Fact-checked by Grok 2 weeks ago

FiveThirtyEight

FiveThirtyEight, stylized as 538, was an American website founded by statistician in that specialized in quantitative forecasting and analysis of political elections, sports outcomes, economic trends, and scientific phenomena using statistical models and poll aggregation. It achieved early prominence for its accurate projection of Barack Obama's presidential victory, which relied on probabilistic modeling to outperform traditional punditry. The site expanded under affiliations with (2010–2012), ESPN (2013–2018), and (2018–2025), emphasizing empirical data over narrative-driven reporting, though it faced ownership shifts and Silver's departure in 2023 to pursue independent projects. FiveThirtyEight's defining methodology involved proprietary algorithms like its election forecasting model, which integrated national and state-level polls, economic indicators, and historical data to generate win probabilities, influencing public discourse on electoral odds. Notable achievements included high accuracy in the 2012 U.S. presidential election forecast and contributions to via tools like for baseball projections, originally developed by Silver. However, it drew significant controversy for its 2016 presidential model, which assigned only a 28.6% chance of victory despite his win, prompting critiques of overreliance on polls that underestimated non-college-educated and rural support, as well as potential institutional biases in sampling that mirrored broader polling errors averaging within historical norms but failing to capture populist shifts. The outlet ceased operations in March 2025 when , under , shuttered it as part of widespread media layoffs, ending its role as a key aggregator of polling data and approval ratings amid declining trust in establishment .

Methods and Models

Poll Aggregation and Pollster Ratings

FiveThirtyEight's poll aggregation methodology involved selecting publicly available polls that met specific quality criteria, such as excluding hypothetical matchups, tracking poll overlaps, and surveys with evident flaws like non-probability sampling without proper weighting. Polls were weighted based on factors including sample size (using the square root of the sample divided by the median sample size), recency (de-weighting multiple polls from the same firm within 14 days, e.g., halving weights for two polls), and pollster quality derived from their ratings system. Adjustments were applied for house effects—systematic biases unique to each pollster—via a Bayesian method that shrunk estimates toward the overall average, particularly for pollsters with limited data. Additional refinements included population adjustments using generalized additive models, trendline estimation via a mix of exponentially weighted moving averages and kernel-weighted polynomial regression optimized for historical accuracy, and, as of late 2023, a 2.4-point correction for partisan-sponsored polls shown to overestimate Democratic support. Outlier polls received a minimum weight of 0.05 after enhanced detection using kernel methods, and uncertainty was quantified with 95th-percentile error bands reflecting potential future deviations. These averages powered forecasts by blending polling data with fundamentals like economic indicators, though the model distinguished between published averages (smoother for public display) and internal versions allowing more responsiveness to recent shifts. The site's pollster ratings system, updated in 2023 and refined in 2024, assigned letter grades from A+ to F based on a composite of historical accuracy, methodological transparency, and polling volume. Accuracy was the primary component, measured through predictive plus-minus, which combined simple prediction error (root-mean-square deviation from actual election margins) and bias (systematic partisan lean, e.g., overestimating Democrats or Republicans) adjusted for race difficulty and herding toward consensus forecasts. Predictive error incorporated penalties for opaque methods and convergence to averages, with scores reverting toward the mean for pollsters with fewer than 20 recent polls, yielding provisional grades like A/B. Transparency scores (0-10) rewarded detailed disclosures on sampling, weighting, and response rates, with bonuses for adherence to standards from the American Association for Public Opinion Research (AAPOR) or deposition in the Roper Center. Pollsters failing basic thresholds, such as those involved in data fabrication, received an F and exclusion from aggregates; the system drew from polls since 1998 conducted within 31 days of elections. Unlike earlier versions that factored in pollster consortia like the National Council on Public Polling, the 2023 update emphasized individual performance and predictive modeling, feeding directly into aggregation weights to prioritize reliable firms while downweighting those with persistent errors or biases. This approach aimed to mitigate known polling challenges, such as nonresponse bias favoring Democrats in telephone surveys, though critics noted it could underpenalize pollsters with short-term luck over long-term methodological rigor.

Probabilistic Forecasting Framework

FiveThirtyEight's probabilistic forecasting framework centers on simulation-based models that generate probability distributions for outcomes by integrating empirical polling with structural fundamentals and accounting for multifaceted sources of . Rather than deterministic predictions, the approach employs methods to run thousands to tens of thousands of simulated scenarios, deriving win probabilities as the proportion of simulations in which a or secures victory. This framework, applied across presidential, congressional, and gubernatorial races, uses Bayesian techniques to blend inputs, treating fundamentals as informative priors that adjust polling estimates when is sparse or volatile. Polling data forms the core input, aggregated through multilevel models that weight surveys by pollster historical accuracy, recency, sample composition adjustments (e.g., for likely voters), and house effects to mitigate systematic biases. For instance, national and state-level polls are correlated based on shared demographics, , and history, with errors modeled using fat-tailed distributions to capture outliers like polling misses in or 2020. Non-polling fundamentals—such as economic indicators (e.g., GDP growth weighted over two years with temporal decay), incumbency advantages, (with in-state dollars emphasized fivefold), lean from past results, and generic ballot trends—provide anchors, typically weighted at 17.5-35% depending on the race type to prevent overreliance on noisy polls. In congressional models, tools like the system infer unpolled district outcomes from similar areas, while expert ratings from sources like the Cook Political Report refine "Deluxe" variants. Uncertainty is modeled hierarchically: local errors (e.g., district-specific noise at 75% of total variance in House races), regional correlations, national swings, and temporal drifts (e.g., convention bounces fitted empirically). Simulations incorporate these via for presidential forecasts, propagating errors across states or districts to reflect real-world dependencies, such as correlated polling biases in swing areas. Probabilities emerge directly from the simulation tally—for example, a 60% win chance indicates victory in 6,000 of 10,000 runs—enabling forecasts like distributions or seat projections with credible intervals. This setup evolved from Nate Silver's initial 2008 model, emphasizing empirical over narrative-driven adjustments, though critiques note potential underestimation of tail risks in low-probability events. The framework extends beyond elections to sports and , adapting ratings or models with similar probabilistic simulations, but political applications highlight its strength in handling sparse through strength-borrowing across units. Validation against historical outcomes shows aggregated accuracy, with polling errors averaging around 3.8 points on margins, though systemic challenges like nonresponse persist.

Transparency Practices and Methodological Evolution

FiveThirtyEight emphasized transparency in its analytical processes by publicly detailing methodologies for key components such as poll aggregation and pollster evaluations. For pollster ratings, introduced in 2008, the site assigned grades based on historical accuracy since 1998, incorporating factors like simple polling error (deviation from actual election margins), (systematic partisan skew), (excessive similarity to other polls), and methodological quality, with boosts for adherence to standards from the American Association for Public Opinion Research (AAPOR) Transparency Initiative or deposit at the Roper Center. These ratings were supported by publicly available datasets on , including raw poll data and calculation scripts, enabling external verification. Additionally, the site maintained a polls policy committing to honest, comprehensive accumulation and of polling data, with FAQs explaining inclusion criteria like recency and pollster performance. A dedicated GitHub repository housed data and code underlying articles, graphics, and forecasts, covering topics from election polls to , though updates to sports-related content ceased by June 13, 2023. This release practice facilitated , with specific folders for pollster ratings data allowing users to replicate scores using provided R scripts and historical results. Pollster evaluations explicitly factored in , penalizing firms for opaque practices such as undisclosed weighting or sampling methods. Methodologically, FiveThirtyEight's approaches evolved in response to empirical polling challenges, particularly after the 2016 election where national polls underestimated Donald Trump's support by an average of 1.5 percentage points in final surveys. Initial models under relied heavily on state-level poll aggregation with adjustments for pollster quality and economic indicators; post-2016, the site incorporated lessons on nonresponse bias, enhancing weights for demographics like education and rural voters, while pollsters broadly shifted toward better likely voter screens and online panels. By 2020, the presidential forecast model diverged from 2016 versions by integrating impacts on turnout, expanding simulation ensembles to 80,000 per day, and applying house effects to correct for pollster-specific biases observed in prior cycles. Pollster ratings underwent significant refinement, transitioning from earlier trust-based metrics to a update emphasizing predictive error and adjusted for race difficulty via models and 1,000 bootstrapped simulations to mitigate luck. This iteration introduced a score derived from 10 questions on (e.g., sample composition, weighting details), combined with empirical performance into a POLLSCORE and star ratings (0.5-3.0), replacing subjective elements with Pareto-optimal rankings. Post-2020 analyses, where polls again erred by underestimating margins in key states, prompted further adjustments in forecasts to widen intervals and incorporate correlated errors from past cycles like 2016. These changes reflected a data-driven , prioritizing empirical validation over static assumptions, though critiques noted persistent challenges in capturing shifts in voter .

Founding and Early Development

Origins in Baseball Analytics and 2008 Election

, who earned a in from the in 2000, initially applied statistical methods to baseball analytics while employed as a consultant at . Bored during downtime, he developed , an acronym for Player Empirical Comparison and Optimization Test Algorithm, a system designed to project players' future performance by comparing them to historical "comparable" players based on attributes like age, playing time, and statistical similarities. was first published in 2003 through Baseball Prospectus, an online publication dedicated to , where Silver became a contributor and eventually managing partner. The model gained recognition for its accuracy in forecasting player statistics and team outcomes, outperforming many traditional scouting methods by incorporating empirical data and probabilistic adjustments for variables such as injuries and performance regression. Silver's experience with PECOTA honed his approach to handling noisy data and generating probabilistic forecasts, skills he later adapted to political polling aggregation. As the 2008 U.S. presidential primaries unfolded, Silver began analyzing election polls under the pseudonym "Poblano" on platforms like , applying multilevel regression techniques similar to those used in projections to weigh pollster track records, sample sizes, and historical biases. These early analyses demonstrated high accuracy, such as correctly anticipating Barack Obama's delegate lead over despite conventional media narratives favoring Clinton. On March 7, 2008—three days after Clinton's victories in the and Democratic primaries—Silver launched FiveThirtyEight.com as an independent focused on aggregating and modeling and state-level polling to forecast election outcomes. The site's name referenced the 538 electors in the , emphasizing its emphasis on electoral math over raw vote margins. Silver's model integrated hundreds of polls, adjusting for house effects (systematic biases by polling firms) and economic indicators, yielding forecasts that diverged from by prioritizing empirical aggregation over narrative-driven interpretations. FiveThirtyEight achieved prominence during the 2008 general election, with Silver's final projection on November 4, 2008, giving a 349-189 electoral vote edge over , closely mirroring the actual 365-173 result (after Nebraska's split). The model accurately predicted outcomes in 49 of 50 states and all but one , attributing success to rigorous poll weighting and rather than overreliance on late-deciding voter assumptions common in mainstream analyses. This performance established Silver's reputation for data-driven forecasting, contrasting with outlets that underestimated Democratic turnout based on less systematic methods.

