South Dakota: Rounds ran well ahead of the selected poll
Emerson College Polling put Rounds 8 points ahead. NBC’s November 5 evening snapshot has a 21.00-point lead, with 99% estimated vote in.
What the count shows
Mike Rounds (R, SD Senate); Brian Bengs (I, SD Senate). NBC News projects Mike Rounds as the winner. In the snapshot checked November 5 at 8:48 p.m. Eastern, Rounds had 184,588 votes to 120,509 for Brian Bengs. The reported lead was 64,079 votes, or 21.00 percentage points across all candidates.
The selected Emerson College Polling poll had a 8-point Rounds lead. The reported margin is 13.00 points larger than that published poll margin. The projected winner matches the poll’s leading candidate; the size of the lead is a separate test.
| Candidate | Source party label | Votes | Share of all votes |
|---|---|---|---|
| Mike Rounds | Rep | 184,588 | 60.5014% |
| Brian Bengs | Ind | 120,509 | 39.4986% |
| Total | 305,097 | 100% |
| Measure | Margin in percentage points |
|---|---|
| Selected original poll | 8 |
| All reported candidates | 21.002829 |
| Leading two candidates only | 21.002829 |
What makes this comparison useful
Brian Bengs ran as an independent. Calling this a Republican-versus-Democratic polling error would mislabel the race. Emerson’s initial 50% to 42% question and its 54% to 46% leaner version both gave Rounds an eight-point lead. The reported count is wider under either version.
Keep the original poll intact
Emerson surveyed 803 likely South Dakota voters October 19–21 and reports a ±3.5-point credibility interval. The initial ballot left 8% undecided. The later leaner allocation is a follow-up within the same survey, not a new poll.
The selected survey’s original ballot and methods are reproduced below. It is one published poll, not a polling average or a claim that no later poll existed. Undecided responses, minor candidates, the voter screen and the field dates remain part of the record.
The survey’s stated uncertainty applies to its estimates under its method. It is not an automatic pass-or-fail cutoff for a difference between two candidates, and it cannot turn a provisional count into a final polling-error score.
| Candidate or response | Published percent |
|---|---|
| Mike Rounds | 50 |
| Brian Bengs | 42 |
| Undecided | 8 |
| Stage | Mike Rounds | Brian Bengs | Lead, points |
|---|---|---|---|
| Initial ballot | 50 | 42 | 8 |
| With undecided leaners | 54 | 46 | 8 |
The denominator check
Using all 305,097 reported candidate votes produces the 21.00-point lead. Restricting the denominator to Mike Rounds and Brian Bengs produces 21.00 points. Both calculations are shown so readers can see the effect rather than confuse the two measures.
For a fully two-candidate poll comparison, the published poll shares must also be renormalized. It would be misleading to change only the result denominator and describe that as a like-for-like final error.
Where the evidence stops
NBC estimated 99% of the vote in. This is a dated reported-count snapshot, not a certified final result. The national senate payload was updated Fri, 06 Nov 2026 01:48:30 GMT. That is the source update clock, not the original race-call time or a timestamp for an individual ballot batch.
A final polling audit should repeat the same calculations after the outcome is stable, preserving the original pre-election poll. This article does not change 21CEE’s locked forecast or claim that a polling gap proves which voters switched parties or stayed home.
Sources and calculations
Reporting: the linked original publishers. Calculations and graphics: 21CEE. Poll shares are kept as published; result margins use the exact candidate counts in the accessible table. The source clocks above describe the retained snapshot. No certification or original call time is inferred.