Official Swedish Parliamentary Election Results Data Pipeline

1. Overview and Purpose

This pipeline acquires, parses, normalizes, and validates official historical results for Swedish parliamentary (Riksdag) general elections across the modern era:

• Elections Covered: 2002, 2006, 2010, 2014, 2018, 2022 (national Riksdag elections).
• Authoritative Source: Valmyndigheten (Swedish Election Authority) sole official source.
• Output Datasets:
  1. data/processed/elections/riksdag_election_results_source_parties.csv: Source-party-level table preserving all 265 original party lines across elections.
  2. data/processed/elections/riksdag_election_results.csv: Normalized canonical 10-party table (M, L, C, KD, S, V, MP, SD, FI, OTHER) with a fixed 10-row grid per election.

2. Authoritative Data Sources & Provenance

All raw files are stored without manual modification in data/raw/elections/ alongside a SHA-256 retrieval manifest (data/raw/elections/retrieval_manifest.json):

3. Canonical Party Codes & Historical Normalization

Canonical party schema:

Party Mapping Rules

• M: Moderaterna / Moderata Samlingspartiet.
• L: Liberalerna / Liberalerna (tidigare Folkpartiet) / Folkpartiet / Folkpartiet liberalerna / FP.
• C: Centerpartiet.
• KD: Kristdemokraterna / Kristdemokratiska Samhällspartiet / KDS.
• S: Socialdemokraterna / Arbetarepartiet-Socialdemokraterna / Sveriges Socialdemokratiska Arbetareparti.
• V: Vänsterpartiet / Vänsterpartiet Kommunisterna / VPK.
• MP: Miljöpartiet / Miljöpartiet de gröna.
• SD: Sverigedemokraterna.
• FI: Feministiskt initiativ / F!.
• OTHER: All other valid votes for parties not mapped to the 9 named categories.

Missing-Party Representation

If a canonical party did not participate in a given election (e.g. FI in 2002, prior to its founding in 2005), it is explicitly recorded with votes = 0, vote_share = 0.0, and source_vote_share = 0.0. Absence of official election votes is represented deterministically as zero rather than null.

4. Construction and Provenance of OTHER

To guarantee complete provenance and reconstructability:

1. Source Parties Table (riksdag_election_results_source_parties.csv): Contains every individual minor party line reported by Valmyndigheten (e.g. Piratpartiet, Medborgerlig Samling, Alternativ för Sverige, Partiet Nyans, Enhet, etc.).
2. Canonical Table (riksdag_election_results.csv): $\text{OTHER}$ is constructed as: \(\text{votes}_{\text{OTHER}} = \text{valid\_votes\_total} - \sum_{p \in \{\text{M, L, C, KD, S, V, MP, SD, FI}\}} \text{votes}_p\) This ensures that: \(\sum_{p \in \text{CANONICAL}} \text{votes}_p \equiv \text{valid\_votes\_total}\) with exactly 0 difference across every single election.

5. Vote-Share Conventions

Both calculated and published vote shares are stored:

• vote_share: Exact floating-point percentage calculated directly from integer vote counts: \(\text{vote\_share} = \frac{\text{votes}}{\text{valid\_votes\_total}} \times 100\)
• source_vote_share: Published rounded percentage from Valmyndigheten’s official report.
• Validation ensures that calculated shares agree with published shares within standard rounding tolerance ($\le 0.02\%$).

6. Refresh Commands

The election results pipeline is fully integrated into the project’s Makefile:

# Fetch raw documents from Valmyndigheten, normalize, and validate
make fetch-election-results

# Offline mode: rebuild processed datasets from cached data/raw/elections/ files
make process-election-results

# Run full automated unit test suite (including test_elections.py)
make test-pollofpolls