Attribute is a specific answer to some question. It’s a combination of a question,
a datapoint and a suffix (if the question has suffixes). It’s defined as a standalone
entity for easier manipulation, e.g. to be used as a building block in the platform
audience builder where the audience is based on attributes.
Any effect which artificially distorts the collected research such that it is not a
perfect representation of reality. There are a variety of biases including sample
biases (for example demographic skew) and response biases (for example the ordering
of datapoints for a survey answer).
A namespace that can be analysed in isolation but can also be combined
with parent for wider analysis. A child namespace may also be a
parent namespace as part of the namespace lineage.
The base standard survey that the business sends out to the majority of people
across most countries. In certain countries Core Survey might be changed (e.g.
no alcohol questions in Egypt).
The umbrella term to describe all work done by the custom research team and more
generally to describe work done for a single client by request, rather than for general release.
In a simple case it’s the most granular option to be selected for a question.
Such as “Male” to a question “What gender do you identify with?”. Datapoints
can be created from different types of questions and an actual answer selected
by the respondent might be a combination of, e.g., a suffix and a datapoint, such
as if the previous gender question was so called “Matrix” one, suffix being
“Strongly identify” and Datapoint “Male”.Every question should have at least one datapoint in the platform
The index tells you how much more or less likely members of your audience are to exhibit
a particular trait compared to the base. An index above 100 means they’re more likely
to exhibit that trait, while an index below 100 means they’re less likely.
A metric which indicates how much more (over 100) or less (under 100) likely an
audience or subgroup is to engage in a certain type of behaviour vs. the base sample.See our knowledge base article.
for more on index scores and how to interpret them.
The “family tree” of namespaces which describes the suitability of cross-namespace analysis.
Specifically, rather than describe the namespaces themselves this really describes whether or
not namespaces might share respondents. When respondents are shared across namespaces then it is
considered viable to analyse across namespaces.
An optional numerical value attached to a datapoint to allow aggregations.
For example, for the question “How often do you drink green tea” the datapoints
might be “daily”, “once a week” and “never”, and the associated midpoints “7”, “1” and “0”, respectively.
An entity specifying “where the data are loaded” in our in-house DB chronicle.
This tells calculations where to look for specific questions (e.g. gwi-123.q2).
It’s managed in the Louvre tool. The most important information namespace holds is
its label and what type of data are in it (fresh sample, recontact, add-on) and in
what releases (“waves”). See more in Namespaces, releases and waves | Namespace
The source research that sits at the top of a namespace lineage, this is the namespace from
which others would draw their recontact sample. More specific examples here: GWI Namespaces | Recontact vs. Fresh Sample.Technically, only namespaces have “lineage” (Louvre projects are not having self-reference), but this is often used interchangeably.
One of the main, top, entities in GWI. A question has various forms in different
systems, but most of the time has a “Question text” (“What country do you live in?”)
and then some answer options (“Czechia”, “UK”, “USA”, …) respondents can answer with.
Question come in different types and shapes (Slider question, Multiple choice question, Matrix question, …).
To recontact is to find respondents from any given research and directly target them for
further additional research. Most commonly this is used to give richer insight about
specific individuals - often to provide specific attributes of client interest. More
explanation is available here: GWI Research Projects | Recontact vs. Fresh Sample .
Used to specify in what waves the data is loaded and how that wave should display in the
platform. Therefore, sometimes used interchangeably with wave (but it’s misleading).
Can have multiple parent releases (from a parent namespace). See more in Namespaces,
releases and waves | Wave and Release
The collection of data provided by a single individual in response to a single piece of
research. The response is comprised of the collection of questions and answers.
The ID given to a specific response which is also shared throughout the lineage of a
namespace such that responses can be combined for cross-analysis if applicable.
Formally known as multipliers or segmentation. Are defined per namespace in the labels
management platform under “Splitters” tab. You can mark a namespace’s question as a splitter
and that then displays in the GWI platform as a question by which you might want to
segment/group by your respondents without having to create an audience manually.
Not every question type has a Suffix, but e.g. Matrix type questions do. It’s often used
for “scale” or “sentiment” when answering, such e.g. for a question of “What activity you do often”
with a datapoints “Ski“ and “Swimming”, you might have suffixes “Often”, “Sometimes” and “Never”.
Combination of Suffix and Datapoint is then the respondents answer, that can be understood as an
Attribute “Often do Skiing“ or “Never do Swimming”.
One output sold to multiple different clients with minor or no variations. In our case this often refers to studies like “core” and differentiates from custom work done for a specific client.
Time interval used to specify when the research has happened. Basically used as a
partitioning for our data. Every datapoint is in a specific wave of a specific namespace.
A factor applied to the outcome of a measurement to mitigate the effect of
biases in collected data. Most typically these will exist as demographic weights
(used to mitigate for sampling biases) and response weights (used to mitigate for other methodology biases).
Was this page helpful?
Assistant
Responses are generated using AI and may contain mistakes.