School Board Competitiveness

Question: Which school board races were the most competitive in 2022, and how many were uncontested?

Method

Filter L4 flat export to contests where office_level is school_district or the contest name matches school board keywords. Aggregate precinct-level results to contest-level totals. Compute margins and uncontested rates.

The Query

jq — filter to school board contests

cat flat_export.jsonl \
  | jq -c 'select(.contest_name | test("school board|board of education|school district|school trustee"; "i"))' \
  | jq -r '"\(.state)\t\(.county)\t\(.contest_name)\t\(.candidate_canonical)\t\(.candidate_entity_id)\t\(.votes_total)"' \
  | sort -u \
  > school_board_candidates.tsv

Python — full analysis

import json
from collections import defaultdict

contests = defaultdict(lambda: defaultdict(int))
vote_for = {}

school_keywords = ["school board", "board of education", "school district", "school trustee",
                   "board of ed", "school committee", "school director"]

with open("flat_export.jsonl") as f:
    for line in f:
        r = json.loads(line)
        contest = r.get("contest_name", "")
        if not any(kw in contest.lower() for kw in school_keywords):
            continue
        if "write" in r.get("candidate_canonical", "").lower():
            continue
        key = (r["state"], r.get("county", ""), contest)
        contests[key][r["candidate_canonical"]] += r["votes_total"]
        if key not in vote_for:
            vote_for[key] = r.get("contest", {}).get("vote_for", 1) or 1

# Compute margins
results = []
uncontested = 0
for key, candidates in contests.items():
    state, county, contest_name = key
    n = vote_for.get(key, 1)
    ranked = sorted(candidates.items(), key=lambda x: -x[1])

    if len(ranked) <= n:
        uncontested += 1
        continue

    # Margin between last winner (Nth) and first loser (N+1th)
    last_winner = ranked[n - 1]
    first_loser = ranked[n]
    margin = last_winner[1] - first_loser[1]

    results.append({
        "state": state, "county": county, "contest": contest_name,
        "last_winner": last_winner[0], "last_winner_votes": last_winner[1],
        "first_loser": first_loser[0], "first_loser_votes": first_loser[1],
        "margin": margin, "candidates": len(ranked), "seats": n,
    })

results.sort(key=lambda x: x["margin"])

total = len(contests)
print(f"School board races: {total}")
print(f"Uncontested: {uncontested} ({100*uncontested/total:.1f}%)")
print(f"Contested: {len(results)}")
print(f"\nClosest 15:")
for r in results[:15]:
    seats_note = f" (vote for {r['seats']})" if r["seats"] > 1 else ""
    print(f"  {r['margin']:>5} votes  {r['state']} {r['county']}: {r['contest']}{seats_note}")
    print(f"             {r['last_winner']} ({r['last_winner_votes']:,}) vs {r['first_loser']} ({r['first_loser_votes']:,})")

Results

The Closest School Board Races

Dawson County, Georgia — The Exact Tie

The most striking result in the entire dataset: Dawson County, Georgia’s Board of Education race, a “vote for 3” contest with 6 candidates. The top two candidates each received 25,186 votes — an exact tie.

Because this is a multi-seat contest, the tie occurs between co-winners. Both tied candidates were elected. The meaningful margin — between 3rd place (24,901 votes) and 4th place (24,844 votes) — is 57 votes. The 4th-place candidate, who lost, was 57 votes away from winning a seat.

This illustrates why the vote_for field matters. A naive 1st-vs-2nd margin reports “0 votes” — technically true but misleading. The actual competitive margin is 57 votes at the win/lose boundary.

The 30.8% Uncontested Rate

30.8% of school board races nationally were uncontested in 2022 — fewer candidates filed than seats available.

This is lower than the overall local race uncontested rate of 48.8%, making school boards one of the more competitive local office types. Only city council (10% uncontested) is more consistently contested.

By State (Selected)

School board uncontested rates vary significantly:

Minnesota’s high rate (71.4%) reflects the same pattern seen in its overall uncontested rate — many small school districts in rural areas where recruiting candidates is difficult. California’s low rate (6.5%) reflects larger districts with more political activity and media coverage.

Multi-Seat Complications

School boards are disproportionately multi-seat contests. A “vote for 3” race with 4 candidates is technically contested, but only one seat is competitive. A “vote for 3” race with 3 candidates is uncontested even though it looks like it has plenty of names on the ballot.

The Python recipe above handles this correctly: a race is uncontested if len(candidates) <= vote_for. Margins are computed at the win/lose boundary (Nth place vs N+1th place), not between 1st and 2nd.

When vote_for is missing from the source data, the default is 1 (single-seat). This undercounts uncontested multi-seat races and overestimates competitiveness. The vote_for field is available in MEDSL for most states. NC SBE does not provide it — it must be inferred from contest name patterns like “VOTE FOR 3” or “ELECT TWO.”

Cross-References

• Closest Races in America — all office types, not just school boards
• Uncontested Race Rate — national uncontested analysis with full office-type breakdown
• Office Inventory — what school board districts exist in a given county