Violence, redistricting, and democratic norms in Trump’s America

Bright Line Watch September 2025 survey

In recent months, the Trump admin­is­tra­tion has continued to challenge executive branch norms as well as legal and con­sti­tu­tion­al limits on the powers of the pres­i­den­cy, including dra­mat­i­cal­ly expanding the capacity and activ­i­ties of U.S. Immigration and Customs Enforcement and deploying the military to Los Angeles and Washington, D.C. A Republican effort to change con­gres­sion­al district maps to gain five seats in the U.S. House of Representatives in the 2026 election was enacted in Texas, prompting California Democrats to develop a redis­trict­ing plan to offset those losses. Most recently, the con­ser­v­a­tive activist and com­men­ta­tor Charlie Kirk was shot and killed during a public event at a Utah uni­ver­si­ty, prompting a nation­wide debate over political violence and free speech. 

In this context, we fielded parallel surveys of 703 political science faculty at American colleges and uni­ver­si­ties (whom we refer to as “experts” below) and a rep­re­sen­ta­tive sample of 2,750 Americans (whom we refer to as “the public” below). The expert survey was fielded from September 5–17, 2025 and the public survey from September 5–16, 2025.

Our key findings are the following:

Assessments of demo­c­ra­t­ic performance

• Ratings of American democracy among the public as a whole and among experts are lower since President Trump returned to office, but they have not changed mea­sur­ably since our April 2025 survey. Expectations for 2027 remain pes­simistic among experts.

• Current expert ratings of U.S. democracy are closer to those of a mixed or illiberal democracy than a full democracy or countries often con­sid­ered as relevant com­par­isons such as Great Britain and Canada.

• The partisan gap in ratings of U.S. democracy is at its highest since we started col­lect­ing data on this question in 2017. 

• Expert ratings of U.S. per­for­mance at upholding specific demo­c­ra­t­ic prin­ci­ples are mostly stable since April, but we observe sig­nif­i­cant declines on tol­er­a­tion of peaceful protest, not using gov­ern­ment agencies to punish political opponents, and fair electoral district bound­aries. Among the public, Republican ratings were largely stable, whereas Democrats perceived declines in a number of areas.

Threats to democracy

• Most experts rated the shooting of Charlie Kirk as a threat to democracy. Evaluations of U.S. democracy did not change afterward, but the share of experts who said the gov­ern­ment is effec­tive­ly pre­vent­ing political violence fell substantially.

• Numerous events since April were rated as threats to democracy by 80% of experts or more, including Texas changing its Congressional districts and the National Guard deploy­ment to Los Angeles. Many of these attracted wide­spread support among Republicans. However, only one is among the ten most often rated by experts as posing an extra­or­di­nary or serious threat to democracy since Trump took office again.

• Looking ahead, many potential actions by the Trump admin­is­tra­tion were rated as extra­or­di­nary or serious threats to democracy by 89% of experts or more, including Trump seeking a third term, the sus­pen­sion of habeas corpus, invoking the Insurrection Act, and directing law enforce­ment officials to not to enforce court orders.

Fighting fire with fire

• Experts over­whelm­ing­ly view the Texas redis­trict­ing plan as a threat to democracy and were split over the California plan. They expected to rate U.S. democracy lower in 2027 if only Texas redis­tricts versus a scenario in which both states do so. Republicans tended to support partisan state redis­trict­ing more in general.

• Respondents from both parties say it is inap­pro­pri­ate to violate demo­c­ra­t­ic norms in response to a demo­c­ra­t­ic norm violation when their side is described as com­mit­ting the initial infrac­tion, but reverse their stance when the other party violates norms first.

• Support for a battery of anti­de­mo­c­ra­t­ic actions among Democrats including illegal redis­trict­ing and restrict­ing or shutting down Fox News has increased since 2021, though none attracts majority support. Republican support for these actions is lower or unchanged compared to 2021 except for illegal redis­trict­ing and restrict­ing MSNBC.

Forecasting future threats to democracy

• Human forecasts from the Metaculus pre­dic­tion platform and those from experts are broadly con­sis­tent after adjusting for experts’ prior tendency toward pessimism. AI forecasts from the startup Mantic, which has created the best-per­form­ing AI fore­cast­ing bot to date, rate the like­li­hood of most events lower than either human source.

• Five future events that most experts rate as extra­or­di­nary or serious threats to democracy are rated as more likely to occur than not by a composite of human, pre­dic­tion platform, and AI forecasts: normal pro­ce­dures being over­rid­den in legal cases con­cern­ing both pres­i­den­tial allies and opponents, a military deploy­ment to a third metro area besides Los Angeles and Washington, D.C., an order revoking nat­u­ral­iza­tion based on sub­se­quent conduct, and an order directing law enforce­ment to not enforce a court order. Estimated prob­a­bil­i­ties are much lower for sus­pend­ing habeas corpus, invoking the Insurrection Act, a denial of birthright cit­i­zen­ship, and impris­on­ing a U.S. citizen abroad.

Overall eval­u­a­tions of American democracy

We have asked expert and public respon­dents to rate the overall per­for­mance of U.S. democracy on a 0–100 scale since May 2017 and  September 2017, respec­tive­ly. Our September 2025 survey also asked expert respon­dents what rating they would expect to give the U.S. on the same scale in 2027. The figure below graphs how these ratings have varied over time among experts, the public overall, and Democrats and Republicans in the public sample. We also present projected future ratings of American democracy in 2027 for each group.

