The Case for Calling Mississippi

1. Where the count stands

AP results feed²⁴ and Decision Desk HQ results feed,²⁵ read 6:45 PM CST Nov. 6.

The most complete count

The two feeds load county batches at different times. The "most complete" row takes, for each of the 82 counties, whichever of AP, DDHQ or the county's own published figures reports the most ballots. It differs from AP in five counties:

• a. Lowndes: DDHQ feed, batches of Nov. 4–5 and Nov. 6, 4:23 PM CST.²⁵ The combined affidavit total (Colom 342, Hyde-Smith 172, Pinkins 4) matches the county's affidavit records shown on video by Abiola Agoro, who obtained them in person (X, Nov. 6, 4:05 PM CST).²¹ The county reported "more than 550" affidavits cast (Deputy Circuit Clerk Ann Marie Higgins, Commercial Dispatch, Nov. 4).²¹
• b. Chickasaw: Mississippi Free Press, Nov. 6, 12:40 PM CST: 204 affidavits cast, 105 accepted, Colom 81, Hyde-Smith 23, Pinkins 1.¹⁶ DDHQ loaded the same batch at 4:01 PM CST.²⁵
• c. Washington: Delta Democrat-Times, Nov. 4, 10:10 AM CST, "unofficial results from the Washington County Election Commission," all 19 precincts.²² Neither feed has loaded the commission's figures.
• d. Tippah: county affidavit tally, 147 cast, 71 accepted, Colom 39, Hyde-Smith 31.²³ DDHQ loaded the batch Nov. 6; AP has not.²⁵
• e. George: DDHQ feed, Nov. 6, 6:41 PM CST; unlabelled batch of 110 (Colom 31, Hyde-Smith 69, Pinkins 10).²⁵

Every other county matches AP exactly, or AP already carries the larger figure. Where AP and DDHQ differed earlier today on Oktibbeha, Pike, Marion, Benton, Scott, Holmes, Wilkinson and Choctaw, both have now loaded the same batches.

2. AP's own remainder, at today's county leans

AP's expected-vote parameters total 983,599, so about 30,285 remain.²⁴ The table applies each county's current two-party share to AP's remaining votes for that county. This assumes nothing about affidavits: it treats every remaining ballot as if it votes like the county already has. County shares are from the AP feed.²⁴

AP's estimate shows almost nothing left in three counties whose own documents indicate their absentees have not yet been counted: Hancock (election-night summary equal to the sum of its 26 Election Day precinct reports), Monroe (summary headed "No Affidavit or Absentee Ballots") and Tippah (absentees reported pending).^23,29 The Secretary of State's Nov. 2 report shows 1,798, 1,343 and 718 absentees received in those counties.⁸ Adding them at county leans, the remaining vote still widens Colom's lead by about 2,000.

A check that points the other way. Turnout in these three counties is already high without those absentees. Against each county's 2024 presidential vote, Hancock stands at 79.0%, Monroe at 79.4% and Tippah at 76.9%, at or above the highest of the counties with a similar Hyde-Smith share (highest: 78.5%, 81.6% and 73.9%; medians 71.2%, 75.1% and 70.6%).^24,28 Adding the absentees would put them at 87.5%, 87.8% and 84.3%, well above every similar county. So some or all of these ballots may already be in the count, for example if the counties tallied absentees inside their precinct totals this year. The documents and the turnout disagree, and nothing published settles it. This report counts the absentees as still to come, which costs Colom about 1,270 votes in the table. If they are already counted, his final lead is about 1,270 larger (next table).

The received-mail row covers ballots that arrived after the Secretary of State's Nov. 2 report and are still uncounted, spread like the Nov. 2 unreturned ballots and split at county shares. Hinds' share is left out because it sits inside the Hinds remainder above. AP's remainder may already expect some of these ballots in counties below 99% reporting, so this row may count a little twice; it is small either way. Harrison's mail is not in it: see Section 4b.

