Trafalgar Group

Trafalgar Group is an American opinion-polling and survey company founded by Robert Cahaly and based in Atlanta, Georgia. It first released polling publicly in 2016, during the United States presidential election. The firm became prominent after publishing unusually favorable results for Donald Trump in several battleground states, including Michigan, where it was identified as the only pollster showing Trump ahead. It also came close to predicting Trump's victory in Pennsylvania and, according to contemporary assessments, correctly projected the overall electoral-vote result even though it did not identify every state that produced that outcome. The company presents itself as a polling operation designed to detect forms of response distortion that it describes as social-desirability bias. In addition to asking respondents how they intend to vote, Trafalgar has asked how respondents believe their neighbors will vote. The firm has said that this approach is intended to identify voters who may be reluctant to disclose their preferences directly. Its polling operation has used multiple contact channels, including live telephone calls, interactive voice response calls, text messages, online opt-in platforms, and two proprietary methods. The company generally uses short questionnaires, reportedly limited to nine questions or fewer, and works from a contact pool of approximately 75,000 potential respondents. It contacts members of that pool until it reaches a target sample size and researches characteristics such as voting likelihood. Trafalgar has stated that it applies relatively little weighting to its results. Unlike many major polling organizations, Trafalgar has concentrated on state-level presidential polling rather than national surveys. Cahaly has argued that the national popular vote is not the relevant statistic for determining the presidency. This focus, together with the firm's willingness to publish results that diverged sharply from polling averages, helped establish Trafalgar as a prominent outlier in American election coverage. Trafalgar's record has been mixed. Its 2016 polling successes made it influential among analysts and conservative political audiences. In the 2020 presidential election, FiveThirtyEight calculated that Trafalgar's average statistical error was approximately 2.6 percentage points, better than the average error across pollsters that year. However, the firm predicted that Trump would win the national election and several states that Joe Biden ultimately carried, including Arizona, Georgia, Michigan, Nevada, and Pennsylvania. Its polling for the 2021 Georgia Senate runoffs and that year's Virginia gubernatorial election produced some relatively close or accurate estimates, although it missed the eventual outcome in the Georgia contest involving Kelly Loeffler and Raphael Warnock. The firm's 2022 midterm polling was substantially less successful. Trafalgar projected Republican victories or competitive races in a number of states where Republican candidates ultimately lost by clear margins. These included races involving Mehmet Oz, Herschel Walker, Kari Lake, Tudor Dixon, Adam Laxalt, Blake Masters, Christine Drazan, and Tim Michels. It also portrayed several other Republican candidates as being within striking distance when some, including Doug Mastriano, lost decisively. Trafalgar has faced sustained criticism over transparency and partisan skew. Conventional pollsters and election analysts have questioned the firm's sampling design, its use of fixed contact pools, its treatment of low-engagement voters, and its limited public disclosure of methodological details. FiveThirtyEight has found that Trafalgar generally overestimated Republican performance by about two points. The firm has defended its privacy around proprietary techniques, with Cahaly arguing that the methodology should be evaluated by its results rather than fully disclosed. Its reputation therefore combines innovation and occa…