We pre­vi­ous­ly observed declines in ratings of U.S. democracy among every group, including experts, since Trump returned to office. Expert ratings declined, for instance, from 67 in November 2024 to 55 in February 2025 and 53 in April 2025. Public ratings dropped, too, from 58 in November to 53 in February and 49 in April. Notably, Democrats and Republicans showed a similar decline in April.

By contrast, ratings in this wave remained stable since our last survey. Expert ratings of U.S. democracy today held at 54, sta­tis­ti­cal­ly unchanged since April (53). Similarly, ratings are sta­tis­ti­cal­ly unchanged from April among the public overall (con­sis­tent at 49) as well as Democrats (45 vs. 44) and Republicans (56 vs. 58). 

However, as the next figure shows, the partisan gap in democracy ratings has reached an all-time high. The average dif­fer­ence across all previous surveys was 8 points, and the previous maximum dif­fer­ence was 12 points, in July 2018, when the average rating among Republicans was 63 versus 51 among Democrats. In our September 2025 survey, the gap reached 15 points (58.5 among Republicans versus 43.6 among Democrats). When our respon­dents are asked to project a democracy rating for 2027 on the same scale, their expec­ta­tions diverge by 18 points. 

The current expert rating of 54 continues to place the United States between the scores that experts gave in April 2025 to Israel (49) and Mexico (60) and well below our nominal demo­c­ra­t­ic peers of Great Britain (83) and Canada (88). 

In this survey, we also asked experts to rate three hypo­thet­i­cal countries with differing levels of election com­pet­i­tive­ness, judicial and pros­e­cu­to­r­i­al inde­pen­dence, media freedom, and quality of data and civil service on the same 0–100 scale to provide a benchmark for ratings of the U.S. The countries were presented and rated sep­a­rate­ly in ran­dom­ized order in the survey; they are grouped together in the table below for visual clarity (the descrip­tions in brackets were not shown in the survey). “Country A” was con­struct­ed as a strong democracy with high per­for­mance in each area; “Country B” was con­struct­ed as an illiberal democracy with per­for­mance in each area; and “Country C” was con­struct­ed as a non-democracy with low per­for­mance in each area.

As the graph below indicates, the current U.S. democracy rating of 54 among experts is closest to the 44 rating experts gave to our hypo­thet­i­cal illiberal democracy (“Country B”). Experts put the U.S. at approx­i­mate­ly equal distance from the strong democracy (“Country A”), which received an average rating of 92, and the non-democracy (“Country C”), which received an average rating of 18.

Notably, variation in ratings is greater for the United States than the hypo­thet­i­cal illiberal democracy, which could represent genuine het­ero­gene­ity in demo­c­ra­t­ic per­for­mance for the U.S. (discussed further below) or could suggest that eval­u­at­ing a real country generates greater polar­iza­tion among experts than a hypo­thet­i­cal one.

When asked how they expect to rate U.S. democracy in 2027, experts continue to forecast a decline to 47, which would be its lowest point in our data. In contrast, the public expects slight improve­ments in American democracy in two years — to 49 for the public overall and to 41 and 60 for Democrats and Republicans, respec­tive­ly. These forecasts of future ratings are sta­tis­ti­cal­ly unchanged among experts, the public, and Democrats and Republicans in the public since our April survey (see graph in the appendix).

Assessing U.S. per­for­mance on specific demo­c­ra­t­ic principles

As in past surveys, we asked expert and public respon­dents to rate the extent to which specific demo­c­ra­t­ic prin­ci­ples are currently upheld in the U.S., including four new items on the neu­tral­i­ty and political exploita­tion of the military and law enforce­ment. The full list of all 35 prin­ci­ples measured is provided in the appendix. Experts were asked about a random subset of nine prin­ci­ples; the public was asked about seven.

The figure below shows the per­cent­age of respon­dents who regard the U.S. as fully or mostly meeting each standard (as opposed to meeting it partly or not at all) among political science experts (green) as well as Democrats (blue) and Republicans (red) from our public sample. The dark markers and con­fi­dence intervals represent estimates from September 2025; the light versions represent April 2025 estimates. Asterisks indicate sta­tis­ti­cal sig­nif­i­cance at the p<.05 level. (A version of this graph that presents assess­ments from the experts jux­ta­posed with those of Democrats and Republicans is provided in the appendix.)

Perceptions of demo­c­ra­t­ic per­for­mance in specific areas of concern were rel­a­tive­ly stable among experts; ratings of per­for­mance on 27 of the 31 prin­ci­ples included in the April and September surveys did not change significantly.

Three of the four changes we observe in expert ratings since April are declines in per­for­mance in the following areas: tol­er­a­tion of peaceful protest (from 49% in April 2025 to 38% in September 2025), not using gov­ern­ment agencies to punish political opponents (from 15% to 7%), and unbiased election district bound­aries (from 8% to 3%). These changes are espe­cial­ly notable con­sid­er­ing that the latter two prin­ci­ples had little room to decline further from their previous levels. 

By contrast, experts rated demo­c­ra­t­ic per­for­mance as improving since April on one demo­c­ra­t­ic principle: orga­ni­za­tions such as uni­ver­si­ties and busi­ness­es operating inde­pen­dent­ly of gov­ern­ment pressure (from 10% in April to 20% in September). This change may reflect Harvard’s success in resisting pressure from the Trump admin­is­tra­tion; the April survey took place just after Harvard filed suit against the admin­is­tra­tion and the September survey began just after a district court decision in its favor.