What that means for the race

The table above shows only how the remaining votes would split. Adding that to Colom's current 2,574-vote lead on AP²⁴ gives his final margin:

Even the harshest combination here, the smallest Hinds remainder and no affidavit lean at all, leaves Colom ahead by more than 2,000 votes. All these rows also understate him, because they give every remaining ballot exactly its county's lean; every one of the 15 affidavit batches reported so far ran more Colom than its county (Section 3). The full simulation in Section 5 uses the county's own Hinds figures, the in-hand absentees, the received late mail and the measured affidavit lean, and gives Colom better than a 99.9% chance of winning.

3. Affidavit ballots are breaking for Colom, including in Republican counties

Most of what remains is affidavit (provisional) ballots. The Secretary of State reported 11,202 processed statewide by Friday afternoon, 8,717 accepted (78%) and 2,485 rejected,¹ up from 6,936 processed Thursday night.² For every county that has added affidavits, the table compares Colom's share of the county's regular vote with his share of its affidavits.

"Colom share" is his share of the two-party vote. The county figure excludes the batch itself and comes from the AP and DDHQ feeds.^24,25 *Post-election batches the county has not labelled; their size and lean match affidavits. All 15 batches run more Colom than their county. None runs toward Hyde-Smith.

Hyde-Smith netted votes only in Alcorn, Itawamba and George, counties Trump carried by 50 to 65 points in 2024,²⁸ and even there by less than her overall margin. The Mississippi Free Press noted that across Alcorn, Chickasaw and Itawamba she netted just 71 votes "despite holding substantial overall leads in two of those counties."¹⁶

The pattern has a known cause. Mississippi moved more than 200,000 voters to inactive status before the election and requires inactive voters to vote by affidavit.^19,26 A DeSoto County commissioner said "more than half" of the affidavits she reviewed were from previously inactive voters who were reactivated and whose ballots will count.¹¹ Those voters are not a random sample of their county.

4. Late mail cannot close the gap, counting both kinds

Late mail comes in two pools. The Secretary of State's "outstanding" figure counts only the first: on Nov. 4 it described its count as ballots "that must be received within five business days."³¹ Ballots already in a clerk's office but not yet counted are a second pool, not included in that figure.

a. Not yet returned

• How many: just under 2,900 mail ballots (2,857) were still outstanding Friday afternoon,¹ down from about 3,300 Thursday night.² These are ballots sent out and not yet received. Those postmarked by Election Day can arrive through 5 PM Tuesday, Nov. 10.
• How many will arrive: few. Hinds' certified 2024 results show 18 "5 Day Absentee" ballots⁴ against 1,093 unreturned on the Secretary of State's Nov. 4, 2024 absentee report,²⁷ under 2%. Even at ten times that rate, this pool adds a few hundred votes statewide.
• Where they are: the Secretary of State has not published a county breakdown of the 2,857. Its last county-level report, from Nov. 2, showed 6,587 unreturned ballots, 43% of them in counties Hyde-Smith leads and 57% in counties Colom leads.⁸ If the remaining ballots are spread the same way and split at county leans (AP feed),²⁴ all 2,857 arriving would net Colom about +300.
• The ceiling, with no assumption about where they are: if a quarter of the 2,857 arrived, more than ten times the only measured late-arrival rate we found (Hinds 2024: 18 of 1,093), and every one of them voted for Hyde-Smith, she would net about 714 votes,¹ roughly a quarter of the 2,574-vote lead.

b. Already received, not yet counted

This pool has two kinds of ballots: late arrivals that came in after the Nov. 2 report and have not been counted, and on-time mail never entered, received by Election Day but not yet added to the count. Harrison is the measured example of the second kind, and Rankin's 282 likely includes some. The statewide estimate below sizes only the late arrivals; the named counties cover the rest.