Initial Blogging and Ground-Level Reporting

initiated the FiveThirtyEight blog in late 2007, with its early focus on of Democratic primary polls, expanding to by March 2008. The blog's content emphasized aggregating hundreds of polls, weighting them by sample size, recency, and pollster track records, while incorporating economic indicators and historical voting patterns to generate state-by-state projections. Silver updated forecasts multiple times daily during peak periods, often posting 5-10 times per week, to reflect new data and refine probabilistic models that assigned win probabilities to candidates. This approach contrasted with traditional punditry by prioritizing empirical aggregation over narrative speculation. Early blogging delved into campaign dynamics beyond raw polls, including examinations of models and field operations inferred from available data. For instance, Silver analyzed polls from the 2008 election to quantify the "contact gap," finding that Obama voters were contacted by their campaign at rates 10-15 percentage points higher than McCain voters in key states, correlating with higher turnout among demographics like young and minority voters. He argued this demonstrated the causal impact of ground-level mobilization efforts, such as door-knocking and phone banking, on margins in battleground areas. Such posts highlighted how superior data-driven targeting could amplify small polling edges into electoral advantages. Silver's ground-level insights extended to critiquing underemphasis on mechanics, noting in September 2008 that coverage often fixated on national horse-race numbers while ignoring efficacy metrics like volunteer recruitment rates. He incorporated non-poll data, such as FEC filings on spending allocation to versus ads, to assess strategic resource deployment. These analyses, drawn from public datasets rather than original shoe-leather reporting, positioned the blog as a to anecdotal dispatches, emphasizing verifiable causal links between on-the-ground actions and vote shares. The blog's in , including public disclosure of adjustment formulas, built credibility amid skepticism from established outlets.

Ownership and Organizational Changes

New York Times Partnership (2010-2013)

In June 2010, reached an agreement in principle with to incorporate FiveThirtyEight's content into NYTimes.com, positioning Silver as a regular contributor focused on election forecasting and statistical analysis. The partnership formally launched on August 25, 2010, with the relaunch of the FiveThirtyEight blog under the banner "FiveThirtyEight: 's Political Calculus," hosted on the Times' platform and benefiting from its editorial resources and audience reach. This three-year arrangement allowed Silver to maintain operational independence while leveraging the Times' infrastructure for expanded data-driven political coverage, including midterm elections in 2010. During the partnership, FiveThirtyEight emphasized probabilistic forecasting models, aggregating polls and incorporating economic indicators, demographic shifts, and historical data to predict outcomes, such as House races in the 2010 elections where it projected results based on polling, , and past returns. The site's approach, rooted in Silver's quantitative methods, gained prominence for its transparency in model assumptions and updates, contrasting with more narrative-driven at the Times. However, this data-centric style occasionally clashed with traditional Times sensibilities; the public editor noted in 2013 that Silver's probabilistic framing—treating elections as uncertain rather than foregone conclusions—went "against the grain for some at the Times," particularly among those preferring interpretive reporting over strict . The partnership peaked during the 2012 presidential election, where FiveThirtyEight's final forecast on November 5 gave a 91% chance of victory, accurately predicting his national popular vote margin within 0.9 percentage points and correctly calling the winner in 50 states and the District of Columbia—though critics later debated the model's overconfidence in safe states versus tighter races. Traffic surged, with Silver's posts drawing millions of unique visitors, underscoring the appeal of empirical polling aggregation amid partisan media skepticism. Yet, internal tensions arose over editorial control and Silver's resistance to Times style conventions, such as softening probabilistic language for readability, which some viewed as diluting rigor. The arrangement concluded in July 2013 when Silver announced his departure for and , citing a desire to expand into and secure greater autonomy under a multiyear deal ending his Times contract in August. The move, speculated to stem from frustrations with Times bureaucracy and limited scope for interdisciplinary work, marked the end of FiveThirtyEight's integration with the newspaper after three years of symbiotic growth in .

ESPN and ABC News Acquisition (2013-2023)

In July 2013, ESPN acquired FiveThirtyEight and hired its founder Nate Silver as editor-in-chief, integrating the site into ESPN's digital portfolio while allowing it to maintain its independent branding at fivethirtyeight.com. Silver's role extended to contributions across ESPN and ABC News platforms, with the acquisition enabling expansion into sports, economics, culture, science, technology, and politics through data-driven analysis. The relaunched FiveThirtyEight debuted on March 17, 2014, with an initial team of approximately 20 full-time staff members building out content in , including forecasts for the 2014 midterms and 2016 presidential election. Under ESPN ownership, the site grew its scope beyond politics to emphasize probabilistic modeling in and economic projections, aligning with ESPN's sports focus while retaining election coverage. This period saw FiveThirtyEight produce interactive visualizations and poll aggregations, though some critics noted tensions between its quantitative approach and ESPN's broader entertainment-oriented content strategy. On April 17, , ownership transferred from to , both under , to better align FiveThirtyEight's political and with ABC's news operations amid rising demand for analytical coverage ahead of the 2018 midterms. The move allowed continued sports content via partnership with , while prioritizing enhancements in statistical reporting for ABC's platforms, including regular appearances by Silver and his team. From 2018 to 2023, under maintained its core forecasting models, producing coverage for the and 2022 elections with expanded data tools like presidential approval trackers. In May 2023, appointed from to lead data journalism efforts, integrating 's poll aggregation with ABC's broader analytics. By September 2023, the site's content shifted to a dedicated vertical on abcnews.com/538, reflecting deeper organizational integration while preserving its focus on empirical political analysis.

Nate Silver's Departure and Final Years (2023-2025)

In April 2023, , founder and editor-in-chief of FiveThirtyEight, announced his departure from amid broader layoffs at parent company , stating that his contract was set to expire without renewal. The exit followed tensions over editorial direction and resource allocation, with Silver citing Disney's cost-cutting measures as a key factor impacting the site's operations. Concurrently, Disney reduced FiveThirtyEight's sports coverage and laid off roughly two-thirds of its staff, merging remaining content into ' main platform while diminishing the site's independent identity. Despite these changes, FiveThirtyEight persisted in a scaled-back form through 2023 and 2024, continuing to produce political forecasts, poll aggregations, and tools like presidential approval trackers during the 2024 election cycle. Silver, who had reacquired pre-2023 and forecasting models (excluding the brand name) as part of his original agreement, shifted focus to independent projects, including a on his site natesilver.net. On March 5, 2025, shuttered FiveThirtyEight entirely as part of -wide restructuring affecting approximately 200 employees across its units, with the site's 15 remaining staff informed of layoffs via a scripted announcement from ABC . Silver publicly attributed the closure to mismanagement by executives, arguing that business priorities overrode the site's data-driven mission and that similar cuts had eroded its viability since 2023. Post-shutdown, assets like FiveThirtyEight's poll database were transferred to outlets including , preserving some data continuity amid the site's demise.

Key Personnel and Contributors

Nate Silver's Role and Influence

Nate Silver established FiveThirtyEight in 2008 as an independent blog specializing in probabilistic forecasting and statistical analysis, initially centered on the U.S. presidential election and drawing from his prior work in baseball sabermetrics. As founder and editor-in-chief through 2023, Silver personally constructed the site's core election models, which aggregated data from numerous polls, incorporated economic indicators and historical voting patterns, and utilized thousands of Monte Carlo simulations to produce win probabilities rather than deterministic predictions. This methodology yielded high accuracy, correctly forecasting Barack Obama's Electoral College victory in 49 of 50 states in 2008 and all 50 states plus the District of Columbia in 2012. Silver's leadership positioned FiveThirtyEight as a pioneer in data-driven journalism, shifting public and media attention toward empirical aggregation of polls over individual survey outliers or , a practice that became known as the "Nate effect." His emphasis on transparency—publishing model code, poll weights, and uncertainty intervals—fostered greater methodological scrutiny in forecasting, influencing outlets to adopt similar quantitative rigor while critiquing overconfidence in qualitative punditry. Silver also extended the site's scope beyond politics, integrating like his system for projections and economic modeling, which broadened FiveThirtyEight's audience and demonstrated the applicability of statistical tools across domains. Silver's departure from FiveThirtyEight occurred in April 2023, coinciding with Disney's layoffs at and the expiration of his contract, after which he retained rights to the original models but not the brand. His tenure defined the organization's identity as a hub for causal, evidence-based analysis, though subsequent editorial shifts under highlighted tensions between Silver's apolitical, forecast-centric vision and broader newsroom priorities favoring narrative integration. Despite these changes, Silver's frameworks continued to underpin elements of the site's output until its 2025 closure, underscoring his enduring impact on how statistical uncertainty is communicated in high-stakes reporting.