Though expert ratings have remained largely stable since April, it is important to note the longer-term decline in expert eval­u­a­tions of demo­c­ra­t­ic per­for­mance we have seen compared to post-election ratings from November 2024. As the figure below demon­strates, expert ratings have declined sig­nif­i­cant­ly on 20 of the 31 ratings we have measured across that time period (at the p<.01 level), including gov­ern­ment agencies not being used to punish political opponents (from 68% in November 2024 to 7% in September 2025), not inter­fer­ing with the press (from 76% to 23%), and respect­ing the right to free speech and expres­sion (from 81% to 30%). 

Performance ratings were also largely stable since April among Democrats and Republicans on the 31 demo­c­ra­t­ic prin­ci­ples tracked in both surveys. Democrats were more likely to perceive declines in per­for­mance; eight such declines were sta­tis­ti­cal­ly sig­nif­i­cant at the p<.05 level, including, for example, sanctions for mis­con­duct and con­sti­tu­tion­al limits on the executive branch. Republicans only perceived a sig­nif­i­cant decrease in per­for­mance on district bound­aries not being biased and actually reported improved per­for­mance on judicial limits on the executive (perhaps reflect­ing the cases where court orders have blocked Trump admin­is­tra­tion actions).

The aftermath of the killing of Charlie Kirk

The con­ser­v­a­tive com­men­ta­tor and activist Charlie Kirk, who directed the group Turning Point USA, was shot and killed in an apparent act of political violence during an event at Utah Valley University on September 10. The event attracted enormous political and media attention given Kirk’s fame and the public nature of his murder, which was captured on videos that cir­cu­lat­ed widely on social media. President Trump and his allies called for ret­ri­bu­tion against left-wing groups they blamed for the shooting despite a lack of credible evidence to support these claims. The event’s aftermath also featured a backlash against reactions to Kirk’s death that were seen as inap­pro­pri­ate. Numerous people were fired from their jobs or sanc­tioned by their employers, including teachers and pro­fes­sors as well as ABC late night host Jimmy Kimmel, whose show was tem­porar­i­ly suspended amidst pressure from Brendan Carr, the Trump appointee who chairs the Federal Communications Commission.

Our expert and public surveys were col­lect­ing data at the time of the shooting. As discussed below, 85% of experts rated Kirk’s shooting as a threat to democracy, including 57% who called it an extra­or­di­nary or serious threat. 

In the figure below, we sep­a­rate­ly measure overall expert and public ratings of U.S. democracy as well as eval­u­a­tions of per­for­mance on the demo­c­ra­t­ic principle that “Government effec­tive­ly prevents private actors from engaging in polit­i­cal­ly-motivated violence or intim­i­da­tion” for responses received in the days before the shooting and those received afterward. (We exclude the day of the shooting given the ambiguity of when people learned about the news.)

None of the changes in ratings of U.S. democracy overall reach con­ven­tion­al levels of sta­tis­ti­cal sig­nif­i­cance for any group (experts, the public, and Democrats and Republicans in the public). However, the per­cent­age of experts who say the country is fully or mostly meeting the standard of the gov­ern­ment effec­tive­ly pre­vent­ing political violence or intim­i­da­tion did fall sig­nif­i­cant­ly, declining from 52% to 29% (no changes were sig­nif­i­cant among the public or partisan subgroups). We observe no sta­tis­ti­cal­ly sig­nif­i­cant dif­fer­ence after the shooting among any group in perceived per­for­mance at pro­tect­ing unpopular speech, using law enforce­ment to advance political agendas, or using gov­ern­ment agencies to punish political opponents.

Our survey measures of approval of the Kirk shooting were espe­cial­ly con­cern­ing. Prior surveys by Bright Line Watch and others show that support for hypo­thet­i­cal acts of political violence is typically very low once inat­ten­tive respon­dents are excluded and terms like “violence” are defined more precisely. In February 2025, for instance, we found that only 3% of Democrats and 2% of Republicans endorse violent felonies to advance their political goals. New data collected after the Kirk shooting by the Polarization Research Lab (PRL) shows that only 1% of Americans explic­it­ly endorse partisan murder in a hypo­thet­i­cal scenario (down from 3% previously).

However, the PRL data shows that 3.3% of Americans said it is always accept­able for a person to be happy about the murder of a public figure they oppose and another 8.1% said it was usually accept­able. Among Democrats, 4.3% said it was always accept­able to be happy and 14.9% said it was usually acceptable.

In our public survey data, we asked respon­dents a more direct question — whether they approved or dis­ap­proved of various recent events, including that “Conservative com­men­ta­tor Charlie Kirk shot and killed while speaking at a college campus in Utah” (added to our survey on the evening of the shooting). For these estimates, we follow best practices from past research and exclude survey respon­dents who failed a survey attention check or who admit to providing humorous or insincere responses. Overall, Americans over­whelm­ing­ly indicated that they dis­ap­prove of Kirk’s killing, including 88% of Democrats and 95% of Republicans. However, 12% of Democrats approved — an estimate that is dis­tress­ing­ly high for a real-world event. These results suggest that more people may approve of or at least tolerate violence against specific dis­fa­vored figures after the fact than are explic­it­ly willing to endorse political violence in the abstract, espe­cial­ly when the figure in question pre­vi­ous­ly made state­ments that partisans find offensive.

Fighting fire with fire? Redistricting and responses to demo­c­ra­t­ic norm violations

With Republican prospects in the 2026 U.S. House of Representatives elections looking unfa­vor­able, Texas Republicans redrew the district bound­aries in their state in an unusual mid-decade redis­trict­ing intended to gain five seats for the party. A similar plan was enacted in Missouri to yield another Republican seat, though an effort is underway to force it to be approved through a ref­er­en­dum first. In response to the Texas move, Democrats in California enacted a plan to put an ini­tia­tive up for a special election that would redis­trict their state to yield five addi­tion­al seats for Democrats. Other states may soon enter the fray.

