Network Composer 

Type

Free-form network builder

Creates

Nodes of a single type, edges of multiple types, and node and edge attribute data

Uses prompts

No

Configure in

• Architect ClassicArchitect Classic (not available)
• Architect

Run in

• Interviewer ClassicInterviewer Classic (not available)
• Interviewer
• Fresco

Requires protocol schema 8 or later.

The Network Composer is a free-form, single-screen interface for building a whole personal network in one place. Where most Network Canvas Interfaces lead through one task at a time — naming alters, interpreting them, then drawing ties — the Network Composer brings these together onto a single canvas, so you can create alters, record their attributes, draw relationships of several types, and reposition everything as the picture develops.

It is best thought of as a network "notepad". Because it is dense and fast — with a tool palette, an inspector, keyboard shortcuts, and undo/redo — it is designed to be driven by a researcher or interviewer, either alongside a participant or working alone to sketch, pilot, or model a network, rather than handed to an unassisted participant.

How it works

The Interface presents a single canvas with a floating tool palette. Each tool stays active until you switch to another, so you can repeat an action without selecting it again:

• Select / Move selects alters and ties, and repositions alters by dragging. Selecting one opens the inspector, where you can fill in its attributes; clicking an empty part of the canvas clears the selection.
• Add Node opens a name field beside the palette. Type a name and press Enter to add the alter — it is placed on the next free spot of a grid (or arranged by the automatic layout, when that is on) — and the field stays open so you can add several in a row.
• Draw Edge draws relationships. A single button opens a menu of the stage's edge types; pick one — the button adopts that type's color while it is active — then tap one alter and then another to connect them. Tapping the same pair again removes that tie, and tapping the first alter a second time cancels a half-drawn one. Ties are drawn in their type's color, and the tool is only shown when the stage defines at least one edge type.
• Auto-layout is a switch that turns a force-directed arrangement of the network on or off at any point during the interview. Its starting state is a default you set in Architect — unlike the Sociogram, the mode is not fixed at design time — and whichever way it is switched, that choice is remembered for the stage.
• Groups shades the alters that share a value of a single categorical variable — chosen in Architect — as colored convex hulls behind them, making clusters such as "household" or "friendship group" easy to see. The hulls are always visible once the variable is configured, and the automatic layout gently pulls members of the same group together. The tool lists the groups alongside their colors; pick one — the button adopts that group's color while it is active — then tap an alter to toggle its membership.
• Undo / Redo reverse and replay any change — adding or removing alters and ties, editing attributes, group membership, and repositioning — and are also available with the usual keyboard shortcuts.

To assign groups quickly, you can select several alters at once — by shift-clicking them, or by drawing a freehand lasso around them — and then add the whole selection to a group in one step, using the color-coded buttons that appear beneath the canvas. The same multi-selection can also be deleted together.

Configuring

Everything the tools operate on is set up in Architect when you create the stage:

The name variable

The Add Node field fills a text variable you choose — the same quick-add mechanism as the Name Generator using Quick Add.

The layout variable

Each alter's position on the canvas is stored as normalized coordinates in a layout variable, exactly as the Sociogram stores them.

Attribute forms

The inspector shows one optional attribute form for the alters, and one for each edge type. Each field pairs a codebook variable with the input control used to collect it, plus a label and an optional hint to caption the field. Every field needs a label. When you choose the attribute, Architect fills in the label with the attribute's name, so change it to the words participants should read. The label can be translated. Unlike other form-based Interfaces, the input control is chosen on the stage rather than on the codebook variable, so the same variable can be collected with different controls on different stages. A stage whose form has no fields simply shows that there are no attributes to edit.

Edge types

Each relationship type the Draw Edge tool can create is added to the stage, with a label, and each type may optionally carry its own attribute form. Participants choose the type by its label in the Draw Edge menu, and the label is the title of its form. It starts as the edge type's name, and you can translate it like any other text (see Translating Your Protocol). The tool only appears when the stage defines at least one edge type, and each edge type can be added once.

Groups

An optional categorical variable is drawn as the Groups convex hulls. The tool toggles an alter's membership — a value of that variable — and the automatic layout gently pulls members of the same group together.

Background

The canvas background is either concentric circles or a custom image, shared with the Sociogram and Narrative Interfaces — see Responsive SVG Backgrounds for producing images that scale cleanly across devices.

Automatic layout

The design-time setting only chooses the starting state: unlike the Sociogram, the force-directed layout can be switched on and off during the interview, and the live choice is remembered for the stage.

Customize built-in wording

The Customize built-in wording section of the stage editor holds the words the interviewer reads on the canvas. Every Network Composer stage shows the same words, so they are shared by every stage of this kind: a change to one applies to every Network Composer stage in the protocol, and is saved as soon as you leave the field. Each starts as Network Canvas words it, in every language it has wording for, and you can change and translate it like any other text. Your protocol keeps only the wording you change (see Words Network Canvas supplies).

• Name box placeholder is the hint shown in the Add Node name field.
• Warning about an overtaken edit is shown in the dialog that asks whether to discard changes, when an undo or redo changed an answer while the interviewer was editing it, so their edit was not saved. The protocol holds it only while a Network Composer stage has a form for nodes, or for an edge type it draws.
• Groups tool button is the name of the tool that puts nodes in groups, shown in its tooltip. The protocol holds it only while a Network Composer stage groups alters by a variable.
• Add node tooltip, Automatic layout tooltip and Draw edge tooltip name the canvas tools. The protocol holds the Draw edge tooltip only while a Network Composer stage has at least one edge type. The tools are icons, so a tooltip is how the interviewer learns what each does. The select and undo and redo tools have no tooltip, because their icons are familiar.
• Connection label is the name the inspector gives an edge of a type the stage does not draw, such as one made on another stage. The protocol holds it only while the codebook has an edge type that a Network Composer stage does not draw.

Best Practices

Try to Avoid