• Named by the counties: Jackson County Circuit Clerk Randy Carney said 280 absentee ballots have been received, with 138 still outstanding.¹² Rankin County received 282 mail-in absentee ballots that commissioners still need to process, per Circuit Clerk Michelle Adcock.¹⁰ Hinds says its post-election work includes "eligible mail-in ballots received after Election Day";⁶ it had 639 not returned on Nov. 2.⁸ Rankin's 282 is more than the 206 it had not returned on Nov. 2,⁸ so a county's uncounted mail can include ballots that arrived before that report.
• Harrison: Circuit Clerk Justin Wetzel said Nov. 4 that the county's mail-in absentees, processed on Election Day, were not entered before results were reported and "still must be added."³² Harrison's totals on AP, on DDHQ and on the circuit clerk's own page are unchanged since 9:21 PM CST on election night,^24,25,32 so these ballots have not been added. No count has been published. AP's Harrison remainder of about 2,000 may include them. Harrison is close to even (Colom 48.1%), so even 1,300 such ballots at county shares move the margin by about 50 votes.²⁴
• Late arrivals, how many (counted by the state): unreturned ballots fell from 6,587 on Nov. 2⁸ to about 3,600 on Nov. 4,³¹ about 3,300 on Nov. 5² and 2,857 on Nov. 6.¹ About 3,700 ballots therefore arrived after the Nov. 2 report. If the state measures from ballots sent rather than ballots requested (it has not said), the figure is about 3,300.⁸
• Late arrivals, how many still uncounted (our estimate): some of those ballots arrived by Election Day and were counted that night. No statewide figure says how many. Four counties publish absentee ballots cast by type, which can be set against what they had received by Nov. 2: Madison 5,796 v 5,764; Lauderdale 1,975 v 1,951; Lafayette 3,162 v 3,144; Pearl River 2,014 v 2,005.^8,33 Those 83 extra ballots are about a quarter of the roughly 350 arrivals these four counties' share of the statewide total implies, so about three quarters of their arrivals are still uncounted (60–81% by county). Applying 55–80% to the 3,300–3,700 arrivals gives roughly 1,800–3,000 ballots received but not yet counted. This is an estimate, not a count, and it likely runs low: the four counties all entered their mail on election night, while a county like Harrison has entered none of its mail.
• Why the estimate does not decide anything: four counties are a small sample, and a county like Rankin may handle its mail differently. The ceiling below does not use the four-county rate at all. It treats every one of the roughly 3,700 arrivals as uncounted, including all of Jackson's 280 and Rankin's 282, and puts them in the most Republican counties that had ballots out. Any real county's practice, Rankin's included, sits inside that worst case.
• Where they are: these ballots can only come from counties that had ballots out on Nov. 2. Spread like those ballots and split at county leans (AP feed),^8,24 1,800–3,000 of them net Colom about +190 to +310. Jackson's 280 and Rankin's 282 at their own county leans net Hyde-Smith about 40 and 70.
• The ceiling: count all of the roughly 3,700 arrivals as uncounted, and place them in the most Republican counties first, up to each county's Nov. 2 unreturned count, and split each at its county's vote. Hyde-Smith nets about 600 at most, whatever the pool's size, because those counties had too few ballots out to supply more.^8,24 Harrison's on-time mail adds about 50 at most, since Harrison is close to even.

c. Both pools together

Both worst cases at once, which would need every not-yet-returned ballot that arrives to go to Hyde-Smith and every arrived ballot to sit in her strongest counties, net her about 1,350 votes, including Harrison's on-time mail, about half of the 2,574-vote lead. The two ceilings also overlap, since a ballot cannot be in both pools. At county leans, the two pools together most likely net Colom a few hundred. The simulation in Section 5 includes both, with Jackson's and Rankin's ballots as named blocks.

