What Is 'Connection Score' Methodology in Investigative Journalism — and How to Apply It to Academic Research

# What Is "Connection Score" Methodology in Investigative Journalism — and How to Apply It to Academic Research

Dateline: December 1, 2025 By Nobel Pardon Prize Research Desk

---

WASHINGTON — When you open the Epstein Files Emails Index (EFE) at nobelpardonprize.org/efe, the first thing you see is a ranked table. Ghislaine Maxwell sits at position one with a connection score of 2,295. Sarah Kellen is second at 1,222. Jean-Luc Brunel is third at 879. Each score combines two directional metrics: emails sent to Jeffrey Epstein (→ JE) and emails received from Jeffrey Epstein (← JE).

That composite score is not arbitrary. It reflects a methodology with roots in computational journalism, graph theory, and social network analysis — a methodology that students can understand, critique, and apply in their own research.

Where "Connection Score" Comes From

The concept of scoring relational connections by documented frequency appears in multiple research traditions:

Investigative journalism has used connection graphs since at least the mid-2000s. Academic research published in the IEEE Bulletin of the Technical Committee on Data Engineering describes systems like ConnectionLens, which integrates heterogeneous data sources into a queryable graph and allows journalists to identify connections between entities across document sets. The core challenge identified in that research is the same one the EFE index addresses: "there is no single, well-behaved score function" — so composite scoring (combining inbound and outbound links) is a practical approximation. Social network analysis formalizes this as degree centrality: the count of direct connections a node has. The EFE's connection score is functionally equivalent to undirected degree centrality in a bipartite email graph, weighted by message volume. Citation analysis in academic publishing uses a similar logic: papers that are cited more are treated as more influential. The EFE score treats email references as citations — each documented email reference is a data point contributing to the score.

What the Score Measures — and What It Doesn't

The EFE connection score measures volume of documented email contact with Jeffrey Epstein across court-authenticated records. It does not measure:

• The nature or content of those communications
• The legal significance of the contact
• Whether any individual engaged in wrongdoing

This is a critical methodological distinction that any research paper using EFE data must state explicitly. Being named in a court document does not imply any allegation of misconduct — this was explicitly noted in the January 2024 unsealing order for Giuffre v. Maxwell.

How to Apply Connection-Score Thinking to Your Own Research

The connection-score methodology is adaptable to many research contexts:

Political science: Score legislators by the number of documented campaign contributions from a defined donor class. Higher scores = tighter documented financial relationship. Sociology: Score institutions by the number of shared board members across a defined sector (interlocking directorates). This is a standard method in elite network research. Communications/Media: Score public figures by the number of times they are co-mentioned in a defined set of news articles. This is a form of co-occurrence analysis used in media framing studies.

In each case, the methodology requires:

• A clearly defined population (who counts as a node)
• A clearly defined relationship type (what counts as an edge)
• A clearly defined data source (where the edges come from)
• An explicit interpretive limit (what the score cannot tell you)

The EFE index is exemplary in all four dimensions: nodes are named court-referenced individuals, edges are authenticated email references, the source is DOJ-released case records, and the index notes (via the ⚠ symbol) which individuals have disputed or contested connections.

Peer-Reviewed Literature on Scoring Methods

For your literature review, consult:

• Saez-Trumper et al. (2013) on social influence scoring in news networks
• The PNAS study on measuring investigative journalism (2021), which develops a classifier-based "investigativeness score" using text features and document influence
• Ronald Burt's Structural Holes (1992) on betweenness-based power scoring in social networks

---

How to Cite This for Your Assignment

APA (7th ed.)

Nobel Pardon Prize Research Desk. (2025, December 1). What is "connection score" methodology in investigative journalism — and how to apply it to academic research. Nobel Pardon Prize. https://nobelpardonprize.org/efe

MLA (9th ed.)

Nobel Pardon Prize Research Desk. "What Is 'Connection Score' Methodology in Investigative Journalism — and How to Apply It to Academic Research." Nobel Pardon Prize, 1 Dec. 2025, nobelpardonprize.org/efe.

---

Research Hub

The Epstein Files Emails Index at nobelpardonprize.org/efe provides a live demonstration of connection-score methodology applied to a real, court-authenticated dataset. Explore the ranked table, understand how the scores are computed, and use it as a template for your own quantitative research design.