We surveyed both experts and the public to under­stand how they evaluated these actions. We consider first the actions taken by Texas and California, which were the most widely pub­li­cized, before dis­cussing the more general scenario of tit-for-tat redis­trict­ing and measuring how different groups evaluate violating demo­c­ra­t­ic norms in response to perceived norm violations.

The experts are concerned about Texas Republicans redrawing their state’s con­gres­sion­al map before the 2026 election — in total, 92% rate it as a threat to democracy — but are divided on the California plan to redraw their districts. California Democrats describe their actions as a response to the Texas plan, but only 32% of experts see the California redis­trict­ing proposal as a benefit to democracy, whereas 38% see it as a threat. One possible inter­pre­ta­tion of these findings is that experts are concerned about the prospect of a spiral of esca­la­tion between the parties.

To estimate the extent of the perceived harm, we asked both expert and public respon­dents to provide the ratings on a 0–100 scale that they expect to give U.S. democracy in 2027 under two scenarios: (1) Texas redraws its con­gres­sion­al map bound­aries and California does not versus (2) California and Texas both redis­trict. These ratings are plotted below for experts, the public overall, and Democrats and Republicans in the public sample.

Experts, who indicated earlier in the survey, on average, that they expected to rate U.S. democracy as 47 in 2027, expected to rate U.S. democracy as 43 on average in 2027 if only Texas redis­tricts, a sig­nif­i­cant decline. However, con­di­tion­al on California respond­ing with its own redis­trict­ing, their expected ratings rebound somewhat to 49, a sig­nif­i­cant improve­ment versus the Texas-only scenario. We observe similar patterns of higher ratings for the off­set­ting redis­trict­ing scenario among the public overall (38 versus 44) and among Democrats (24 versus 40), whereas Republicans expect to rate U.S. democracy somewhat lower if both states redis­trict (49) than if Texas redis­tricts uni­lat­er­al­ly (55).

As the figure below illus­trates, this pattern of dif­fer­en­tial approval of redis­trict­ing among Republicans largely persists when we asked respon­dents about ran­dom­ized vari­a­tions of a scenario in which one party redis­tricts in one unspec­i­fied state and the other party redis­tricts in another. 

As in the specific Texas/California case above, Republicans largely approve (64%) of the GOP redis­trict­ing first in one state and dis­ap­prove (40% approval) when Democrats redis­trict in response. Democrats, unsur­pris­ing­ly, dis­ap­prove of Republicans redis­trict­ing first (just 8% approve) but they approve (62%) of Democrats redis­trict­ing in another state in response. 

However, Republicans tend to be more sup­port­ive of redis­trict­ing overall. For instance, when Democrats move first, most party members do not approve. In total, only 34% of Democrats approve of their party redis­trict­ing uni­lat­er­al­ly, which is not mea­sur­ably different than the 33% approval we observe among Republicans in the same scenario. By contrast, a majority of Republicans (64%) approve of their party moving first.

We also asked a random subset of respon­dents to consider a more abstract scenario in which the political parties are not named. They were instead just told about “one party” ini­ti­at­ing redis­trict­ing and the “other party” respond­ing. Here again, a majority of Republicans (52%) approves of the party ini­ti­at­ing the process and a near-majority approves the second party respond­ing in kind (47%). Democrats, by contrast, over­whelm­ing­ly dis­ap­prove ini­ti­at­ing a redraw (just 13% approve) but, con­di­tion­al on one party having started the process, they typically approve of the second party respond­ing (61%). 

Finally, we consider the principle of “fighting fire with fire” in response to perceived demo­c­ra­t­ic norm vio­la­tions (as California Governor Gavin Newsom and other Democrats have put it). We asked members of the public the following question: “When politi­cians from {the Democratic Party, the Republican Party, one party} violate demo­c­ra­t­ic norms to gain advantage, is it accept­able for the {Republican Party, Democratic Party, other party} to violate demo­c­ra­t­ic norms in response?” Respondents could answer by saying “Always accept­able — you have to fight fire with fire,” “Sometimes accept­able — only to restore balance,” or “Never accept­able — two wrongs don’t make a right.” 

We observe notable dif­fer­ences in support for respond­ing to demo­c­ra­t­ic norm vio­la­tions with a demo­c­ra­t­ic norm violation. In a scenario where the other party is described as com­mit­ting the first norm violation, 67% of Republicans and 64% of Democrats say it is never accept­able for the other party to violate norms in response. In the opposite scenario, however, 52% of Republicans and 73% of Democrats say it is always or sometimes accept­able for their party to violate norms in response to a norm violation by the other party. The greater endorse­ment among Democrats may reflect their status as the party out of power or the sequenc­ing of the Texas and California redis­trict­ing ini­tia­tives. (Partisan responses to the generic “one party”/“other party” scenario fall somewhere in between these two extremes.)

Experts were only asked the “one party”/“other party” question variant. Seven in ten said violating norms in response to a norm violation is sometimes accept­able to restore balance compared to 23% who said it is never accept­able and 6% who said it is always acceptable. 

