PENNER STRATEGYRESEARCH
  1. Research

Washington late-ballot rejections: 2018-2024 general elections and 2024-2026 primaries

General elections: 2018, 2020, 2022 and 2024. Primaries: August 2024 and August 2026. Detailed 2026 records dated August 25.

This study examines ballots rejected for missing Washington's return deadline across several elections. Federal survey records cover four general elections; separate state records and news reporting cover two primaries.

The sources below cover different elections and count different things. Federal survey files report Washington's late-rejected mail ballots for the November general elections in 2018, 2020, 2022 and 2024. A separate comparison concerns the August 2024 and August 2026 primaries, but its late counts come from a news report and lack retained first-party confirmation. The detailed August 2026 return-method analysis uses a state ballot-status snapshot dated August 25, 2026. These sources should not be treated as one continuous trend.

Explanation updated September 12, 2026. The dates above describe the records studied.

ballot-status file
Washington's statewide record of how each returned ballot was processed, including whether it was accepted or rejected and why.
Too Late
A rejected-ballot code for missing Washington's legal deadline. A timely postmark can make a mailed ballot valid even if it arrives after election day.
EAVS
Election Administration and Voting Survey, a federal survey of local election officials. Its returned-ballot and counted-ballot totals are different measures.
Denominator
The total used to calculate a rate. Dividing late rejections by returned ballots gives a different rate from dividing them by counted ballots.
Primary and general election
The primary takes place before the November general election. Figures from these different elections describe different sets of ballots.

Principal limit The EAVS deadline-rejection rate can be computed on two different published columns of the same federal file, mail ballots returned or mail ballots counted, and which one is used changes the published rate.

Answer

Washington's federal EAVS files show mail ballots rejected for arriving late, across all 39 county jurisdictions, at 3.009 per 1,000 counted ballots in the 2018 general election, 0.611 in 2020, 2.730 in 2022 and 1.177 in 2024; these use the file's counted-ballot column, and its separate returned-ballot column gives slightly different rates (2.963, 0.606, 2.697, 1.166), since both are different columns of the same file. For the August 2024 and August 2026 primaries, officially reported counted totals of 1,994,096 and 1,892,873 combine with news-reported late counts of 16,131 and 20,267, which are not confirmed in any official state record, to give rates of 8.089 and 10.707 per 1,000, an increase this article calculates at 32.4 percent. The state's own August 25, 2026 snapshot separately records 20,265 Too Late and 1,892,281 accepted, which differ from those reported figures for reasons the retained evidence does not explain.

Material qualifications

  • The EAVS deadline-rejection rate can be computed on two different published columns of the same federal file, mail ballots returned or mail ballots counted, and which one is used changes the published rate.
  • Figures drawn from different years and different elections, the 2018-2024 general elections or the August 2024 and August 2026 primaries, cover different populations of ballots and should not be read as one continuous trend.
  • The late-ballot counts for the two primaries come from a single news article, not from a retained first-party state record.

Current answer from edition 0001 · Dated September 12, 2026 · This dated edition

Evidence covers sources through September 12, 2026.

On this page

Findings

How often were mail ballots rejected as late in Washington's federal general elections, 2018-2024?

In Washington's 39 county jurisdictions, the EAVS-reported deadline-rejection rate for mail ballots was 3.009 per 1,000 counted ballots in the 2018 general election, 0.611 in 2020, 2.730 in 2022 and 1.177 in 2024. On the returned-ballot denominator, the same years read 2.963, 0.606, 2.697 and 1.166 per 1,000. Each year is a separate federal general election with its own population of ballots, not points on one continuous trend.

Washington's EAVS deadline-rejection rate across four general elections

In Washington's 39 county jurisdictions, the EAVS-reported rate of mail ballots rejected for arriving after the deadline was 3.009 per 1,000 counted ballots in 2018, 0.611 in 2020, 2.730 in 2022 and 1.177 in 2024. On the returned-ballot denominator the same years read 2.963, 0.606, 2.697 and 1.166 per 1,000. These are four separate general elections, not a single continuous series, and the rate depends on which of the two published denominators is used.