5. Hyde-Smith's path, tested

With about 30,300 votes left²⁴ and Pinkins taking about 2%, Hyde-Smith needs 54.3% of the two-party remainder. In the affidavit batches so far (Section 3) she has won 35%. She has exceeded 54% only in Alcorn (63%), Itawamba (74%) and George (69%), three small counties that together hold under 500 of the remaining votes.²⁴

What would have to be true for Hyde-Smith to win

The table comes from a county-by-county simulation (200,000 draws) of every outstanding pool: each county's measured or estimated affidavit count, acceptance rates centred on the measured county and Secretary of State figures,^1,10,16,20 late mail of both kinds (not yet returned, and received but not yet counted),^1,8,31 and named blocks such as the Hancock, Monroe and Tippah absentees (included even though they may already be counted, which favors Hyde-Smith), and Jackson's and Rankin's received mail.^10,12,29 Each row changes one thing: how the remaining affidavits lean compared with their own county. The fourth column reruns each row with the Hancock, Monroe and Tippah absentees treated as already counted (Section 2), which removes about 3,850 ballots that favor Hyde-Smith. The last column says whether any of the 15 reported batches (Section 3) has behaved that way.

Read the last row as a boundary, not a scenario. To reach even odds, Hyde-Smith needs affidavit voters statewide to be markedly more Republican than the counties they live in. Fifteen counties have reported, across every region of the state and every partisan lean, and all fifteen went the other way. If the Hancock, Monroe and Tippah absentees are already counted, even that boundary row gives Colom 65%. The conservative case is the second row, no affidavit lean at all, and it is not close either way.

6. Hinds County: AP's 7% outstanding is very likely too high, and it does not matter for the call

AP shows Hinds at 93% reporting: 77,465 counted against 83,296 expected, so about 5,800 out. AP's expected-vote parameters for Hinds are 74,951 Election Day in-person, 8,083 advance or absentee, and only 262 provisional.²⁴ In other words, AP's remainder assumes thousands of regular ballots are still uncounted, and almost no affidavits. The county's records point the other way on both.

a. The precinct data show absentees are already counted

Hinds' 2024 certified precinct report splits every precinct into Election Day and absentee votes.⁴ Absentee shares ranged from 1.6% to 31.5%. If 2026 absentees were missing from the election-night report,³ precincts heavy with absentee voters would look badly short against 2024. They do not:

Turnout relative to 2024 is flat at 86–91% in every group. That is what counted absentees look like. Missing absentees would show as a steady decline down the table. Raymond 1, the most absentee-heavy precinct (31.5% in 2024), stands at 81% of its 2024 turnout; for its absentees to be uncounted, its 2026 Election Day turnout would have to exceed the 2024 presidential election by 18%. A regression across all 107 matched precincts, controlling for partisanship, puts the share of 2026 absentees already counted at 100% (90% interval 84–123%).

b. State rule: absentee scanning starts the moment polls close

"After the close of the polls at 7:00 p.m., the Resolution Board will open the ballot box containing the accepted absentee ballots … and shall immediately begin processing the accepted ballots through the central scanner. … The totals shall then be combined with the precinct tabulation totals for the unofficial vote count."

Mississippi Administrative Code, Secretary of State Rule 731c, Part 17 Rule 3.2 (e)–(f).⁵ Envelope review may begin when the polls open (3.2(a)). The rule also lets the board "recess as necessary and resume" before certification (3.2(g)), so it sets the expected practice rather than compelling it. The precinct data are the evidence that Hinds followed it this year.

c. The county described its post-election work as late mail and unreadable ballots

"The Hinds County Resolution Board began reviewing and processing absentee ballots at noon Nov. 4. That process includes eligible mail-in ballots received after Election Day, as well as ballots that cannot be properly read by scanning equipment."

WLBT, Nov. 4, 2026, reporting the Hinds County Election Commission's statement through board attorney Tony Gaylor.⁶ Hinds had 8,070 absentees returned by Nov. 2 and 639 not returned;⁸ mail "received after Election Day" can only come from the latter.

d. What is actually left in Hinds

Affidavits, which the county has not yet released. Commissioners have given counts for three of five districts: about 650 in District 5, 387 in District 3, and 136 from 5 of District 4's 14 precincts.⁷ Scaled to all five districts, that is roughly 2,300 cast. District 1 Commissioner Kidada Brown has said the county will not "turn a number in until everyone is complete."⁷ At the acceptance rates seen elsewhere, Hinds' affidavits yield about 1,300–1,800 votes. Add Hinds' late mail, which cannot exceed the 639 ballots it had not returned on Nov. 2,⁸ and the realistic remainder is 1,500–2,500, not 5,800. It is also far more than the 262 provisionals in AP's Hinds parameters.

