News Coverage of Chauvin Trial
News Coverage of Chauvin Trial
Paper submitted to the 74th Annual Conference of the International Communication Association
Gold Coast, Australia 20–24 June 2024
Abstract
Abstract: On 26 May 2020 the world learned of the death of George Floyd during an arrest in
Minneapolis. Two information sources framed the death: a police press release reporting a
deadly “medical incident during police interaction” and a bystander’s video showing the
infamous restraint determined as a murder committed by Derek Chauvin and three other police.
Our research, based on a human-coded content analysis of the national, cable and local news
coverage of protests following the release of the video, the trial and trial verdict, shows that these
three news sources differed significantly in tone, framing (Iyengar: episodic/thematic), level of
detail, accuracy and relevance of detailed analyses, and W. Lance Bennett’s four information
biases: authority-disorder, dramatization, personalization, and fragmentation. We question
whether detailed, thematic news stories (and the expected attributions of responsibility to
authorities) are desirable if the details are inaccurate, tangential and geared to Bennett’s
information biases.
1
Local, National, and Cable News Coverage of the Protests and the Trial of
Derek Chauvin for the Murder of George Floyd
On May 25, 2020, at around 9:00 p.m., George Floyd—an African-American man—died
when police arrested him on suspicion of attempting to use a counterfeit $20 bill at a
neighborhood grocery store in Minneapolis. No news broadcasters covered the event that day.
Shortly after midnight on May 26, the Minneapolis Police Department reported a “medical
incident” resulting in a death. The statement indicates that Floyd resisted arrest, appeared to be
suffering “medical distress,” and “was transported to Hennepin County Medical Center by
ambulance where he died a short time later” (City of Minneapolis, 2020). The statement fails to
mention that officers subdued Floyd by kneeling on his neck, back, and leg as he repeatedly said,
“I can’t breathe,” while bystanders begged Officer Derek Chauvin and others to render aid. An
entirely different scenario was painted by the cell phone recording made by seventeen-year-old
Darnella Frazier, which showed that Floyd died as Chauvin knelt on his neck for nine-and-a-half
minutes. Frazier posted that video to Facebook and Instagram at 1:46 a.m. on May 26. Frazier’s
video resulted in protests about Floyd’s murder in Minneapolis (as well as in other cities in
America and worldwide). On April 20, 2021, Chauvin was tried and found guilty of second– and
anticipating and then reacting to the verdict that dominated the news that day.
Our content analysis of the national, cable, and local news coverage of protests following
the release of the video, the trial, and the trial verdict shows that these three news sources
differed significantly in tone, framing (Iyengar 1991: episodic/thematic), level of detail, accuracy
and relevance of detailed analyses, and the four information biases described by Bennett (2016).
We question whether detailed, thematic news stories (and the expected attributions of
2
responsibility to authorities) are desirable if the details are inaccurate, tangential, and exhibit
Previous Research
Local News
Local news coverage of world, national, and, indeed, local happenings is an under-
researched communication medium. Although the use of TV as a main source of news declined
in the second decade of the twenty-first century, local TV news has remained a favorite choice
for those who watch broadcast news and a primary source of election news. In 2020, local TV
news viewership outpaced network and cable (Pew Center for the People and the Press, 2021).
Given the continued popularity of local TV news as a source of political information, we may
ask: What is on the news? Focusing on news coverage of Election 2000, Stevens et al. (2006)
showed that local news is not local. Results from studies of election news coverage for 2008 and
2016 show small changes in election coverage since 2000, confined mainly to a further decline in
coverage of local events (Allen et al. 2023). As the Floyd murder and Chauvin trial were local
events for the Minneapolis stations considered in this study, we have an opportunity to examine
how local news may have differed from national representations of these events. As a foundation
for this paper, we may ask more specifically: In previous research, how are race, matters of racial
The crime narrative found in local news broadcasts not only overrepresents Black persons
as perpetrators of crimes, but underrepresents this group as victims of crimes; White persons, by
contrast, are overrepresented as victims, and Latinx persons are generally absent from news
crime narratives (Dixon et al., 2003; Dixon & Linz, 2000). Among White viewers, exposure to
3
the racial dimension of this crime narrative increased negative attitudes about Black Americans
as well as support for punitive methods as a central feature of crime controlling policies, while
the opposite was the case for Black viewers (Gilliam & Iyengar, 2000).
Exposure to network news also increases endorsement of many racial stereotypes (Dixon,
2008). In experiments varying the race of perpetrators of a crime and arresting police officers in
a story embedded in a newscast, Dixon (2007) found that respondents to an experiment were
more likely to identify an unidentified perpetrator as Black and an unidentified officer as White.
Respondents who were heavier TV news viewers were more likely than lighter news users to
make this racialized identification of an unidentified officer, and heavier consumers of news
were also more likely than lighter news consumers to have positive views of the unidentified
officers, but not of the Black officers. Fridkin et al. (2017) show that framing matters. A law-and-
order frame of a violent confrontation between a White police officer and a Black woman
increased support for the officer and decreased support for the woman, while a brutality frame
shifted support in the opposite direction for each story character. Research on cable and network
news coverage from 2008 through 2012 (Dixon & Williams, 2015) shows new patterns of
defining the “other,” in which Muslims are also extremely overrepresented as terrorists.