What is EAVS? EAVS is the U.S. Election Assistance Commission's Election Administration and Voting Survey, a federal file states submit after each general election; this article uses Washington's 2018, 2020, 2022 and 2024 public-release files covering all 39 county jurisdictions.

The EAVS deadline-rejection rate can be computed on two different published columns of the same federal file, mail ballots returned or mail ballots counted, and which one is used changes the published rate.

How has Washington's EAVS deadline-rejection rate moved across four general elections?

Bar chart of eight deadline-rejection rates across four federal general elections.
Figure 1. Washington's deadline-rejected mail-ballot rate per 1,000, in 2018, 2020, 2022 and 2024, on both the reported-counted and the returned denominators.Download chart (PNG)

The rate per 1,000 counted mail ballots was 3.009 (2018), 0.611 (2020), 2.730 (2022) and 1.177 (2024); each year is a separate election, and the rate differs slightly by denominator.

The EAVS deadline-rejection rate can be computed on two different published columns of the same federal file, mail ballots returned or mail ballots counted, and which one is used changes the published rate.

View data table

Scroll horizontally to see all columns.

Figure 1 data. How has Washington's EAVS deadline-rejection rate moved across four general elections? Each quantity appears rounded to two decimals and again exactly as recorded. Exact plotted rows: assets/data/eavs-rate-figure.csv
SeriesRate (per 1,000)Rate as recorded (per 1,000)
2018 per 1000 counted3.013.008923
2018 per 1000 returned2.962.962575
2020 per 1000 counted0.610.610689
2020 per 1000 returned0.610.605989
2022 per 1000 counted2.732.730325
2022 per 1000 returned2.702.697407
2024 per 1000 counted1.181.176835
2024 per 1000 returned1.171.165991

Sources and calculations:

Read the full finding Sources and calculations

  • Not resolved here
  • Why the rate moves between years is not addressed by these figures alone.

Why does the rate differ depending on which total is used?

The federal EAVS file publishes two different columns that can serve as a denominator: mail ballots returned (C1b) and mail ballots counted (C3a in 2018 and 2020, C8a in 2022 and 2024). These are two different columns of the same file, not two estimates of the same thing, and a third figure, returned minus all mail rejections, is a derived residual rather than a published field. This article reports the rate calculated on both denominators rather than selecting one as standard.

EAVS rates use two different denominators

The federal EAVS file publishes both a returned-ballots column (C1b) and a separate reported-counted column (C3a in 2018 and 2020, C8a in 2022 and 2024), and these are not the same figure. A third quantity, returned minus all mail rejections, is a derived residual and not itself a published field. This article reports the rate on both published denominators rather than choosing one.

Why does the article give two rates for each EAVS year instead of one? Because the federal file has two different published columns, mail ballots returned and mail ballots counted, that can each serve as a denominator, and the rate changes depending on which one is used.

The EAVS deadline-rejection rate can be computed on two different published columns of the same federal file, mail ballots returned or mail ballots counted, and which one is used changes the published rate.

Read the full finding Sources and calculations

  • Not resolved here
  • Which denominator other publishers or the state itself treat as standard is not stated here.

Does the 2018 EAVS file's own math add up?

In the 2018 EAVS file, returned mail ballots (C1b) minus all mail rejections (C4a) exceeds the file's own reported-counted total (C3a) by 15,611 statewide. Seven county rows, led by Whitman County at 13,555, carry the entire residual. The retained evidence does not establish why these fields disagree in this one file.

2018 EAVS totals differ by 15,611

In the 2018 EAVS file, returned mail ballots minus all mail rejections exceeds the file's own reported-counted total by 15,611 statewide. Seven counties, led by Whitman County at 13,555, carry this entire difference. The retained evidence does not explain why these federal fields disagree in this one file.

Why doesn't the 2018 EAVS file add up? Returned mail ballots minus all mail rejections exceeds the file's own reported-counted total by 15,611 statewide in 2018, concentrated in seven counties, and the retained evidence does not explain why.

In the 2018 EAVS file, returned mail ballots minus all mail rejections exceeds the file's own reported-counted total by 15,611, a residual concentrated in seven counties whose cause the retained evidence does not explain.

