Terms used throughout webENA, in alphabetical order. Older names are shown in parentheses where the interface has changed.
A concept, behavior, or feature you are tracking, stored as a column in your data. Values are usually 0 or 1 (absent or present) but may be weighted. ENA models how often codes co-occur. See Building a Model.
The main plot, showing every unit as a point in the ENA space, with group means and the network for the selected units. See Plots & Units.
A code that is present on a line when two other codes, A and B, both are. By default A and B are replaced by their prime codes, the lines with only one of them. See Compound & Nested Codes.
A pair of codes that co-occur within a window. The strength of a connection for a unit is how often that pair co-occurs in the unit's data.
A matrix of code pairs you can switch off so they are left out of the model. See Advanced Options.
A method for modeling the structure of connections among codes in discourse or activity data, and for comparing those structures across units and groups.
Leaves a unit out of the model. Its lines are still used as context for the other units' connections, but it gets no point and does not affect the rotation or the means. Compare with Hide.
A rotation by regression that extends means rotation to more than two groups and can control for nesting. The web tool regresses each unit's network on the group column and the intermediate unit columns (with interactions) and builds the X axis from the group term alone; with only two unit columns it equals a means rotation. See Advanced Options › Rotations.
Correlations between distances in the plot and distances between the underlying networks. High values mean the plot is a faithful picture of the data. See Statistics.
The average position of a group's units, drawn as a square. It always includes hidden units; exclude a unit to leave it out of the mean.
Stops a unit's point from being drawn. A hidden unit is still in the model and still counts toward its group's mean. Compare with Exclude.
Previously called Conversation. The boundary within which connections can be made. Lines in different horizons are never connected. See Building a Model.
One row of your data, for example one utterance, post, or log event.
A rotation that places the first axis through the means of two groups, so the plot shows their difference as clearly as possible.
A code marked as a more specific kind of another code, its parent. The parent is replaced in the model by its subcodes and a residual code. See Compound & Nested Codes.
A code drawn in the network. Node positions are fixed for the whole model, so networks can be compared directly.
A variant of ENA that models directed connections (A → B is different from B → A). Choose it with the Ordered option on the Codes tab.
Parsimonious Removal with Interpretive Alignment, the method behind Simplify Model. See Simplify Model.
The lines with a parent code but none of its nested subcodes. It stands in for the parent in the model. See Nested subcode.
How the high-dimensional network space is turned to produce the 2-D plot. See Advanced Options.
A dataset together with the model settings built from it. Sets live in projects and folders. See Data & Sets.
Scales each unit's network to the same length, so units are compared on the pattern of their connections rather than on how many connections they made.
The default rotation when no groups are chosen; it finds the axes that explain the most variance between units.
An extension of ENA for data with several modalities (for example speech and log events), where each modality can have its own window and horizon rules. See Transmodal Analysis.
The entity whose network you are modeling, such as a person, team, or document. Each unit gets one network and one point. See Building a Model.
Previously called Stanza. How lines within a horizon are grouped when counting connections, for example a moving window of the last 4 lines. See Building a Model.