Racialized news discourses occur within an institutional media context (van Dijk, 2023)
including processes of news production—how news is gathered, from what expert or community
source, by persons of what background and identity. Modest increases in the number of Persons
of Color working as U.S. journalists have not changed patterns of news discourses that produce
Entman and Rojecki (2001) show racialized representation exceeds crime coverage and
news in general. Research on the racialized news framing of prominent Black figures such as
4
Barack Obama (Ramasubramanian & Martinez, 2017) and Colin Kaepernick (Doehler, 2023)
finds that tokenism, symbolic racism, and elevated attention to race in media coverage reinforce
race stereotypes in the United States. News viewers apparently have difficulty distinguishing
facts presented in the drama of news from fictional narratives of entertainment media (Carlson &
Trichtinger, 2001). From print to broadcast, news and entertainment media significantly promote
racial bias.
Representations of Black Lives Matter (BLM) are no exception (Banks, 2018). Unlike depictions
of the “Arab Spring” (2010-2013) as a movement demanding government reform and regime
change, BLM protests had been viewed as riots and the founders and individual participants
depicted as racist, corrupt, and “anti-police.” Media rhetoric initially described BLM and protests
in ways that underscored Black participation (Bonilla-Silva, 2012), raising the specter of racial
marginalized or blamed victims of police killings (Mills, 2017). As a result, news stories about
anti-racism demonstrations, including BLM, are more likely to be framed negatively than other
mass social movement activism (Brown & Mourão, 2019, 2021). While FOX News is most
generally associated with such coverage (Mills, 2017; Brown & Mourão 2019, 2021), narratives
that promote racial stereotypes, elevate disruption, and discredit BLM demonstrations—while
nonviolent antithesis to BLM demonstrations—are not unique to FOX News (Brown & Mourão,
2022; Banks, 2018). Participants in the protests noted the exaggerated depictions of violence in
5
the protests and said that they trusted the videos posted on social media more than the
Based on their content analysis of online news programming, Brown and Mourão (2022)
find that FOX News covered the Floyd murder protests the most, followed by Associated Press
News and CNN, with MSNBC covering protests at about one-third the rate of FOX News. The
authors furthermore find that all of these sources, with the significant exception of MSNBC,
emphasized disruptions, police confrontations, and lawlessness. MSNBC also signaled these
facets of the protests, but it more frequently presented the protest’s aims and included coverage
legitimizing the protests. Despite the generally negative protest coverage on CNN, Brown and
Mourão’s (2022) survey data show that viewing CNN’s coverage was associated with a shift
toward greater protestor support and the opinion that police responses to the protests had been
too severe, while FOX News consumption was associated with a shift toward less protestor
support and the belief that police were too lenient. Viewing the more positive, legitimizing
coverage of protests on MSNBC did not change opinions about protestors or police responses to
the demonstrations.
Some pundits suggest that the protests led to increasingly negative views of BLM, but
closer inspection of the data tells a more complex story. Reny and Newman (2021) find that
politically liberal and low-prejudice individuals as well as consumers of liberal media (namely
users of MSNBC but not FOX) showed decreased favorability toward the police and increased
Van Dijk (2023) argues that an awareness of racist discourses in news coverage also
offers opportunities for creating an anti-racist discourse. Such opportunities may be greatest for
6
local news, where journalists may be able to inform with accurate facts of a local situation.
While trust in journalism and news has declined during this period, local news has suffered less
than national network or cable broadcasters in this regard—particularly among Black Americans.
In polling just prior to the death of George Floyd, studies from the Pew Research Center (Atske
et al., 2019) found that, on average, Black Americans place greater trust in local news than do
White Americans and greater trust in local broadcasters than in national ones.
Over the fifty years of its evolution, “framing” has come to indicate diverse phenomena
in communications theory, political psychology, media studies, and related disciplines. Here, we
focus on framing understood as a narrative type, including the concepts of episodic and thematic
framing studies by Iyengar (1991). What frames were used to construct the Chauvin trial verdict,
and, in particular, how were episodic and thematic frames deployed by different TV
broadcasters?
Iyengar defines the “episodic” frame as an “event-oriented report” that “depicts issues in
terms of a concrete example.” The “thematic” frame “places public issues in some more general
or abstract context” (p. 14). The former focuses on individual instances of a concern, while the
latter details the context, cause, history, or other broad background features of a story that give a
general account of conditions or outcomes. Iyengar shows that the two types of narratives
influence how news audiences attribute responsibility for events under consideration. Episodic
framing tends to lead viewers to attribute the responsibility for an outcome to the person or other
entity depicted in a concrete instance. When viewers receive thematic frames, which often
broaden the narrative by showing background contexts such as statistical trends or historical
7
trajectories, they tend instead to attribute responsibility to broader institutional causes and the
Related to the idea of framing is Bennett’s (2016) classification of four news narrative
Personalization, which invites the viewer to identify with the particularities of the people in the
story, is akin to Iyengar’s definition of episodic framing, which often features individuals
experiencing a news event. Bennett argues that personalization bias distracts viewers and shifts
attention from broader institutional contexts to personal tropes that make analysis and analytical
thinking difficult. Dramatization focuses on a narrative arc that moves through a crisis cycle
making distressing events seem inevitable. We note that the arc of dramatic narratives also
presents a linear movement from an established beginning to a crisis point into denouement and
finally definitive closure. When the story ends, the problem ends. Disorder, chaos, and crisis are
staples of dramatic news. Accordingly, the third of Bennett’s biases, the authority–disorder
frame, stresses the need for normal order and teaches that only authority can restore it.
