The shows the average sentiment reaching each factor: blue where sources say a cause helps it, red where they say a cause harms it. But an average hides disagreement. A factor sitting near neutral could mean two very different things: sources agree it is neutral, or sources are split, with some reporting a strong positive effect and others a strong negative one. Those are opposite findings that the average paints the same colour.

Sentiment discrepancy separates them. Set Factor Colour to Sentiment discrepancy in Map Formatting and each factor is shaded orange by how much its incoming links disagree. It counts the spread of the individual incoming sentiments rather than averaging them, so a factor where sources contradict each other stands out even when the average is unremarkable.

• A deep-orange factor is contested: many sources say a cause pushes it up, and many say the same or another cause pushes it down.
• Pale or white means the sources broadly agree, or there are too few links for disagreement.
• A factor with many links scores higher than one with only a couple, so the orange marks contested factors that much of the data speaks to rather than one-off outliers.

Why this is worth looking at:

• Contested factors are often the interesting ones. Disagreement about whether something helps or harms usually points to a trade-off, a context that changes the effect, or subgroups living the same change in opposite ways.
• Treat it as a lead to follow up. Once an orange factor catches your eye, read the underlying quotes, or split the map by group (gender, region, role), to find out who reports the positive story and who reports the negative one.
• It complements the mean. Use plain sentiment to read the headline direction, and sentiment discrepancy to find where that headline is really an argument.

See also#

• for following up on who disagrees.