Add insights page

Author: 3bobchen

data/processed/sample_queries

@@ -0,0 +1,48 @@
+1. Vote split distribution on constitutional cases
+
+  SELECT maj_votes || '-' || min_votes AS vote_split, COUNT(*) AS cases
+  FROM cases WHERE primary_issue_sub_area = 'Constitutional law'
+  GROUP BY maj_votes, min_votes ORDER BY cases DESC;
+  Finding: 42% unanimous (7-0), only 6.5% are 4-3 nail-biters. The court reaches very strong consensus on constitutional matters most of the time.
+
+  ---
+  2. Most contested constitutional topics by dissent rate
+
+  SELECT primary_issue, COUNT(*) AS cases,
+    ROUND(100.0 * SUM(CASE WHEN min_votes >= 2 THEN 1 ELSE 0 END)/COUNT(*), 1) AS contested_pct
+  FROM cases WHERE primary_issue_sub_area = 'Constitutional law'
+  GROUP BY primary_issue HAVING cases >= 5 ORDER BY contested_pct DESC;
+  Finding: Corporations power is by far the most contested (89% contested), while judicial power and separation of powers are remarkably uncontested (<10%). The
+  Corporations power data clusters heavily around 2006 — clearly the WorkChoices case cluster generating maximum disagreement.
+
+  ---
+  3. Decision direction by Chief Justice era (constitutional cases only)
+
+  SELECT chief_argument, COUNT(*) AS cases,
+    ROUND(100.0 * SUM(CASE WHEN decision_direction='liberal' THEN 1 ELSE 0 END)/COUNT(*),1) AS liberal_pct
+  FROM cases WHERE primary_issue_sub_area = 'Constitutional law'
+  GROUP BY chief_argument ORDER BY MIN(year_decision);
+  Finding: Sharp shift — Brennan court was 69% liberal, Gleeson dropped to 42%, then French/Kiefel hovered around 48-51%. The Gleeson era was distinctly conservative on
+  constitutional matters.
+
+  ---
+  4. Implied freedom of political communication timeline
+
+  SELECT year_decision, COUNT(*) AS cases,
+    SUM(CASE WHEN decision_direction='liberal' THEN 1 ELSE 0 END) AS liberal,
+    SUM(CASE WHEN decision_direction='conservative' THEN 1 ELSE 0 END) AS conservative
+  FROM cases WHERE primary_issue LIKE '%political communication%'
+  GROUP BY year_decision ORDER BY year_decision;
+  Finding: After the early liberal decisions (1996-97), the court became almost exclusively conservative on implied freedom cases — 2013-2019 saw 11 conservative outcomes
+   vs 2 liberal. The freedom has been heavily curtailed.
+
+  ---
+  5. Panel ideology vs liberal outcome (constitutional cases)
+
+  SELECT ROUND(proportion_liberal_panel, 1) AS panel_liberal_prop,
+    COUNT(*) AS cases,
+    ROUND(100.0 * SUM(CASE WHEN decision_direction='liberal' THEN 1 ELSE 0 END)/COUNT(*),1) AS liberal_outcome_pct
+  FROM cases WHERE primary_issue_sub_area = 'Constitutional law'
+  GROUP BY ROUND(proportion_liberal_panel, 1);
+  Finding: There's a correlation but it's noisy — panels with 70%+ liberal justices deliver 72% liberal outcomes. But a 30% liberal panel still delivers 50% liberal
+  outcomes, suggesting constitutional law constrains ideological voting.