Editorial Team and Notable Analysts

G. Elliott assumed the role of Editorial Director of Data at FiveThirtyEight in May , following Nate Silver's departure, overseeing the site's models and data-driven political during its final years under ownership. , previously a data journalist, managed adjustments to election models amid criticisms of prior underestimations of conservative . Nathaniel Rakich served as Senior Editor and Senior Elections Analyst, specializing in polling aggregation, race ratings, and historical electoral data; his work included detailed breakdowns of and dynamics up to the 2024 cycle. Rakich's analyses often incorporated state-level polling averages and demographic shifts, contributing to the site's proprietary forecasts. Harry Enten, a Senior Political Writer and Analyst, gained prominence for his accessible explanations of polling discrepancies and trends, frequently highlighting over- or under-sampling of key voter groups in national surveys. Enten's commentary, delivered through articles and podcast appearances, emphasized empirical variances between polls and outcomes, such as in the 2020 and 2022 elections. Amelia Thomson-DeVeaux acted as a Senior Editor, focusing on quantitative assessments of policy impacts and candidate viability through models and . Her contributions extended to visualizations of economic indicators tied to political events. In , Neil Paine held the position of acting sports editor, building on the site's origins in with coverage of player projections and league trends using advanced metrics like and ratings. Galen Druke hosted the FiveThirtyEight , coordinating discussions among analysts on methodological debates and real-time forecast updates from onward. The editorial team contracted significantly after 2023 layoffs, dropping from approximately 35 staff to 15 by the site's closure on March 5, 2025, amid broader reductions. This downsizing followed earlier cuts that affected contributors in and roles.

Content Scope and Expansion

Political Analysis and Election Coverage

FiveThirtyEight's political analysis emphasized data-driven probabilistic forecasting, aggregating hundreds of opinion polls while adjusting for methodological differences such as house effects—systematic biases in polling firms' samples—and incorporating fundamentals like economic growth, incumbent approval, and candidate spending. The site's models simulated election outcomes tens of thousands of times daily, producing win probabilities and expected vote margins rather than point predictions, with updates reflecting new data releases. This approach extended beyond presidential races to congressional, gubernatorial, and state legislative contests, often via "Lite" models for lower-data environments. Election coverage began with Nate Silver's independent blog during the cycle, where his model forecasted Barack Obama's national popular vote share at 51.6 percent—actual: 52.9 percent—and projected 307 electoral votes, closely aligning with the final 365 after late swings. By , under partnership, the forecast gave Obama a on the eve of the election, accurately projecting 332 electoral votes against Mitt Romney's 206, with the popular vote margin off by less than one point. Coverage featured frequent updates, state-by-state breakdowns, and visualizations of polling trends, alongside explanatory articles on factors like turnout models and undecided voter behavior. Post-2012, analysis diversified to include generic congressional ballots, where FiveThirtyEight tracked Democratic leads averaging 3-5 points in even-year cycles, correlating them with seat projections via efficiency gaps and partisanship metrics. Midterm coverage, such as 2010 and , highlighted wave dynamics through historical analogies and pollster ratings, rating firms on and past errors. The site also produced non-forecast content, such as approval trackers and issue polling deep dives, using weighted averages from multiple pollsters to assess public sentiment on topics like healthcare policy shifts or foreign interventions. In later cycles, models evolved to weight recent polls more heavily and simulate correlated errors across battlegrounds, as in the 2024 forecast incorporating national and state-level data with 80,000-plus simulations per update. Coverage integrated multimedia, including podcasts dissecting swing-state dynamics and interactive maps displaying forecast densities, aiming to convey uncertainty through ranges rather than binaries. This quantitative focus distinguished FiveThirtyEight from narrative-driven outlets, prioritizing empirical aggregation over qualitative punditry.

Sports and Economics Extensions

Following its acquisition by in July 2013, FiveThirtyEight significantly expanded into , integrating statistical modeling with coverage of major professional leagues including the , NBA, MLB, NHL, and soccer. , drawing on his earlier work in baseball , emphasized this shift as a return to sports-oriented , with the relaunched site in 2014 featuring predictive models for game outcomes, player evaluations, and league trends. Examples included the Soccer Power Index (), which forecasted match probabilities and team strengths based on historical performance data, and analyses of dominance in viewer preferences spanning 1937 to 2017. The sports section grew to encompass retrospective evaluations, such as identifying the most consistently mediocre teams across five decades of data from 1968 onward, and decade-end summaries of achievements in the across multiple leagues. Coverage also addressed broader issues like Title IX's impact on equality, using enrollment and participation statistics to highlight disparities in NCAA and . Even after FiveThirtyEight's partial shift to in 2018, collaboration with persisted on sports content until ESPN discontinued the site's sports forecasting models in 2023. In , FiveThirtyEight's extensions focused on data-informed commentary rather than proprietary forecasting models comparable to those in politics or sports, with articles analyzing , policy effects, and expert consensus. Key topics included public perceptions of recessions, where surveys showed a majority of Americans believed the U.S. was in one despite official metrics in 2022; economists' concerns over post-COVID recovery risks like spending hesitancy and viral resurgence in June 2020; and debates on duration, with projections varying widely among specialists in November 2021. Additional reporting examined bailout precedents from the crisis in the context of 2023 events and tracked indicators like GDP growth and through visualizations and economist interviews. This coverage prioritized empirical aggregation of economic data over novel predictive simulations, reflecting the site's broader emphasis on verifiable trends amid institutional forecasts.

Other Topics and Data Visualization Innovations

FiveThirtyEight expanded its coverage to include science, health, and environmental topics, applying statistical analysis to issues such as perceptions and . For instance, in February 2019, the site published an analysis showing increased public worry about preceding Donald Trump's presidency, drawing on Gallup polling data from 1989 to 2018 that indicated a rise in concern from 35 percent to 62 percent among Americans. Similarly, articles examined rural Americans' growing skepticism toward scientific institutions, linking it to cultural identity shifts amid events like the , with surveys revealing lower trust in science among rural respondents compared to urban ones. The outlet also addressed technology and health-related subjects, critiquing concepts like "sound science" as a rhetorical to undermine , as detailed in a December 2017 piece that referenced historical uses in and chemical industries to delay regulations. Coverage extended to public understanding of , with a March 2019 article arguing that Americans demonstrated factual knowledge on topics like and astronomy, countering narratives of widespread scientific illiteracy using survey data. Environmental analyses included recommendations for relocation amid risks, such as a September 2019 evaluation favoring Michigan's Upper Peninsula for lower exposure to sea-level rise and based on NOAA and FEMA datasets. In data visualization, FiveThirtyEight innovated with custom, interactive graphics to convey complex probabilities and trends across topics. A notable technique was the "snake" plot for , introduced around 2016, which animated state-by-state win probabilities as a serpentine path to illustrate path dependency in the without overwhelming users with static maps. The site produced annual compilations of distinctive charts, such as 33 visualizations in 2022 covering policy shifts post-Roe v. Wade and demographic trends, often using histograms, scatter plots, and dashboards built with open-source code shared via their data portal. These visualizations emphasized transparency by releasing underlying datasets and or scripts, enabling replication and scrutiny, as seen in their 2020 retrospective of 40 unconventional charts blending science, policy, and societal data like impacts and environmental metrics. This approach distinguished FiveThirtyEight from traditional journalism by prioritizing empirical clarity over narrative flair, though it occasionally drew critique for simplifying multivariate causal relationships in favor of accessible formats.

Forecasting Accuracy and Performance

Successes in 2008 and 2012 Elections

FiveThirtyEight, founded by as a in March 2008, achieved early prominence through its model for the U.S. that year. The model aggregated national and state-level polls, adjusted for historical biases and house effects, and incorporated economic indicators and demographic shifts to simulate thousands of election outcomes. On , November 4, 2008, Silver's forecast correctly identified the winner in 49 of the 50 states, erring only on , which ultimately carried by 1.0 . The model projected Obama's national popular vote share at approximately 52.1 percent, aligning closely with the actual result of 52.9 percent, a margin within one . This accuracy contrasted with many traditional pundits and outlets that underestimated Obama's margins in key battleground states like and , where FiveThirtyEight's simulations assigned Obama probabilities exceeding 90 percent in the final days. The site's methodology emphasized empirical polling data over narrative-driven analysis, earning Silver recognition for demystifying electoral uncertainty through simulations that accounted for and late-deciding voters. By correctly forecasting Obama's 365 electoral votes against John McCain's 173—near the actual 365-173 split—FiveThirtyEight demonstrated the value of systematic aggregation over isolated polls. Building on this foundation, FiveThirtyEight's 2012 election coverage, now hosted by , refined the model to include polls-plus adjustments for turnout models and economic variables. Silver's final forecast on November 5, 2012, gave Obama a 90.9 percent chance of reelection, projecting exactly 332 electoral votes to Obama's 206 for , matching the certified outcome. The model accurately called the winner in all 50 states, including narrow victories in states like (Obama by 5.4 points) and (Obama by 0.9 points), where probabilities hovered between 70 and 85 percent. These results validated the site's emphasis on probabilistic ranges rather than point predictions, as the popular vote remained competitive (Obama 51.1 percent to Romney's 47.2 percent), yet the electoral map favored the decisively. FiveThirtyEight outperformed many competitors by weighting recent polls more heavily and correcting for Republican-leaning biases observed in 2010 midterms, fostering trust in data visualization tools like interactive maps and vote simulators that illustrated . The successes in both cycles established FiveThirtyEight as a for , influencing subsequent journalistic standards for in model assumptions and against historical data.

