Evidence review · Media: 30 September–6 October 2026 IST · election baseline: May 2026
TVK, DMK, ADMK and NTK: what Tamil Nadu's numbers show
TVK has the strongest observed position among these four parties: the largest May seat tally and the most party-naming videos in our latest completed week. DMK is closer in news coverage, while ADMK retains a substantial electoral base. NTK's smaller presence needs careful reading. The evidence supports those comparisons; it does not yet support odds for the next election.
The result and the next contest are different questions
Tamil Nadu voted on 23 April, and counting took place on 4 May. Vijay was sworn in as Chief Minister on 10 May. This October assessment therefore starts with an election result, then asks what the coverage recorded since then adds to our understanding.
The state government's gazette scheduled by-elections in Madurantakam and Dharapuram for 6 October. The May figures below describe the general election, not an updated count of today's Assembly membership or a call on those separate contests. Our media window ends at midnight after 6 October, India time. This review compares four parties, TVK, DMK, ADMK and NTK. It does not cover BJP, Congress, PMK or other parties, which TamilTick tracks on their own pages.
TVK's May lead was clear, but short of a standalone majority
The Election Commission records 108 seats and 34.92% of the vote for TVK. That put it 49 seats ahead of DMK and 61 ahead of ADMK. It was ten seats below 118, the arithmetic majority in a full 234-seat Assembly. These numbers establish the scale of the result; later support arrangements require separate evidence.
DMK's 24.19% and ADMK's 21.21% show why both remain substantial comparators. The gap between them was 2.98 percentage points and 12 seats. NTK received almost two million votes, or 4.00%, without winning a seat. A statewide vote total and a seat total measure different things in a constituency election.
May 2026 general-election results
| Party | Votes | Vote share | Seats |
|---|---|---|---|
| TVK | 17,226,209 | 34.92% | 108 |
| DMK | 11,929,144 | 24.19% | 59 |
| ADMK / AIADMK | 10,462,146 | 21.21% | 47 |
| NTK | 1,972,537 | 4.00% | 0 |
Source: ECI May 2026 party results and published vote shares, checked 7 October. Party totals, not alliance totals. Other parties won the remaining 20 seats.
The vote-share baseline
Official May vote shares. These bars are not October polling or future win probabilities.
The margins reveal a competitive map beneath the totals
Across the 234 constituency results held in our ECI-derived archive, 103 were decided by five percentage points or less; 57 of those were within two points. A close result describes the previous contest. It does not mean the seat will flip.
TVK's median winning margin was 8.36 points, compared with 3.77 for DMK and 4.96 for ADMK. TVK also finished second in 74 constituencies; DMK did so in 80 and ADMK in 44. NTK was neither winner nor runner-up in this result set. These are more useful descriptions of the May contest than assuming that a statewide percentage will translate uniformly into seats.
| Party | Won | Second | Wins within 5 pp | Median winning margin |
|---|---|---|---|---|
| TVK | 108 | 74 | 31 | 8.36 pp |
| DMK | 59 | 80 | 36 | 3.77 pp |
| ADMK | 47 | 44 | 24 | 4.96 pp |
| NTK | 0 | 0 | 0 | Not applicable |
Derived from all 234 ECI constituency records saved on 28 September. Margin in percentage points = 100 × (winner minus runner-up votes) divided by all votes including NOTA. Five-point counts include the two-point contests.
TVK's coverage advantage is larger on YouTube than in news
For 30 September through 6 October, TamilTick's stored headline matches identify 2,763 distinct retained videos naming TVK. DMK appears in 1,316, ADMK in 710 and NTK in 191. TVK's count is about 2.1 times DMK's. These are videos from media coverage, not a count of unique people watching, and not a measure of approval.
The news comparison is less lopsided: TVK appears in 468 distinct archived reports and DMK in 374, a ratio of about 1.25 to one. ADMK appears in 175 and NTK in 21. Those totals must be read with the collection record: four of the seven news days are marked partial, and daily outlet coverage ranges from 28 to 75. The video archive passes our internal coverage checks on all seven days, but it is still a selected source panel.
Videos whose headlines name each party
30 September–6 October IST. A video can name more than one party. Official party channels, film clips and duplicate uploads are excluded by the existing headline pipeline; classification can still miss or misidentify names.
The separate news record
| Party | Distinct reports | Stored outlet identities |
|---|---|---|
| TVK | 468 | 42 |
| DMK | 374 | 38 |
| ADMK | 175 | 27 |
| NTK | 21 | 11 |
Same publication window; four news days have partial coverage. There are 905 distinct reports naming any tracked party. Rows overlap when one report names multiple parties; they must not be summed.
A large lead is not the same as an accelerating lead
TVK's video count rose from 2,444 in 23–29 September to 2,763 in 30 September–6 October. But the number of distinct videos naming any tracked party also rose, from 3,616 to 4,124. TVK therefore appeared in 67.6% of that pool in the earlier week and 67.0% in the later week. Within this archive, the sensible description is a video-coverage lead that held across both weeks, not evidence of a fresh surge in voter support. In news the lead is newer: in 23–29 September DMK had more distinct reports than TVK, 388 to 378.
The leader comparison points in the same direction without being interchangeable with party coverage. Vijay appears in 3,636 videos, M. K. Stalin in 799, Edappadi K. Palaniswami in 299 and Seeman in 191 in the completed week. Udhayanidhi Stalin appears in another 200, but those must not simply be added to M. K. Stalin's total: the same video can name both. Vijay's roles as Chief Minister, party leader and public figure also prevent us from treating every mention as a TVK endorsement.