Actual and potential threats to democracy

We asked our experts to assess the impact of a list of recent events (including the Kirk shooting) and potential future events on American democracy. The set of recent events that experts were asked to consider was the following (each rated a random subset of 8 from the full list):

Full list of recent events

• After its federal research grants were frozen, Columbia University pays a $200 million fine to settle alle­ga­tions from the Trump admin­is­tra­tion that it failed to do enough to stop the harass­ment of Jewish students.
• The Trump admin­is­tra­tion freezes federal research grants to UCLA and demands a $1 billion set­tle­ment over alleged civil rights violations.
• President Trump tries to fire Federal Reserve Governor Lisa Cook after his admin­is­tra­tion accuses her of mortgage fraud
• Federal Reserve Governor Lisa Cook sues the Trump admin­is­tra­tion after he tries to fire her for alleged mortgage fraud.
• Texas Republicans redraw their state’s con­gres­sion­al map before the 2026 election to be more favorable to Republicans.
• In response to Texas GOP redis­trict­ing, California Democrats schedule a special election on a ballot measure to redraw their state’s con­gres­sion­al maps before the 2026 election to be more favorable to Democrats.
• The Trump admin­is­tra­tion deploys over 800 National Guard troops to Washington, D.C.
• Over 1,100 National Guard troops from Republican-led states are deployed to Washington, D.C., at the request of the Trump administration.
• President Trump fires Bureau of Labor Statistics Commissioner Erika McEntarfer after accusing her of mis­rep­re­sent­ing jobs data.
• President Trump issues an executive order requiring political appointees to approve federal research grants which were pre­vi­ous­ly managed by career sci­en­tists and civil servants.
• The Trump admin­is­tra­tion announces an inves­ti­ga­tion into New York Attorney General Letitia James for alleged mortgage fraud.
• U.S. Immigration and Customs Enforcement (ICE) dra­mat­i­cal­ly expands its enforce­ment and detention capacity.
• CBS parent company Paramount agrees to $16M set­tle­ment of Trump lawsuit over a Kamala Harris interview on “60 Minutes.”
• The Federal Communications Commission approves merger of CBS parent company Paramount after $16M set­tle­ment with Trump over a Kamala Harris interview on “60 Minutes.”
• The Supreme Court issues a decision that allows the Department of Homeland Security to deport nonci­t­i­zens to third countries without notice while lit­i­ga­tion proceeds.
• Columbia graduate Mahmoud Khalil is released from ICE detention by court order.
• The Supreme Court allows Trump to remove inde­pen­dent agency members without cause.
• After a scuffle outside an immi­gra­tion detention facility, Newark Mayor Ras Baraka is arrested for tres­pass­ing and Congresswoman LaMonica McIver is charged with assault­ing law enforce­ment officers.
• The United States accepts the donation of a Boeing 747‑8 jet from Qatar for use as Air Force One.
• A Supreme Court decision allows the Trump admin­is­tra­tion to proceed with sweeping layoffs of federal workers.
• The Trump admin­is­tra­tion deploys National Guard troops to Los Angeles in response to protests.
• The Supreme Court holds that federal courts cannot issue universal or nation­wide injunctions.
• The admin­is­tra­tion ends direct federal funding to NPR and PBS.
• A gunman kills Minnesota state Rep. Melissa Hortman and her husband and wounds another lawmaker and their spouse in a polit­i­cal­ly motivated attack.
• FBI agents raid the home of former Trump adviser turned critic John Bolton.
• President Trump claims that he has the authority to use military force against suspected drug smugglers outside the U.S. as if they were wartime combatants.
• A federal judge orders the Trump admin­is­tra­tion to reverse its cuts of more than $2.6 billion in research funding for Harvard University.
• The District of Columbia sues to stop President Trump’s deploy­ment of National Guard in Washington.
• President Trump directs the Secretary of Defense to create a standing National Guard quick reaction force.
• Conservative com­men­ta­tor Charlie Kirk is shot and killed while speaking at a college campus in Utah.

Experts were first asked whether each event would benefit, threaten, or not affect American democracy. Those who selected benefit or threat were then asked if the benefit or threat was little, moderate, serious, or extra­or­di­nary. The graph below is ordered by the per­cent­age of experts who identify the event as a threat to democracy. 

In total, twenty events were rated as a threat to democracy by 80% of experts or more, including the killing of Minnesota state leg­is­la­tor Melissa Hortman and her husband (95%), Trump’s firing of the Bureau of Labor Statistics com­mis­sion­er Erika McEntarfer (93%), and Texas changing its con­gres­sion­al districts mid-decade to advantage Republicans in the 2026 election (92%). The events that were rated as extra­or­di­nary or serious threats by the greatest pro­por­tion of experts were the National Guard deploy­ment to Los Angeles (77% extraordinary/serious), the Supreme Court allowing the Department of Homeland Security to deport non-citizens without notice (70% extraordinary/serious), Trump directing the Department of Defense to create a National Guard quick reaction force (65% extraordinary/serious), and the dramatic expansion of the enforce­ment and detention capa­bil­i­ties of U.S. Immigration and Customs Enforcement (ICE; 61% extraordinary/serious). Six other events are rated as threats to democracy by 60% of experts or more. By contrast, four events are rated as benefits to democracy by 60% or more of experts, including the Harvard court decision ordering the reversal of gov­ern­ment funding cuts (81% say it is a benefit) and Mahmoud Khalil being released from ICE detention (79%). Finally, experts are split over the California redis­trict­ing plan (as discussed above), with 38% rating it as a threat to democracy, 32% saying a benefit, and 30% saying it will not affect democracy.

To see how the public views these events, we asked members of the public whether they approved of them or not. The figure below presents approval among Democratic and Republican members of the public (including leaners).

