Data projects, publishing and results
| Term | Meaning in VoxDash |
|---|---|
| Banner | A wide brand or header image uploaded to a profile or a data project; 1200x400 is recommended. Not a crosstab banner and not a notification bar. |
| Crosstab (cross-tabulation) | A table that breaks one question down by another. The crosstab is the object you build; cross-tabulation is the technique it applies. |
| Data license | The licence terms under which a published data project may be reused, such as CC-BY-SA or CDLA. The licence names themselves stay in Latin script; in a Creative Commons licence, NC marks the NonCommercial condition. |
| Data owner | The party that owns a data project and may claim or retain ownership of it. A role on the project, distinct from the data provider. |
| Data profile | A saved interest card — topics, keywords, geographies, sectors — used for tagging and search. Not a summary of a dataset, and not your user profile. |
| Data project | The top-level published object: a survey or set of waves, its files, its metadata and its access rules. This is the only name VoxDash uses for it, and it is not a software project. |
| Data provider | The main role on a data project: the party that publishes it and grants access to it. It is a role, not a place — the console providers work in is VoxDash Studio. |
| Data role | The function a vendor or person performed on a data project: data collector, data QC, funder. A credit, not a permission. |
| Dataset | One of the file types a data project can hold (Dataset, among Document, Audio, Video and Other). The published object itself is a data project. |
| DOI | Digital object identifier — the permanent citation identifier minted for a published data project. Choose a licence before requesting a DOI. |
| Geography | A place entity in VoxDash’s geographic catalogue: a country, province or city. Plain region in an address, a time zone or a demographic level is the ordinary word, not this entity. |
| Map | As a noun, a geographic map of results. As a verb, to assign an uploaded column to a role. |
| Metadata | The descriptive fields recorded about data, a variable, a question or a DOI — not the data itself. |
| Standard | A hosted standards or guidelines document with its own page. Used as an adjective, standard simply means default, and standard users means analysts. |
| Survey | A hosted instrument together with its responses, inside a data project. The whole data-collection undertaking is also called a survey; a questionnaire authored in the Questionnaire Builder is a different object. |
| Survey question | An item of a hosted survey, with its own page and results. |
| Topic | A curated subject tag attached to a survey question, data project or institution, used for categorising and search. Not a discussion thread. |
| Trend | A curated link between comparable questions across surveys or waves, kept as an object you can open. Not a fashion trend. |
| VoxDash Studio | The provider console where data providers manage their data projects. The name stays as written in every language. |
| Wave | One round of a repeated survey, compared over time in a trend. |
Organizations, vendors and people
| Term | Meaning in VoxDash |
|---|---|
| Individual | A person profile, such as a researcher or an expert. Not a respondent. |
| Key person | The named principal contact on a vendor or institution profile, with its own public page. |
| Organization | The account workspace, with its users, groups, data projects and subscription. Not the same thing as a vendor profile. |
| Title | A heading or a name. A person’s job title is their role; the honorific (Mr, Ms, Dr) is the separate Name prefix field. |
| User | An account holder, whatever their role. A respondent answering a survey is not a user. |
| Vendor | A research organisation profile credited on data — a polling firm, institute, agency or association — which may hold a data role such as data collector or data QC. Not a seller or supplier. |
Access, governance and roles
| Term | Meaning in VoxDash |
|---|---|
| Access | Permission to view or use data, granted through public access tiers and ownership levels. Nothing to do with accessibility. |
| Access level | How open a data project is to others: open, restricted or gated; public, restricted or private. Distinct from an ownership level, which is one person’s access to private data. |
| Access request | A request to a data provider for access to private data, carrying a status of Pending, Approved or Rejected. Also written as a request for access. The HTTP request named in a technical error message is a different thing entirely. |
| Admin (organization role) | The highest organization role: it can manage members and settings. Separate from Owner, which is an ownership level on data. |
| Analyst (organization role) | The organization role that analyses data without administering the organization. This is what standard users refers to. |
| Catalog (ownership level) | The lowest ownership level: the holder sees only the catalogue entry. It names a level, not the public catalogue itself. |
| Crosstab (ownership level) | The ownership level that permits crosstab analysis of private data. It names a level, not the crosstab table. |
| Data classification | The confidentiality level assigned to data according to the impact of disclosing it. A governance label. |
| Download (ownership level) | The ownership level that permits downloading private data. It names a level, not the download button. |
