Tow Center audience study: Reader perspectives on local partisan news sites
More information and background on partisan news networks and how they function can be found here .
Executive Summary and Key Findings
The Tow Center for Digital Journalism started researching the phenomenon of political actors co-opting the appearance of local news outlets in 2019. Building on landmark reporting by the Lansing State Journal in October of that year, Tow Center research has shone a light on the emergence , expansion , and operating practices of Metric Media, a vast network of more than 1,200 websites that purports to fill the “growing void in local and community news after years of steady disinvestment in local reporting by legacy media.”
Previous Tow Center research has examined the network’s “creation of partisan outlets masquerading as local news organizations.” The New York Times said the “operation is rooted in deception, eschewing hallmarks of news reporting like fairness and transparency,” while heavy reliance on low-cost, automated story generation has resulted in the sites being labeled “ pink slime ” journalism. NewsGuard summarizes Metric Media as “a network of websites that falsely present themselves as locally based news sites. The sites do not disclose their conservative agenda, and much of the content is created by algorithms.” It advises readers to “proceed with caution,” as the network “severely violates basic journalistic standards.”
Such coverage often alludes to news consumers’ experiences with these sites. For example, the Lansing State Journal story that first uncovered the Metric Media network said the sites “present a challenge for readers navigating a digital media environment that has unlimited space for publishing stories that are hard to distinguish as journalism, advocacy or political messaging.”
To date, however, little attention has been paid to understanding how local news audiences negotiate and interpret these sites.
In an effort to address this, we conducted a six-day audience study with 90 participants based in 35 US states and the District of Columbia. Participants were assigned a local news website operating on Metric Media’s technology stack and asked to complete a three-part exercise:
an initial survey about their local news consumption habits and needs, and an assessment of their first impressions of their assigned website;
a daily diary exercise in which they detailed their experience of using their assigned website and other local news outlets over five consecutive days; and
a final reflection about their assigned website, addressing issues such as how, if at all, they improved their understanding of local issues, and how trustworthy they perceived the sites to be.
Sign up for CJR's
Finally, participants were asked to seek information about who owned and operated their assigned site and reflect on the process of establishing provenance.
A sizable minority of participants—more than a third—formed positive initial impressions of their assigned sites, highlighting the operators’ success at mimicking the appearance of traditional news websites. These positive impressions were typically based on superficial aspects such as design (“clean,” “professional”), lack of paywalls, mobile responsiveness, or similarities to established news sites. However, other participants immediately highlighted aspects they found concerning, such as a lack of bylines on articles, or obvious giveaways that sites lacked basic familiarity with the areas they purported to cover (e.g. outlet names with uncommon references to local geography).
While a majority of participants had formed negative impressions of their sites by the end of the study, it was notable that their most commonly cited drawbacks resulted from repeat visits, such as frustration at the lack of updates to the homepage, the predominance of uncontextualized automated stories, and a failure to cover newsworthy local stories or events.
Despite numerous misgivings, a majority of participants reported at the end of the study period that political information had been valuably covered by their assigned sites.
Perceptions of trustworthiness and bias
Despite the overwhelmingly negative final impressions of the websites, perceptions around the sites’ trustworthiness and bias were much more mixed. While a majority of participants rated the outlets as untrustworthy—though by a much smaller margin than with overall assessments on quality—a very narrow majority rated coverage as fair and balanced.
By the end of the study, roughly a third of participants claimed there was conservative bias in their assigned outlet’s editorial coverage, and a smaller number suspected that a partisan organization was responsible. These perceptions of bias grew over the course of the study; only approximately a dozen participants observed it at the beginning.
The network’s heavy reliance on low-cost automated articles drawing upon public datasets—quintessential examples of so-called pink slime journalism—produced a surprising range of opinions in relation to audience perceptions of trustworthiness and bias. (Participants were not explicitly prompted to view or comment on these articles, but almost all did, such is their prominence across the network’s sites.) At one end of the spectrum, some participants opined that the reliance on automation and/or the lack of human bylines created a sense of “fakeness” or “robot” content. At the other end, however, some participants pointed to these articles as evidence that their sites appeared trustworthy, fair, and/or ideologically neutral. These participants typically interpreted the extensive use of quantitative data to mean output was inherently trustworthy and/or free of bias. The high volume of automated stories was also seen by some as evidence their site was well resourced and/or producing particularly hyperlocal and/or comprehensive coverage.