Read the full finding Sources and calculations

  • Not resolved here
  • The cause of the 2018 partition residual is unexplained.

Washington’s ballot deadline

Under RCW 29A.40.110, a ballot must be received by 8:00 p.m. on the day of the primary or election, or postmarked no later than that day. If the postmark is missing or illegible, the date on the voter's signed ballot declaration determines validity. The state ballot-status file assigns a Too Late code to a rejected ballot found to have arrived after this deadline, but the file has no postmark field of its own.

Washington's statutory deadline rule for mail ballots

Under RCW 29A.40.110, a ballot must be received by 8:00 p.m. on election day or postmarked no later than election day. If the postmark is missing or illegible, the date on the voter's signed ballot declaration determines validity. The state's ballot-status file records a Too Late code for ballots found to have arrived after this deadline, but the file itself has no postmark field.

What counts as 'too late' under Washington law? Under RCW 29A.40.110, a ballot must be received by 8 p.m. on election day or postmarked no later than that day; if the postmark is missing or illegible, the signed ballot declaration date is used instead.

The state ballot file has no postmark field, so none of these rates indicate when or why a ballot was mailed or arrived late.

Read the full finding Sources and calculations

  • Not resolved here
  • How counties determine the postmark or declaration date in practice is not described here.

Late-ballot counts in the 2024 and 2026 primaries

A news article reported 16,131 late-rejected ballots for the August 2024 primary and 20,267 for the August 2026 primary. Against the officially reported counted totals of 1,994,096 and 1,892,873, that gives rates of 8.089 and 10.707 per 1,000 counted, a 32.4 percent increase by this article's own arithmetic, close to the 32 percent figure the news account attributes to state officials. The two counted totals are official state records; the two late counts are news-reported and not confirmed in any retained first-party state record.

Late-ballot counts for two primaries come from a news report, not an official record

A news article reported that late-rejected ballots were 16,131 for the August 2024 primary and 20,267 for the August 2026 primary. Divided by the official counted totals of 1,994,096 and 1,892,873, those figures give rates of 8.089 and 10.707 per 1,000 counted, an increase this article calculates at 32.4 percent, close to the 32 percent the news account attributes to state officials. No retained first-party state record confirms the late counts or states the method behind the reported percentage.

Are the primary late-ballot counts an official government figure? No. The 16,131 and 20,267 late counts come from a single September 2, 2026 news article and are not confirmed in any retained first-party state record.

The late-ballot counts for the two primaries come from a single news article, not from a retained first-party state record.

How do the two primaries' reported late-ballot rates compare?

Bar chart of the reported late-rejection rate for two primaries.
Figure 2. The reported late-rejected rate per 1,000 counted for the August 2024 and August 2026 primaries.Download chart (PNG)

News-reported late counts give 8.089 per 1,000 counted in August 2024 and 10.707 in August 2026, a 32.4 percent increase by this article's arithmetic; the late counts are not from an official record.

The late-ballot counts for the two primaries come from a single news article, not from a retained first-party state record.

View data table

Scroll horizontally to see all columns.

Figure 2 data. How do the two primaries' reported late-ballot rates compare? Each quantity appears rounded to two decimals and again exactly as recorded. Exact plotted rows: assets/data/primary-rate-figure.csv
PrimaryRate (per 1,000 counted)Rate as recorded (per 1,000 counted)
August 2024 primary8.098.089380
August 2026 primary10.7110.707005

Sources and calculations:

Read the full finding Sources and calculations

  • Not resolved here
  • The method or denominator behind the reported 32 percent figure is not stated in any retained first-party record.

August 2026 ballot-status records

Washington's final statewide ballot-status file for August 25, 2026 covers all 39 counties with 1,919,356 total ballot records, 1,892,281 accepted and 27,075 rejected. Of those, 20,265 carry a Too Late code across all return channels and 19,141 within the mail channel, out of 592,576 mail-channel returns. These totals differ from the separately reported figures of 20,267 late and 1,892,873 counted, for reasons the retained evidence does not explain.