e. Why the higher estimate is understandable

• 2023 precedent. On election night in the 2023 governor's race, Hinds reported only 447 absentees and counted the rest the next day.¹⁷ A model trained on that behavior would expect the same in 2026. The 2026 election-night report uses a different template with no vote-type columns, and the precinct test above shows the absentees are in.
• The "exceeded 2020" statement. The commission said preliminary figures indicated participation exceeded the November 2020 presidential election.⁶ Hinds cast 100,944 ballots in 2020.³⁰ Even adding every absentee and affidavit to the current 77,465 would not reach that. The statement cannot be literal and should not be read as evidence of a large uncounted pool.
• Turnout against the rest of the state. Statewide, the 2026 count stands at 77.6% of the 2024 presidential vote.^24,28 Hinds is at 89.2%, right at the median for comparable Black-majority counties. Another 9,000 votes would put Hinds at 99.6% of its presidential-year turnout, the highest of all 82 counties.

f. Either way, Hinds helps only Colom

If AP's 5,800 were right²⁴ and split like the rest of Hinds (Colom 80–19), Colom's final margin would rise to roughly +1.10 to +1.18 points. If the county's own data are right, Hinds adds about 1,000–1,500 net for Colom. Neither scenario offers Hyde-Smith anything. A model that keeps the race uncalled because 7% of Hinds is outstanding is holding the call on votes that can only extend the leader's margin.