As the figure indicates, numerous recent events that a majority of experts rate as extra­or­di­nary or serious threats to democracy receive wide­spread support from Republicans. For instance, 87% of Republicans approve of the National Guard deploy­ment to Los Angeles, which 77% of experts rated as an extra­or­di­nary or serious threat to democracy. Besides Charlie Kirk’s murder, only two events of this type attract support from fewer than 70% of Republicans: the Texas redis­trict­ing (69%) and the FBI raid on the home of former Trump adviser turned critic John Bolton (63%). 

In addition to these ratings of recent events, experts were also asked to rate whether the  following possible future events would present a threat or benefit to American democracy:

Full list of future events

• Habeas corpus is suspended for any group of people within the United States before 2027.
• A federal, state, or local gov­ern­ment denies or refuses to recognize the cit­i­zen­ship status of one or more U.S.-born children of non-citizen parents before 2027.
• U.S. author­i­ties transfer a current or former U.S. citizen to a prison abroad before 2027.
• The U.S. gov­ern­ment seeks a judicial order to revoke the cit­i­zen­ship of a nat­u­ral­ized American for conduct that occurred after nat­u­ral­iza­tion before 2027.
• The U.S. president formally invokes the Insurrection Act before 2027.
• The President or another executive branch official directs federal law enforce­ment to refuse to enforce a federal court order before 2027.
• Senior DOJ leaders override normal pro­ce­dures to open or accel­er­ate a case against a high-profile opponent or critic of the president before 2027.
• Senior DOJ leaders overrule normal DOJ practices and pro­ce­dures to block, stall, or close a case involving the President or his family, sup­port­ers, or allies before 2027.
• The Senate passes a rec­on­cil­i­a­tion bill where Byrd Rule com­pli­ance is not deter­mined with input from the Congressional Budget Office before 2027.
• The U.S. pres­i­den­tial admin­is­tra­tion alters or delays the public release of a major federal economic indicator before 2027.
• Federal pros­e­cu­tors indict one or more Obama admin­is­tra­tion officials in con­nec­tion with the Trump/Russia inves­ti­ga­tion before 2027.
• Citing concerns about crime, the President deploys National Guard troops to other cities besides Los Angeles and Washington, D.C., before 2027.
• The federal gov­ern­ment conducts a new census before the regularly scheduled 2030 census.
• Citing concerns about voter fraud, the President deploys military forces or federal law enforce­ment outside polling places in the U.S. during the 2026 midterm elections.
• Harvard reaches an agreement with the Trump admin­is­tra­tion before 2027.
• The Supreme Court sig­nif­i­cant­ly limits key pro­vi­sions of the Voting Rights Act before 2027.
• Trump attempts to stay in power beyond the end of his term in office in January 2029.

The figure below again shows expert ratings of these potential future events ranked by the per­cent­age of experts who rate them as a threat to democracy.

Experts are virtually unanimous on most of the 17 potential events surveyed. In total, 13 were rated as threats to democracy by 89% or more of experts, such as Trump seeking a third term in office in violation of the 22nd Amendment to the Constitution (98%, including 97% who rate it as an extra­or­di­nary or serious threat), the sus­pen­sion of habeas corpus (98%; 96% extraordinary/serious), and invoking the Insurrection Act (98%; 92% extraordinary/serious).

With so many democracy-threat­en­ing events taking place or being con­tem­plat­ed as pos­si­bil­i­ties, it is important to consider the full scope of Trump’s second term. We therefore list the top 10 past events and the top 10 potential future events in the figure below that experts most often rated as posing extra­or­di­nary or serious threats to democracy (items from the most recent survey are rep­re­sent­ed in bold; others are drawn from the February and April surveys).

As the figure indicates, the National Guard deploy­ment to Los Angeles is the only event since April 2025 that experts have rated among the top threats to democracy during Trump’s second term (though experts may have adjusted their standards as events have developed since January). However, experts rate eight potential events from the most recent survey as among the ten most sig­nif­i­cant potential threats to democracy since Trump took office for his second term, sug­gest­ing they are even more appre­hen­sive about future developments.

Assessing the like­li­hood of future threats to democracy

Assessing the like­li­hood of threats to democracy is a core task of Bright Line Watch. We have therefore asked experts to assess the like­li­hood of potential threats in our surveys since 2018 and have, since December 2024, adjusted those forecasts to account for an observed tendency toward pessimism (discussed below). To further address concerns about the accuracy of expert forecasts, we have partnered with the pre­dic­tion platform Metaculus in recent months to generate com­par­i­son pre­dic­tions from fore­cast­ers, which we will assess for accuracy in the future. 

Following past surveys, we asked experts and fore­cast­ers to assign a prob­a­bil­i­ty to potential future events that are relevant to the state of democracy in the U.S. As part of our part­ner­ship with Metaculus, both groups provided forecasts for forty questions — the 39 questions in the new U.S. Democracy Threat Index that we developed with Metaculus (see overview here) plus one on future expert ratings of U.S. democracy measured by Bright Line Watch. The full list of questions and linked demo­c­ra­t­ic prin­ci­ples is available here. (We focus in this report on forecasts for the 2025–2026 period; the U.S. Democracy Threat Index forecasts on Metaculus also include pre­dic­tions for 2027–2028.)

Finally, to provide another point of com­par­i­son for our experts, we also solicited forecasts from Mantic, a new AI fore­cast­ing startup that recently placed eighth out of more than 550 human and AI entrants in a fore­cast­ing com­pe­ti­tion held by Metaculus. The platform described the per­for­mance of Mantic’s AI forecast bot as “the best rank a bot has ever achieved competing with human forecasters.” 