Some participants took surprising—and potentially concerning—cues when forming opinions about integrity, uncritically accepting a site’s own claims about impartiality at face value, or immediately interpreting nonprofit status as evidence of neutrality and/or virtuous intentions.
Attitudes toward (a lack of) transparency
Despite the growing emphasis on media literacy in recent years, basic skills and knowledge, such as the importance of identifying who owns and operates a given news source to ascertain trust, were found to be lacking in some quarters. A sizable majority of participants—including many with misgivings about their assigned sites—did not attempt to seek information about ownership until prompted. Some admitted it did not—and typically would not—cross their mind to do so when encountering an unfamiliar news source.
Of those who researched ownership of their assigned sites when prompted, a significant portion were unable to find any information. Most were only able to find a name (e.g. Metric Media or LGIS) but no other information about the group, pointing to a significant lack of transparency around the networks’ ownership and operations. On discovering limited, vague, or missing ownership information (e.g. About pages for at least four sites from the Record network were broken at the time of the study), participants’ attitudes varied considerably, ranging from indifference to mild annoyance to severe alarm. Some expressed no strong opinion or gave no indication that this lack of information affected their view of the site’s integrity.
Among participants who said they found their assigned sites to be trustworthy, fair, and without bias, none indicated that their subsequent findings about the lack of transparency around ownership affected their assessment.
Notably, the minority of participants who had researched ownership of their assigned sites of their own volition tended to articulate the strongest feelings, describing the lack of transparency using terms such as “shocking,” “unsettling,” “odd,” and “worrying.”
For more than a decade, researchers and journalists have documented the economic crisis in the local news industry . In recent years, entire fields of research and new methodologies have emerged to track and quantify the effects of this crisis on newsrooms and local communities . Concurrently, efforts have been made to explore and identify potential solutions and develop promising case studies . Some of the most high-profile examples have included the founding of new nonprofit media organizations by wealthy benefactors and philanthropic groups . But in the past few years, a new crop of organizations has been accused of co-opting the rhetoric of these initiatives, and exploiting higher levels of trust in local news, to promote partisan causes.
While groups on both sides of the political aisle have engaged in related practices, the most sprawling networks, which are the subject of this study, have ties to conservative organizations and causes, according to reports . Building on the initial reporting by the Lansing State Journal in October 2019, the Tow Center for Digital Journalism’s research has tracked the nebulous connections among networks such as Metric Media, Locality Labs, Franklin Archer, the Record Inc., and Local Government Information Services (LGIS), as well as their dramatic growth in the lead-up to the 2020 presidential election. In December 2018, Tow research identified 450 sites across these networks, 189 of which had only been set up in the previous year by Metric Media. By August 2020, that total number of outlets had increased threefold to more than 1,200 sites. In September 2020, Metric Media owner Brian Timpone told the Deseret News that he planned to establish another 15,000 websites within the next year.
With generic names and simple layouts, the Metric Media sites mimic the appearance of traditional local news outlets. Boilerplate “About us” pages say the sites “provide objective, data-driven information without political bias.” But Tow’s most recent research has revealed Metric Media’s undisclosed financial ties to conservative groups and donors, and the ways in which it uses an online tool, Community Newsmaker, to allow its clients to pitch stories with a political agenda. Similarly, a 2020 New York Times investigation based on internal emails and interviews with employees of the outlets found “many of the stories are directed by political groups and corporate P.R. firms to promote a Republican candidate or a company, or to smear their rivals.” While the sites “generally do not post information that is outright false,” the Times reported, “the operation is rooted in deception, eschewing hallmarks of news reporting like fairness and transparency.” According to the Deseret News , Timpone disputes these claims and said that his websites had been “misrepresented” and “have no political leaning.”
While the extent of these “pay-for play” networks and the lack of transparency around their political motivations and funding has been well documented, less attention has been paid to the ways in which news consumers interpret these sites. The majority of content published by the best-documented networks are algorithmically generated articles based on publicly available data and slotted into identical design templates. This low-cost process has been dubbed “pink slime” journalism, a 21st-century take on the sensationalized “yellow journalism” of the 19th century.