Failures and Adjustments in 2016 and 2020

In the 2016 presidential election, FiveThirtyEight's polls-plus forecast model assigned a 28 percent chance of victory on the eve of the election, higher than most competitors who offered odds below 10 percent. Despite this margin for error, prevailed, exposing shortcomings in underlying polls that underestimated his support in states like , , and by 4 to 6 points. Key factors included nonresponse bias among low-education, white working-class voters—a demographic core to 's coalition—and late-deciding voters breaking heavily for him (59 percent in ). FiveThirtyEight defended the forecast by emphasizing its incorporation of historical polling errors (averaging 2 points since 1972) and correlated state-level misses, arguing it avoided overconfidence unlike outlets predicting landslides. Nonetheless, the outcome prompted critiques that the model underweighted fundamentals like economic discontent in deindustrialized areas and over-relied on aggregates prone to systematic undercounting of enthusiasm. Post-2016, FiveThirtyEight and affiliated pollsters adjusted methodologies to address these gaps, prioritizing empirical corrections over narrative-driven interpretations. Pollsters introduced education-based weighting to correct underrepresentation of non-college-educated respondents, who skewed ; for instance, firms like Hart Research boosted high school-or-less samples. Sampling shifted toward address-based methods and higher cellphone penetration (e.g., from 80 to percent in some surveys) to reach rural and low-propensity voters, while geographic balancing targeted urban-rural disparities. The forecasting model amplified uncertainty parameters, simulating more scenarios for regional errors and undecided voters, reflecting causal insights from 2016's late swings (e.g., FBI Director Comey's letter). These changes aimed at causal realism by treating polling misses as non-random, often favoring Republicans due to differential turnout and response rates. By the 2020 election, these refinements yielded a polls-plus model giving Joe Biden an 89 percent chance of winning the Electoral College, with Trump at 10 percent, acknowledging a potential 3-point polling error in Trump's favor akin to 2016. Biden won as projected, securing 306 electoral votes, but national polls overestimated his margin by about 3.5 points (averaging +8 percent versus actual +4.5 percent popular vote), with errors flipping calls in Florida and North Carolina. Persistent issues included turnout misestimation amid pandemic-induced mail voting—low-propensity Trump voters surged late—and residual nonresponse among conservatives, though fewer undecideds (4.8 percent versus 12.5 percent in 2016) stabilized aggregates. Critics noted the high probability conveyed undue certainty, mirroring pre-2016 hubris, as house effects (firm-specific biases) again undervalued Republican strength in battlegrounds. Adjustments for 2020 built on by embedding correlated error simulations and education-race weighting, but empirical data revealed incomplete fixes: Republican underestimation recurred, suggesting deeper challenges like in surveys or institutional reluctance to probe Trump-aligned sentiment. FiveThirtyEight incorporated COVID-specific variables, such as mail-ballot enthusiasm gaps, yet post-election analysis affirmed that while winner accuracy improved, margin errors highlighted ongoing causal disconnects between polled samples and actual electorate composition. These cycles underscored the limits of aggregate-dependent models in capturing volatile voter mobilization, prompting further scrutiny of alternatives like precinct-level fundamentals over polls alone.

Post-2020 Models, 2024 Coverage, and Empirical Critiques

Following Nate Silver's departure from FiveThirtyEight in November 2023, the site, now under , shifted its forecasting approach with assuming leadership of data journalism efforts. The post-2020 models retained core elements like polling aggregation and simulation-based projections but incorporated adjustments from the 2020 election review, which highlighted strengths in presidential forecasting (e.g., correctly favoring with an 89% win probability) while noting overconfidence in some congressional races. For the midterms, these models demonstrated improved calibration, with national polling errors averaging under 2 percentage points and correctly anticipating Democratic overperformance in the popular vote despite House losses. However, the ABC-era emphasized economic fundamentals and historical patterns more heavily in simulations, diverging from Silver's poll-centric methodology. In the lead-up to the 2024 presidential election, FiveThirtyEight suspended its forecast in July 2024 amid the Biden-Harris transition, resuming in late August with a rebuilt model that integrated over 1,000 polls, economic indicators, and 80,000-plus simulations per update. The model weighted recent polls higher but applied adjustments for historical biases, such as nonresponse among Republican voters, and incorporated state-level correlations. Coverage focused on tight national margins, with Harris leading slightly in aggregates (e.g., +1.5 points as of October 2024), but the final forecast on November 5, 2024, assigned her a 50% chance of winning the Electoral College and 48% for the popular vote, framing the race as a coin flip despite swing-state volatility. House and Senate projections similarly showed narrow Republican edges, with Republicans at 52% for House control. Empirical critiques of the post-2020 models centered on persistent underestimation of performance in presidential races, echoing 2016 patterns. In 2024, while raw polls erred by about 1.8 points nationally (underestimating Donald 's +1.6 popular vote margin), the model's even odds failed to capture 's decisive victory (312-226), as simulations overweighted Democratic fundamentals like incumbency analogs despite polling trajectories favoring in key states like and . , analyzing the model independently, argued it underweighted polls early on—assigning Harris 58% odds in August despite averages showing a tie—and produced illogical outputs by prioritizing structural factors over data, potentially reflecting institutional pressures at . Independent evaluations, including assessments of probabilistic accuracy, ranked the 2024 forecast below Silver's standalone model, which gave 52% final odds via similar simulations but with tighter poll reliance. These shortcomings contributed to broader skepticism of aggregate-based under media ownership, with critics noting unaddressed shy- effects and sample biases in online/opt-in polls comprising 40% of inputs. The site's shutdown in March 2025 curtailed post-mortems, leaving unresolved debates on whether model tweaks sufficiently mitigated partisan nonresponse.

Controversies and Criticisms

Polling Biases and Underestimation of Conservative Outcomes

FiveThirtyEight's polling aggregates and probabilistic forecasts have exhibited a pattern of underestimating Republican and conservative outcomes in U.S. presidential elections, particularly those involving , across multiple cycles. This discrepancy manifests as consistently outperforming national and state-level polling averages by 2 to 4 percentage points, leading to forecasts that assigned him low probabilities of victory despite actual results. In the 2016 election, FiveThirtyEight's final model gave a 29 percent chance of winning the on the eve of voting, reflecting polling averages that underestimated his national popular vote share by about 2 points and missed key states by larger margins, such as 3.1 points in and 3.4 points in . Post-election analyses highlighted insufficient adjustments for nonresponse among rural and less-educated voters, who skewed , contributing to the surprise outcome. The 2020 cycle showed similar issues, with exceeding polls in battleground states like (by 3.5 points) and (by 2.6 points), narrowing Biden's projected leads and preserving competitiveness in the despite FiveThirtyEight's 89 percent probability for Biden in late forecasts. Aggregates overrepresented Democratic support in these regions, a pattern attributed to persistent sampling challenges rather than random error, as Republican performance aligned with pre-2016 baselines when adjusted for turnout. This trend repeated in 2024, where won the popular vote by 1.5 percentage points and the by 312 to 226, outperforming final polling averages by an average of 2.2 points nationally and more in swing states like (2.6 points). FiveThirtyEight's October forecast treated the race as a coin flip, with Harris at roughly 50 percent odds, yet Trump's margins exceeded expectations amid late voter shifts. Pollsters, including those aggregated by FiveThirtyEight, acknowledged undercapturing Trump's support, with errors directional rather than symmetric. Explanations for these biases center on nonresponse and reluctance among conservative respondents, who participate less in surveys due to distrust of institutions associated with and academia, leading to underrepresentation in samples. FiveThirtyEight has explored this in its own analyses, proposing theories like differential response rates—where Republicans are 5-10 percent less likely to complete interviews—and the "shy voter" effect, where supporters underreport preferences due to perceived social pressures, supported by list experiment data showing hidden conservative leanings. Adjustments for "house effects"—systematic biases in individual pollsters, often Democratic-leaning—have been applied but appear insufficient to fully correct the aggregate skew, as evidenced by consistent directional errors favoring Democrats. Critics, including empirical reviews, argue that polling methodologies, reliant on telephone and online panels with low response rates (often under 5 percent), fail to proportionally capture working-class and rural conservatives, exacerbating errors in low-propensity voter segments. While FiveThirtyEight incorporates polling error simulations (e.g., assuming a 3-4 point national standard deviation), these models have not fully anticipated the partisan asymmetry, prompting calls for enhanced weighting by , turnout history, and validation against alternative data like consumer records. This underestimation raises questions about the representativeness of pollster samples, given the left-leaning composition of many survey firms and their ties to urban, higher- demographics.

Internal Model Disputes and Editorial Interference

Nate Silver, FiveThirtyEight's founder and long-time editor-in-chief, departed the organization in June 2023 amid Disney's layoffs at ABC News, citing a contract expiration with "profound mutual disinterest" in renewal, though he emphasized that Disney had rarely interfered in editorial processes during his tenure. Post-departure, Silver publicly critiqued FiveThirtyEight's revised 2024 presidential election model for overweighting economic and political fundamentals (85 percent of the forecast) over recent polling data (15 percent), arguing this produced inconsistent and misleading probabilities, such as favoring Joe Biden in swing states despite Trump leading in aggregated polls by margins exceeding historical errors. Silver contrasted this with his own Silver Bulletin model, which allocated a 70/30 split to polls versus fundamentals, contending that fundamentals like incumbency advantages had become less predictive in a polarized electorate and carried wide error bars (e.g., ranging from Trump +20 to Biden +30 points absent polling input). These methodological differences highlighted broader philosophical tensions within FiveThirtyEight's forecasting approach, evolving from earlier models that adjusted post-2016 for polling underestimation of conservative support but shifted further after 2020 toward incorporating non-polling factors like impacts while de-emphasizing approval ratings to avoid overreacting to short-term volatility. Silver's critiques implied that the post-2023 model risked amplifying establishment-favoring assumptions, such as overvaluing incumbency in low-approval environments, potentially at the expense of empirical polling trends that had repeatedly missed strength in recent cycles. Such disputes underscored challenges in balancing data-driven rigor with interpretive layers, where heavier fundamental weighting could inadvertently embed analyst priors over raw empirical signals. Editorial decisions also drew scrutiny for potentially overriding model continuity. In late July 2024, following Biden's withdrawal, FiveThirtyEight suspended its presidential forecast—a tool it had maintained through prior transitions—citing the need for "stable" polling on , a move Silver described as politically motivated to await potentially favorable Democratic data rather than adhering to consistent probabilistic updates. This pause, lasting until mid-August, contrasted with historical practice and fueled perceptions of interference prioritizing narrative alignment over uninterrupted analysis, especially given ' parent company's institutional incentives amid a competitive media landscape prone to left-leaning interpretive biases in election coverage. While FiveThirtyEight defended the suspension as methodologically prudent to avoid noisy early Harris surveys, it exemplified how oversight could disrupt the site's foundational commitment to transparent, real-time probabilistic modeling.

