results are worth anything.
Privacy & compliance
The obligations that come with collecting answers from people.
Anonymous survey
An anonymous survey is one in which nothing collected can be linked back to an identifiable individual — not by the person running the survey, and not by anyone who later receives the data.
Data Protection Impact Assessment (DPIA)
A Data Protection Impact Assessment (DPIA) is a documented assessment of the data protection risks in a processing activity, carried out before that processing begins.
Data residency
Data residency is the geographic location where data physically sits at rest, and by extension the jurisdiction whose law applies to it most directly.
WCAG
The Web Content Accessibility Guidelines (WCAG) are the World Wide Web Consortium’s standard for making web content usable by disabled people.
How to use this glossary
Most survey vocabulary is borrowed from social research and then bent by whichever tool a team happened to use first. The result is that two people can say "branching" and mean genuinely different mechanisms, or say "response rate" and be quoting numbers with different denominators. Each entry here therefore starts with a definition tight enough to disagree with, and then says where the term is used loosely.
Three conventions run through the set. Where a term has a strict technical meaning and a looser everyday one — Likert scale is the clearest case — both are given, and the difference is explained rather than ignored. Where a definition depends on how a particular tool implements it, the entry says which tool it is describing. And where NumoForms does not do something, the entry says so plainly instead of leaving the impression that it does: see the note on jump targets in branching, or on disqualify actions in screener questions.
Nothing here is a substitute for legal advice. The DPIA, data residency and WCAG entries summarise obligations accurately as far as they go, but the obligation sits with whoever is running the survey, and the details of your case matter.
Definitions become decisions in the builder.
Every term defined here describes something you can put in a survey: a logic rule, a matrix grid, a screener, a hidden field. Building one is the quickest way to find out which of them you actually need.
- All sixteen question types, matrix and ranking included
- Conditional logic written per question, not as a routing map
- Responses stored in London and exported as CSV