In this report, we focus on the ten forecast questions con­cern­ing unam­bigu­ous­ly negative events for democracy where experts rated the threat to democracy the event would pose. The full survey instru­ment admin­is­tered to experts, which includes all the forecast questions, is provided here; the U.S. Democracy Threat Index is here. Figures reporting the full set of expert, fore­cast­er, and AI forecasts are in the appendix.

Notably, we have found in past research with the political scientist Andrew Little that experts tend to over­es­ti­mate the prob­a­bil­i­ty of negative events for democracy. We therefore apply the pessimism cor­rec­tion we developed with Little based on the observed incidence of events during the first Trump term to create adjusted forecast prob­a­bil­i­ties for each item, which are plotted below against the raw expert estimates. 

The pessimism adjust­ment most heavily down­weights events around the middle of the prob­a­bil­i­ty scale. As a result, the sus­pen­sion of habeas corpus, which experts rate as having a 50% chance of taking place, has an adjusted forecast prob­a­bil­i­ty of 14%. We similarly see sub­stan­tial decreases in the prob­a­bil­i­ty of Trump invoking the Insurrection Act (from 65% to 30%) and impris­on­ing U.S. citizens abroad (70% to 41%). The adjust­ment is smaller for events seen as most likely; for instance, experts estimate the prob­a­bil­i­ty of economic data being altered or delayed to be 80% compared with 65% after the adjustment.

The figure below then compares our pessimism-adjusted expert forecasts for these potential events with human fore­cast­ers on Metaculus.

The human fore­cast­ers and experts are closely aligned after the pessimism adjust­ment to the expert forecasts; the median dif­fer­ence in forecasts is 9 per­cent­age points. Alignment is espe­cial­ly close for the Trump admin­is­tra­tion revoking nat­u­ral­iza­tion for sub­se­quent conduct (53% experts, 60% Metaculus), ordering gov­ern­ment officials to disregard court orders (53% experts, 63% Metaculus), and the gov­ern­ment altering or delaying economic data (65% experts, 59% Metaculus). Metaculus fore­cast­ers see a much greater like­li­hood than experts of normal pro­ce­dures being over­rid­den in a case against a critic or opponent of the president (41% experts, 67% Metaculus) but are more skeptical about the chances of a U.S. citizen being impris­oned abroad (41% experts, 23% Metaculus). Notably, however, experts are not always more pes­simistic after the adjust­ment; Metaculus fore­cast­ers rate the prob­a­bil­i­ty of six of the ten events higher than do experts.

The figure below instead compares expert forecasts (with the pessimism adjust­ment) to the AI forecasts from Mantic. (A graph comparing Metaculus human forecasts and Mantic AI forecasts is provided in the appendix.)

As the figure indicates, experts thought the events in question were more likely to occur than the AI forecasts from Mantic even after the pessimism adjust­ment. Divergences were espe­cial­ly wide for a U.S. citizen being impris­oned abroad (41% experts, 7% Mantic) and nat­u­ral­iza­tion being revoked for sub­se­quent conduct (53% experts, 18% Mantic). This gap may reflect the emphasis in fore­cast­ing (both human and AI) on prior prob­a­bil­i­ties of outcomes taking place (“base rates”), which are typically low in the U.S. given its rel­a­tive­ly high levels of demo­c­ra­t­ic per­for­mance and adherence to the rule of law in recent decades. The reasoning that Mantic’s AI provides for its estimates repeat­ed­ly describes his­tor­i­cal prece­dents and event frequencies.

Finally, the figure plots the three event prob­a­bil­i­ty forecasts — experts, pre­dic­tion platform fore­cast­ers, and the AI forecasts from Mantic — against the per­cent­age of experts who view the potential event as posing an extra­or­di­nary or serious threat to democracy.

Among the events rated by most experts as an extra­or­di­nary or serious threat to democracy (all but the manip­u­la­tion of economic data), the median forecast among experts, pre­dic­tion platform forecasts, and Mantic’s AI forecast for 2025–2026 is a 50% or greater like­li­hood of normal pro­ce­dures being over­rid­den in legal cases con­cern­ing pres­i­den­tial allies and opponents (66% and 62%, respec­tive­ly), a military deploy­ment to a third metro area besides Los Angeles and Washington, D.C. (68%), an order revoking nat­u­ral­iza­tion based on sub­se­quent conduct  (53%), and an order directing law enforce­ment not to enforce a court order (53%). Forecast prob­a­bil­i­ties are lower for other events rated as very serious threats to democracy: invoking the Insurrection Act (median forecast: 22%), impris­on­ing a U.S. citizen abroad (23%), denying birthright cit­i­zen­ship (13%), and sus­pend­ing habeas corpus (14%).

Support for anti­de­mo­c­ra­t­ic actions

Participants were asked about their support for practices that undermine the public good and could damage democracy using a scale drawn from prior research. Specifically, we asked whether they agree or disagree that their co-partisans should take the following set of actions:

• should do every­thing they can to hurt the [other party], even if it is at the short-term expense of the country;
• should do every­thing in their power within the law to make it as difficult as possible for [the other party] to run the gov­ern­ment effectively;
• should redraw districts to maximize their potential to win more seats in federal elections, even if it may be tech­ni­cal­ly illegal;
• should use the Federal Communications Commission to heavily restrict or shut down Fox News [shown to Democrats] / MSNBC [shown to Republicans] to stop the spread of fake news
• it’s OK to sacrifice U.S. economic pros­per­i­ty in the short-term in order to hurt [the other party’s] chances in future elections.

We pre­vi­ous­ly measured support for these anti­de­mo­c­ra­t­ic actions in our November 2021 survey; we revisit them now to assess whether support levels have changed nearly four years later with a switch in party control of the White House and President Trump taking many actions that experts perceive as threat­en­ing democracy. The results are presented in the figure below.

