South Carolina was not on The Cook Political Report’s Toss Up list. It still is not. But a state that Cook classified as Solid Republican only a month ago has now moved twice toward the competitive end of the Senate map: first to Likely Republican on September 23, then to Lean Republican on October 20.
That distinction matters as Republican Sen. Darline Graham faces Democratic pediatrician Annie Andrews in the November 3 election. Several recent polls show margins of one to four points. The question is not whether Cook has called the race a coin flip—it has not—but what the shift in its rating says about the evidence analysts are now weighing.
Two Cook changes in four weeks, neither a Toss Up
The chronology is unusually clear. On September 23, Cook moved South Carolina from Solid Republican to Likely Republican, a change its Senate editor Jessica Taylor described in a September 24 election outlook. On October 20, Cook made another adjustment: Likely Republican to Lean Republican. Its October 20 Senate analysis and public announcement confirmed the second move.
The October 20 Cook ratings list places South Carolina in Lean Republican, separate from its seven Toss Up races. The official South Carolina race page also gives the state’s Cook Partisan Voting Index as R+8, a measure of its Republican-leaning political baseline. Cook’s ratings are expert judgments of competitiveness—not polling figures or guarantees of an election result.
Cook’s published explanation focuses partly on candidate-specific factors alongside the national political environment. Graham was appointed to the seat after the death of her brother, Lindsey Graham, and subsequently won the Republican primary. Andrews is attempting to contest a seat in a state where Republicans have maintained a long record of success in federal statewide elections.
What six public polls actually show
Rather than treat every fresh headline as a turning point, it helps to compare published surveys, their fieldwork periods and the gap in each one. The table below shows six polling waves from September and October. It uses voting-intention percentages as published, not normalized two-candidate shares.
| Fieldwork | Pollster | Graham (R) | Andrews (D) | Margin |
|---|---|---|---|---|
| Oct 16–18 | Trafalgar Group | 48.6% | 45.6% | Graham +3.0 |
| Oct 5–8 | Rasmussen Reports | 49% | 45% | Graham +4 |
| Sep 28–29 | InsiderAdvantage | 46% | 45% | Graham +1 |
| Sep 23–25 | Trafalgar Group | 43.4% | 42.3% | Graham +1.1 |
| Sep 14 | Rasmussen Reports | 48% | 43% | Graham +5 |
| Sep 8–9 | InsiderAdvantage | 45.1% | 43.2% | Graham +1.9 |
Sources: Trafalgar October release, Rasmussen, InsiderAdvantage, and Call the Map’s sourced poll archive. Dates refer to interviewing, not necessarily the date each survey was published. Percentages rounded only when reported that way by the source.
These surveys repeatedly put Graham numerically ahead, but not by the kind of margin that the state’s Republican presidential voting history alone might suggest. Crucially, they do not show one consistent late swing: within Rasmussen, the Republican margin was five points in September and four in October; within Trafalgar it rose from about one point in late September to three in mid-October. InsiderAdvantage’s September waves moved from a roughly two-point Graham lead to one point.
The October RealClearPolling average—based on the latest Trafalgar, Rasmussen and InsiderAdvantage polls with fieldwork between September 28 and October 18—stands at 48.0% Graham to 45.3% Andrews, a 2.7-point numerical difference. That is an aggregation of selected polls, not a Cook rating and not a prediction of the result.
Why the rating changed despite a Republican polling lead
The reason these facts can coexist is that race ratings and polling averages answer different questions. A poll captures answers from a sampled electorate at a particular time. A rating incorporates polling but also considers candidate performance, campaign organisation, partisan history, fundraising and the broader environment.
For context, Cook’s R+8 partisan index still describes South Carolina as a Republican-leaning state. The recent single-digit polling gaps are therefore noteworthy relative to its broader electoral history. But the margin alone cannot establish a change in the underlying electorate, especially when surveys come from different firms and use different collection methods.
Pollster affiliations also warrant transparency. Some surveys in this race have been conducted or sponsored by organisations identified with a party. Pollsters also differ on turnout screens, likely-voter definitions and how third-party candidates or undecided voters are presented. The six-wave comparison is not a probability model and should not be read as six consecutive daily tracking measurements.
Other survey estimates—including campaign-sponsored releases earlier in the year—have shown closer or different results. Votendi has limited the table above to a traceable group of published September and October toplines with identifiable fieldwork periods; a reported additional On Point Polling measurement is not used here because publicly circulating sample descriptions were not consistent enough to treat it as independently verified.
Why the final campaign is attracting national attention
The national parties’ behaviour adds context, not a substitute for polling. President Donald Trump was scheduled to campaign in South Carolina for Graham on October 21, while Republican organisations have directed new spending into the state, according to Reuters. Associated Press has also reported heightened spending and national interest in the contest.
Those developments reflect the attention campaign organisations are giving the race. They do not prove which candidate has gained or lost support. Turnout and any new high-quality surveys will help explain whether the close margins documented in September and October persist through the final days.
What to watch before November 3
The central question is whether future comparable surveys keep the contest within a few points, or whether different pollsters begin to show a more consistent direction. Analysts will also examine turnout among habitual Republican voters, Democratic voters and independents rather than extrapolating directly from national generic-ballot numbers.
For now, the verified description is precise: South Carolina is rated Lean Republican by Cook, not Toss Up. Cook’s movement from Solid R to Likely R to Lean R is newsworthy because it documents a change in the analyst’s assessment. The published polls help explain the closer scrutiny; they do not settle the outcome.
Data and ratings checked October 21, 2026. Election day: November 3, 2026. Polls measure sampled voting intention, not official election results. Cook Political Report ratings are an external analytical classification and are not Votendi forecasts.
More from Votendi: Why the national generic ballot differs from individual Senate contests · 2026 U.S. midterms election hub.
