EuropeScopeEU

Methodology

Importance Score

A transparent, decomposable 0–100 indicator that helps citizens, journalists and researchers navigate large volumes of European Parliament voting data by surfacing votes with higher public, political or analytical relevance.

What this score does NOT do

  • It does not judge whether a vote is good or bad.
  • It does not rank MEPs morally.
  • It does not infer hidden political intentions.
  • It is a prioritisation tool, not an objective measure of democratic value.

Levels

Key
80–100
Major votes
High
60–79
Important
Medium
40–59
Relevant
Low
0–39
Routine / procedural

Formula

score = vote_type
      + legislative
      + topic
      + group_division
      + country_division
      + close_vote
      + turnout
      + rebel_votes

score = min(100, score)

Every assignment carries the full per-factor breakdown and a human-readable explanation, so the score is fully auditable. The current methodology version is importance-v0.1.

The 8 factors

Vote type

max +25

Final/main votes weigh most. Resolutions, budget and discharge votes get full weight. Amendments and paragraph votes get partial weight. Procedural votes get the least.

  • Main vote (final adoption)+25
  • Budget / estimates / discharge+15
  • Motion for resolution+15
  • Amendment+8
  • Paragraph / split vote+5
  • Procedural / agenda / request+3

Legislative weight

max +15

Approximates the legislative significance of the procedure from title keywords. (Native procedure metadata is not exposed by the upstream source; this is a deterministic approximation.)

  • Legislative procedure (heuristic from main vote)+15
  • Budgetary procedure (budget, discharge, estimates)+12
  • Institutional decision (appointment, election)+8
  • Non-legislative resolution+6

Topic importance

max +15

Each topic in the public taxonomy carries a weight. The vote gets the MAX of its topics (not the sum), so multi-tagged votes are not artificially inflated.

  • Rule of law, Defence, Ukraine, Russia+15
  • Budget, Migration, Climate, Energy, Digital, AI, Agriculture, China, Moldova+12
  • Human rights, Media freedom, Disinformation, Corruption, Democracy, Enlargement+10
  • Foreign affairs, Trade, Health, Security, Civil liberties, Environment, Gender equality+8
  • Other topics+5–6

Political group division

max +15

Measures how divided the European political groups were. A unanimous vote scores 0; a vote where groups split into clear opposing blocs scores high.

  • Two or more large groups internally split+15
  • Three or more distinct majority positions across groups+12
  • Two distinct majority positions across groups+10
  • One large group internally split+6

National delegation division

max +15

Measures how divided national delegations were — both between countries (East vs West patterns) and inside each delegation.

  • Three position blocs AND many split delegations+15
  • Three or more distinct majority positions across countries+12
  • Two majority positions AND several split delegations+10
  • Two distinct majority positions across countries+8
  • Several split delegations+6

Close vote

max +15

Closer votes are politically more consequential. Margin is |For − Against| relative to expressed votes.

  • Margin under 1% of cast votes+15
  • Margin 1–3%+12
  • Margin 3–5%+10
  • Margin 5–10%+6
  • Margin 10–20%+3

Turnout

max +5

High participation signals the vote was politically anticipated. Low participation often indicates a procedural or routine vote.

  • Turnout ≥ 85%+5
  • Turnout 75–85%+4
  • Turnout 60–75%+2

Rebel votes

max +10

Counts how many MEPs voted against their own group's majority. High rebel share signals intra-group conflict — politically interesting.

  • ≥ 10% of votes were rebel+10
  • 5–10%+7
  • 2–5%+4

Limitations

  • Some procedurally important votes are hard to detect from title metadata alone.
  • The legislative-procedure inference is heuristic; native procedure type is not exposed by the upstream data source.
  • Amendment votes can sometimes carry more importance than the score reflects.
  • The methodology can change in future versions; scores are versioned and recalculated when it does.

Implementation: open source under MIT. The full classifier is available in the project repository at pipelines/compute_importance/.