Over-Reliance on Aggregates vs. Alternative Predictors

FiveThirtyEight's models have historically placed primary emphasis on aggregating and adjusting polls, weighting them by factors such as pollster track records, sample size, and recency, while incorporating secondary elements like economic indicators and historical vote patterns. This aggregate-centric approach assumes that averaging multiple polls mitigates individual errors and provides a robust signal of voter intent, but it has drawn for underemphasizing systemic flaws in polling data, including non-response among conservative-leaning voters and methodological where pollsters converge on similar results to avoid status. In the 2016 presidential , for instance, FiveThirtyEight's final forecast assigned a 29% chance of winning the based largely on national and state-level poll aggregates showing ahead, yet Trump secured victory, highlighting how aggregates failed to capture late shifts or hidden support. Critics argue that this over-reliance amplifies polling's vulnerabilities, such as left-leaning sample compositions in academic and media-affiliated surveys, which empirical post-mortems have linked to consistent underestimation of performance since 2016. A 2020 analysis by and colleagues decomposed polling errors into bias (systematic undercounting of certain demographics) and variance (random noise), concluding that 2016's misses stemmed more from bias than aggregation alone could correct, as pollsters struggled with turnout modeling for low-propensity voters. FiveThirtyEight has adjusted its models post-2016 by applying house effects and uncertainty intervals, but detractors contend these tweaks remain too poll-dependent, ignoring causal drivers like economic sentiment or cultural backlash that polls undermeasure due to respondent reluctance. Alternative predictors, particularly prediction markets, have demonstrated superior calibration in several cycles by aggregating decentralized information from bettors with financial skin in the game, incentivizing accuracy over narrative conformity. In 2016, platforms like implied Trump win probabilities around 40% on the eve of the election—higher than poll-based aggregates—aligning closer to the outcome than FiveThirtyEight's estimate. Similarly, in 2024, Polymarket odds favored at over 60% in the final weeks, outperforming traditional poll aggregates that depicted a toss-up, with post-election reviews confirming markets' edge in capturing swing-state dynamics and voter enthusiasm. Blockchain-based markets like Polymarket further reduced risks through transparent trading, yielding forecasts that integrated polls with real-time signals like betting volume, which traditional models overlooked. Financial derivatives, such as options pricing, also eclipsed poll aggregates in 2024 by embedding macroeconomic expectations, predicting Trump's win margin more precisely. These alternatives underscore a key limitation of aggregate-heavy models: their reduced sensitivity to emergent causal factors, like incumbency fatigue or partisan mobilization, which markets price dynamically without relying on potentially biased survey responses.

Impact, Recognition, and Legacy

Influence on Data-Driven Journalism

FiveThirtyEight exerted substantial influence on data-driven journalism by pioneering the integration of statistical modeling into mainstream political reporting, starting with its founding by in March . The site's early adoption of poll aggregation—weighting surveys by factors such as pollster track record, sample size, and recency—provided a systematic alternative to selective or unweighted poll citations, encouraging journalists to prioritize empirical synthesis over . This approach gained traction after the site's accurate 2008 presidential election forecast, which projected securing 349 electoral votes (actual: 365), demonstrating the value of probabilistic outputs in conveying uncertainty. The "Nate Silver Effect" prompted news organizations to elevate polling data in coverage, with aggregators like and drawing over 200 million visits in October 2016 alone, reshaping how elections were framed through averages rather than isolated results. Outlets began emulating these methods, such as requiring multiple polls before reporting findings or incorporating historical pollster ratings to assess reliability, as evidenced by increased polling volume from approximately 37,000 surveys in 2012 to 48,600 by 2016. This shift extended beyond politics, adapting tools like Silver's system—originally for player projections—to broader journalistic applications, fostering a ""-style emphasis on data over narrative intuition. FiveThirtyEight's emphasis on , expressing outcomes as percentages rather than certainties, trained audiences and reporters to grapple with uncertainty, influencing competitors like The New York Times' The Upshot and academic blogs such as The Monkey Cage. By , the site's successful state-by-state predictions further solidified this model, leading to wider adoption of data visualization and in across media. However, this proliferation highlighted challenges, as many outlets lacked in-house statistical expertise, often deferring to aggregators without independent validation, which amplified reliance on potentially flawed inputs like non-response biases in surveys.

Awards, Metrics of Reach, and Cultural Penetration

FiveThirtyEight garnered recognition for its innovative approach to data visualization and statistical analysis in journalism. In 2016, the site was awarded Data Journalism Website of the Year by the Global Editors Network, honoring its exceptional use of data in storytelling across politics, sports, and science. It also earned finalist status in the Online Journalism Awards for Investigative Data Journalism (large newsroom category) in 2016 for its Flint water crisis coverage and for Digital Video Storytelling (medium form, large newsroom) in 2022. Earlier accolades included Bloggie Awards for Best Political Coverage in 2008 and Best Weblog about Politics in 2009, reflecting its early influence on political blogging. The site's reach expanded rapidly, particularly during high-stakes election cycles, driven by its models. On November 8, 2016, FiveThirtyEight recorded an all-time high of 16.5 million unique visitors, with a peak hourly count of 3.6 million, surpassing previous benchmarks amid intense in presidential race projections. Monthly unique visitors grew from approximately 2.8 million in April 2014 to 10.7 million by January 2016, coinciding with heightened election coverage under ownership. Post-acquisition by in 2018, integration into broader platforms sustained visibility, though traffic metrics fluctuated with partisan scrutiny of its models; for instance, the site handled record loads without downtime during the 2020 election. Following its shutdown on March 5, 2025, as part of ABC/ cost-cutting measures affecting 15 staff, archived content continued to draw modest traffic, around 281,000 visits monthly as of late 2025. FiveThirtyEight penetrated popular culture by mainstreaming data aggregation and uncertainty quantification in election analysis, shifting journalistic norms away from binary predictions toward ensemble modeling. This "Nate Silver Effect" encouraged outlets to prioritize poll averaging to mitigate house effects and selection biases, influencing coverage at competitors like The New York Times' The Upshot. Its forecasts became cultural touchstones, referenced in late-night television, podcasts, and policy debates, embedding terms like "polling averages" and "swing state probabilities" in public lexicon. Critics, however, noted that its emphasis on quantitative rigor sometimes overlooked qualitative political dynamics, yet the site's legacy endures in the proliferation of probabilistic tools across media, even after Silver's 2023 departure and the outlet's closure.

Shutdown Implications and Long-Term Assessment

The shutdown of FiveThirtyEight occurred on March 5, 2025, when , under parent company , laid off its remaining staff of approximately 15 employees as part of broader cost-cutting measures affecting about 6% of and Networks personnel, totaling around 200 jobs. This decision reflected ongoing media industry pressures, including declining ad revenue and the challenges of maintaining niche operations within large conglomerates, leading to the consolidation of ABC's news magazine shows like 20/20 and . For ABC, the implications included streamlined operations and potential short-term financial savings, but at the cost of specialized capabilities that had previously distinguished its and polling coverage. The closure marked the effective end of FiveThirtyEight as an independent entity following its partial integration into after founder Nate Silver's departure in 2023 and the redirection of its domain traffic in September 2023. Silver attributed the site's vulnerability to Disney's lack of investment in and product strategy, which allowed talent poaching by competitors and prevented sustainable growth. Immediate fallout included the disruption of ongoing tools like polling averages and approval trackers, leaving a gap in accessible public data resources until successors emerged. In the long term, FiveThirtyEight's shutdown underscored the precariousness of data-driven journalism within corporate media structures, where profitability often overrides niche expertise, contributing to a broader contraction in specialized political analysis outlets. Its legacy endures through the standardization of and poll aggregation, which influenced outlets like to launch replacement tools, such as a freely available database tracking polls including President Trump's job approval ratings, independently collected to maintain continuity for researchers and journalists. Silver's subsequent Silver Bulletin has extended this work with public pollster ratings and approval metrics, demonstrating that core methodologies persist outside media. Overall assessments highlight FiveThirtyEight's role in elevating empirical political coverage since its founding—evidenced by its accurate Obama election forecast—but also reveal limitations in scalability and resilience, as corporate priorities ultimately prevailed over its contributions to causal, data-centric analysis.