| Organizer (ownership level and organization role) | The ownership level below Owner, and the organization role of the same name. It is the same role in both lists. |
| Owner (ownership level) | The highest ownership level on private data, above Organizer. Distinct from the Admin organization role. |
| Ownership | Who holds a data project, and at what ownership level. A governance relation, not legal title to property. |
| PII | Personally identifiable information — data that identifies a person, as flagged by “Yes - collects PII”. |
| Private (public access) | The public access tier in which a data project is not public at all. Also the plain adjective, as in “Private - only you can view this”. |
| Public (public access) | The public access tier in which a data project is listed and fully available. |
| Public access | Who can see a data project, set as one of three tiers: Public, Semi-Private or Private. Not accessibility. |
| Semi-Private (public access) | The public access tier in which a data project is listed in the catalogue but the data itself is gated. |
| User group | A named set of users granted access to data together. Distinct from a quota group. |
| User (organization role) | The lowest organization role. Here User is a role label, not the general word for an account holder. |
| Viewer (ownership level) | The ownership level that permits viewing private data. Not a questionnaire collaborator. |
Plans, credits and billing
| Term | Meaning in VoxDash |
|---|---|
| Add-on | A paid extra bought on top of a subscription: AI Analysis top-ups, extra storage, extra user seats. A purchase, not a software plug-in. |
| AI credits | The prepaid balance spent on AI tasks and calls. Often shortened to credit — it is the same balance, and nothing to do with a credit card. |
| Data project bundle | Several data projects sold or granted together as one package. |
| One-time purchase | Buying a whole data project outright for a single price, instead of reaching it through a subscription. |
| Plan | A subscription plan or tier. |
| Plan limit | A ceiling the subscription plan puts on a countable resource. Going over it restricts access until the plan is upgraded. |
| Seat | One user licence included in, or bought for, a subscription, which a team member occupies. |
| Subscription | An organization’s recurring plan, with its limits and add-ons. The billing object. |
| Top-up | A purchased addition to the AI credit balance. Unlike a manual grant, a top-up is refunded through the payment gateway. |
Samples, populations and statistics
| Term | Meaning in VoxDash |
|---|---|
| Base | The number of respondents a percentage or estimate is computed over, as in percent of base, comparison base or off base. Not a base map. |
| Cluster sampling | A probability design that samples groups of units — households, areas, schools — and then units within them. Often multi-stage. |
| Confidence level | The statistical confidence attached to an estimate, usually 95%. Not a trust or permission level. |
| Convenience sampling | A non-probability design that uses whoever is readily reachable. |
| Design effect | The factor by which the sampling design inflates variance compared with simple random sampling. Abbreviated DEFF. |
| General population | Everyone in the territory, as opposed to a specific target group. |
| Margin of error (MOE) | Half the width of the confidence interval around an estimate, at a stated confidence level. |
| Non-probability sampling | A design in which selection is not random, so the chance of any unit being chosen is unknown. Paired with probability sampling. |
| Panel | In research, a standing group of recruited people a study draws its respondents from — an online panel, a panel survey, a panel count. In the interface, panel also means a region of the screen, such as the Tags panel. The repeating question group in the Questionnaire Builder is a dynamic panel, a third sense. |
| Population | Every unit the survey aims to describe, and from which the sample is drawn. The target population is that same group as the screening rules define it. Not the inhabitants of a place. |
| Probability sampling | A design in which every unit has a known, non-zero chance of selection. Paired with non-probability sampling. |
| Purposive sampling | A non-probability design in which the researcher selects units by judgement. Also called judgment sampling. |
| Quota sampling | A design that fills preset counts for defined subgroups without random selection. Despite the word quota it is non-probability sampling, and it is not the same as a quota used inside a probability design. |
| RDD | Random digit dialling — telephone numbers generated at random rather than taken from a list, as in “RDD or List type sample”. |
| Sample | The subset of a population a survey’s results are based on, as reported in the metadata. In AI-CATI the word means the uploaded contact list instead; see sample (AI-CATI). |
| Sample size | The number of units in the sample, written n. |
| Sampling error | The variation in an estimate that arises because it comes from a sample rather than the whole population. Distinct from non-sampling error. |
| Sampling frame | The list or source the sample is drawn from: address-based, telephone-based, panel. |
| Sampling method | The design used to select the sample: simple random, cluster, stratified, quota. |
| Simple random sampling | A probability design in which every unit of the frame has an equal chance of selection and every sample of a given size is equally likely. Abbreviated SRS, and the benchmark the design effect is measured against. |