Fragmentation bias concerns discontinuities across news stories in a broadcast and the brevity
and incoherence of a news account. Although the 24/7 phenomenon of cable news might be
expected to eliminate the problem of diffuse superficial coverage, the problem of fragmentation
bias remains. All broadcast forms, including the 24/7 news channels, are prone to fragmented
coverage given their penchant for personalized, dramatic stories depicting a narrative of chaos.
biases in the coverage analyzed. We also document instances of the “FOX effect,” a salient
aspect of research on framing differences among cable news outlets. A recent study by Dutta et
8
al. (2022) shows significant differences in the language (parsed at the sentence level) used by
FOX News, CNN, and MSNBC to describe police protection after the Floyd murder, during the
resulting protests, and during the Chauvin trial and the reading of the trial verdict. FOX News,
the study shows, stands in sharp opposition to CNN and MSNBC alike in its positive
constructions of the police within discourses covering these news events. Likewise, a Signal AI
(2020) analysis of 1,880,507 news items in the two weeks of protest following Floyd’s murder
finds that in contrast to liberal and center news outlets, conservative outlets used the terms “riot”
and “looting” and avoided the terms “racism” and “police brutality” in their protest coverage.
Reporting George Floyd’s Death, the Protests, and Derek Chauvin’s Trial
It may be surprising to learn that Darnella Frazier’s cell phone recording did not produce
a common framing of facts. Even the Hennepin County medical examiner’s June 2020 ruling that
the manner of Floyd’s death was homicide and the cause of death included “law enforcement
subdual, restraint and neck compression” failed to lead news organizations to a consensus; nor
has the jury finding Chauvin guilty of murder yielded a common characterization of Floyd’s
death.
common. Official documents allowed for diverse interpretations of Floyd's death. For instance,
the initial criminal complaint made a case for Chauvin's guilt but also mentioned Floyd's
underlying health conditions and possible intoxicants in his system as factors contributing to his
death. Some media outlets ignored the inherent danger of the officers' restraint and highlighted
these factors as alternative causes of Floyd’s death, which, in the view of medical experts,
“weaponize medical language” from a scientific report (Crawford-Roberts et al., 2020, p. 1).
9
Journalists must guide their audience through the specialized scientific language and help them
comprehend the meaning of words in this context. This involves conveying how medical experts
analyze autopsy reports and, in this case, the significance of evidence like Frazier's video and
body camera footage in reaching conclusions. The presentation of these facts exhibited
To cover Floyd’s murder and Chauvin’s trial, different news organizations engaged many
experts and frames to distinguish a wide range of facts from falsehoods. On the eve of Chauvin’s
trial, FOX News’s Tucker Carlson called Floyd’s death a “carefully concocted myth” of
“Democratic partisans'' and asserted that there was “no physical evidence that George Floyd was
murdered by a cop,” falsely claiming the autopsy revealed fentanyl as the cause of death
(Carlson, 2021). During the trial and on the eve of the verdict, CNN reporters and host Don
Lemon, among others, likened Floyd’s death and the events that followed to revelations and
actions emanating from the 1955 torture and lynching of fourteen-year-old Emmett Till in
Mississippi (Lemon, 2021). Drawing parallels between the widely circulated photographs of
Till’s mutilated body and the filmed account of Floyd’s arrest and death, journalists, as well as
Till and Floyd family members, saw the Floyd protests and calls for dramatic shifts in public
policy as a nascent “racial reckoning.” These examples of narrative framing are dissimilar—the
FOX narrative is a falsehood, and the CNN narrative is a conjecture beyond facts—but even the
arguably benign narrative of Lemon’s analogy leads viewers to an orientation to events that
exceeds the facts of reporting in its editorializing. Do viewers discern the difference between fact
and editorial?
10
Consider further what was reported as facts. FOX News did not cover Floyd’s death until
Bret Baier’s May 27 Special Report on the protests, “a day after tear gas and flash grenades” left
Minneapolis “a city on edge.” Coverage was sporadic in the weeks that followed, with an
emphasis on protests in U.S. cities, the “ambiguity” of the “Floyd killing,” and what Tucker
Carlson called a “pandemic” of “craziness” and “hysteria” that started in Minneapolis “when a
man named George Floyd died” (Baier, 2020). CNN immediately referred to other killings of
Persons of Color by police, including the 2016 shooting of Philando Castile during a traffic stop
in the Minneapolis-St. Paul metropolitan area, yet the conversation shifted to coverage of
protests and the memorialization of the site of Floyd’s death by mid-June. MSNBC also
contextualized Floyd’s death in the killings of other Persons of Color, highlighting the plea “I
can’t breathe.” Coverage of the protests as well as Floyd’s death decreased considerably on
MSNBC, with only sporadic stories that featured Floyd family members and national political
figures. Similarly, within a month, CBS and NBC primarily focused on the protests occurring in
major cities worldwide, rather than Floyd’s death, as the protests ebbed. In addition to covering
protests, ABC turned to policy responses, such as the proposed “George Floyd Justice and
Policing Act,” and officials’ views of peaceful Juneteenth celebrations. In July, ABC coverage
looked at matters of race in the U.S. military, business, and other institutions with stories on
workplace diversity and inclusion programs, as well as incidents of White supremacy in the
Local News
While national networks broadened the lens, local Minneapolis news showed longer
staying power for focusing on the arrest and Floyd’s specific case. The local-oriented coverage
11
started with detailed descriptions of Frazier’s video. CBS affiliate WCCO framed the situation as
a “deadly arrest” and offered biographical details of Floyd and his family. They provided a
detailed timeline of arrest events and covered the spontaneous memorializing of the site of the
arrest as well as the “protest,” “looting,” and “demonstrations,” which they termed “the pain that
is very real and . . . spilling out into the streets.” In the weeks that followed, WCCO scrupulously
confrontation” [with police], presented a detailed breakdown of the video, noting the bystanders
pleading with officers to attend to Floyd’s welfare, and showed the memorialization of the site.
