Finding — Where TVK lost, and what unites those districts
TVK won 108 of 234. The 126 they lost cluster into clear caste/regional pockets. Christians did not drive the TVK wave — the most-Christian district (Kanniyakumari, 47%) was a TVK shutout.
Hypothesis 1: "Christians voted for TVK"
This is testable from the data. Verdict: not supported.
| Cut | n ACs | Median Christian % | TVK median vote share | TVK win rate |
|---|---|---|---|---|
| Top decile (most Christian) | 29 | 11.5% | 34.4% | 41.4% |
| Bottom decile (least Christian) | 25 | 1.5% | 32.8% | 44.0% |
| Pearson r (Christian% × TVK%) | — | — | — | +0.017 |
TVK did slightly worse in Christian-heavy ACs. The correlation is statistically indistinguishable from zero.
The killer case: Kanniyakumari — TN's most-Christian district at 47% Christian — TVK won 0 of 6 ACs there. The hypothesis fails most loudly exactly where it should have been strongest.
Other Christian-heavy districts and TVK performance
| District | Christian % | TVK wins | Win rate |
|---|---|---|---|
| Kanniyakumari | 46.9% | 0/6 | 0% |
| Thoothukudi | 16.7% | 3/6 | 50% |
| The Nilgiris | 11.5% | 0/3 | 0% |
| Tirunelveli | 11.1% | 3/5 | 60% |
| Tiruchirappalli | 9.0% | 6/9 | 67% |
No pattern. Christian % does not predict TVK performance — caste dynamics, candidate quality, and alliance arithmetic dominate.
Hypothesis 2: TVK lost in specific caste/regional pockets
This is supported. Sixteen districts where TVK's win rate was 0% or under 30%:
| District | TVK wins | Win rate | Dominant community | Region |
|---|---|---|---|---|
| Thiruvarur | 0/4 | 0% | Kallar, Vellalar, Muslim | Cauvery delta |
| The Nilgiris | 0/3 | 0% | Badaga, Toda, mixed | Hill district |
| Villupuram | 0/7 | 0% | Vanniyar, SC | Northern |
| Tenkasi | 0/5 | 0% | Maravar, Nadar, Christian | Southern |
| Nagapattinam | 0/6 | 0% | Kallar, Vellalar, Muslim, Christian | Cauvery delta |
| Kanniyakumari | 0/6 | 0% | Nadar, Christian (heavy) | Southern tip |
| Cuddalore | 1/9 | 11% | Vanniyar, SC, Muslim | Eastern |
| Tiruvannamalai | 1/8 | 13% | Vanniyar | Northern |
| Dindigul | 1/7 | 14% | Thevar (Kallar), Naicker | Southern |
| Dharmapuri | 1/5 | 20% | Vanniyar, Kongu Vellala | Western |
| Tirupathur | 1/4 | 25% | Vanniyar, Muslim | Northern |
| Ramanathapuram | 1/4 | 25% | Maravar (Thevar), Muslim | Southern coastal |
| Karur | 1/4 | 25% | Kongu Vellala Gounder | Kongunadu |
| Thanjavur | 2/8 | 25% | Kallar, Vellalar | Cauvery delta |
| Kallakurichi | 1/4 | 25% | Vanniyar | Northern |
Five clear clusters emerge:
Cluster 1 — Cauvery Delta (DMK's historic heartland)
Thiruvarur, Nagapattinam, Thanjavur, Mayiladuthurai. TVK won 2 of 22 ACs here (9%). These are Kallar / Vellalar / mixed-caste delta districts where DMK has held majority since the 1960s. The TVK wave did not crack them. DMK's most concentrated remaining strength.
Cluster 2 — Vanniyar belt (PMK heartland + AIADMK NDA)
Villupuram, Cuddalore, Tiruvannamalai, Kallakurichi, Dharmapuri, Tirupathur. TVK won 4 of 41 ACs in these six districts (10%). Vanniyar voters in northern TN have a long association with PMK, which fought 2026 in the AIADMK-led NDA alliance. The Vanniyar bloc held against the TVK wave — caste-bloc voting alive and well.
Cluster 3 — Thevar/Mukkulathor belt (AIADMK historic)
Dindigul, Theni, Sivaganga, Madurai (partial), Ramanathapuram. TVK won 1 of 4 in Ramanathapuram; 1 of 7 in Dindigul. Thevar voters were historically AIADMK's southern fortress. They held — partially.
Cluster 4 — Kongunadu Gounder belt
Coimbatore, Erode, Tiruppur, Karur, Salem (partial). TVK results were mixed here — better than the other clusters but uneven. Karur 1/4 (25%) is the lowest in Kongu. The landed-industrial Gounder community had Republican alliances historically; TVK got some but not the bloc.
Cluster 5 — Christian-Nadar (Kanniyakumari)
Kanniyakumari is a distinctive 6-AC district. 47% Christian, dominated by Nadar (a backward-classified but socially-mobile community), historically a BJP-Congress-DMK battleground. TVK lost all 6 here. A combination of: (a) Vijay's appeal less resonant in southern-tip Christian-Nadar voter base, (b) Congress and DMK had strong local candidates, (c) BJP also had pockets.
What the data does NOT say
This analysis can describe the regional/caste pattern of TVK losses but cannot cause it from this data alone:
- We don't have per-AC caste composition. The dominant-community tags above are qualitative from TN political-geography common knowledge (Wikipedia, Lokniti analyses, academic literature on TN caste politics). They are not from a structured CSV.
- We can't say why Vanniyars stayed with PMK or Thevars with AIADMK — only that the geographic pattern of TVK losses correlates with these well-known community heartlands.
- A Lokniti-style post-poll with self-reported caste would convert this from "geographic association" to "individual-level vote choice."
The structural takeaway
TVK's wave was broad but not bloc-busting. Where caste-affinity political alliances were already deep (Vanniyar-PMK, Thevar-AIADMK, Cauvery delta-DMK), the TVK swing was sharply muted. Where no clear caste-bloc alternative existed (most urban centres, mixed-caste districts in the middle of the state), TVK won.
The interpretation that fits all the data:
- TVK's median statewide vote share (~34%) was achieved everywhere — the wave was broad.
- The conversion of vote share to seats was uneven — TVK won 46% of contests overall but 0% in 5 districts.
- The 0% districts share a property: a pre-existing caste-bloc alliance that voted as a bloc against TVK.
Cross-check
- Source:
s3://tnelection2026/results/curated/year=2026/ac=*/candidate_totals.parquet - District religion mix:
s3://tnelection2026/demographics/curated/district_religion.parquet - Computation:
pipelines/tvk_loss_map.py - Community overlay: hand-built table from TN political geography (file embeds source notes — explicitly marked as qualitative / analyst-knowledge, not from structured CSV)