🏛️ Congressional Cosponsorship Network Analysis

Degree Centrality Comparison Across Party Affiliation in the U.S. Congress

---

📌 Project Overview

This project presents a high-level analysis plan for building and analyzing a bill cosponsorship network from the United States Congress. Using data acquired through the Congress.gov API (Library of Congress), the plan describes how to construct a network where legislators are connected by shared cosponsorship of legislation — and then compare degree centrality across categorical groups, primarily party affiliation (Democrat, Republican, Independent).

🎯 Central Research Question: Does a legislator's party affiliation predict how broadly they collaborate with colleagues through bill cosponsorship?

---

🔗 Network Definition

Element	Description
🟢 Nodes	Members of Congress (~535 legislators per session)
🔵 Edges	Cosponsorship ties — two legislators share an edge if they cosponsored the same bill
⚖️ Edge Weight	Number of bills cosponsored together (stronger weight = deeper collaboration)
📊 Degree Centrality	Fraction of all other legislators a given member is directly connected to

---

📂 Repository Contents

📁 congressional-cosponsorship-network/
│
├── 📄 Congressional_Cosponsorship_Network_Analysis_Plan.docx
│       └── Full project plan document (data source, loading steps,
│           network definition, hypothetical outcomes)
│
├── 📄 Video_Presentation_Script.docx
│       └── Narrated walkthrough script for video presentation
│
├── 📄 README.md
│       └── This file
│
└── 📁 code/ (future)
        └── Jupyter notebooks and Python scripts for implementation

---

🌐 Data Source

Congress.gov API — maintained by the Library of Congress

Detail	Info
🔑 Authentication	Free API key from api.data.gov
⚡ Rate Limit	5,000 requests/hour
📦 Format	JSON responses with pagination support
📅 Coverage	Bill and member data from the 81st Congress (1949) to present

Key Endpoints Used

Endpoint	Purpose
GET /v3/member	Retrieve all legislators with party, state, and chamber
GET /v3/bill/{congress}/{type}	Retrieve all bills for a Congressional session
GET /v3/bill/{congress}/{type}/{number}/cosponsors	Get cosponsor list for each bill

---

🏷️ Categorical Variables

The primary grouping variable for degree centrality comparison:

Variable	Categories	Why It Matters
🟥🟦 Party (primary)	Democrat · Republican · Independent	Does one party collaborate more broadly?
🏛️ Chamber	Senate · House	Does chamber structure affect connectivity?
🗺️ Region	Northeast · South · Midwest · West	Are regional caucuses more insular?
📍 State	50 states + territories	Do large delegations cluster together?

---

🔬 Hypothetical Outcomes

The plan describes five predicted outcomes from comparing degree centrality across party groups:

🔮 Outcome A — Minority Party Shows Higher Centrality

The minority party must build cross-aisle coalitions to advance legislation, forcing broader cosponsorship networks and inflating degree centrality.

🔮 Outcome B — Independents as Extreme Outlier Bridges

The 2–3 Independent legislators show dramatically high centrality because they cosponsor freely across both parties, acting as structural bridges.

🔮 Outcome C — Chamber Matters More Than Party

Senate vs. House differences dominate party differences — the filibuster's 60-vote threshold forces Senators into broader collaboration patterns.

🔮 Outcome D — Bimodal Distribution in the Majority Party

A leadership-backbencher divide emerges: committee chairs and party leaders have very high centrality while rank-and-file members cluster low.

🔮 Outcome E — Swing-State Legislators Collaborate More Broadly

Electoral pressure drives collaboration — legislators from competitive states cosponsor more broadly to demonstrate bipartisan appeal to voters.

---

🛠️ Technology Stack

Tool	Role
🐍 Python 3.10+	Primary language
🌐 requests	API calls with authentication and pagination
🕸️ NetworkX 3.x	Graph construction and centrality computation
🐼 pandas	Data wrangling and group aggregation
📊 scipy / scikit-posthocs	Kruskal-Wallis, Dunn's post-hoc, Mann-Whitney U
🎨 matplotlib / seaborn	Visualization (box plots, violin plots, network graphs)

---

📋 Data Loading Pipeline (High Level)

graph LR
    A[🔑 API Auth] --> B[👥 Fetch Members]
    B --> C[📜 Fetch Bills]
    C --> D[🤝 Fetch Cosponsors]
    D --> E[🔗 Build Edge List]
    E --> F[🕸️ Construct NetworkX Graph]
    F --> G[📊 Compute Degree Centrality]
    G --> H[📈 Group by Party & Compare]

Loading

Step-by-step:

1. Environment Setup — Install requests, networkx, pandas, scipy, seaborn
2. Fetch Members — Pull ~535 legislators with party, state, chamber from /v3/member
3. Fetch Bills & Cosponsors — For each bill, retrieve the full cosponsor list
4. Build Edge List — Generate pairwise combinations from each bill's collaboration group
5. Construct Graph — Load into NetworkX with party as a node attribute
6. Compute Centrality — nx.degree_centrality(G) → group by party → compare
7. Statistical Testing — Kruskal-Wallis across parties, Dunn's post-hoc for pairwise
8. Cache Locally — Save graph as .graphml, DataFrames as .csv

---

👩🏽‍💻 Author

Candace Grant M.S. Data Science — CUNY School of Professional Studies

---

📄 License

This project is submitted as coursework for the CUNY SPS Data Science program. The data is sourced from the public domain Congress.gov API.

---

Built with 🐍 Python · 🕸️ NetworkX · 🏛️ Congress.gov API