In most projects, the data contains many repeated causal claims with the same cause and the same effect (often across many sources). We call these bundles (or co-terminal link bundles).

This extension is about:

• turning many individual coded claims into a smaller set of bundle rows, and
• being clear about what we mean when we say “a link” on a map.

Why bundling is useful (for practitioners)#

• Simpler maps: instead of 200 separate “A → B” rows, you see one “A → B” bundle with counts.
• Clearer evidence signals: you can read “how widely shared” (sources) vs “how often said” (citations).
• Better reporting: it’s easy to state things like “7 sources mentioned A → B”.

What you get (in plain terms)#

Instead of one row per coded claim, you get one row per unique cause→effect pairing (after whatever label transforms you applied).

Each bundle row can show:

• a readable key like cause >> effect
• source count (how many distinct sources made at least one claim of this form)
• citation count (how many coded claims / rows are in the bundle)
• optional summaries like mean sentiment (if you use sentiment)

How to read a “link” on a map (interpretation rule)#

On maps and in tables we often still say “link”, but it usually means:

a bundle representing “many similar claims that cause influences effect”.

So if a link label says “7 sources / 12 citations”, read it as:

• 12 coded claims were bundled together, coming from
• 7 distinct sources.

Practical cautions#

• Bundling happens after transforms: if you zoom/collapse/combine opposites/cluster first, you are bundling the transformed labels, not the raw labels. That’s often what you want, but be deliberate.
• Counts are evidence volume, not effect size: a frequent bundle means “often claimed”, not “strong causal effect”.

Formal notes (optional)#

The filter operates on the current links table (one row per coded claim / citation) and produces a derived table with fewer rows by grouping on the current (possibly transformed) labels.

• Bundle key: (cause label, effect label)
• One output row = one bundle (one unique cause→effect pair)

The bundled output adds aggregate columns such as:

• bundle: a readable key like cause >> effect
• citation_count: number of underlying link rows in the bundle
• source_count: number of distinct sources contributing at least one link row to the bundle

Optional further summaries can be computed from the underlying rows (e.g. mean_sentiment, per-tag counts, per-group counts).