67.0% means 2,763 of 4,124 stored videos naming at least one tracked party also name TVK. It is not a share of Tamil Nadu voters, all YouTube videos, unique viewers or positive sentiment. Names can overlap, so party percentages need not sum to 100%.
What the evidence says about each party
TVK: the strongest observed position. The May vote lead, 108-seat result, wider median winning margin and current coverage lead all point in that direction. They support taking TVK seriously as the leading force in this comparison. They cannot tell us whether its support has risen since May, or whether the next constituency contest will follow the statewide pattern.
DMK: the closest comparator here. It was ahead of ADMK in both May votes and seats and is ahead in both recorded media types this week. Its news presence is considerably closer to TVK's than its YouTube presence. Its 80 runner-up finishes also show how much competition the statewide winning-seat tally alone leaves out. None of this is a measured recovery in voting intention.
ADMK, formally AIADMK: a smaller attention count than DMK does not erase 10.46 million votes and 47 May wins. Its current recorded video coverage is also spread across 34 channels, almost the same breadth as DMK's 35, despite fewer videos. Volume and source breadth answer different questions. Alliance choices, candidates and local contests cannot be inferred from those counts.
NTK: electoral support with weak seat conversion and a smaller recorded media footprint. Its 4.00% May vote share is a real electoral result; zero seats does not mean zero support. Its 191 party-naming videos and 21 news reports are much smaller samples. Alias recognition and uneven source coverage matter particularly at that scale. They provide no basis for declaring its voters gone or transferring their support to another party.
What we have in the archive, and what is still missing
Our 7 October census holds 169,835 stored video records from 186 channel identities and 108,916 news rows, representing 100,709 distinct stored article keys. The publication dates in both archives run from 1 June to 7 October in India time. Archive size is not the size of a representative political sample. The weekly comparisons above use the separate retained headline-mention records.
Neither archive contains material published before the April poll. Upstream historical recovery does hold older material, but it has not produced a balanced, analysis-ready campaign panel: none of our original ten requested news channels passes the upstream readiness gate. TVK's official campaign channel is ready while comparison channels remain unevenly processed. Comparing only the ready records would build that collection imbalance into the conclusion.
We can describe the result declared in May and what our sources covered later. We cannot use October view counters or retrospectively collected campaign clips to reconstruct what voters had seen before polling, or to establish why Vijay won.
The forecast status on 7 October
TamilTick does not currently have a validated four-party electoral forecast. The repository's scenario calculations have not been shown to produce calibrated win probabilities on elections held out from development. There is therefore no defensible percentage chance of victory to publish for TVK, DMK, ADMK or NTK from this system today.
The evidence-based assessment is narrower and still useful: TVK leads the May electoral baseline and this week's recorded media attention; DMK is the closest of these rivals on both measures; ADMK retains a substantial electoral base; and NTK's statewide votes have not translated into May seats. A future forecast needs a named contest, current candidate and alliance information, representative voter evidence and validation against results. Media attention can supplement that work, but cannot replace it.
A record to return to
This article freezes the evidence available on 7 October. A later review should compare the next official results, changes measured on a stable source panel and any new representative polling against this dated baseline. The snapshot and its limits should remain visible even if the political picture changes.
How this was measured
- ECI May 2026 party seats and vote shares were checked on 7 October. Constituency margins derive from the 234-seat ECI archive saved on 28 September; the margin in percentage points is 100 × (winner minus runner-up votes) divided by total votes including NOTA.
- Media publication window: 30 September 00:00 through 7 October 00:00 IST, end exclusive. Comparison window: 23 September through 30 September 00:00 IST. The read-only database snapshot was captured at 08:46 IST on 7 October.
- A news count is a distinct stored article key, with item ID as fallback, per entity over the whole week. A video count is a distinct retained video ID. Source counts are distinct stored outlet/channel identities within each media type.
- The stored matcher reads headlines, including available translated news titles. It does not count video-description boilerplate. The existing pipeline excludes bulletins, identified duplicates, film clips and official-party video channels.
- Party and leader matches are separate and nonexclusive. The video-attention fractions use all distinct retained videos naming any tracked party, not only these four. News and YouTube are never added into a single support score.
- The downloadable snapshot preserves counts, dates, source coverage, calculation definitions and forecast status. The cover is an AI-generated editorial illustration; all charts use the recorded numerical evidence.
What these numbers cannot tell you
- This is post-election coverage, not a forecast issued before the May result. May party results are not current Assembly membership or alliance totals.
- Four of seven news days have partial coverage. Video coverage checks passing do not establish that every publisher or political video was observed.
- The headline matcher can miss names or produce false matches. Low counts, especially for NTK, must not be interpreted as absence of electoral support.
- Multiple outlets or clips can report one event. A mention need not be favourable; views need not be unique people, residents or voters.
- No representative voting-intention survey, causal estimate or calibrated party win probability is established by these records.
Referenced in this analysis
Sources
- Frozen data and methodology · 7 October 2026 (JSON)
- Election Commission · May 2026 Tamil Nadu party results
- Election Commission · May 2026 Tamil Nadu vote shares
- District government · 2026 polling and counting schedule
- Lok Bhavan · Vijay sworn in on 10 May 2026
- Tamil Nadu Government Gazette · 2026 by-election notifications
- Earlier fixed snapshot · Vijay's Dharapuram YouTube coverage, 30 September
Observation window: Media: 30 September–6 October 2026 IST · election baseline: May 2026. Each figure uses the date and measure stated alongside it. Media coverage and engagement are not measures of voter support.