August 2026 ballot-status totals

Washington's final statewide ballot-status file for the August 4, 2026 primary, in its August 25, 2026 final snapshot, covering all 39 counties, records 1,919,356 total ballot records, 1,892,281 accepted and 27,075 rejected. Of these, 20,265 were coded Too Late across all return channels and 19,141 were coded Too Late within the mail channel, out of 592,576 mail-channel returns. These totals differ from the separately news-reported figures of 20,267 late and 1,892,873 counted, for reasons the retained evidence does not explain.

What does 'Too Late' mean in the state's ballot-status file? It is the challenge reason code the state assigns to a rejected ballot found to have arrived after the statutory deadline.

The county-level ballot status snapshot's own Too Late and Accepted totals differ from the separately reported 20,267 and 1,892,873 figures for reasons the retained evidence does not explain.

Read the full finding Sources and calculations

  • Not resolved here
  • Why the snapshot's own totals differ from the separately reported figures is not resolved.

Is distance from a mail facility related to a county's late-ballot rate?

Across Washington's 39 counties, the mail-channel Too Late rate per 1,000 mail ballots correlates with straight-line distance to the nearest of three curated USPS facility locations at 0.570; the all-channel rate correlates at 0.394. The facility list is curated from web research and geocoding rather than an official postal routing dataset, and straight-line distance is not postal routing or transit time. This is an association across county aggregates, not a statement about any individual ballot or its cause.

County-level association between distance to a mail facility and late rates

Across Washington's 39 counties, the mail-channel Too Late rate per 1,000 mail ballots correlates with straight-line distance from the county's population center to the nearest of three curated USPS facility locations at 0.570, and the all-channel rate at 0.394. These facility locations are curated from web research and geocoding, not an official postal routing dataset, and distance is not postal routing or transit time. This is a county-aggregate association, not a statement about any individual ballot or its cause.

Does the distance measure show actual postal routing? No. It is straight-line distance from a county's population center to one of three curated facility locations built from web research and geocoding, not an official postal routing dataset or transit time.

The three-facility list used to compute distance is built from web research and geocoding, not an official postal routing dataset.

Does distance from a mail facility relate to a county's late rate?

Bar chart of 39 counties' mail-channel Too Late rate, ordered by distance.
Figure 3. Each county's mail-channel Too Late rate, ordered by straight-line distance to the nearest of three curated USPS facility locations.Download chart (PNG)

The mail-channel Too Late rate correlates with distance to a curated facility list at 0.570 across 39 counties; this is an aggregate association, not a postal-routing measurement or a cause.

The three-facility list used to compute distance is built from web research and geocoding, not an official postal routing dataset.

An association between distance and late-ballot rates across county aggregates does not establish that distance, or any other factor, caused any ballot to arrive late.

View data table

Scroll horizontally to see all columns.

Figure 3 data. Does distance from a mail facility relate to a county's late rate? Each quantity appears rounded to two decimals and again exactly as recorded. Exact plotted rows: assets/data/county-figure-data.csv
CountyDistance to nearest facility (miles)Too Late rate, mail channel (per 1,000 mail returns)Too Late rate, mail channel as recorded (per 1,000 mail returns)
Clark7.4114.5514.5450
Spokane7.9019.0619.0620
Pierce8.9820.2620.2617
King18.1919.5619.5649
Kitsap26.2023.6423.6384
Thurston31.4821.6921.6893
Skamania32.4811.6611.6618
Mason33.2441.0040.9982
Lincoln37.8011.2411.2402
Cowlitz41.4924.2124.2059
Snohomish44.1553.5753.5655
Stevens47.5173.9473.9397
Pend Oreille47.6259.6759.6745
Lewis52.6467.8167.8118
Jefferson54.5414.4214.4222
Whitman58.7648.6148.6094
Wahkiakum60.9914.1814.1844
Island64.0331.7931.7896
Grays Harbor70.9622.0322.0264
Klickitat71.2245.0245.0225
Clallam75.6744.4844.4773
Garfield79.3879.5079.4979
Skagit80.9559.0859.0810
Kittitas81.3561.4061.4035
Ferry83.120.000.0000
Asotin87.6553.2653.2630
Adams88.2860.7760.7676
Pacific91.1235.2335.2326
Columbia92.2323.6023.5988
San Juan93.4842.7142.7112
Chelan94.0159.2459.2367
Grant97.0851.0751.0708
Douglas101.4155.3555.3483
Yakima104.6290.3390.3253
Whatcom105.8535.1035.0969
Okanogan111.7160.0660.0649
Walla Walla114.3471.5571.5499
Franklin118.5477.7377.7311
Benton127.0860.4660.4614