| Stratified sampling | A probability design that divides the frame into strata and samples within each. Stratification is that dividing-up, done so every stratum is represented. |
| Systematic sampling | A probability design that takes every kth unit from an ordered frame, after a random start. |
| Weight variable | The variable holding each respondent’s weight, applied when results are reprocessed. |
| Weighting | Adjusting the influence of respondents so the sample reflects the population — correcting unequal selection probabilities, non-response and coverage, including post-stratification to population benchmarks. Not physical weight. |
Fieldwork, dispositions and rates
| Term | Meaning in VoxDash |
|---|---|
| AAPOR | The American Association for Public Opinion Research, whose standards define the disposition codes and response-rate formulas VoxDash reports. |
| Break-off | An interview the respondent starts and abandons before the end. AAPOR records it as a partial interview; it is not a dropped connection. |
| Completes | Completed interviews, counted as a noun: target completes is the number you want. |
| Contact rate | The proportion of all eligible units in which a responsible member was reached, as AAPOR defines it. |
| Cooperation rate | The proportion of all units contacted that yielded an interview, as AAPOR defines it. |
| Disposition | The AAPOR outcome code recording how a contact attempt ended: completed, refused, no answer. |
| Eligibility | Whether a contacted person qualifies for the survey under the screening rules, including AAPOR’s unknown eligibility. Not account permission. |
| Field | As a noun, one input on a form, or one column in a data file. As a verb, to put a questionnaire or a sample into fieldwork — “ready to field”, “before you go to field”. |
| Field period | The span of dates over which fieldwork runs. Here field is the fieldwork activity, not a form field. |
| Fieldwork | The activity of collecting survey data in the field. In AI-CATI, the object that carries it out is a study. |
| Incidence rate | The share of contacted people who qualify for the survey: a screening and eligibility rate, nothing to do with epidemiology. |
| Interview | One administration of the questionnaire to a respondent — by telephone in AI-CATI. Not a job interview. |
| Non-response | Failure to obtain an interview or an answer from a selected unit. It covers non-contact, refusal and inability to participate. |
| Quality control (QC) | Verification of collected interviews, typically by calling a share of respondents back. QC is also a data role credited on a project, and a review step — never software quality assurance. |
| Refusal | A contact in which the person declines to be interviewed. The AAPOR outcome, not the rejection of an access request and not a declined payment. |
| Refusal rate | The proportion of all units in which a respondent refused, as AAPOR defines it. |
| Respondent | The person interviewed, or answering a survey. |
| Response rate | Complete interviews divided by the eligible units in the sample, as AAPOR defines it. Cases of unknown eligibility enter the denominator with an estimation factor. |
(I+P)+R+O+NC+(UH+UO) — among them NC for a non-contact, UH for an unknown household and UO for unknown other. They are formula names and symbols rather than words, which is why they are never translated.
Data collection modes
| Term | Meaning in VoxDash |
|---|---|
| AI-CATI | AI-moderated telephone interviewing: an automated voice interviewer places the call and conducts it. A data collection mode, not a shift staffed by human interviewers. |
| AI-CAWI | AI-assisted web interviewing: a self-completion web mode with AI assistance. Distinct from AI-CATI, and not a chatbot. |
| CAPI | Computer-assisted personal interviewing — face to face with an interviewer. |
| CATI | Computer-assisted telephone interviewing. |
| CAWI | Computer-assisted web interviewing. |
| Data collection mode | The channel an interview is conducted through: web, telephone, face-to-face, SMS, IVR, AI-CATI or mixed. Not an editing or preview mode. |
| IVR | Interactive voice response — a telephone mode in which recorded prompts are answered on the keypad or by voice. Automated, with no live interviewer. |
| SMS-only | A data collection mode conducted entirely by text message. The mode itself, not an SMS notification or reminder. |
Questionnaire Builder
| Term | Meaning in VoxDash |
|---|---|
| Admin questions | Questions or variables carried for routing, operations or metadata rather than to measure anything. Administrative items, not questions put to an administrator. |
| Closed-ended | Of a question: offering fixed answer options. Paired with open-ended. |
| Custom variables | Variables the author defines to hold respondent attributes, computed values or URL parameters rather than answers. The author-defined kind, not a question variable. |
| Dummy data | Synthetic respondent data generated in the Questionnaire Builder to test a questionnaire. Fake test data — never a placeholder value and never a dummy variable. |
| Dynamic panel | The repeating question group in the Questionnaire Builder: one block of questions asked once per item. Not a region of the interface. |
| Emotions rating | A question type on which the respondent rates a feeling or sentiment, usually on a pictorial or affective scale. The question type, not a sentiment score derived from text. |
| Logic rules | The branching and skip conditions authored on a questionnaire, deciding which question comes next. |
