Finding — The postal-vote anti-TVK gap of 10.6 percentage points
The institutional electorate (armed forces, govt employees, elderly/disabled postal voters) preferred DMK over TVK by 5.3 pp, while the general EVM electorate preferred TVK over DMK by 10.9 pp. The TVK wave was almost entirely an EVM phenomenon.
The data — TN 2026 statewide postal vs EVM
| Party | % of EVM votes | % of postal votes | Postal − EVM |
|---|---|---|---|
| TVK | 35.04% | 24.46% | −10.58 pp |
| DMK | 24.13% | 29.43% | +5.30 pp |
| AIADMK | 21.20% | 22.26% | +1.06 |
| NTK | 3.98% | 5.38% | +1.40 |
| INC | 3.36% | 4.43% | +1.07 |
| BJP | 2.97% | 3.66% | +0.69 |
| PMK | 2.17% | 2.54% | +0.37 |
| DMDK | 1.19% | 1.43% | +0.24 |
Postal votes are only 1.06% of total (521,903 out of 49,324,121) but their composition is demographically distinctive: armed forces, govt employees, polling-day duty staff, elderly (80+) voters, persons with disabilities, and overseas. This is the closest TN gets to a stratified institutional electorate.
What this says
The headline
TVK got 10.6 pp less of postal votes than EVM votes. Conversely DMK got 5.3 pp more. Both gaps are larger than for any other major party. The institutional/older electorate was demonstrably less TVK-leaning than the general electorate.
If you took only the postal electorate as a proxy for "voters who couldn't easily vote in person — armed forces, govt employees, elderly, disabled", DMK would actually have come close to or beaten TVK statewide.
This is consistent with — and quantifies — the qualitative analyst commentary that "TVK's appeal skews young, urban, and male; older and institutional voters were less swayed." Yashwant Deshmukh (The Federal): "As you move from urban to rural, his votes come down." That hold-out segment shows up cleanly in postal-vs-EVM here.
Why this matters
- DMK's losses were not uniform. Where the electorate skewed institutional/older, DMK retained share. The youth-wave thesis isn't decoration — it's measurable.
- TVK's mandate has a generational bias. If postal voters were the only electorate, the result would be DMK ≈ TVK. The 108-seat TVK plurality was driven by under-60, in-person, EVM-casting voters.
- Future elections: if TVK's young/urban base ages into postal status (overseas voters, eventually elderly), how does the coalition hold?
Caveat — 1.06% of the electorate
This is not a population-weighted statement. Postal is 1% of votes. Within that 1%, TVK ran 10.6 pp behind their EVM share. We don't have the absolute size to claim postal changed the outcome anywhere. What we can say:
- The composition of the postal electorate is meaningfully different from the EVM electorate (older, more institutional).
- Within that segment, TVK underperformed and DMK overperformed.
- The magnitude (10.6 pp) is too large to be noise even at this sample size.
What we still can't see
- Postal vs EVM per AC — interesting but messy because per-AC postal counts are small (often <1000). Could surface "DMK won AC X on postal but lost on EVM" type patterns.
- Age-cohort exit polling — Lokniti will eventually publish; not yet available.
- Booth-level patterns within urban metros — would let us see if "young urban" is a real cohort or a stand-in for "Chennai".
Cross-check
All numbers from s3://tnelection2026/results/curated/year=2026/ac=*/candidate_totals.parquet. Compute via pipelines/deep_dive.py:
df.group_by('party_norm').agg(
pl.col('evm_votes').sum().alias('evm_total'),
pl.col('postal_votes').sum().alias('postal_total'),
)