Sources Bar

🌻 Sources Bar#

📄 What you can do here: Choose which source documents (e.g. interviews or reports) from your current project you want to focus on. You can select one or more sources. Use this to narrow your analysis to specific interviews, reports, or other source materials. - The text of the selected source is shown below in the Create Links panel. - Selecting these sources also fetches only their links and no others, starting off the [Links Pipeline](../links-pipeline/): only the links from the currently selected sources are available for further filtering, and are finally shown in the output tabs.

Factor label set (global)#

Treat label sets like different versions of the same document: you keep a default “main” version (the usual factor labels on each link) and you can spin up extra named versions to experiment—try a different coding scheme, run AI Recode on a copy, or compare schemes—without changing the main story until you choose to. Most people finish by copying one experimental version back into the default so the map and tables show that version as the “real” one, while the other versions stay in the project if you need them later.

Which version am I using? The Label set button sits under the Sources bar and shows the version you are on. Open it for the list of every version in this project; the one you are using has a tick. Pick another to view and edit that version’s labels in the Links Pipeline, map, links table, and link editor. Switching refreshes those outputs so they match.
New experimental version: choose Duplicate “” at the bottom of the menu, give it a name, and confirm. The app fills a new column-pair for every link with a copy of the version you were on, then switches you to it. Handy before Recode, so the AI writes into the experiment rather than the default. There is no way to make an empty version, and that is on purpose: a version with no labels shows blank for every link.
Delete a version: the bin icon on its row in the menu. The default has none, because it cannot be deleted. Anything stored only in that version goes with it, and a saved view or bookmark that used it will show nothing.
Done experimenting — make it “official”: choose Promote to default, which appears only when you are on a non-default version. That copies the version you are on into the default columns for every link. Your other named versions are not deleted. What the default held before is not kept anywhere, so duplicate it first if you might want it back.
What each version is: the menu shows the name you gave it with a line beside it saying when it was made, which version it was copied from, what wrote it and how many links it covered. A version made before the app recorded that shows its stored name and nothing else.
Renaming a version is not possible yet, though the name is now stored separately from the columns, so it is a small change rather than a rewrite of every link. Duplicate it under the name you want and delete the old one.
Temporary Factor Labels (old filter in the pipeline): deprecated for new work—use this control instead; old bookmarks may still reference the filter.

👉🏼 Sources (multi-select dropdown): pick one or more sources from the current project (typing searches the list).
👉🏼 Empty selection: means “all sources”.
Selected sources appear as pills; when two or more are selected, a pen-to-square icon appears to the right of the pill for whichever source is currently open in the Source Text Viewer (not shown when only one source is selected).

Default behavior when switching projects:

When you load a project via the Project Dropdown or Projects Panel, and nothing else specifies which sources to load, the app auto-selects the first source so the Source Text Viewer shows something immediately.
When loading from a URL/bookmark, we do not change the sources selection. An empty Sources selection means all sources (by design).
Source loading/viewing behavior is the same for coded and uncoded sources (having 0 links must not block selector state, text loading, or source navigation).

Source Groups sub-panel#

What this does: Filter your analysis by participant demographics or document characteristics, using the [custom columns](../custom-columns/) you have defined for your project. For example, show only responses from "women aged 25-35" or interviews from "urban areas." Perfect for comparing how different groups see causal relationships.

Controls:

👉🏼 Field (dropdown): choose which source metadata/custom column to group by (includes built-in fields like title/projectname).
👉🏼 Value (multi-select dropdown): choose one or more values (list is filtered by Field).
👉🏼 Random N (buttons): load a deterministic random sample of sources from the whole project (e.g. Random 5 / 10 / 20).
👉🏼 Random N / Group (buttons): after choosing Field, load up to N deterministic random sources per value of that Field.
👉🏼 Clear (button): reset the Source Groups selector.

The sampling buttons make a random selection, but deterministically, so the same sources are chosen again if you click the same button again.

The second, "Value" dropdown is filtered to show only valid values for the selected field. Previous/next buttons cycle through values of the selected group.

The effect is to retain only links where the selected custom column has the selected values.

This dropdown is NOT a filter and it does NOT get saved/restored in URL. It is a loader: when we click it, the app automatically loads corresponding sources into the sources selector. These sources then DO form part of the links pipeline and ARE restored from the URL.

There is a similar filter in the Analysis Filters.

Factor label set (global)#

Sources Dropdown#

Source Groups sub-panel#