The websites also feature a smaller number of stories that appear with bylines. Research by Asa Royal and Philip M. Napoli found that over the course of a 78-day observation period, “nearly two-thirds of [Metric Media] outlets did not publish a single human-written article … 97% of Metric Media’s content is autogenerated.” After running comparisons between the articles in various strains of uniform story templates, the researchers found that “most auto-generated stories within a strain varied by at most a few words.”
Automated content can “co-opt the language, design and structure of news organizations … [and] can be used to build credibility,” Tow researcher Priyanjana Bengani noted in one report . “It can also make a news organization look far more prolific than it is.” Metric Media’s sites tend to have a minimal social media presence, and the size of their readership is unknown. But some of the earlier created sites perform well in Google Search results, appearing just below government pages, “potentially adding to [their] credibility,” according to Bengani. The utility of the sites to their backers, she wrote, may be in “focusing attention on issues such as voter fraud and energy pricing, providing the appearance of neutrality for partisan issues, or to gather data from users that can then be used for political targeting.” Davey Alba and Jack Nicas, in their investigation for the Times, noted that the sites may have a “modest reach in the national conversation. But with the focus on small towns, less readership is needed to make an impact.” In a Poynter article about philanthropic support for nonprofit local news organizations, Jennifer Preston, the former head of journalism funding for the Knight Foundation, said that she was “deeply concerned about websites masquerading as local news. … They have an impact on highly uninformed [but] highly engaged [news consumers].”
Many journalism experts and critics have warned that the sites’ lack of transparency and skirting of basic journalism conventions like mastheads and bylines make it challenging for readers to fairly assess them or to distinguish if what they are reading is “journalism, advocacy or political messaging,” according to the Lansing State Journal. “The question is not about bias—it’s about journalistic standards,” Josh Pasek, a communications professor at the University of Michigan, told the student newspaper, the Michigan Daily , in an article about the networks. “It’s okay to have outlets that have varying different views out there, but there’s a certain point at which [that] … oversteps how journalism is supposed to operate. And once that occurs, now there becomes a substantive question as to whether what you’re observing is in fact news, or is instead a disinformation campaign.” In an article for The Atlantic about the disinformation campaigns surrounding the 2020 election, McKay Coppins agreed, writing that “readers are given no indication that these sites have political agendas—which is precisely what makes them valuable.”
These characterizations of the sites’ political intentions, utility, and obfuscation have thus far gone untested with readers. How do news consumers navigate and respond to these problematic local news websites? Considering the “void in community news after years of decline in local reporting by legacy media,” as Metric Media writes in the sites’ About pages, do local news audiences feel these sites meet their information needs and deepen their understanding of local issues?
Existing research has documented the connections between some of these sites and conservative groups and donors, but do news audiences infer any bias in the reporting—and how do they formulate perceptions of trustworthiness? Do they suspect any political motivations behind the sites’ ownership—and what views, if any, do they have about their levels of transparency?
These are among the questions explored in this study. The findings are based on the responses of 90 participants assigned to assess one of 73 sites from, or with links to, Metric Media and associated networks over the course of six days. As the first study to incorporate the public’s perception of these sites, this research aims to further the emerging scholarship around this still poorly understood localized phenomenon.
This audience research study used dscout , a qualitative research platform through which “scouts” (participants) complete researcher-assigned “missions.” Over the course of six days in June 2021, participants recorded their local news habits, daily reviews of an assigned local news website, and final reflections on their experiences using their assigned sites. Participants were told at the outset of the study that the purpose of the project was to “deepen our understanding of US news audiences’ attitudes towards emerging local news websites,” but were not informed that the websites had been accused of being partisan and/or inauthentic. At the conclusion of the study, participants were given a disclosure form that explained Tow’s previous research on these networks, the real purpose of the project, and the necessity for deception in order to gain their unbiased assessments. They were given the option to withdraw their participation at that point; none did.
How the study was conducted
After an initial screening process, participants were selected based on their interest in local news and with the aim of having the final group be as demographically and geographically diverse as possible. As qualitative research, the sample is not representative of any broader populations, and we make no claims to generalizability. A total of 106 participants were selected, of whom 90 completed the study. They were drawn from 35 states plus the District of Columbia.
Each participant was assigned a site from Tow’s database of outlets using Metric Media’s technology stack. Prior to commencement of the study, a significant challenge to the daily diary design was discovered: the publication frequency of each site had declined dramatically between the 2020 election and the inauguration of President Biden. The pages showing in RSS feeds for each site were nowhere to be found on the corresponding homepages. Using both manual and computational methods, any outlet that had not published new content within one week of the start of the study was removed from the list. During this process it was discovered that outlets from several of the networks—Locality Labs, Florida New Network, and American Catholic Tribune Media Network—had gone dormant and in some cases no longer had active websites.