References

  1. [1]
    Polling wizard Nate Silver went from blogger to Disney-backed ...
    Aug 15, 2023 · Nate Silver isn't just the polling guru who famously forecasted the states that would elect—and reelect—Barack Obama as president in 2008 ...
  2. [2]
    A eulogy for FiveThirtyEight - The Daily Campus
    Sep 29, 2023 · Founded by Nate Silver in 2008, FiveThirtyEight wasn't a typical news website. Silver placed a heavy emphasis on data-driven analysis, academic research and ...
  3. [3]
    Disney and ABC News cuts include the shutdown of FiveThirtyEight
    Mar 6, 2025 · The data-driven site, founded by Nate Silver in 2008, gained prominence for its near-perfect forecast of Barack Obama's victory that year.
  4. [4]
    https://www.wsj.com/articles/nate-silvers-fivethirtyeight-is-moving-to-abc-news-1523986333
  5. [5]
    https://www.vanityfair.com/news/story/nate-silver-on-the-edge
  6. [6]
    The Polls Missed Trump. We Asked Pollsters Why. | FiveThirtyEight
    Nov 9, 2016 · The polls missed Donald Trump's election. Individual polls missed, at the state level and nationally (though national polls weren't far off) ...Missing: accuracy | Show results with:accuracy
  7. [7]
    FiveThirtyEight and the Big Data Fail: Election 2016
    Apr 4, 2017 · Statistics junkie Nate Silver uses data to predict everything from internet slang to Oscar winners to the US Presidential election.Missing: controversies | Show results with:controversies
  8. [8]
    Political poll news site 538 to close amid larger shuttering across ...
    Mar 5, 2025 · Disney is reportedly cutting staff across ABC News Group and its entertainment network as media layoffs continue.
  9. [9]
    A few words about FiveThirtyEight - by Nate Silver
    including Trump approval ratings. But the media just lost an extremely ...
  10. [10]
    How Our Polling Averages Work | FiveThirtyEight - Politics News
    Jun 28, 2023 · References Presidential Approval Tracker / 2024 Republican Presidential Primary Polls / 2024 Presidential Candidate Favorability Trackers ...
  11. [11]
    How (most of) our polling averages work - ABC News
    Trump leads in swing-state polls and is tied with Biden nationallyHow 538's new 2024 presidential general election polling averages work. Almost since its ...
  12. [12]
    How Our Pollster Ratings Work | FiveThirtyEight - Politics News
    Mar 10, 2023 · The Details Longtime readers of FiveThirtyEight are probably familiar with our pollster ratings: letter grades that we assign to pollsters ...
  13. [13]
    How 538's pollster ratings work - ABC News
    Jan 25, 2024 · 538 rates the 'best' pollsters in America538's Elliot Morris discuss how to rate the "pollsters" in America using their "historical track record ...
  14. [14]
    How 538's 2024 presidential election forecast works - ABC News
    Jun 11, 2024 · How Trump's conviction in hush money case affects public opinionIn this installment of the 538 Politics podcast, Galen and Elliott talk about ...
  15. [15]
    How FiveThirtyEight's House, Senate And Governor Models Work
    In presidential races, forecast and polling errors are highly correlated from state to state. This also holds in congressional races, though to a considerably ...
  16. [16]
    2024 presidential election model methodology update - Silver Bulletin
    Jun 26, 2024 · ... FiveThirtyEight from 2008 through 2023.1 So with relatively few exceptions, the model is identical to the one I ran at FiveThirtyEight in 2020.Missing: framework | Show results with:framework
  17. [17]
    https://github.com/fivethirtyeight/data/tree/master/pollster-ratings
  18. [18]
    Polls Policy And FAQs | FiveThirtyEight - Politics News
    Jan 9, 2022 · At FiveThirtyEight, we strive to accumulate and analyze polling data in a way that is honest, informed, comprehensive and accurate.
  19. [19]
    Data and code behind the articles and graphics at FiveThirtyEight
    See the index for a list of the data and code we've published and their accompanying stories. As of June 13, 2023, sports predictions and forecasts are no ...
  20. [20]
    What Pollsters Have Changed Since 2016 — And What Still Worries ...
    Oct 13, 2020 · But as FiveThirtyEight wrote in the run-up to the 2016 election, Donald Trump was always a normal polling error behind Hillary Clinton. And ...
  21. [21]
    How FiveThirtyEight's 2020 Presidential Forecast Works
    Aug 12, 2020 · Our presidential forecast, which launched today, is not the first election forecast that FiveThirtyEight has published since 2016.
  22. [22]
    My comments on Nate Silver's comments on the Fivethirtyeight ...
    Jul 19, 2024 · In 2020, the Fivethirtyeight forecast gave wider uncertainties for state and national vote predictions, leading to win probabilities that were ...<|control11|><|separator|>
  23. [23]
    Nate Silver - The University of Chicago Magazine
    Nate Silver can't hit a fastball, but he's an ace at dissecting baseball stats. And Major League Baseball has taken notice. Nate Silver, age 30, ...Missing: origins analytics
  24. [24]
    How Nate Silver Went From Forecasting Baseball Games to ...
    Oct 9, 2008 · Furthermore, Silver himself invented a system called PECOTA, an algorithm for predicting future performance by baseball players and teams. (It ...Missing: origins | Show results with:origins
  25. [25]
    Reintroducing PECOTA: They're Here! - Baseball Prospectus
    Feb 7, 2011 · Last year Nate Silver wrote a piece in the NY Times making exactly your point: "The key difference in Pecota, the forecasting system that I ...
  26. [26]
    What is PECOTA, and should you hate it? - ESPN UK
    Feb 7, 2017 · Player Empirical Comparison and Optimization Test Algorithm, or PECOTA, is the projection system created by Nate Silver for Baseball Prospectus ...
  27. [27]
    A Conversation with Nate Silver. Thursday, February 28, 2013
    Sep 2, 2016 · Bored at work, Silver began to develop a forecasting system for baseball statistics. This project eventually became PECOTA, or the Player ...Missing: origins analytics<|separator|>
  28. [28]
    [PDF] BIO
    Founder, FiveThirtyEight.com. Nate Silver began by predicting 2008 primary election results with stunning accuracy – and often in opposition to the better-known ...Missing: origins analytics
  29. [29]
    Welcome (and Welcome Back) to FiveThirtyEight
    Aug 25, 2010 · FiveThirtyEight.com premiered on March 7, 2008, three days after Hillary Rodham Clinton won the Democratic presidential primaries in Texas ...
  30. [30]
    Frequently Asked Questions, Last Revised 8/7/08 | FiveThirtyEight
    Mar 1, 2008 · 538 is the number of electors in the electoral college. What is the mission of this website? Most broadly, to accumulate and analyze polling and ...
  31. [31]
    America's hottest pollster gives his final verdict as US elections ...
    Nov 3, 2008 · Nate Silver made his name making uncannily accurate baseball predictions. Now his poll-crunching website is providing possibly the best forecast available of ...
  32. [32]
    Today's Polls and Final Election Projection: Obama 349, McCain 189
    Nov 4, 2008 · It's Tuesday, November 4th, 2008, Election Day in America. The last polls have straggled in, and show little sign of mercy for John McCain.
  33. [33]
    Finding Fame With a Prescient Call for Obama - The New York Times
    Nov 9, 2008 · 9, 2008. At 9:46 p.m., blogging on his site FiveThirtyEight.com, Nate Silver called the presidential election for Barack Obama. The ...
  34. [34]
    The Contact Gap: Proof of the Importance of the Ground Game?
    Nov 8, 2008 · One of the more interesting questions posed on this year's exit polls was whether the voter had been contacted by the Obama and McCain ...
  35. [35]
    The Media Has Been Thrown Off Its Game | FiveThirtyEight
    Sep 13, 2008 · One of the truisms of political reporting is that it is exceptionally results-oriented. When a campaign wins, essentially every aspect of ...
  36. [36]
    FiveThirtyEight to Partner with New York Times
    Jun 3, 2010 · Some exciting news this morning: We have reached agreement in principle to incorporate FiveThirtyEight's content into NYTimes.com.Missing: 2010-2013 | Show results with:2010-2013
  37. [37]
    NYTimes.com to Host Nate Silver's Blog on Politics and Polls
    Jun 3, 2010 · The New York Times said that it would begin hosting FiveThirtyEight and make its founder, Nate Silver, a regular contributor to the ...
  38. [38]
    Welcome (and Welcome Back) to FiveThirtyEight - U.S. Politics
    Aug 25, 2010 · Mrs. Clinton was already well ahead in the polls in the next big primary contest, in Pennsylvania. From reading news reports or watching the ...
  39. [39]
    Behind Nate Silver's war with The New York Times - POLITICO
    Aug 15, 2019 · The paper's onetime data guru can't stop kicking his ex-employer for its 2020 polling analysis and supposed capitulation to Trump.
  40. [40]
    FiveThirtyEight Forecast - Election 2010 - The New York Times
    Dec 10, 2010 · Projected Election Results. Based on polling, expert forecasts, fundraising, past election returns and other indicators, ...Missing: partnership 2010-2013
  41. [41]
    Nate Silver Went Against the Grain for Some at The Times
    Jul 22, 2013 · Why did Nate Silver decide to leave The New York Times and accept an offer from ESPN? That's the cause of great speculation in media circles at the moment.
  42. [42]
    Nate Silver quits New York Times to take up 'dream job' at ESPN
    Jul 22, 2013 · Nate Silver is joining ESPN from the New York Times after a stunning three-year run in which he coolly nailed the biggest election calls.
  43. [43]
    Nate Silver Went Against the Grain for Some at The Times
    Jul 22, 2013 · Why did Nate Silver decide to leave The New York Times and accept an offer from ESPN? That's the cause of great speculation in media circles at the moment.
  44. [44]
    Nate Silver Is Leaving The New York Times For ESPN | TechCrunch
    Jul 19, 2013 · Silver's three-year contract with The Times ends in August, around the time that Olbermann's show is set to launch. The New York Times bought ...
  45. [45]
    Nate Silver, Star Statistician, Leaves The New York Times for ESPN
    Jul 19, 2013 · Nate Silver, famous for his eerily accurate election predictions, is dumping the Gray Lady for the network of Keith Olbermann.