| LOI | Length of interview — how long a questionnaire takes to complete, as in “~12m LOI” or “Target LOI achieved”. Never a letter of intent. |
| Multi-select matrix | A grid question allowing more than one answer per row. |
| Multiple choice | A closed question from which the respondent may pick more than one answer. In VoxDash it always means multi-answer; a pick-one list is single choice. |
| Non-substantive only | An export filter keeping only routing, display, administrative and computed items — the ones that measure nothing. The opposite of substantive only. |
| Open-ended | Of a question: inviting free text rather than fixed options. Paired with closed-ended. |
| Question metadata | The export-side settings on a question: variable name, labels, formats, missing values and value labels. Not the wording of the question. |
| Question order | Whether a questionnaire keeps its authored question sequence or randomises it to reduce order effects. The setting, not a question’s index number. |
| Question types | The categories a questionnaire’s items fall into: single choice, matrix, rating scale, open ended. A classification of items, not a count of questions. |
| Questionnaire | The instrument authored and versioned in the Questionnaire Builder. An AI-CATI study binds one version of it. A hosted survey inside a data project is a different object. |
| Questionnaire Builder (QB) | VoxDash’s questionnaire-authoring module. The name is fixed everywhere, and its abbreviation QB stays in Latin script. |
| Rating scale | A question offering an ordered scale: agreement, satisfaction, importance, likelihood. The class that includes Likert and NPS — not a star widget and not a score. |
| Response | A respondent’s answer to a question, and the stored record of it. Not one of the offered choices, which is a response option, and not a reply to a message. |
| Response option | One of the fixed choices offered for a closed-ended question. Never the answer a respondent gave, which is a response. |
| Screener | The question or questions at the start of a questionnaire that decide whether a respondent qualifies. Tagged Screener in the Questionnaire Builder. |
| Single choice | A closed question from which the respondent picks exactly one answer. One answer permitted — not a single question, and not the only choice available. |
| Single-select matrix | A grid question allowing exactly one answer per row. |
| Substantive only | An export filter keeping only the questions that measure attitudes, behaviour, characteristics or outcomes. The opposite of non-substantive only. |
| Version history | The list of saved and published versions of a questionnaire, which can be previewed and compared. |
Quotas and targets
| Term | Meaning in VoxDash |
|---|---|
| Cell | One box of a grid: a quota cell defined by a combination of answers, or a day-and-daypart box in the calling-window grid. Not a spreadsheet cell and not a phone cell. |
| Cells to consider | A quota routing setting that limits evaluation to a given number of the least-filled eligible cells. A routing control, not a count of cells. |
| Hard quota | A quota that enforces its target and turns further respondents away once it is full. The opposite of a soft quota. |
| Least-filled routing | Quota routing that sends each qualifying respondent to whichever eligible cell is furthest from its target, to even out fill. |
| Linked questions | The questions a quota group is defined over, supplying its dimensions and its eligibility. Linked in the quota sense — not hyperlinked and not piped. |
| Quota | A required count of completes for a defined subgroup, built from quota groups and cells. Not a sales quota. |
| Quota group | A set of quota cells defined over one or more questions in the Questionnaire Builder. |
| Quota management | The page where quota groups, cells, targets, routing and fill status are created and reviewed. The feature, not a quota itself. |
| Selected dimensions | The questions chosen as the axes of a quota group, whose answer codes generate its cells. Quota dimensions, not crosstab dimensions. |
| Soft quota | A quota that tracks progress against its target but does not turn respondents away once it is reached. The opposite of a hard quota. |
| Target | The wanted number of completes, or the wanted per-answer counts; in a quota, the min, max and total per cell. |
| Total cells | The number of quota cells defined across the selected quota groups. A count of quota cells, not of table cells and not of targets. |
| Total target | The number of completed interviews a quota group is aiming for in total. The goal figure — not a target audience and not the target population. |
AI-CATI studies and calling
| Term | Meaning in VoxDash |
|---|---|
| Attempt limit | The maximum number of dial attempts allowed on one sample record before it is closed as out of attempts. |
| Callable records | The sample records still eligible to be dialled: not yet completed, not out of attempts, not excluded. |
| Callback | A re-attempt scheduled for a later time, after an earlier call did not reach or complete the interview. The telephone-research sense, not a software callback. |
| Calling engine | The service that places and runs the AI calls. It is the component named when a call fails or an action is refused. |
| Calling hours | The hours of the day in which calls may be placed, in each respondent’s own time zone. |
| Calling rules | The settings governing how a study dials: attempt limit, calling windows, retry spacing and callbacks. |