But the station also permitted on-air interview subjects to use the terms “excessive force,” with
KSTP (ABC) led with “A Black man in handcuffs dead and four police fired,” describing
events as a “death in custody” in a detailed explanation of the video, and they then turned to the
response of local and federal authorities and the "swift and strong" community response. Such
stories, as well as those covering free Covid tests for protestors (few of whom tested positive)
and businesses rebuilding in neighborhoods subjected to violent protests, aired on all local news
stations.
KMSP, the local independent broadcaster, explained that “George Floyd died…after
officers made an arrest,” continuing, “it has created an incredibly incendiary situation in
downtown Minneapolis.” In interviews with Floyd’s family; local and national authorities; and
protestors, in addition to dissecting the video, KMSP—like KARE—challenged the idea that
Chauvin used authorized restraint techniques. Within weeks, KMSP adopted the language of
12
“police killing,” referring to the “police who killed George Floyd”; the station also noted the
Hennepin County Board of Commissioners resolution calling racism a public health crisis and
covered the State of Mississippi debate over removing the Confederate Stars and Bars from the
state’s contemporary flag. In all, local coverage of Floyd’s death and the protests offered a
picture of unauthorized police practices, a beloved but troubled man beset by a drug addiction,
peaceful demonstration as well as violent and destructive protests, and businesses and
Local stations portrayed that picture with a focus on local facts. On the evening of May
31, a KARE news reporter broadcasted helicopter feed of protests that were “spreading.” The
protests included a lot of “profanity,” explained the anchor who later paused commentary so
viewers could “listen in” to the protesters’ chants of justice for George Floyd. If a resident
switched channels to CBS, they would be greeted with an anchor’s monologue of “protests
erupting” across the nation, where Floyd’s death represented a “match in the firestorm of
inequality” in America. Different broadcasters fit facts into different editorializing frames.
Research on strong, contesting frames such as a right to peaceful assembly and the need
for secure public order suggests that each consideration is important and can be elevated by
framing to encourage not only that a given consideration prevails but also thoughtful debate in
the process of decision making (Chong & Druckman, 2007). Adding nuance to this finding,
however, is research that shows false claims in some types of political communication confuse as
much as they mislead those who are exposed to them (Allen & Stevens, 2019). We consider
research findings on these and other aspects of the frames, biases, narrative forms, and accuracy
(2) Framing Black Lives Matter and the Floyd Murder Protests
Within hours of Darnella Frazier’s posting of the video recording George Floyd’s death, a
local story became a global phenomenon. The protests offered an opportunity for thematic stories
on issues of great significance and a case constituted by dramatic episodes of local concern—
including the need for information relevant to public safety. Thus, local coverage may focus on
episodic developments, while national coverage runs with more thematic narratives. That trend is
observed in a comparison of local and national coverage of Floyd’s murder and related protests
across 231 news stories (Norquist, 2021), showing that compared to local stories, national stories
are more likely to be thematic, over five times more likely to mention other victims of police
brutality, and almost three times as likely to be coded as having a racial justice focus. Before
concluding that these differences mark the superiority of national coverage, we looked more
to assess polarity in FOX and MSNBC from May to July 2020. The average tone of the
transcripts for each month was developed and interpreted using the LIWC-22 emotional tone
analysis, a single summary variable, constructed such that “the higher the number, the more
positive the tone,” with numbers below 50 suggesting “negative emotional tone.” The average
Despite their differing ideologies, the two stations both covered the protests with a
generally negative average tone. These findings are consistent with the concept of the “protest
dismiss and derogate protests that threaten the status quo (a score of 50 or below indicates a
Here, we focus particularly on how local and national coverage differed from cable news
in framing style and information biases. Since 1991, researchers have generally presumed
thematic framing to be superior to episodic stories. Thematic frames are characterized by depth,
data, and analysis. In light of the four information biases, we question if “more” is actually
better, asking what kinds of information constitute the “depth” of thematically framed stories.