  46. [46]
    Nate Silver - Renowned Statistician, Author and Founder of ...
    Jul 22, 2013 · July 22, 2013. Nate Silver – Renowned Statistician, Author and Founder of FiveThirtyEight – Joins ESPN in Multi-Faceted RoleMissing: details | Show results with:details
  47. [47]
    Nate Silver Sold FiveThirtyEight To ESPN - ADWEEK
    Jul 22, 2013 · FiveThirtyEight creator Nate Silver didn't just leave the New York Times to join ESPN (and ABC News), he sold his company to them.
  48. [48]
    The New FiveThirtyEight Website Relaunches On ESPN - HuffPost
    Mar 17, 2014 · Nate Silver introduced his site with its first article Monday morning in a much anticipated relaunch equipped with a Fox logo, 20 full time ...
  49. [49]
    Prediction: Change Coming for ESPN's FiveThirtyEight
    Jan 25, 2018 · ESPN acquired FiveThirtyEight during an expansion that included the start of the websites The Undefeated and Grantland. Grantland, which ...Missing: details | Show results with:details
  50. [50]
    FiveThirtyEight joins ABC News - The Walt Disney Company
    Apr 17, 2018 · ABC News announced it has reached an agreement in principle to acquire the renowned data journalism organization FiveThirtyEight from ESPN.Missing: details | Show results with:details
  51. [51]
    Nate Silver's FiveThirtyEight Moving From ESPN To ABC - Deadline
    Apr 17, 2018 · Nate Silver's data-driven news site FiveThirtyEight is changing teams, with ABC News agreeing to acquire the site from ESPN.
  52. [52]
    FiveThirtyEight is shutting down as part of broader cuts at ABC and ...
    Mar 5, 2025 · Though Nate Silver left in 2023, FiveThirtyEight still offered election forecasts, a presidential approval tracker, and other tools.<|control11|><|separator|>
  53. [53]
    ABC News Hires The Economist's G. Elliott Morris To Lead Data ...
    May 19, 2023 · Elliott will lead our data journalism efforts for ABC News and FiveThirtyEight, including launching a full suite of poll-aggregation and ...<|separator|>
  54. [54]
    Welcome to the new 538 website - ABC News
    Sep 18, 2023 · 538 content will be published at a new vertical on the ABC News website, www.abcnews.com/538, where we will be focusing on coverage of elections, politics and ...
  55. [55]
    Nate Silver, FiveThirtyEight Founder, Expects to Depart ABC News ...
    Apr 25, 2023 · Mr. Silver, who founded the data-driven news site in 2008, said on Twitter that layoffs at the Walt Disney Company had “substantially ...<|separator|>
  56. [56]
    FiveThirtyEight's Nate Silver Exits ABC News Amid Disney Layoffs
    Apr 25, 2023 · FiveThirtyEight founder Nate Silver is exiting ABC News as The Walt Disney Company ramps up layoffs, Variety has confirmed.
  57. [57]
    Nate Silver - X
    Apr 25, 2023 · Disney layoffs have substantially impacted FiveThirtyEight. I am sad ... My contract is up soon and I expect that I'll be leaving at the end of it ...<|separator|>
  58. [58]
    Disney puts final dagger in 538 after axing sports content in 2023
    Mar 5, 2025 · 538 founder Nate Silver speaks out on the site's undoing after Disney announced layoffs across ABC News this week.
  59. [59]
    Goodbye FiveThirtyEight - The Lego Mirror
    Mar 10, 2025 · The very next week, Disney laid off two-thirds of the staff, including Silver, and merged the website into the primary ABC News one.
  60. [60]
    ABC News boss 'read from script' as he laid off FiveThirtyEight workers
    Mar 6, 2025 · Almin Karamehmedovic took no questions and offered no explanations as to why corporate parent Disney decided to shut down the site, ...
  61. [61]
    Nate Silver Blames Disney Bosses For 538's Demise - Yahoo Finance
    Mar 5, 2025 · Nate Silver blamed the higher-ups at Disney for the business failures of the political data analysis site he founded, FiveThirtyEight, ...
  62. [62]
    The New York Times picks up the shuttered FiveThirtyEight's poll ...
    like a presidential approval rating ...
  63. [63]
    Nate Silver - MIT Sloan Sports Analytics Conference Speaker
    Nate Silver is the founder and former editor-in-chief (2008-2023) of FiveThirtyEight, which was published by The New York Times, ESPN, and ABC News.
  64. [64]
    Nate Silver Says We're Bad at Making Predictions - Freakonomics
    Oct 27, 2023 · Data scientist Nate Silver gained attention for his election predictions. But even the best prognosticators get it wrong sometimes.<|control11|><|separator|>
  65. [65]
    The 'Nate Silver Effect' Is Changing Journalism. Is That Good?
    Oct 5, 2017 · Sites like FiveThirtyEight have drilled into readers the importance of averaging across polls as a corrective to people's tendencies to “pick ...
  66. [66]
    What Was Nate Silver's Data Revolution? | The New Yorker
    Jun 13, 2023 · Jay Caspian Kang writes about the pollster Nate Silver, of FiveThirtyEight, and about how probabilities are misunderstood by the lay reader.
  67. [67]
    Nate Silver - Renowned Statistician, Author and Founder of ...
    Jul 22, 2013 · Leading statistician and best-selling author Nate Silver will join ESPN later this year, bringing his unique brand of creativity, journalism and statistical ...
  68. [68]
    FiveThirtyEight is gone. Its legacy will endure. - The Washington Post
    Mar 7, 2025 · The chief political analyst at the New York Times is a data whiz named Nate (Cohn) who joined the paper essentially as Silver's replacement.
  69. [69]
    ABC News Hires Replacement for Nate Silver: G. Elliott Morris.
    May 19, 2023 · ABC News has hired a replacement for FiveThirtyEight's Nate Silver: Data journalist G. Elliott Morris.
  70. [70]
    538's former top numbers guy to launch data journalism site - Semafor
    May 11, 2025 · After ABC decided in 2023 to cut costs at 538, resulting in layoffs and the departure of founder Nate Silver, it shocked the political ...
  71. [71]
    Nathaniel Rakich – FiveThirtyEight - Politics News
    Nathaniel Rakich is a senior editor and senior elections analyst at FiveThirtyEight. Sep. 13, 2023 The Senate Is Losing One Of Its Few Remaining Moderate ...
  72. [72]
    Harry Enten – FiveThirtyEight - Politics News
    Harry Enten was a senior political writer and analyst for FiveThirtyEight.Missing: history | Show results with:history
  73. [73]
    https://fivethirtyeight.com/videos/video-fivethirtyeights-politics-crew-analyzes-the-debate/
  74. [74]
    Masthead | FiveThirtyEight - Politics News
    G. Elliott Morris: Editorial Director of Data Analytics ; Nathaniel Rakich: Senior Editor and Senior Elections Analyst ; Amelia Thomson-DeVeaux: Senior Editor and ...
  75. [75]
    Neil Paine – FiveThirtyEight - Politics News
    Neil Paine was the acting sports editor at FiveThirtyEight. He previously wrote for ESPN Insider, The New York Times and Sports-Reference.com and consulted ...
  76. [76]
    FiveThirtyEight Politics (Podcast Series 2016– ) - IMDb
    FiveThirtyEight Politics: With Nate Silver, Galen Druke, Claire Malone, Jody Avirgan. FiveThirtyEight Politics is a politics podcast that covers the latest ...
  77. [77]
    ABC Shuttering FiveThirtyEight Amid Disney Layoffs
    Mar 6, 2025 · The statistical-analytics site that dabbled in sports, politics, and economics was shut down by ABC News as part of a wave of layoffs Wednesday.Missing: changes | Show results with:changes
  78. [78]
    How 538's 2024 presidential election forecast works - ABC News
    Aug 23, 2024 · 538 has built a new presidential election forecast model. This article explains how we did it: the ingredients that go into the forecast, how we process them.
  79. [79]
    Numbers nerd Nate Silver's forecasts prove all right on election night
    Nov 7, 2012 · His final forecast gave Obama a 90.9% chance of victory. Silver also forecast 332 electoral college votes for Obama against 206 for Romney – the ...
  80. [80]
    2012 Presidential Election – FiveThirtyEight - Politics News
    FiveThirtyEight uses statistical analysis - hard numbers - to tell compelling stories about elections, politics and American society.
  81. [81]
    26 charts that helped explain 2024 in politics - ABC News
    Dec 26, 2024 · Here are 26 charts that encapsulate the themes of the year, from our flagship election forecast to visuals that show where Americans stood on the issues that ...
  82. [82]
    A Guide to FiveThirtyEight's Comprehensive Election Day Coverage
    Nov 4, 2016 · As the 2016 U.S. presidential election reaches its final days, today FiveThirtyEight outlines its Election Day coverage for Tuesday, Nov.
  83. [83]
    Super Model - Columbia Journalism Review
    Mar 7, 2025 · It was a slow and staggered end for FiveThirtyEight, the site that made everyone into armchair experts on the art of data modeling.
  84. [84]
    FiveThirtyEight To Be Relaunched in Partnership with ESPN
    Aug 18, 2013 · FiveThirtyEight will be relaunched at this address in the impending months in partnership with ESPN. The new version of FiveThirtyEight will ...Missing: acquisition | Show results with:acquisition
  85. [85]
    FiveThirtyEight is Dead; Long Live Public Soccer Projection Models
    Aug 2, 2023 · Silver and FiveThirtyEight have been a major part of American political coverage growing increasingly driven by data, polling, and evidence.
  86. [86]
    How Massive The NFL Really Is, In 4 Charts | FiveThirtyEight
    Feb 6, 2023 · Relative share of respondents who said each sport was their favorite to watch among top-ranked sports, 1937-2017. FOOTBALL. BASEBALL.Missing: coverage | Show results with:coverage
  87. [87]
    We Checked 50 Years Of Sports History To Find The Team That ...
    Aug 28, 2018 · According to our algorithm, no team in major pro sports has been more consistently mediocre over the past five seasons than the Flyers.
  88. [88]
    The Best — And Worst — Of Pro Sports In The 2010s | FiveThirtyEight
    Dec 31, 2019 · We wanted to highlight the best (and worst) of the past 10 years in five major professional leagues: the NBA, WNBA, NFL, NHL and MLB.
  89. [89]
    Title IX And The Fight For Women's Equality In Sports | FiveThirtyEight
    Mar 22, 2022 · With March Madness in full swing and Women's History Month underway, NCAA basketball star Sedona Prince and Olympic gymnastics legend ...<|separator|>
  90. [90]
    Nate Silver's FiveThirtyEight Moving From ESPN to ABC News