Democrats con­trolled the White House and both chambers of Congress in 2021 but now control none of these, plus they have very negative expec­ta­tions about the future of American democracy. Accordingly, they are more sup­port­ive of each of these norm-violating tactics than they were nearly four years ago, though none receives majority support from Democratic partisans. It is important to note that the inter­pre­ta­tion of some items may differ in the current context versus 2021 (e.g., the appro­pri­ate­ness of “do[ing] every­thing in their power within the law to make it as difficult as possible for [the other party] to run the gov­ern­ment effec­tive­ly”). However, Democratic approval increased even for “tech­ni­cal­ly illegal” redis­trict­ing (from 12% in 2021 to 27% in 2025) and for restrict­ing or shutting down Fox News (from 37% to 50%). Republican support for these actions is lower or unchanged except for redis­trict­ing (from 10% to 16%), possibly reflect­ing support for the Texas plan, and restrict­ing or shutting down MSNBC (an increase from 38% to 47%), poten­tial­ly in response to President Trump repeat­ed­ly calling for gov­ern­ment actions against tele­vi­sion networks.

Appendix

Bright Line Watch conducted its twenty-sixth survey of academic experts and its twenty-third survey of the general public from September 5–17 and September 5–16, respec­tive­ly. Our public sample consisted of 2750 par­tic­i­pants from the YouGov panel who were selected and weighted to be rep­re­sen­ta­tive of the U.S. adult pop­u­la­tion. We also surveyed 703 political science experts across a diverse range of subfields. Our email list was con­struct­ed from the faculty list of U.S. insti­tu­tions rep­re­sent­ed in the online program of the 2024 American Political Science Association conference.All estimates shown in the report used weights provided by YouGov. Our expert sample is unweight­ed because we do not collect demo­graph­ic data to protect anonymity. Error bars in our graphs represent 95% con­fi­dence intervals. Data are available here.

35 democratic principles

• Government officials are legally sanc­tioned for misconduct
• Government officials do not use public office for private gain
• Government agencies are not used to monitor, attack, or punish political opponents
• All adult citizens enjoy the same legal and political rights
• Government does not interfere with jour­nal­ists or news organizations
• Government effec­tive­ly prevents private actors from engaging in polit­i­cal­ly-motivated violence or intimidation
• Government protects indi­vid­u­als’ right to engage in unpopular speech or expression
• Political com­pe­ti­tion occurs without criticism of opponents’ loyalty or patriotism
• Elections are free from foreign influence
• Parties and can­di­dates are not barred due to their political beliefs and ideologies
• All adult citizens have equal oppor­tu­ni­ty to vote
• All votes have equal impact on election outcomes
• Elections are conducted, ballots counted, and winners deter­mined without pervasive fraud or manipulation
• Executive authority cannot be expanded beyond con­sti­tu­tion­al limits
• The leg­is­la­ture is able to effec­tive­ly limit executive power
• The judiciary is able to effec­tive­ly limit executive power
• The elected branches respect judicial independence
• Voter par­tic­i­pa­tion in elections is generally high
• Information about the sources of campaign funding is available to the public
• Public policy is not deter­mined by large campaign contributions
• Citizens can make their opinions heard in open debate about policies that are under consideration
• The geo­graph­ic bound­aries of electoral districts do not sys­tem­at­i­cal­ly advantage any par­tic­u­lar political party
• Even when there are dis­agree­ments about ideology or policy, political leaders generally share a common under­stand­ing of relevant facts
• Elected officials seek com­pro­mise with political opponents
• Citizens have access to infor­ma­tion about can­di­dates that is relevant to how they would govern
• Government protects indi­vid­u­als’ right to engage in peaceful protest
• Law enforce­ment inves­ti­ga­tions of public officials or their asso­ciates are free from political influence or interference
• Government sta­tis­tics and data are produced by experts who are not influ­enced by political considerations
• The law is enforced equally for all persons
• Incumbent politi­cians who lose elections publicly concede defeat
• Universities, busi­ness­es, and pro­fes­sion­al orga­ni­za­tions operate inde­pen­dent­ly of gov­ern­ment pressure
• Military lead­er­ship remains polit­i­cal­ly neutral and does not favor or support any political party or candidate
• Officeholders do not use the military to advance political agendas or intim­i­date political opponents
• Law enforce­ment remains polit­i­cal­ly neutral and does not favor or support any political party or candidate
• Officeholders do not use law enforce­ment to advance political agendas or intim­i­date political opponents

Approval of recent and future events

Approval of future events and future threat to democracy

Political systems

Trump / GOP supporter?

Since late 2022, we have asked each respon­dent who iden­ti­fies as a Republican (or leans toward that party) whether they regard them­selves as more a supporter of the Republican Party or more a supporter of Donald Trump. Through the summer of 2023, more iden­ti­fied primarily with the party. By early 2024, when Trump had locked up the party’s pres­i­den­tial nom­i­na­tion, a narrow majority iden­ti­fied primarily with Trump, and the share of Trump-over-party Republican respon­dents grew further, sta­bi­liz­ing at about 60% of Republican respon­dents across four surveys through late 2024 and early 2025. In September 2025, the share dropped to 53% more Trump to 47% more GOP.

Biden rightful winner

Forecasting future events

The full list and wording of forecast questions are available in the survey instru­ment. All Mantic AI forecasts are available here.

Adjusted expert forecasts of future threats to democracy

Evaluations of U.S. demo­c­ra­t­ic performance