7. What comes next

Sources and documents

1. Secretary of State Michael Watson, X, Nov. 6, 2026, 4:43 PM CST: 75 of 82 counties uploaded; just under 2,900 mail ballots out; 11,202 affidavits processed, roughly 78% accepted. x.com/MichaelWatsonMS/status/2118836746565910576. The exact figures (8,717 accepted, 2,485 rejected, 2,857 mail outstanding) are from Queen City Newsfeed citing the SOS office: x.com/bert_on_spec/status/2118844952482615364.
2. Mississippi Free Press, "Secretary of State: Around 3,300 Absentee Mail Ballots Outstanding, 6,936 Affidavit Ballots Processed," Nov. 5, 2026. mississippifreepress.org. SOS midday update Nov. 6 (9,242 processed; 3,139 mail out): x.com/MississippiSOS/status/2118778412059984223.
3. Hinds County, 2026 General Election Unofficial Precinct Report (election night, 11/03/2026). co.hinds.ms.us/pgs/results/Unofficial Precinct Report.pdf · Summary: Unofficial Summary Report.pdf
4. Hinds County, 2024 General Election certified precinct report (Election Day, absentee, affidavit and 5-day columns). co.hinds.ms.us/pgs/elections/20241105_Precinct Results Report.pdf
5. Mississippi Secretary of State, Administrative Code Rule 731c, Part 17 (absentee processing, Rules 3.1–3.2). sos.ms.gov/adminsearch/ACCode/00000731c.pdf
6. WLBT, "Hinds County election results expected to be certified by Nov. 13 after historic turnout," Nov. 4, 2026. wlbt.com
7. Mississippi Free Press, "Hinds County Makes Slow Progress on Affidavit Ballots, Still With Little to Share," Nov. 6, 2026 (District 3, 4 and 5 counts; Commissioner Brown quote). mississippifreepress.org
8. Mississippi Secretary of State, 2026 General Election Absentee Report, Nov. 2, by county (Hinds: 8,070 received, 639 not returned; statewide 6,587 not returned). sos.ms.gov
9. Mississippi Free Press, "Harrison County Received 1,448 Affidavit Ballots, Won't Count Until Nov. 10," Nov. 6, 2026. mississippifreepress.org
10. Mississippi Free Press, Rankin County entries, Nov. 5–6, 2026: 1,667 affidavits, plus 282 mail-in absentee ballots for commissioners to process (Circuit Clerk Michelle Adcock); about 20% rejected so far, wrong precinct the main reason; about half processed, scanning begins 5 PM Nov. 10.
11. DeSoto County: public notice, review and count Nov. 11, 3 PM; affidavit process; MFP, about 2,100 to be counted; MFP, Commissioner Edler on reactivated inactive voters.
12. Mississippi Free Press, "Jackson County Received More Than 1,400 Affidavit Ballots," Nov. 5, 2026 (Circuit Clerk Randy Carney: 280 absentee ballots received, 138 outstanding). mississippifreepress.org
13. Oktibbeha County, General Election Results with Affidavit Ballots, 11/06/2026 12:41 PM (503 affidavits: Colom 401, Hyde-Smith 96, Pinkins 5). oktibbeha.ms.gov/DocumentCenter/View/536 · MFP: 80-point margin
14. Pike County, Summary Results "with absentee, affidavits," 11/06/2026 11:30 AM. pikecountyelectionresults.org/2026_general/DW.pdf
15. Mississippi Free Press, "Lafayette County's Affidavit Ballots Break for Scott Colom 69%-29%." mississippifreepress.org
16. Mississippi Free Press, "Scott Colom Is Overperforming Affidavit Ballots in North Mississippi—Even in Hyde-Smith Friendly Counties," Nov. 6, 2026 (Alcorn, Chickasaw, Itawamba, Benton, Tishomingo, Calhoun, Pontotoc). mississippifreepress.org
17. Hinds County 2023 election-night unofficial results (archived Nov. 9, 2023; Election Day 64,232, absentee 447). web.archive.org
18. Mississippi Free Press, "Mississippi Counties Have a Hard Deadline to Certify Election Results." mississippifreepress.org
19. NPR, "No one knows how many ballots are outstanding in Mississippi's Senate race," Nov. 6, 2026 (inactive-voter rule; estimates of the affidavit pool). npr.org. Magnolia Tribune, "Thousands of affidavit ballots could decide Mississippi Senate race," Nov. 6, 2026. magnoliatribune.com
20. U.S. Election Assistance Commission, Election Administration and Voting Survey 2020–2024 (Mississippi provisional ballots by county and reason: 2024 cast 21,421, counted 14,685; 365 cast for lack of ID). eac.gov
21. Lowndes County: Commercial Dispatch, "It may be more than a week before we know who won the Mississippi Senate seat," Nov. 4, 2026 (Deputy Circuit Clerk Ann Marie Higgins: "more than 550" affidavits; 39 accepted by 2 PM Wednesday "only rejecting a handful"). cdispatch.com. County affidavit totals (Colom 342, Hyde-Smith 172, Pinkins 4) per the county's provisional ballot records, shown on video by Abiola Agoro, "LOWNDES COUNTY MISSISSIPPI Ballot Cure update," X, Nov. 6, 2026, 4:05 PM CST: x.com/ahbeeolahh/status/2118827230918807661. They match the sum of DDHQ's two Lowndes batches exactly.