Trafalgar Group was founded by Robert Cahaly and is based in Atlanta, Georgia. The company first entered public view as an election-polling organization in 2016. Its early public identity was shaped by a willingness to publish findings that differed sharply from the polling consensus, particularly in the U.S. presidential election. Trafalgar was the only cited polling firm showing Donald Trump ahead in Michigan and was regarded as nearly alone in correctly indicating a Trump victory in Pennsylvania. The firm also reportedly matched the eventual electoral-vote total for the two candidates, although it did not correctly identify every state contributing to that total. The company's approach relies on a mixture of conventional and proprietary voter-contact methods. These include live telephone interviews, interactive voice response calls, text messages, online opt-in recruitment, and two additional methods that Trafalgar treats as proprietary. Rather than relying exclusively on open or broadly distributed sampling, it has used a recurring contact pool of roughly 75,000 potential respondents. The firm researches people in that pool for characteristics such as likely turnout and contacts respondents until it obtains a target sample. Questionnaires are generally short, commonly consisting of no more than nine questions. Trafalgar has said that it applies limited weighting to its results. A defining feature of the firm's methodology is its effort to account for what it calls social-desirability bias. Trafalgar argues that some respondents may be unwilling to tell a pollster their actual voting preference, particularly when they believe the preference may be socially disfavored. Its surveys have therefore asked both how respondents intend to vote and how they think their neighbors will vote. The company has also described the use of lifestyle and engagement data to locate voters who might otherwise be missed by traditional approaches. In presidential elections, Trafalgar has generally emphasized state-level polling rather than national surveys, arguing that the national popular vote is not what determines the presidency. Trafalgar remained an outlier in later election cycles. In Georgia's 2018 gubernatorial race, its final poll showed Brian Kemp with a 12-point advantage even though most other polls showed a much closer contest. Kemp won, but by less than two points. Before the 2020 presidential election, Trafalgar predicted a Trump victory and projected Trump wins in Arizona, Florida, Georgia, Michigan, Nevada, North Carolina, and Pennsylvania. FiveThirtyEight later calculated that Trafalgar's average error was about 2.6 points, better than the average error among pollsters in that unusually difficult cycle. Nevertheless, the firm missed the winning candidate in several battleground states, limiting the value of its overall election call. The 2021 Georgia Senate runoffs produced another mixed result. Trafalgar initially showed Republican incumbents Kelly Loeffler and David Perdue ahead, then later showed both trailing. Its final pre-election forecast reversed those movements by predicting narrow leads for Loeffler and Perdue. Jon Ossoff defeated Perdue, while Raphael Warnock defeated Loeffler. In the 2021 gubernatorial contests, Trafalgar's Virginia poll closely approximated Glenn Youngkin's narrow victory, while its New Jersey poll correctly projected a Phil Murphy win but overstated his final margin. Trafalgar's standing deteriorated after the 2022 midterm elections. It projected Republican wins or close races in numerous contests that Republicans ultimately lost, including races involving Mehmet Oz, Herschel Walker, Kari Lake, Tudor Dixon, Adam Laxalt, Blake Masters, Christine Drazan, and Tim Michels. It also characterized several other Republican campaigns as competitive when some candidates, including Doug Mastriano, lost by large margins. The firm has faced continuing criticism because it does not fully explain its sampling and weighting procedures. The New York Times and other commentators have described the available methodology as insufficiently specific, while FiveThirtyEight has reported that Trafalgar generally overestimates Republican performance by approximately two points. Trafalgar has defended its decision to keep parts of its process private, arguing that its proprietary techniques should be judged by performance. FiveThirtyEight's ratings have fluctuated considerably, from C-minus before the 2020 election to A-minus afterward, then B in March 2023, and 0.7 out of 3 in September 2024. These changes reflect the firm's uneven record and the difficulty of evaluating a pollster whose methods are not fully reproducible from public documentation.

1. 2024Later FiveThirtyEight rating

   FiveThirtyEight listed Trafalgar with a rating of 0.7 out of 3 as of September 17, 2024.

   Source
2. 2022Substantial misses in the midterm elections

   Trafalgar's forecasts overestimated Republican performance across numerous Senate, gubernatorial, and other statewide contests.

   Source
3. 2021Mixed performance in Georgia runoffs and gubernatorial elections

   The firm missed the winners of both Georgia Senate runoffs but came close to the Virginia gubernatorial result and correctly predicted a Murphy victory in New Jersey.

   Source
4. 2020High numerical accuracy but incorrect overall presidential forecast

   Trafalgar predicted a Trump victory, while post-election analysis found a relatively low average polling error but incorrect winner calls in several states.

   Source
5. 2018Outlier Georgia gubernatorial poll

   The firm's final Georgia gubernatorial poll gave Brian Kemp a 12-point lead; Kemp won by less than two points.

   Source
6. 2016First public polling and recognition in the presidential election

   Trafalgar publicly released polling and gained attention for showing Donald Trump ahead in Michigan and nearly correctly identifying his Pennsylvania victory.

   Source

An unconventional American political polling and survey firm positioned around state-level election research, alternative respondent-recruitment methods, and attempts to detect voters who may not state their preferences directly. Its public profile is strongly associated with Republican-leaning election forecasts and with debate over polling transparency.