scripts/build_insights_data.py

@@ -0,0 +1,206 @@
+"""
+Build constitutional insights data from hca.db for the Insights dashboard tab.
+"""
+import json
+import sqlite3
+from pathlib import Path
+
+DB_FILE = "data/processed/hca.db"
+OUTPUT = "website/public/insights.json"
+
+
+def strip_prefix(issue):
+    """Strip 'Public Law—Federal constitutional law—' prefix."""
+    for prefix in [
+        "Public Law—Federal constitutional law—",
+        "Public Law—Federal constitutional law— ",
+    ]:
+        if issue.startswith(prefix):
+            return issue[len(prefix):].strip()
+    return issue.strip()
+
+
+def get_insights_data():
+    conn = sqlite3.connect(DB_FILE)
+    conn.row_factory = sqlite3.Row
+    c = conn.cursor()
+
+    # Unnamed cases (no hca_citation) are unmatched sub-matters of multi-matter decisions
+    # that are already represented by the named lead case. Exclude them from all stats
+    # so we count distinct decisions rather than inflating counts with duplicate sub-matters.
+    CITED = "hca_citation IS NOT NULL AND hca_citation != ''"
+
+    # 1. Vote split distribution on constitutional cases
+    c.execute(f"""
+        SELECT
+            maj_votes || '-' || min_votes AS split,
+            COUNT(*) AS count
+        FROM cases
+        WHERE primary_issue_sub_area = 'Constitutional law'
+          AND {CITED}
+        GROUP BY maj_votes, min_votes
+        ORDER BY (maj_votes + min_votes) DESC, maj_votes DESC
+    """)
+    vote_splits = [{'split': r['split'], 'count': r['count']} for r in c.fetchall()]
+
+    # 2. Topic breakdown table
+    c.execute(f"""
+        SELECT
+            primary_issue AS topic,
+            COUNT(*) AS cases,
+            ROUND(100.0 * SUM(CASE WHEN min_votes = 0 THEN 1 ELSE 0 END)/COUNT(*), 1) AS unanimous_pct,
+            ROUND(100.0 * SUM(CASE WHEN min_votes >= 2 THEN 1 ELSE 0 END)/COUNT(*), 1) AS contested_pct,
+            ROUND(100.0 * SUM(CASE WHEN decision_direction='liberal' THEN 1 ELSE 0 END)/COUNT(*), 1) AS liberal_pct,
+            ROUND(100.0 * SUM(CASE WHEN decision_direction='conservative' THEN 1 ELSE 0 END)/COUNT(*), 1) AS conservative_pct,
+            ROUND(100.0 * SUM(CASE WHEN party_winning LIKE 'appealing%' THEN 1 ELSE 0 END)/COUNT(*), 1) AS appellant_win_pct,
+            ROUND(AVG(CAST(maj_votes AS REAL) / (maj_votes + min_votes)), 2) AS avg_majority_share
+        FROM cases
+        WHERE primary_issue_sub_area = 'Constitutional law'
+          AND {CITED}
+        GROUP BY primary_issue
+        HAVING cases >= 3
+        ORDER BY cases DESC
+    """)
+    topic_breakdown = []
+    for r in c.fetchall():
+        topic_breakdown.append({
+            'topic': strip_prefix(r['topic']),
+            'cases': r['cases'],
+            'unanimous_pct': r['unanimous_pct'],
+            'contested_pct': r['contested_pct'],
+            'liberal_pct': r['liberal_pct'],
+            'conservative_pct': r['conservative_pct'],
+            'appellant_win_pct': r['appellant_win_pct'],
+            'avg_majority_share': r['avg_majority_share'],
+        })
+
+    # 3. Decision direction by Chief Justice era (constitutional cases only)
+    c.execute(f"""
+        SELECT
+            chief_argument AS chief,
+            COUNT(*) AS cases,
+            ROUND(100.0 * SUM(CASE WHEN decision_direction='liberal' THEN 1 ELSE 0 END)/COUNT(*), 1) AS liberal_pct,
+            ROUND(100.0 * SUM(CASE WHEN decision_direction='conservative' THEN 1 ELSE 0 END)/COUNT(*), 1) AS conservative_pct,
+            ROUND(100.0 * SUM(CASE WHEN decision_direction='unspecifiable' THEN 1 ELSE 0 END)/COUNT(*), 1) AS unspecifiable_pct,
+            ROUND(100.0 * SUM(CASE WHEN party_winning LIKE 'appealing%' THEN 1 ELSE 0 END)/COUNT(*), 1) AS appellant_win_pct,
+            MIN(year_decision) AS from_year,
+            MAX(year_decision) AS to_year
+        FROM cases
+        WHERE primary_issue_sub_area = 'Constitutional law'
+          AND chief_argument IS NOT NULL
+          AND {CITED}
+        GROUP BY chief_argument
+        ORDER BY MIN(year_decision)
+    """)
+    direction_by_era = [dict(r) for r in c.fetchall()]
+
+    # 4. Implied freedom of political communication — year-by-year
+    c.execute(f"""
+        SELECT
+            year_decision AS year,
+            COUNT(*) AS cases,