For instance, research on political advertising claims (Allen & Stevens 2019)—many of which
likewise included data and an analytical logic engaging facts—shows that false claims can be
responsibility from individuals featured in the media narrative to authorities and leaders of
15
relevant institutions. But an authority-disorder bias directs attention to what authorities are doing,
bypassing ordinary community members and their responses and responsibilities in confronting
public matters. Personalization biases further elevate private and personal responsibilities over
public and community duties and capacities. We must examine the basis of a thematic frame as it
“leftist woke conspiracy” (both tropes entered the news analysis of the Chauvin trial), the claims
Our content analysis covers framing and other characteristics of 267 news stories about
the Derek Chauvin trial broadcast on the day of the trial verdict, April 20, 2021, on ABC, CBS,
CNN, FOX News, MSNBC, KARE, KMSP, KSTP, NBC, and WCCO news programs. These
programs included major national broadcasts, branded cable news shows, news breaks, and
regular news programming on local stations. The first story in our data aired at 6 a.m., the last at
11:56 p.m. Approximately 30% of the 267 stories aired before the verdict was announced at 3:05
p.m. Following the jury’s decision, regular programming ceased in the Twin Cities Media
Market, and it was replaced with live coverage, including celebrations of the verdict and press
conferences with officials and the Floyd family. Network news continued to cover events
through the speeches and resumed regular news programming thereafter in the Eastern and
Pacific time zones. In the Central time zone of Minneapolis, however, local news anchors
preempted programming through ABC News Tonight, CBS Evening News, and NBC Nightly
News.
The story is our unit of analysis for coverage on the day of the Chauvin verdict. Expert
analysts read each transcript and identified the starting and ending points of each relevant story
in each broadcast. We selected relevant stories when they concerned any aspect of George Floyd
16
or Derek Chauvin’s biographies and behaviors as they were used to frame the actions
surrounding Floyd’s alleged crime, arrest, and death; the Minneapolis protests; the Chauvin trial;
and the Chauvin verdict. Pairs of trained coders received a news broadcast transcript covering
the events (and a video of that broadcast where possible), with enumerated selected stories which
they content analyzed using a digital form to input information for each transcript related to the
study’s variables. Expert coders reconciled differences between paired coders for final data entry.
Prior research on the accuracy of claims in political advertising (Allen and Stevens, 2019;
Allen and Stevens, 2015; Allen et al., 2016; Stevens et al., 2008; Stevens, 2009) guided our
approach to assess the accuracy of claims within news stories. We employed two key measures.
Firstly, we evaluated the depth of detail presented in the narrative, ranging from minimal to
significant depth. Secondly, we examined whether the news narrative extended beyond factual
statements to encompass statements of cause and effect. This measure ranged across: (1) a simple
(3) plausible sequences depending on imputed motives; (4) implausible narratives with no factual
basis. To assess accuracy, we used the truth score, validated in the analysis of political
advertising claims (Allen & Stevens 2019). This measure approach allowed us to categorize the
story as presenting gross untruths, major distortions, omissions, and misleading representations
of facts; mostly accurate facts with some logical inferences; or entirely accurate facts. We also
adapted Iyengar’s framing types, rating stories on a five-point scale from entirely or mostly
episodic to mostly or entirely thematic narrative forms, with a midpoint where the narrative was
framing, tone, bias, and accuracy in reporting the Chauvin trial verdict. These metrics address
17
three key questions: First, do cable stations use more thematic framing than network and local
broadcasters in covering the Chauvin trial verdict, similar to their approach to covering the
protests? Second, if national/cable stations used thematic framing, on what themes did they
focus? Third, we explore the implications of these findings for when individuals or stations more
distant from the events engage in thematic framing compared to local reporters. This aspect
sheds light on how the public perceives events like Floyd's murder and Chauvin's trial, which
carry profound implications for the United States' historical reckoning with its contemporary
events.
Another aspect of framing is questions of who controls the narrative. Bennett’s focus on
authority and institutions gets at this point. Using LIWC-22, we asked simply who gets to speak.
Of cases where a commentator other than a host or anchor is part of the story, 170 (71%)
featured national authority figures, 11 (5%) local authority figures, and 58 (24%) were victims of
police brutality. What these figures suggest is that national authorities told the story and leaders
and people on the scene were much less likely to be part of the narrative.
The level of detail, framing (episodic or thematic), and accuracy of the stories underpin
the causal basis of narrative claims. Examining news coverage on the day of the Derek Chauvin
trial verdict, over 80% of stories provided minimal to moderate detail, with only around 16%
stories offering significant detail. About 75% of the coverage was mostly or entirely episodic,
while only one in eight stories was mostly or entirely thematic. The majority of stories were
mostly accurate, but roughly 25% were largely inaccurate, and only 15% were wholly accurate.
18
For the tone of coverage, we focus on a further three variables: tone of frame, content,
and portrayal of George Floyd. (We also coded the tone of the portrayal of Derek Chauvin, but
there was too little variation in tone for comparisons among stations, particularly after the
verdict, when most coverage was negative.) The tone of frame and tone of content were evenly
distributed among positive, negative, and balanced/neutral stories (which we combine into a
middle category for purposes of analysis). While tone of frame and tone of content are correlated
(Stevens et al. 2006), 25% of stories differed on these two measures. On the day of the verdict,
the tone of George Floyd's portrayal was mostly positive or neutral/balanced, with few negative
portrayals.
independent local broadcaster, KMSP. We combine national and local stations in this analysis
because local affiliates took over much of the live coverage after the verdict, blurring the
We estimated models in which we predicted the different indicators of content and tone
as a function of broadcaster and time, using NBC/KARE (boasting the largest Minnesota
audience) as the reference variable. Content and tone could be affected by when the coverage
aired. We include a count variable in minutes from 0, when the first story in our data was aired at
6 a.m., to 1076, when the last story aired at 11:56 p.m., as a control to account for this possibility.
