Corrections and updates

Every correction is listed here with its date: what was wrong and what is right. Updates record what was added and when. To report a possible error, or anything you think breaks our neutrality standard, write to support@spotbookr.com; we answer here.

• Update 2026-11-01: Baseline report: the plain-English Summary grew to about 5,200 words: numbered source citations with a source list, a How to read this box for each part (what the figures mean, why it matters, what not to conclude), why the report matters, a worked example reading two counties, a plain guide to election night, how to use the interactive study, questions readers ask, and a state-by-state table. It can be downloaded as a PDF. The Added Nov 1 tab is now called Addendum and lists additions under the date each was added.
• Update 2026-11-01: Baseline report: the Summary tab is now a full long read in plain English (about 3,200 words) with a proper headline: how to read the numbers, the national picture in 2024, how 2024 compared with 2020 (closest states, largest and smallest moves), who turned out, every one of the 23 measures in plain words with whether the pattern holds inside most states, counting people instead of counties, the newer figures, campaign money, what the report cannot tell you, how it was kept neutral, and a glossary. Written from fixed sentences filled in from the figures.
• Update 2026-11-01: Baseline report: added a Summary tab, first in the bar, that explains the report in plain English: how to read margins and patterns, why the study describes places and not people, where things stood after 2024, the clearest patterns between what places are like and how they voted, why counting people instead of counties changes some results, turnout and campaign money, and what the report cannot tell you. It is written from fixed sentences filled in from the figures, like the rest of the report.
• Update 2026-11-01: Campaign money views now name their office: 'U.S. Senate, by state' (states only; districts are off because Senate races are statewide; states with no 2026 Senate election are striped), 'U.S. House, by district' and 'U.S. House, by state'. Previously one view showed the Senate for states and the House for districts without saying so. The baseline addendum adds two money maps (Senate by state, House by district) and a count: Democratic candidates raised more in 20 of 35 Senate races and 237 of 436 House districts, Republican candidates in 15 and 199.
• Update 2026-11-01: Added a second campaign money view, 'U.S. House, by state', summing every House district so all 50 states and DC are shown (16 states have no U.S. Senate election in 2026). DC's delegate race now carries its FEC figures.
• Update 2026-11-01: Added a map view, 'Campaign money 2026: which party's candidates raised more': each state with a 2026 Senate race and each House district coloured by the party whose candidates reported raising more to the FEC (candidate committees only; party and outside spending not included), with the same comparisons, charts, summaries and CSV download as the other views.
• Update 2026-11-01: Published a Methods page: every source, how each figure is built (margins, districts with new lines, reliability ratings, recent figures, turnout), how comparisons and findings work (correlation, weighting, uncertainty ranges, thirds and fifths, similar places, neighbours) and what the study does not do. Linked from the study and the baseline report.
• Update 2026-11-01: Baseline addendum: every correlation with the 2024 presidential margin is now also shown weighted by county population (each resident counts equally, not each county), with a 95% range for the unweighted figure (Fisher's z, read as a guide since counties are not independent). The largest difference is for population change since 2020: -0.03 with each county counted once, -0.33 weighted by population. Original sections unchanged.
• Update 2026-11-01: Baseline report: added a dated addendum tab ('Added Nov 1') with the more recent figures (September 2026 unemployment and its change from a year earlier, population change to July 2025) written up the same way as the original measures, two charts, and a summary of campaign money reported to the FEC. The original sections and their data file are unchanged. Tab labels shortened so all tabs fit on one line.
• Update 2026-11-01: Added campaign money: for each state's U.S. Senate race and each House district, the totals every candidate's campaign committee reported to the Federal Election Commission for 2025-2026 (raised, spent, cash on hand), through each candidate's latest report, largest raised first, with candidates on the state's official ballot marked. Money spent by parties and outside groups is not included. Also fixed: links to research.spotbookr.com typed without https:// now open the page (the redirect was broken since this morning).
• Update 2026-11-01: Added more recent figures beside the Census 2020-2024 averages, so recent change in a place shows: the unemployment rate in September 2026 and its change from September 2025 (Bureau of Labor Statistics, preliminary, not seasonally adjusted, so the same month is compared a year apart), and population change from July 2024 to July 2025 and from April 2020 to July 2025 (Census Bureau estimates). States are summed from counties; districts are built from the counties they contain, weighted by residents, and marked as our estimate. They are available in the chart's extra measures and every area's report. The baseline report published before the vote is unchanged.
• Update 2026-11-01: Added a 'Limits of this baseline' section to the preliminary report: the Census figures are 2020-2024 averages and may not show changes of the last year or two, and campaign finance and polls are not included yet. More recent figures (ACS 1-year 2025, July 2025 population estimates, monthly county unemployment) and pre-election FEC filings will be added as dated addenda, with the original text kept.
• Update 2026-11-01: Ready for election night: once votes are counted, the 2026 map shows who is ahead so far in each state's U.S. Senate race (Governor where there is none) and each House race, marked unofficial; we do not call races or show anyone else's calls. The page fetches new counts every two minutes without a reload and shows when results were last read from the states. Every state's panel links to its official results; states whose results are not read here say so.
• Update 2026-11-01: Change to the neutrality standard, section 8: it said two readers with different political views would review every screen before the first edition. We are not set up to do that yet, so the standard now says what we do: every page is checked automatically against the list of words we do not use and read against the standard by the Spotbookr team before it is published. The rest of the standard is unchanged.
• Update 2026-10-31: The income filter's range now follows the places still in play: with a 'Compare with' highlight on, it spans only the highlighted places (for example, the counties in the highest fifth for poverty run from $22,188 to $80,263), so it never offers an income none of them has. Also: a new header, and search-engine files (robots.txt, sitemap.xml, llms.txt) with structured data so the study can be found in search and dataset search.