Sources and calculations:

Read the full finding Sources and calculations

  • Not resolved here
  • Whether the association reflects actual postal routing, rather than the curated facility list used here, is not settled.

USPS sources on postmark timing

A quotation appearing in a January 21, 2026 U.S. Senate press release and a September 2, 2026 news article states that the postmark date does not necessarily align with when the Postal Service first took possession of a mailpiece. The news account also describes mail from post offices more than 50 miles from a distribution center now being transported to that center once a day rather than twice, with postmarking happening at the more distant facility. Neither retained document is a Postal Service operational directive, and both place the change at approximately December 2025 without fixing a precise date.

A postmark-timing statement quoted in a Senate release and a news account, not a directive

A quotation appearing in a U.S. Senate press release dated January 21, 2026 and in a September 2, 2026 news article states that 'the postmark date does not inherently or necessarily align with the date on which the Postal Service first accepted possession of the mailpiece.' The news account also describes mail from post offices more than 50 miles from a distribution center now being transported to that center once a day rather than twice, with postmarking happening at the more distant facility. Neither retained document is a Postal Service operational directive, and neither fixes a precise date for when this practice began, though both place it at approximately December 2025.

Does the USPS quotation come from an official USPS policy directive? No. It is quoted in a U.S. Senate press release and a news article; neither retained document is a Postal Service operational directive.

Neither retained USPS-related document is a Postal Service operational directive, and neither fixes a precise date for the practice it describes.

Read the full finding Sources and calculations

  • Not resolved here
  • The exact effective date and operational scope of the described practice are not established by these documents.

Distance and past vote share in the county model

An ordinary least squares fit across the 39 county rates associates distance with 0.361 additional Too Late ballots per 1,000 mail ballots per mile (p=0.0006), with a historical two-ticket vote-share covariate included in the same fit but not separable from distance in these 39 rows. A parallel Poisson count model with an exposure offset gives a distance log-rate coefficient of 0.00987 (p<0.001). The covariate is one ticket's recorded county votes divided by the sum of both tickets' votes in the November 2024 presidential election, leaving out other candidates and write-ins.

A county-level model fitting distance alongside a 2024 vote-share covariate

In an ordinary least squares fit across the 39 county rates, the distance term is associated with 0.361 additional Too Late ballots per 1,000 mail ballots per mile (p=0.0006), with a historical two-ticket vote-share covariate also in the fit but not separable from distance in these 39 rows. A parallel Poisson count model gives a log-rate coefficient for distance of 0.00987 (p<0.001). Both are county-aggregate, ecological specifications that identify no cause and say nothing about any individual ballot or voter.

What is the 'historical two-ticket share' used in the county-level model? It is one ticket's recorded county votes divided by the sum of both tickets' recorded votes in the November 2024 presidential election, from the state's official county results, leaving out other candidates and write-ins.

The county-level models use 39 county aggregates, not individual voters or ballots, and identify no cause and no individual behavior.

Read the full finding Sources and calculations

  • Not resolved here
  • These are ecological, county-aggregate estimates that identify no individual cause or voter behavior.

How sensitive is the distance association to how the model is built?

Dropping the Portland facility location, which reassigns six county centers, leaves the distance term at about p 0.098, no longer conventionally significant. A 5,000-shuffle permutation test finds the observed distance coefficient exceeded in only 3 of 5,000 shuffles, but that procedure shuffles the whole outcome and so tests a null with no association to either variable, not one conditioned on the covariate. A 50-mile threshold indicator, covering 26 of the 39 counties, is also associated with the rate but measures county-center geometry rather than any postal route or pickup schedule.