Figure 1 shows coefficient plots for the stations for each indicator of content with the marker
representing the coefficient and the horizontal error bars the 95% standard error intervals around
each estimate (NBC/KARE is at zero, the red vertical line in the plots). Overlapping standard
19
errors do not preclude statistically significant differences between variables (Krzywinski &
The first two charts in Figure 1 indicate slightly more detail and thematic framing on the
24-hour cable news stations, likely due to their much larger newsholes and resources. FOX and
CNN provided more detail than NBC/KARE, and KMSP than ABC/KSTP (at p<.10). Coverage
on CNN and MSNBC was more thematic than any network coverage; FOX’s framing on this
variable differed only from CBS/WCCO (at p<.10). While more detail and thematic framing are
often seen as desirable attributes in media coverage, when we assess accuracy, cable news fell
short in providing solid factual evidence for its narratives compared to national and local news.
FOX, in particular, was significantly less likely to offer accurate evidence compared to all other
networks. CNN and MSNBC were also less accurate than all networks except CBS/WCCO,
although the differences were less pronounced. To illustrate, we used the model estimates to
predict a story’s truth score knowing only the broadcaster, with time at its mean for stories in the
Figure 2 shows the analysis of tone. FOX was a clear outlier again in its relative
negativity. But the cable stations in general tended to adopt a more negative tone than the
networks, though not in their portrayal of George Floyd, where FOX was on its own. For the
tone of frame or content, the estimates imply that, all else being equal, for every three stories,
two would be negative in tone and one would be neutral or balanced. For MSNBC or CNN, by
contrast, for every three stories, on average the mix would be closer to one negative, one neutral
or balanced, and one positive, while for the networks, it would be two neutral or balanced and
one positive (or two positive and one negative). In sum, there was much more negative content in
Figure 1: Coefficient Plots for Content of Coverage on the Day of the Chauvin Verdict
Level of Detail (ordered probit coefficients) Episodic vs. Thematic (regression coefficients)
Figure 2: Coefficient Plots for Tone of Coverage on the Day of the Chauvin Verdict
Differences in the portrayal of George Floyd were generally less pronounced between
cable and network stations, except for FOX, which displayed stark differences. The estimates in
Figure 2 are derived from a selection model, where we first predict the likelihood of mentioning
Floyd in stories related to the verdict (just over half of the stories in our data mentioned George
Floyd). Then we predict the tone of the portrayal. FOX was notably less likely than most stations
(with exceptions being MSNBC and ABC/KSTP) to mention Floyd in its coverage. Among
stories that did mention him, FOX was far less likely than all other stations to portray Floyd
positively. In Wald tests, the differences between FOX and other stations were statistically
significant at p<.05, while the differences between CNN and MSNBC and the networks were not
statistically significant. The coefficients in Figure 2 equate to a roughly one in two chance of a
negative portrayal of George Floyd on FOX, whereas this probability was less than 1 in 20 on all
other stations. There was an exceedingly low probability that FOX portrayed Floyd positively,
while about three in ten stories did so on other stations, with the remaining stories being neutral
or balanced.
Our data from the day of the verdict also enable us to examine how coverage evolved in
content or tone throughout the day. We reestimated the six models, allowing for interactions
between time (using our count variable) and each station. A statistically significant interaction
implies that coverage became systematically more or less detailed, episodic, negative, and so on
as the day progressed. These estimates (available upon request) show that, for most stations,
there was no change in the characteristics of coverage as the day advanced. The exception again
was FOX, whose coverage became less detailed, less accurate, and more negative over time.
Figure 3 shows plots for level of detail, accuracy and tone of content over the course of
the day for FOX, using CNN as an example of a rival cable station and ABC/KSTP as an
23
Figure 3: Coverage of the Chauvin Verdict over the Course of the day
0.3
Probability majority/wholly
0.8
0.25 0.7
6am 10am 2pm 6pm 10pm 6am 10am 2pm 6pm 10pm
Tone of content
Tone (1=negative, 3=positive)
2.5
Fox
CNN
ABC/KSTP
1.5
example of a network station. The graphs show FOX's initial higher level of detail diminished as
the day progressed, and its accuracy, already lower than other stations, dropped by nearly 50%.
Notably, CNN also saw a decline in accuracy over time, indicating a potential factual looseness
on cable television. However, CNN's decline started from a much higher baseline than FOX.
throughout the day. There were also differences in tone. Most stations adopted positive or neutral
framing for the guilty verdict of Derek Chauvin, considering it a fair application of the rule of
law—a relief, as polls showed the public had not been confident a guilty verdict would be
returned. Opinion polls indicated that most Americans agreed with the verdict. For FOX, the
verdict was associated with a slight decrease in the positivity of its content framing.