Using Google Maps to determine the outlet whose stated geographic coverage area was closest to their location, each participant was assigned a site from the list of remaining outlets. Tow researchers identified sites in Texas and Illinois that appeared to publish content to their homepages more frequently than the majority of other sites. To compensate for the anticipated lack of fresh content, a disproportionate number of participants in those two states (roughly a third of all participants) were selected and assigned four sites that received regular updates. In total, 73 sites were assigned.
Participants were then instructed to complete a three-part “mission” over the course of six days: (1) an initial survey about their local news consumption habits and needs, as well as their first impressions of their assigned website; (2) a daily diary exercise in which they discussed their engagement with their assigned website and other local news outlets over five consecutive days; and (3) a final reflection about their assigned websites, addressing issues such as how, if at all, they improved their understanding of local issues; and how trustworthy they perceived the sites to be. Finally, participants were asked to seek information about who owned and operated their assigned site.
At the conclusion of the study, participants received a disclosure form about the intended purpose of the project.
In this section, we present findings drawn from our participants’ interpretations of their assigned website. This discussion is broken down into three parts. We begin with an overview of participants’ general observations about these sites, paying particular attention to their all-important first impressions. This is followed by a discussion of how participants formulated assessments of trustworthiness and bias. Finally, we delve into perceptions of transparency.
4a. General Observations
None of the participants had encountered their assigned website before the study.
By the end of the study, more than two-thirds of them had formed a negative impression of their assigned website. The remaining participants were divided almost evenly between those who had a positive impression of their site and those whose opinion was mixed. Of those who had recorded a positive impression at the beginning of the study, half still had a positive impression at its conclusion. The majority of participants said their assigned website compared unfavorably to other local news sites they already use.
The most common reason for negative reviews was the lack of updated, relevant content. Some readers cited misgivings about perceived political bias and the lack of transparency around ownership. These factors are discussed in more detail below, but the overarching conclusion from the vast majority of the participants was that, despite the need and opportunity for news sources to fill increasing local news vacuums, the sites did not succeed in deepening their understanding of their local areas. In the words of Bob, a 37-year-old from Seattle (participants’ names have been changed), “There was no new content published throughout the entirety of the mission, and even looking back through older articles there were no fresh perspectives or unique insights that I was able to gather from my assigned site. With the mix of local news content I already consume there are certainly opportunities that could be filled. But this site did not do anything to address those.”
More than a third of participants initially formed a positive or somewhat positive opinion of their assigned website. While this proportion had decreased considerably by the end of the study, these initial responses are noteworthy. Crucially, too, they contextualize the extent to which the broadly negative reactions detailed later in this report resulted from repeat visits. In many cases, participants formed first impressions based on aspects such as design or the volume of published content—the “feel and look” of their website, as one participant put it. These responses, while often based on surface-level factors, offer insights into how readers might assess these unfamiliar sites if they were stumbled upon organically via social media or a search engine result.
Many of these positive first impressions were in response to the sites’ design and layout, which most participants agreed was intuitive and visually appealing. For several participants, the sites’ layout was similar to local news sources they already used, “which is nice,” said Caroline, a 27-year-old from Chicago, “because a standardized website is easy to follow.” For others, their design was actually superior. “Most local sites are very frustrating to navigate between ads and paywalls,” said Jane, a 28-year-old from Houston, Texas. “This was so easy to use.” Suman, a 33-year-old in Gainesville, Florida, thought his website, the NC Florida News , “definitely looks a lot more sophisticated and elegant than some of the other local news websites that I’ve seen.”
Beyond providing a good user experience, these design elements more importantly connotated a degree of legitimacy for previously unfamiliar sites. Miranda, a 27-year-old in Harrisonburg, Virginia, who had praised the Shenandoah Valley News for being mobile friendly and responsive, said, “At first glance, it kind of looks legitimate.”
Besides website design, the elements that first stood out to participants were the same ones that would echo throughout the rest of the study and are explored in more depth later in this report: the appearance of outdated content and the abundance of articles based on data sets. These automated articles, which slot information scraped from a wide range of publicly available data sets into identical design templates, are central hallmarks of what has been dubbed “pink slime” journalism. For many participants, this unusual content struck them as “weird,” “generic,” or “irrelevant”—a sentiment that grew over the course of the study.