| Capacity tier | The band that sets how many calling lines a study may use at once, and the platform price per call. Not a subscription tier. |
| Daypart | A named part of the day in the calling-window grid, enabled or disabled per weekday. |
| Hold | An amount of credit reserved against a study and later released or settled — a credit hold or a spend cap hold. Never a call placed on hold. |
| IPA | The International Phonetic Alphabet, used to correct how the synthetic voice pronounces a word, as in “IPA phoneme overrides”. |
| Pacing | How fast the calling engine dials, throttled automatically from live answer and connect rates. |
| Pre-test | A small trial run of a study, placed before the production run, to check the script on real calls. |
| Production run | The main launch of a study, as opposed to the pre-test run. |
| Quiet hours | Hours in which nobody is called, not even for a scheduled callback. They do not count as enabled calling time. |
| Record | One row of the sample: a contact to call. Never an audio recording. |
| Recording consent | The question in the questionnaire that asks the respondent’s permission to record the call. |
| Retry ladder | The ordered list of waiting intervals between dial attempts, one interval per attempt. |
| Run | One launch of a study by the dialling engine: a pre-test run, or the production run. |
| Sample (AI-CATI) | The uploaded contact list that is cleaned and dialled in an AI telephone study — the people to call. Not the statistical sample. |
| Self-test | A call you place to your own number to hear the AI interview as a respondent would, before launching. |
| Spend cap | The authorised maximum a study may spend on calls, set before launch. Not a plan limit and not an estimate. |
| STT | Speech to text — transcribing what the respondent said, as in “STT confidence”. Paired with TTS. |
| Study | One configured AI telephone data-collection job: a questionnaire version, a sample, calling rules and a completion target, with a draft, launch, pause and close lifecycle. |
| TTS | Text to speech — the synthetic voice that reads a question aloud. Paired with STT. |
| Voice line | One concurrent outbound call channel. A call waits for a free line, and is abandoned if none frees up. |
Data files, variables and SPSS
| Term | Meaning in VoxDash |
|---|---|
| Codebook | The document listing every variable with its codes and value labels. |
| Data map | An Excel map of a dataset’s variables, labels, types, value labels and codes, produced for analysis and data processing. A codebook, not a geographic map. |
| Data type | How a variable is stored for export: string or numeric. The SPSS storage type, not the measurement level and not the question type. |
| Measurement level | The SPSS scale of a variable: Nominal, Ordinal or Scale. Those three label names are fixed. |
| Missing value | A code marking a non-answer such as Don’t know or Refused, excluded from analysis. Often written missing-value code. |
| Numeric decimals | The number of decimal places configured for a numeric variable on export. Not rounding applied to a figure on screen. |
| Numeric format | The SPSS format code of a numeric variable, such as F for standard numeric. Not the decimal count and not a display format. |
| Numeric width | The storage or display width allocated to a numeric variable on export. Not an on-screen column width and not a string width. |
| SPSS | The IBM statistics package whose .sav file format VoxDash reads and writes. |
| SPSS data file | An SPSS or CSV file carrying respondent records together with the variable dictionary. An SPSS shell, by contrast, has no records. |
| SPSS format | The SPSS variable format on export: format code, width and decimal places. A storage format, not a file format and not page formatting. |
| SPSS shell | An SPSS .sav file holding only the variable dictionary, labels and formats, with no respondent records. Produced before fieldwork. Never a command shell. |
| String width | The maximum number of characters allocated to a text variable on export. Not a numeric width and not an on-screen column width. |
| Value label | The text attached to a numeric code of a variable, as in 1 = “Strongly agree”. |
| Variable | A column of the dataset produced by a question, with its own name, type and value labels. Not a programming variable. |
| Variable label | The descriptive text documenting what a variable means in a data file or codebook. Not a value label, and not a field label in the interface. |
AI analysis of open-ended answers
| Term | Meaning in VoxDash |
|---|---|
| AI | Artificial intelligence. In VoxDash it is the modifier in product names: AI Analysis, AI credits, AI-CATI. |
| AI Analysis | The paid feature that runs AI tasks over open-ended answers: coding, sentiment and translation. |
| Auto code | A value or classification the system assigns by itself, through logic or coding, normally hidden from the respondent. Coding open-ended answers automatically is one case of it. Not an access code and not a promo code. |
| Codeframe | The set of codes used to classify open-ended answers. Built before coding, and reported with the data. |
| Sentiment analysis | The AI task that classifies the tone of open-ended answers. One of the named task types, alongside coding and syntax correction. |
| Syntax correction | The AI task that fixes spelling and grammar in open-ended answers without changing their meaning. |
| Verbatim | An open-ended answer recorded in the respondent’s own words, as in “code thousands of verbatims”. A noun here, not the adverb meaning word for word. |