The pre-trial verdict was framed in four distinct ways: 1) as a foregone conclusion due to
the threat of riots if the "wrong" verdict was reached (FOX News); 2) as a racial reckoning
(MSNBC, ABC, CNN); 3) as a dramatic and astonishing change in justice administration (CBS,
NBC); 4) as a moment for relief, solemn reflection, and muted celebration (local stations).
Leading up to the verdict, there was a prevailing sense of suspense across all broadcasters. An
exchange between Gayle King, a Black female co-host on CBS, and Jamie Yuccas, a White
morning of the verdict, King openly expressed her fear, as she had been warned by the National
Guard about the particular risk she faced as a Black woman. She also highlighted the area's
Minnesotans as resilient and able to deal with protests and police reform like they tackled
dangerous winter weather: “It’s just another time the city will have to get through.” Her response
25
scarcely acknowledged King's fears. These contrasts were echoed by respondents in our focus
group.
We conducted nine focus groups to gather responses from local residents regarding our
broadcast coverage. Two groups were comprised of Persons of Color and White respondents all
living within 3 miles of the site of Floyd’s murder and the protests. Two groups were entirely of
Persons of Color who lived within blocks of Floyd’s murder and the protests. Four groups were a
mix of Persons of Color and White participants in a rural location 45 miles from Minneapolis.
Three participants in the all-Persons of Color group asked to continue their conversation in what
constituted a ninth focus group, with an African American female also participating in an in-
were particularly frustrated because they were witnesses to unfolding events that they did not
think were being portrayed accurately: one participant captured this sentiment when saying, “I
can see with my own eyes. I have context because I have lived through it.” It was striking
nevertheless that participants consumed a range of different media, both traditional and online.
Participants objected to the use of shocking imagery and what they regarded as “action scene”
framing because they argued that it distracted from the larger systemic issues surrounding
Floyd’s death. The groups consisting of exclusively Persons of Color, as opposed to the mixed-
White and BIPOC groups, described the coverage of the murder as gratuitous and reflective of a
“white gaze.” They objected to what they saw as the conflation of violent riots and arson with the
entirety of the Black Lives Matter protest, and to being blamed for “burning everything down.”
These groups further argued that broadcasters were profiting from Black suffering and said that
as viewers they were numbed and traumatized by the repeated showing of the murder. A Black
26
participant in the rural location 45 miles from Minneapolis also described themself as
traumatized. Native American participants noted that despite having a community close to where
the murder took place, they were completely ignored by the media.
With respect to the Chauvin trial, Black and Native American participants, in particular,
strongly criticized the tendency to frame the trial in the context of an impending protest crisis.
They felt that this was especially true of Fox, which they saw as misrepresenting the words of the
Minneapolis police chief Medaria Arradondo. They also observed that cable channels rarely
mentioned the initial protest's cause, while local television consistently highlighted Floyd's death
and its impact on the community. By the day of the verdict, they felt that Floyd had been largely
forgotten in national and cable broadcasts. When shown a comparison of national, cable, and
local news recaps of the verdict, participants in one BIPOC group believed that in being
incomplete, the coverage was also inaccurate. It neatly wrapped up the narrative without
addressing the core issue in their view: racism. They were pleased with the verdict but also
Some of the participants felt that they had noticed an increase in Persons of Color
reporting on the news after the murder. There is some evidence of an increase among news
editors but little evidence of an increase in the newsroom (Gottfried et al. 2022). Participants’
observations reflect opinions about coverage of Black people on the news in general: it employs
harmful stereotypes and misses important information (PEW Research Center 2023). For
participants who were Persons of Color, this kind of representation made them more likely to
Conclusion
The need for context introduces the challenge of framing, choice of experts, and, in
general, the representations of facts. “George Floyd died,” while a factual statement, connotes
“during an arrest,” “at the hands of police,” or by “officer Derek Chauvin.” Contesting frames
likewise characterized events of the days that followed as the “demonstrations,” “protests,” and
We examined media coverage surrounding George Floyd's death, the subsequent protests
triggered by Darnella Frazier's video, and the day of Derek Chauvin's guilty verdict for Floyd's
murder. We compared cable news coverage and media coverage on local and national broadcast
stations at these three timepoints, using qualitative and quantitative methods. Broadcast media
differed, with local news emerging as those offering more accurate coverage.
Our findings question the assumed superiority of thematic framing. Local news coverage
was less detailed and more episodic but offered relevant facts; cable news offered thematic
framing, but details did not signal greater accuracy. Before concluding that episodic framing
lacks essential information, we should assess the kind of information that is genuinely valuable
and pertinent, and distinguish opinions that may exceed the bounds of relevance and accuracy.
Our focus group participants noted few policy changes reflecting a “racial reckoning”; a result,
they felt, of a dramatizing, personalizing news narrative and an unremitting focus on the “chaos”
of protests, instead of the protests’ cause: the police murder of an unarmed Black man.