    Apr 17, 2018 · Nate Silver's FiveThirtyEight Moving From ESPN to ABC News. The site will continue working with ESPN on sports content, said executive vp ...
  91. [91]
    Economics – FiveThirtyEight - Politics News
    FiveThirtyEight uses statistical analysis - hard numbers - to tell compelling stories about elections, politics and American society.Missing: finance | Show results with:finance
  92. [92]
    A Majority Of Americans Think We Are In A Recession | FiveThirtyEight
    if the U.S. is ...
  93. [93]
    What Economists Fear Most During This Recovery | FiveThirtyEight
    Jun 23, 2020 · Economists think there are still significant economic risks despite the tentative rebound, including a second wave of COVID-19, an unwillingness to spend from ...
  94. [94]
    Economists Don't Know How Long High Inflation Will Last ... - YouTube
    Nov 15, 2021 · ... article titled “'I Blew It.' Maybe It's Time ... Economists Don't Know How Long High Inflation Will Last l FiveThirtyEight Politics Podcast.
  95. [95]
    Bank Bailout Politics 2.0 | FiveThirtyEight
    Mar 20, 2023 · In part 2 of this installment of the FiveThirtyEight Politics podcast, the crew looks back at the bailout of American banks during the ...
  96. [96]
    Economic Indicators – FiveThirtyEight
    FiveThirtyEight uses statistical analysis - hard numbers - to tell compelling stories about elections, politics and American society.Missing: finance | Show results with:finance<|separator|>
  97. [97]
    Economists – FiveThirtyEight
    FiveThirtyEight uses statistical analysis - hard numbers - to tell compelling stories about elections, politics and American society.
  98. [98]
    Americans Were A Lot Less Worried About Climate Change Before ...
    Feb 6, 2019 · FiveThirtyEight has seen its own uptick in traffic on climate change stories in the last six months, and science editors at BuzzFeed and ...
  99. [99]
    Why Being Anti-Science Is Now Part Of Many Rural Americans' Identity
    Apr 25, 2022 · Monica Potts is a senior politics reporter at FiveThirtyEight. @MonicaBPotts. Filed under. COVID-19 (438 posts) Climate Change (108) ...
  100. [100]
    There's No Such Thing As 'Sound Science' | FiveThirtyEight
    Dec 6, 2017 · Christie Aschwanden was a lead science writer for FiveThirtyEight. Her book “Good to Go: What the Athlete in All of Us Can Learn from the ...
  101. [101]
    Americans Are Smart About Science | FiveThirtyEight - Politics News
    Mar 28, 2019 · Technology · Lifestyle · Live · Video · Shop · Health · Weather · Sports ... Pitch FiveThirtyEight · Advertise With Us · About Nielsen Measurement.Missing: environment | Show results with:environment
  102. [102]
    Here's The Best Place To Move If You're Worried About Climate ...
    Sep 19, 2019 · The U.P. is the place. The Yoopers are the people. Maggie Koerth was a senior reporter for FiveThirtyEight. @maggiekb1 ...
  103. [103]
    33 Cool Charts We Made In 2022 | FiveThirtyEight - Politics News
    Dec 20, 2022 · In 2022, FiveThirtyEight's visual journalists covered the midterm elections, the end of Roe v. Wade and sports stories ranging from the ...
  104. [104]
    FiveThirtyEight: Our Data
    We're sharing the data and code behind some of our articles and graphics. We hope you'll use it to check our work and to create stories and visualizations of ...
  105. [105]
    The 40 Weirdest (And Best) Charts We Made In This Long, Strange ...
    Dec 21, 2020 · Over the course of 2020, FiveThirtyEight's visual journalists covered a historic election, an unprecedented year in sports, ...
  106. [106]
    Data Visualization – FiveThirtyEight - Politics News
    FiveThirtyEight uses statistical analysis - hard numbers - to tell compelling stories about elections, politics and American society.
  107. [107]
    Nate Silver on the Pitfalls of Prediction | On the Media - WNYC Studios
    Nate Silver is something of an authority on political forecasting. In 2008, his blog FiveThirtyEight correctly predicted the outcome of the presidential ...<|separator|>
  108. [108]
    How FiveThirtyEight calculates the data of a divided nation - PBS
    Nov 4, 2018 · Nate Silver's FiveThirtyEight blog burst onto the political scene in 2008, when he forecasted the popular vote for president within one ...Missing: origins | Show results with:origins
  109. [109]
    Nate Silver On Why FiveThirtyEight Forecasts A Clinton Win
    Oct 11, 2016 · Before the 2012 presidential election, statistician Nate Silver correctly predicted the presidential winner in all 50 states, proving wrong ...Missing: accuracy | Show results with:accuracy
  110. [110]
    Final Election Update: There's A Wide Range Of Outcomes, And ...
    Nov 8, 2016 · Throughout the election, our forecast models have consistently come to two conclusions. First, that Hillary Clinton was more likely than not ...
  111. [111]
    Why FiveThirtyEight Gave Trump A Better Chance Than Almost ...
    Nov 11, 2016 · Based on what most of us would have thought possible a year or two ago, the election of Donald Trump was one of the most shocking events in ...
  112. [112]
    Biden's Favored In Our Final Presidential Forecast, But It's A Fine ...
    Nov 3, 2020 · FiveThirtyEight has issued its final presidential forecast. There hasn't been a lot of change over the past 24 or 48 hours, as most of the ...
  113. [113]
    The Polls Weren't Great. But That's Pretty Normal. | FiveThirtyEight
    Nov 11, 2020 · The only states where presidential polling averages got the winners wrong will be Florida, North Carolina and the 2nd Congressional District in Maine.Missing: evolution | Show results with:evolution
  114. [114]
    The polls underestimated Trump's support—again | Brookings
    Nov 13, 2024 · Combining the findings from the three most recent presidential elections, I conclude that today's polling instruments and techniques are not ...
  115. [115]
    2024 polls were accurate but still underestimated Trump - ABC News
    Nov 8, 2024 · How Trump won over votersThe 538 team discusses how Donald Trump won the election despite being disliked by a majority of Americans. Here at ...
  116. [116]
    Polling Error in 2016-2020: Look Out for Wisconsin
    Sep 12, 2024 · Final polling did generally overstate Democrats in both the 2016 and 2020 elections in these states, with Wisconsin standing out.
  117. [117]
    Once again, polls underestimated Trump. Experts only have a hunch ...
    Nov 20, 2024 · Following Republican President-elect Donald Trump's victory in this month's election, some U.S. pollsters are scrambling to understand why ...
  118. [118]
    Why Did Republicans Outperform The Polls Again? Two Theories.
    Mar 2, 2021 · But why are some Republican voters more reluctant to take surveys? As the director of polling at the Cato Institute, I, as well as other ...
  119. [119]
    Who Are The People Who Don't Respond To Polls? | FiveThirtyEight
    Oct 26, 2022 · Can we trust election polls?” is a question that has reached a fever pitch in political junkie circles dating back to the 2016 election.
  120. [120]
    Biased polls: investigating the pressures survey respondents feel
    Jul 29, 2024 · Democratic bias in polls reached record highs in recent elections. The Shy Trump Voter Hypothesis claims some voters feel pressure to hide plans to vote for ...
  121. [121]
    Why Did Republicans Outperform the Polls Again? Two Theories.
    Mar 2, 2021 · But why are some Republican voters more reluctant to take surveys? As the director of polling at the Cato Institute, I ...
  122. [122]
    Why I don't buy 538's new election model - by Nate Silver
    Jul 18, 2024 · It barely pays attention to the polls. And its results just don't make a lot of sense.Missing: date | Show results with:date<|control11|><|separator|>
  123. [123]
    FINAL Silver Bulletin 2024 presidential election forecast
    Nov 4, 2024 · And an archive of the Biden-Trump forecast can be found here. The polls: who's ahead right now? The Silver Bulletin ...Missing: framework | Show results with:framework
  124. [124]
    Our Election Forecast Didn't Say What I Thought It Would
    Aug 17, 2020 · Why we factored in COVID-19, why we ignored approval ratings, and other deep cuts from our election model.
  125. [125]
    Nate Silver says FiveThirtyEight suspended its forecasts to help ...
    Aug 6, 2024 · Nate Silver accused FiveThirtyEight of suspending its presidential forecasts because it is waiting for Kamala Harris' polling numbers to ...
  126. [126]
    https://projects.fivethirtyeight.com/2024-election-forecast/
  127. [127]
    [PDF] Disentangling Bias and Variance in Election Polls
    We conclude by discussing how these results help explain polling failures in the 2016 U.S. presidential election, and offer recommendations to improve polling ...Missing: conservative | Show results with:conservative
  128. [128]
    Betting Markets Favor Trump. But Their Record of Accuracy Is Mixed.
    Oct 25, 2024 · Proponents believe that having real money on the line, and a large crowd of investors, encourages a more accurate election forecast than polling data can ...
  129. [129]
    Are Betting Markets Better than Polling in Predicting Political ... - arXiv
    Jul 11, 2025 · Overall, findings suggest that Polymarket was superior to polling in predicting the outcome of the 2024 presidential election, particularly in ...
  130. [130]
    Election results show potential of prediction markets, blockchain
    Dec 6, 2024 · A prediction market was more accurate in forecasting the 2024 presidential election than traditional polls and pundits.Missing: 2016 2020
  131. [131]
    [PDF] Forecasting the 2024 U.S. Presidential Election: Financial Markets ...
    The analysis reveals that election forecasts derived from options prices on the. S&P 500 index consistently outperformed other methods in predicting Trump's win.<|separator|>
  132. [132]
    Prediction Markets + Polls + Economic Indicators: Better Election ...
    Oct 31, 2024 · In 2016, polls showed Hillary Clinton beating Donald Trump. One reason forecasting U.S. elections is so challenging is that the presidency isn't ...Missing: 2020 | Show results with:2020
  133. [133]
    https://www.wsj.com/business/media/disney-to-cut-nearly-6-staff-across-abc-news-disney-entertainment-networks-8af04b40