22. Delta Democrat-Times, "Franklin wins Washington County Court judge race," Nov. 4, 2026, 10:10 AM CST, carrying "unofficial results from the Washington County Election Commission": U.S. Senate Colom 8,722, Hyde-Smith 2,382, Pinkins 163. ddtonline.com
23. Tippah County: Southern Sentinel, "Tippah County general election results, Nov. 3, 2026" (southern-sentinel.com) and Tippah News, Nov. 4 (tippahnews.com), both noting absentees and affidavits pending on election night. County affidavit tally Nov. 6: 147 cast, 71 accepted, Colom 39, Hyde-Smith 31, relayed from county officials and matched exactly by the DDHQ batch.
24. Associated Press election results feed, Mississippi U.S. Senate, race 20261103MS26692 (county vote counts, reporting percentages and expected-vote parameters by county and vote type). interactives.apelections.org/…/MS/20261103MS26692/detail.json. Read at 6:45 PM CST Nov. 6, 2026; county figures as displayed on AP's published results pages.
25. Decision Desk HQ results feed, Mississippi U.S. Senate, race 294688 (county vote counts with per-county update times). votes.decisiondeskhq.com/api/races/294688. Read at 6:45 PM CST Nov. 6, 2026. CNN's county feed (politics.api.cnn.io) matched AP throughout.
26. WLOX, "Voter roll updates could mean more affidavit ballots cast," Nov. 3, 2026. wlox.com
27. Mississippi Secretary of State, Absentee Ballot Summary Report BP-035, Federal General Election 11/05/2024, run 11/04/2024 (Hinds County subtotal: 13,453 requested, 13,452 sent, 1,093 not returned, 12,360 received; statewide 209,037 / 208,886 / 11,134 / 197,903). PDF: content.govdelivery.com/…/11-4-2024 BP-35.pdf; SOS release page: sos.ms.gov/node/1678.
28. Mississippi Secretary of State, Official 2024 General Election Results, statewide recapitulation by county (used for 2024 county presidential totals and Trump margins). PDF: sos.ms.gov/elections/electionresults/2024 Official Statewide Results.pdf; results index with CSV: sos.ms.gov/…/elections_results_2024_general.aspx; per-county certified reports at …/2024General/County Results/Hinds.pdf (same pattern for each county).
29. Uncounted absentees by county documents. Hancock County: election-night Unofficial Results summary (hancockcounty.ms.gov/DocumentCenter/View/1224), equal to the sum of its 26 Election Day precinct reports (DocumentCenter folder 164). Monroe County: cumulative results headed "UNOFFICIAL RESULTS — No Affidavit or Absentee Ballots," Election Summary 11/04/2026 12:10 AM (monroems.com/circuitclerk/results). Tippah County: source 23. Jackson County: WLOX, "Jackson County to count St. Martin precinct votes Wednesday," Nov. 4, 2026 (wlox.com).
30. Hinds County, November 3, 2020 General Election Official Summary Report (100,944 ballots cast). co.hinds.ms.us/pgs/elections/20201124_Official Summary Report.pdf
31. WJTV (Thao Ta), "Hinds County continues ballot count after general election," Nov. 4, 2026, quoting the Secretary of State's Office: about 3,600 outstanding absentee-by-mail ballots "that must be received within five business days." Syndicated copy: aol.com/articles/hinds-county-continues-ballot-count-031112000.html
32. Sun Herald (Margaret Baker), "Don't expect quick resolution to Mississippi's deadlocked Senate race, officials say," Nov. 4, 2026 (Harrison Circuit Clerk Justin Wetzel: mail-in absentees "still must be added"). Syndicated copy: yahoo.com/news/politics/articles/don-t-expect-quick-resolution-202725914.html. Harrison County Circuit Clerk Facebook page, election-night results (Senate 27,470 / 29,638 / 1,742; read Nov. 6–7, unchanged): facebook.com/profile.php?id=61575047374506
33. County election summaries with absentee ballots cast by type (in-person + mail): Madison, Summary Results Report run Nov. 5, 2026, 10:37 AM (5,509 + 287), madison-co.com/elections/election_results.pdf; Lauderdale, Election Summary Nov. 4, 10:06 AM (1,677 + 298), lauderdalecounty.org; Lafayette, unofficial results, affidavits not counted (2,995 + 167), lafayettems.com; Pearl River, Election Summary Nov. 3, 11:05 PM (1,904 + 110), pearlrivercounty.net. Nov. 2 received figures from source 8. Each county's share of the statewide arrivals is taken in proportion to its Nov. 2 unreturned ballots.

Counts are from the AP and Decision Desk HQ results feeds, read at 6:45 PM CST Nov. 6, 2026. AP's remaining-vote figures are its published expected-vote parameters by county. The precinct comparison uses the Hinds County 2026 unofficial precinct report and 2024 certified precinct report (107 of 108 precincts matched; one precinct's line is unreadable in the county's PDF). The simulation draws affidavit counts, acceptance and lean per county with statewide shared shifts; its assumptions are stated in each table.