• Update 2026-10-31: Added a median household income filter to the map: switch it on and drag the two handles to choose a range; places outside it fade, the vote keeps its colours, and a line counts how the places in range voted. It combines with the 'Compare with' highlight, and its ends follow the places in view (a chosen state's counties, or every place at the level). Census estimates, 2020 to 2024.
• Update 2026-10-31: Every page now has a bar back to the interactive study and to Spotbookr's case studies, the spotbookr.com footer and the Spotbookr icon, and pages have shorter addresses (for example /standard instead of /standard.html; old links still work). The baseline report's address is /baseline-2026-midterm. This layout change also changes the fingerprint of the file findings-plan.html logged earlier today (abd648f6...); the plan's wording did not change. So that layout changes can never affect the check, the plan's text is now also published as a plain text file, /data/findings-plan.md, with SHA-256 fingerprint 6c05f12519ea963f0c96b1ec95bff4b66220fc1b609ea9cd4fd6ff1a95fd5e21. Any change to the plan's wording will be dated and listed here.
• Update 2026-10-31: Published before any 2026 vote is counted: 'What we will report' (findings-plan.html), which fixes the questions, methods and wording rules for our findings, and the preliminary report 'Before the vote: the baseline' (baseline.html), which applies them to 2024 with maps and charts for every Census measure. SHA-256 fingerprint of findings-plan.html as published: abd648f6249669c8474860bf3283dd45d561e219ed58fbe682798cbd497a1b5b. Anyone can download the page and check this fingerprint; any change to the plan will be dated and listed here.
• Correction 2026-10-31: The map highlight could pick out far more than a fifth of places when many share the lowest figure (often 0%, for example small population groups in rural counties): every place at that figure counted as the highest fifth. Now only places above it count, and the key says 'above' instead of 'or more'. Figures and text elsewhere were not affected.
• Update 2026-10-31: Added: who voted in 2024, by group, for the whole country and every state (U.S. Census Bureau, Current Population Survey November 2024 supplement): the share of citizens who reported voting by race and Hispanic origin and by age (and, for the whole country, by education), with margins of error; groups with a margin of 10 points or more are marked unreliable. It shows who voted, never for whom.
• Update 2026-10-31: The map comparison is now a highlight instead of circles: choose a 'Who lives here' figure and the places in its highest (or lowest) fifth keep their vote colours while the rest fade, with a count of how each fifth voted. This compares places, not people: votes are counted by place, and no official count says how members of a group voted. The chart's fourth measure now draws a second chart of the same vote against that figure, beside the first.
• Update 2026-10-31: Changed how comparisons work. 'Show' now offers election results only and colours the map in the same red and blue whatever it is compared with. 'Compare with' offers only who lives in a place (population, age, race and ethnicity, degrees, poverty), shown as circles on the map in fifths of the places in view. Other figures about the place (unemployment, income, commuting, broadband, housing costs, median age) are available as the chart's dot size and in the written report. The chart's dots now take each place's vote colour, as on the map. The neutral two-measure colour grid and the Swap button were removed. Older shared links open with their measures in the new places.
• Update 2026-10-31: Added written reports, generated by formula the same way for every place and comparison. Under the chart: the relationship in words with its r (strength terms defined), the trend per step, averages by thirds, the highest and lowest places on each measure, the places furthest from the trend line, the chosen place against the trend, and the extra colour and size measures by thirds. In each area's panel, a full report: its rank on every measure nationally and within its state, the figures that set it apart, its neighbours (from shared borders) with their margins and incomes, the five places most like it on eleven Census figures (votes not used to find them) and, for a state, what is inside it. Each report has a Copy button that adds the source and licence. 'Compare with' now lists Census figures first, and a Swap button exchanges the two measures.
• Update 2026-10-31: Added: the scatter chart can carry two more measures (dot colour and dot size), with a table of how every pair of the chosen measures moves together, over the same places as the chart. The map stays at two measures. The two-measure map key was redrawn: each third is labelled with its range, vote margins read 'more Republican' and 'more Democratic', and the key notes when a measure is a count of people rather than a share.
• Update 2026-10-31: Added: a full breakdown for every county and district (its Census figures beside its state and the whole country, presidential results beside its state and the country, and for a county the 2026 House districts it lies in with the share of its residents in each; for a district, the counties it covers). Comparisons now follow the place you choose: with a state picked, the scatter chart, the map's colour thirds and the 'In brief' comparison use that state's own counties or districts. 'In brief' now starts from what the map is showing, ranks each county and district within its state as well as nationally, and the chart and legend say which way a vote margin runs (below 0: toward the Republican candidate). National Census figures (ACS 2020-2024, United States) and county-by-district population shares (Census ACS county parts; 2020 census blocks for the nine states with new lines) were added as sources.
• Update 2026-10-31: Added: compare any two measures on one map (for example the change in presidential margin from 2020 to 2024 with median household income), with a scatter chart and correlation; the new measure 'President: change in margin, 2020 to 2024'; and an 'In brief' summary for every state, county and district, written automatically from its figures, the same way for every place.
• Update 2026-10-31: Map display fixed: on first load the state shapes were not filled, so only the borders between states showed and coastlines were missing. No figures were affected. State outlines and borders are now drawn in grey so they show on every state. Also added: zoom (buttons, pinch, drag; Ctrl or Cmd with the scroll wheel) and a scrolling area panel.
• Update 2026-10-31: Preview published: the November 3 ballot as published by the states that have released it, past results (2018 to 2024) and Census figures for every state, county and congressional district. Results will be added from election night.
• Update 2026-10-31: Neutrality standard adopted and published.