Distance results vary across model specifications

Preliminary finding.

Dropping the Portland facility location, which reassigns six county centers, leaves the distance term at about p 0.098, no longer conventionally significant. A 5,000-shuffle permutation test finds the distance coefficient exceeded in 3 of 5,000 shuffles (fraction 0.0006), but the shuffle randomizes the whole outcome rather than testing a null conditioned on the other covariate. A 50-mile threshold indicator, covering the 26 of 39 counties beyond that distance, is also associated with the rate but measures county-center geometry, not any postal route or pickup schedule.

Is the distance finding the same in every version of the model tested? No. Dropping one facility location leaves the distance term at about p 0.098, which is not conventionally significant, showing the result depends on how the model is specified.

The distance estimate stays positive in every specification tested, but it is not conventionally significant in all of them: dropping the Portland facility location, which reassigns six county centers, leaves the estimate at 0.187 per 1,000 mail ballots per mile with a conventional p of about 0.098.

Read the full finding Sources and calculations

  • Not resolved here
  • Whether the association would hold under an official routing dataset or a different facility set is not settled.

How do Too Late rates compare across the different ways ballots were returned?

In the August 25, 2026 snapshot, Drop Box returns show 1,055 Too Late out of 1,316,978 (0.801 per 1,000), while the Mail label alone shows 19,111 of 591,239 (32.324 per 1,000). This article's mail channel combines the Mail and Non-Standard Mail labels, giving 19,141 of 592,576, which is why its mail figures differ slightly from the Mail label alone. The field records the return method a county recorded, not how or when a ballot physically arrived, and the file has no postmark field.

Too Late rates differ by the return method a county recorded

Across all return methods in the August 25, 2026 snapshot, Drop Box returns show 1,055 Too Late out of 1,316,978 (0.801 per 1,000), while Mail alone shows 19,111 of 591,239 (32.324 per 1,000). This article's mail channel combines Mail and Non-Standard Mail, giving 19,141 of 592,576. The field records the return method a county recorded, not how or when the ballot actually arrived, and the file has no postmark field.

What return methods does the state track, and how do their late rates compare? The snapshot labels methods including Mail, Drop Box, Email, In Person, Non-Standard Mail, Non-Standard Dropbox, Fax and Undeliverable; Drop Box shows 0.801 Too Late per 1,000 while Mail shows 32.324 per 1,000.

This article's mail channel combines the state file's Mail and Non-Standard Mail return-method labels, which is why its mail totals, 19,141 of 592,576, differ from the Mail label alone, 19,111 of 591,239.

Read the full finding Sources and calculations

  • Not resolved here
  • What causes the difference in rates between return methods is not addressed by this data.