28
Reference
Allen, B., Lawrence, E., Stevens, D., & Sullivan, J. (2016). Partisanship and perceptions of
fairness: Ignoring the facts. Journal of Experimental Political Science, 3(1), 32–43. Link
Allen, B, & Stevens, D. (2015). What is negative about negative ads? In New perspectives on
negative campaigning: Why attack politics matters, edited by Alessandro Nai and
Allen, B, & Stevens, D. (2019). Truth in advertising? Lies in political advertising and how they
Allen, B., Stevens, D., Fox-Arnold, D., Holtey, B., Vincent, M., & Woollen, B. (2023). “Local
News in a Social Capital Capital Sixteen Years Later.” Paper presented at the 2023
Annual Conference of the Elections, Public Opinion & Political Parties (EPOP)
Atske, S., Barthel, M., Stocking, G., & Tamir, C. (2019). 7 facts about Black Americans and the
Banks, C. (2018). Disciplining Black activism: Post-Racial rhetoric, public memory, and
decorum in news media framing of the Black Lives Matter movement. Continuum, 32(6),
709-720. Link
Bennett, W. L. (2016). News: The politics of illusion, 10th edition. University of Chicago Press.
Bonilla-Silva, E. (2012). The invisible weight of Whiteness: The racial grammar of everyday life
Baier B. (2020, May 27). “Special Report,” FOX News. Transcript provided by News Data
Service.
29
Brown, D. K., & Mourão, R. R. (2019). Media effects and marginalized ideas: Relationships
among media consumption and support for Black Lives Matter. International Journal of
Brown, D. K, & Mourão, R. R. (2021). Protest coverage matters: How media framing and visual
communication affects support for Black civil rights protests. Mass Communication and
Brown, D. K., & Mourão, R. R. (2022). No reckoning for the right: How political ideology, protest
tolerance, and news consumption affect support Black Lives Matter protests. Political
Carlson, T. (2021, February 10). “Tucker Carlson Tonight,” FOX News. Transcript Novak
Archive. Link
racial incident: An intensive analysis. The Communication Review, 4(2), 253-278, Link
Chong, D., & Druckman, J. N. (2007). Framing public opinion in competitive democracies.
City of Minneapolis. (2020, May 25). Man dies after medical incident during police interaction.
Link
Crawford-Roberts, A., Shadravan, S., Tsai, J., Barceló, N. E., Gips, A., Mensah, M., Roxas, N.,
Kung, A., Darby, A., Misa, N., Morton, I., & Shen, A. (2020, June 6). George Floyd’s
Dardis, F. E. (2006). Marginalization devices in U.S. press coverage of Iraq War protest: A
Dixon, T. L. (2007). Black criminals and White officers: The effects of racially misrepresenting
law breakers and law defenders on television news. Media Psychology, 10(2), 270-291.
Link
Dixon, T. L. (2008). Network news and racial beliefs: Exploring the connection between national
Dixon, T. L., Azocar, C. L., & Casas, M. (2003). The portrayal of race and crime on television
network news. Journal of Broadcasting & Electronic Media, 47(4), 498-523. Link
Dixon, T. L. & Linz, D. (2000). Race and the misrepresentation of victimization on local
Dixon, T. L., & Williams, C. (2015). The changing misrepresentation of race and crime on
Doehler, S. (2023). Taking the star-spangled knee: The media framing of Colin Kaepernick. Sport
Dutta, S., Li, B., Nagin, D. S., & KhudaBukhsh, A. R. (2022). A murder and protests, the Capitol
riot, and the Chauvin trial: Estimating disparate news media stance, Proceedings of
Entman, R. M., & Rojecki, A. (2001). The Black image in the White mind: Media and race in
Fridkin, K., Wintersieck, A., Courey, J., & Thompson, J. (2017). Race and police brutality: The
Link
31
Gilliam, F. D., Jr., & Iyengar, S. (2000). Prime suspects: The influence of local television news on
the viewing public. American Journal of Political Science, 44(3), 560-573. Link
Gottfried, J., Mitchell, A., Jurkowitz, M., & Liedke, J. (2022). Journalists give industry mixed
reviews on newsroom diversity, lowest marks in racial and ethnic diversity, Pew Research
Iyengar, S. (1991). Is anyone responsible?: How television frames political issues. University of
Chicago Press.
Krzywinski, M., & Altman, N. (2013). Error bars. Nature Methods, 10, 921-922. Lemon, Don.
(2021, April 19) “Don Lemon Tonight,” CNN. CNN Transcripts. Link.
Mills, C. E. (2017). Framing Ferguson: Fox News and the construction of US racism. Race &
Norquist, L. (2021). A Matter of Perspective: Exploring Differences in Local and National News
Coverage of the George Floyd Murder and Subsequent Protests. Thesis Accepted for
Pew Center for the People and the Press. (2023). Local News Fact Sheet. Link
Pew Research Center. (2023). Black Americans’ Experience with the News. Link
Powell, A. (2022). Two-step flow and protesters: Understanding what influenced participation in
Ramasubramanian, S., & Martinez, A. R. (2017). News framing of Obama, racialized scrutiny,
Reny, T. T., & Newman, B. J. (2021). The opinion-mobilizing effect of social protest against
police violence: Evidence from the 2020 George Floyd protests. American Political
Signal AI. (2020). Media bias in the coverage of George Floyd. Link
Stevens, D., Alger, D., Allen, B., & Sullivan, J. L. (2006). Local news coverage in a social capital
Stevens, D., Sullivan, J., Allen, B., & Alger, D. (2008). What’s good for the goose is bad for the
Van Dijk, T. A. (2023). (Anti)racist discourse. In Paul Gee & Michael Handford Eds., Handbook