• Election opinion pollsPolitical polling2016

  Trafalgar's principal public offering is state-level election polling, especially for presidential, Senate, gubernatorial, and other statewide contests. The firm combines telephone, interactive voice response, text, online opt-in, and proprietary contact methods. Its surveys are generally brief and seek to measure both direct voting intentions and perceptions of how a respondent's neighbors may vote.

• Survey and voter researchMarket research

  The company conducts broader opinion and voter research using a researched contact pool and respondent characteristics such as likely turnout. Its stated methodological emphasis is identifying low-engagement or socially reluctant respondents who may be underrepresented in conventional polling designs.

• 2020Retention of proprietary methodological detailsStrategy

  Analysts criticized the limited public explanation of Trafalgar's sampling, weighting, and social-desirability-bias procedures.

  What changed. Robert Cahaly defended keeping the firm's methods private and argued that the approach should be assessed by its predictive results.

  Aftermath. The decision preserved the firm's proprietary positioning but sustained criticism concerning transparency and independent reproducibility.

  Source
• 2016Emphasis on state-level rather than national presidential pollingStrategy

  Presidential elections are decided through the Electoral College, making state outcomes more directly relevant than the national popular vote to the firm's stated purpose.

  What changed. Trafalgar concentrated its presidential polling on individual states and generally did not conduct national presidential polls.

  Aftermath. This approach helped distinguish the firm from national pollsters and contributed to its visibility during the 2016 and 2020 elections.

  Source

• 2024FiveThirtyEight gives Trafalgar a substantially lower later rating

  As of September 17, 2024, FiveThirtyEight listed Trafalgar with a rating of 0.7 out of 3, reflecting the continuing debate over its polling performance and reliability.

  Other
  Source
• 2023FiveThirtyEight revises its assessment of Trafalgar Group

  FiveThirtyEight's rating of Trafalgar changed over time, moving from C-minus before the 2020 election to A-minus after the election and then to B as of March 9, 2023.

  Other
  Source
• 2022Trafalgar's 2022 midterm forecasts miss multiple Republican losses

  The firm's polling predicted Republican victories or highly competitive races in a series of contests where Republican candidates ultimately lost, sometimes by substantial margins. The results renewed criticism of Trafalgar's sampling and partisan bias.

  Other
  Source
• 2021Trafalgar's Georgia runoff forecasts are mixed

  On the eve of Georgia's Senate runoffs, Trafalgar projected narrow leads for David Perdue and Kelly Loeffler. Ossoff ultimately defeated Perdue by approximately 1.2 points, while Warnock defeated Loeffler by approximately 2.1 points.

  Other
  Source
• 2021Trafalgar records contrasting results in the 2021 Virginia and New Jersey gubernatorial races

  The firm forecast a narrow Glenn Youngkin victory in Virginia, close to the final margin. In New Jersey it predicted a Phil Murphy win but understated the eventual margin.

  Other
  Source
• 2020Trafalgar forecasts a Trump victory in the 2020 presidential election

  Before the election, Trafalgar predicted that Trump would win the presidency and several closely contested states. Post-election analysis found that its average numerical error was comparatively low, but it missed the eventual winner in multiple states and therefore received less credit for the overall forecast.

  Other
  Source
• 2020Criticism intensifies over Trafalgar's undisclosed polling methodology

  Conventional pollsters and media analysts criticized the limited methodological explanation available for Trafalgar's surveys. The firm's use of proprietary methods and its treatment of social-desirability bias became central to debate over its results.

  Other
  Source
• 2018Trafalgar publishes an outlier poll in the 2018 Georgia governor's race

  Trafalgar's final poll showed Brian Kemp leading by 12 percentage points, substantially more than other polling organizations. Kemp won, but by less than two points.

  Other
  Source
• 2016Trafalgar Group gains prominence after correctly identifying Trump's 2016 Michigan victory

  The firm became widely known after publishing one of the few polls showing Donald Trump ahead in Michigan and nearly correctly identifying his victory in Pennsylvania. Its 2016 performance helped establish Trafalgar as a prominent election-polling outlier.

  Other
  Source

1. Trafalgar Group