Study scope

Population
Washington mail ballots and, for the EAVS series, all 39 Washington county jurisdictions reporting to the federal Election Administration and Voting Survey; for the county snapshot and distance table, all 39 Washington counties in the state's August 25, 2026 final ballot-status file.
Place
Washington (ANSI state FIPS 53, boundary 39 counties, 2020 Census boundaries)
Period
Federal general elections 2018-2024 and the August 2024 and August 2026 state primaries: January 1, 2018 to August 25, 2026 (event date)
Units
ballots, ballots per 1,000, percent, miles, people, correlation coefficient
Excludes
Every partisan variable and every predictive or causal model: county partisan-share correlations, joint regression, count models, permutation tests, leave-one-out analysis and any 50-mile threshold group comparison. The claim that both EAVS denominators are equally legitimate conventions in every use. The internal legacy 1,890,736 denominator, its 32.508 percent figure and the 2,137 gap. The specific December 24, 2025 policy date. 2016 EAVS. Any causal statement about why any ballot arrived late. Any individual ballot or voter record.
Definitions
Deadline-rejected mail ballot (EAVS)
A mail ballot a county reports rejecting specifically because it arrived after the statutory deadline, in the EAVS field named for each year: C4b in 2018 and 2020, C9b in 2022 and 2024. Version EAVS 2018/2020/2022/2024 public-release files.
Mail ballots returned (EAVS)
EAVS field C1b: mail ballots returned by voters, regardless of outcome. This is a different, larger figure than the reported-counted total. Version EAVS 2018/2020/2022/2024 public-release files.
Reported counted (EAVS)
The federal file's own counted-ballot column: C3a in 2018 and 2020, C8a in 2022 and 2024. This is not the same as returned minus all mail rejections, which is a derived residual rather than a published field. Version EAVS 2018/2020/2022/2024 public-release files.
Ballots counted (primary)
The certified or officially reported total ballots counted for a Washington statewide primary, as published in the Secretary of State's turnout records. Version 2026-09 edition.
Late-rejected ballots (news-reported)
The number of ballots reported as not counted because they arrived after the deadline, as stated in a September 2, 2026 news account of a state press event. Not confirmed in any retained first-party state record. Version 2026-09 edition.
Too Late (county ballot-status snapshot)
The challenge reason code the state's ballot-status file assigns to a rejected ballot found to have arrived after the statutory deadline in RCW 29A.40.110. Version Statewide ballot status report, 2026-08-25 final.
Mail channel
In the county ballot-status snapshot, ballots returned by the 'Mail' or 'Non-Standard Mail' return method, as distinct from drop box, in-person, fax or email returns. Version Statewide ballot status report, 2026-08-25 final.
Population-weighted county center
The U.S. Census Bureau's 2020 population-weighted mean center for each Washington county: the geographic point where the county's population would balance if every resident weighed the same. Version 2020 Census.
Curated USPS facility list
Three mail-processing facility locations (Seattle, Spokane and Portland, Oregon) identified by web research and geocoded via OpenStreetMap Nominatim. This is a curated list built for this study, not an official USPS routing dataset, and the Spokane location is approximate. Version 2026-09 edition.
Too-late statutory rule
RCW 29A.40.110: a ballot must either be received by 8:00 p.m. on election day or be postmarked no later than election day; where the postmark is missing or illegible, the date on the voter's signed declaration determines validity. Version Effective until 2029-01-01.
USPS postmark-possession statement
A sentence officials attribute to the Postal Service: the postmark date does not inherently or necessarily align with the date the Postal Service first accepted possession of the mailpiece. Quoted in a U.S. Senate press release and a news article; neither document is a Postal Service directive. Version 2026-09 edition.
Retrospective specification
A statistical fit over the 39 county rows of one past election's ballot records. Each joint fit carries two terms: distance in miles, and the county's recorded two-ticket share of the November 2024 presidential vote. The rate fits are ordinary least squares on Too Late ballots per 1,000, so a distance estimate is per 1,000 per mile. The count fits use a log link with the returns as an exposure offset and the quasi-Poisson dispersion convention, so a distance estimate there is a log-rate coefficient. Version 2026-09 edition.
Recorded two-ticket share, November 2024
One ticket's recorded county votes divided by the sum of both named tickets' recorded county votes in the November 2024 presidential election, from the state's official county results export. It excludes every other candidate and every write-in, so it is not a share of all presidential votes, of registered voters or of 2026 primary voters. It is a count of votes already cast, used as a covariate in a retrospective fit. Version 2024 general election export.
County center beyond 50 miles
True when the straight-line distance from the county's 2020 population-weighted mean center to the nearest of the three curated facility locations is more than 50 miles. It is geometry between two coordinates. It locates no post office, assigns no route and observes no pickup schedule. Version 2026-09 edition.

Updates

Full history

Unanswered questions

Ask a question

Follow this question

Last studied September 12, 2026. No update schedule is committed.

Follow updates in your feed reader. Research feed for this question

https://pennerstrategy.com/research/washington-late-ballots/feed.xml

See what changed

Later changes

Updates to this question. Corrections and withdrawals are listed with the current answer.

Use this research

Penner Strategy. How often Washington mail ballots are rejected for arriving late. Version 0001. September 12, 2026. https://pennerstrategy.com/research/washington-late-ballots/editions/0001/. Qualification: The EAVS deadline-rejection rate can be computed on two different published columns of the same federal file, mail ballots returned or mail ballots counted, and which one is used changes the published rate.