+            SUM(CASE WHEN decision_direction='liberal' THEN 1 ELSE 0 END) AS liberal,
+            SUM(CASE WHEN decision_direction='conservative' THEN 1 ELSE 0 END) AS conservative,
+            SUM(CASE WHEN decision_direction='unspecifiable' THEN 1 ELSE 0 END) AS unspecifiable,
+            GROUP_CONCAT(
+                CASE WHEN case_name IS NOT NULL AND case_name != ''
+                     THEN case_name
+                     ELSE hca_citation
+                END,
+                ' | '
+            ) AS case_names,
+            GROUP_CONCAT(hca_citation, ' | ') AS citations,
+            GROUP_CONCAT(maj_votes || '-' || min_votes, ' | ') AS vote_splits,
+            GROUP_CONCAT(decision_direction, ' | ') AS directions
+        FROM cases
+        WHERE primary_issue LIKE '%political communication%'
+          AND {CITED}
+        GROUP BY year_decision
+        ORDER BY year_decision
+    """)
+    implied_freedom = []
+    for r in c.fetchall():
+        implied_freedom.append({
+            'year': r['year'],
+            'cases': r['cases'],
+            'liberal': r['liberal'],
+            'conservative': r['conservative'],
+            'unspecifiable': r['unspecifiable'],
+            'case_names': r['case_names'].split(' | ') if r['case_names'] else [],
+            'citations': r['citations'].split(' | ') if r['citations'] else [],
+            'vote_splits': r['vote_splits'].split(' | ') if r['vote_splits'] else [],
+            'directions': r['directions'].split(' | ') if r['directions'] else [],
+        })
+
+    # 5. Corporations power case list
+    c.execute(f"""
+        SELECT
+            year_decision AS year,
+            CASE WHEN case_name IS NOT NULL AND case_name != '' THEN case_name ELSE hca_citation END AS name,
+            hca_citation AS citation,
+            maj_votes,
+            min_votes,
+            decision_direction AS direction,
+            party_winning
+        FROM cases
+        WHERE primary_issue LIKE '%Corporations power%'
+          AND {CITED}
+        ORDER BY year_decision
+    """)
+    corporations_cases = [dict(r) for r in c.fetchall()]
+
+    # Summary stats for callout cards
+    c.execute(f"SELECT COUNT(*) AS n FROM cases WHERE primary_issue_sub_area='Constitutional law' AND {CITED}")
+    total_const = c.fetchone()['n']
+
+    c.execute(f"SELECT COUNT(*) AS n FROM cases WHERE primary_issue_sub_area='Constitutional law' AND min_votes=0 AND {CITED}")
+    unanimous_const = c.fetchone()['n']
+
+    c.execute(f"SELECT COUNT(*) AS n FROM cases WHERE primary_issue LIKE '%political communication%' AND decision_direction='conservative' AND {CITED}")
+    freedom_conservative = c.fetchone()['n']
+
+    c.execute(f"SELECT COUNT(*) AS n FROM cases WHERE primary_issue LIKE '%political communication%' AND {CITED}")
+    freedom_total = c.fetchone()['n']
+
+    c.execute(f"""
+        SELECT ROUND(100.0*SUM(CASE WHEN min_votes>=2 THEN 1 ELSE 0 END)/COUNT(*),1) AS pct
+        FROM cases WHERE primary_issue LIKE '%Corporations power%' AND {CITED}
+    """)
+    corps_contested = c.fetchone()['pct']
+
+    conn.close()
+
+    return {
+        'summary': {
+            'total_constitutional_cases': total_const,
+            'unanimous_pct': round(100.0 * unanimous_const / total_const, 1),
+            'freedom_total': freedom_total,
+            'freedom_conservative_pct': round(100.0 * freedom_conservative / freedom_total, 1),
+            'corporations_contested_pct': corps_contested,
+        },
+        'vote_splits': vote_splits,
+        'topic_breakdown': topic_breakdown,
+        'direction_by_era': direction_by_era,
+        'implied_freedom': implied_freedom,
+        'corporations_cases': corporations_cases,
+    }
+
+
+def main():
+    print("Building insights data...")
+    data = get_insights_data()
+
+    Path(OUTPUT).parent.mkdir(parents=True, exist_ok=True)
+    with open(OUTPUT, 'w') as f:
+        json.dump(data, f, indent=2)
+
+    print(f"✅ Wrote {OUTPUT}")
+    s = data['summary']
+    print(f"   {s['total_constitutional_cases']} constitutional cases")
+    print(f"   {s['unanimous_pct']}% unanimous")
+    print(f"   {len(data['topic_breakdown'])} topics")
+    print(f"   {len(data['implied_freedom'])} implied freedom data points")
+
+
+if __name__ == "__main__":
+    main()