For other participants, however, the sites’ vast output gave the initial impression of being well resourced. The NC Florida News “comes across as better resourced and organized” than other local news websites, Suman said. Besides being abundant, the novelty of the data articles, many of which are divided by zip code, initially suggested a unique, hyperlocal, and in-depth level of coverage that can be lacking in under-resourced local news outlets. Ludvig, a 30-year-old in Irvine, California, said the Central OC Times “goes more in depth than other sites. … It is more detail-oriented.” John, a 43-year-old in Whiting, Indiana, was also impressed by these data-heavy articles, saying that The Region News “has stories that many of the major newspapers don’t have. It also has news that is very local to me.”
In many cases, the – to quote one participant – “bizarre” nature of these articles, as well as their lack of adherence to journalistic norms like bylines, immediately raised alarms about the trustworthiness of the websites. “It really concerns me that there is no information about who is writing these articles,” said Alice, a 37-year-old in Houston, of her website. “I have no reason to believe this is a legitimate source of unbiased news.” Karen, a 36-year-old in Plano, Texas, said in her first impressions she was “confused by the content” on her website, Dallas City Wire . “It doesn’t seem trustworthy. I don’t understand where the news is coming from. Under ‘business’ I would expect to see actual business news, not boring irrelevant info about licenses expiring. … This does not look like a legit news site.”
Cora, a 25-year-old in Traverse City, Michigan, also noted on the first day of the study that “everything about this website screams extremely fishy and sketchy to me.” In her final reflections, she wrote that “a dead giveaway that this is propaganda in the name of the news outlet [North Michigan News]. No one in Michigan refers to this region as North Michigan. Only northern Michigan or Up North Michigan.” Moreover, she wrote, “everything seems to be slanted towards hating our governor [Democrat Gretchen Whitmer].” Cora was one of roughly a dozen participants who immediately observed a conservative bent in the assigned websites’ coverage, a trend that would grow over the course of the study.
But for some participants, a central aspect of the automated articles—that they reproduce public data sets without any context or analysis—created a perception of ideological neutrality and lack of editorial bias. “It’s just the facts, isn’t it?” said Hannah, a 45-year-old in Buffalo, New York. She found the Buffalo Ledger to be “boring” but trustworthy: “It’s neutral … [and] unbiased and lacking opinions.” Jason, a 59-year-old in Chicago, said, “The website definitely seems fair and trustworthy” of the Cook County Record . “It seems to report the facts without opinion, and I don’t think it leans one way or the other politically.” Tom, a 54-year-old in Boone, Iowa, agreed that Ames Today “present[s] a fairly even-handed approach, so I think they do seem fair and trustworthy.”
The following sections delve into these issues and the participants’ other perceptions of their sites as they became more familiar with their content—an experience that, in many cases, led to changes in their initial assessments. But the immediate impact of surface-level factors like ease of navigation and a plethora of data-heavy content may prove to be some of the most revealing insights into how news consumers may interpret these sites.
Lack of new content
The most common response from participants throughout the five-day diary exercise was frustration at the lack of new content and the prominence of outdated content on the homepage. To illustrate this challenge, consider the homepage for Metric Media’s Denver City Wire on September 14, 2021.
Source: Denver City Wire
The datelines for the articles shown at the top of the site’s homepage range from October 7, 2020 (“Colorado Gov. Jared Polis gets ‘C’ grade for fiscal management”), to February 25, 2021, for the top featured article (“Dominion Voting Systems sues Lindell over statements made about Nov. 3 election”). The article on potential top marginal tax rates under the “Biden plan,” shown in the sidebar, was featured as the leading article across most of the homepages assigned to our participants for the study in June 2021. Its dateline is October 22, 2020.
Scrolling below these featured articles on the homepage is a sidebar titled “Latest News” and a section called “Data Points.”
Source: Denver City Wire
The articles featured under Latest News, however, were among the oldest on the homepage. “DENVER PUBLIC LIBRARY: Virtual Toddler Storytime to be held on January 5, 2021” is a “press release submission” from January 4, 2021. The Data Points section had the site’s freshest content and has been updated on a seemingly weekly basis. (“Catherine Culshaw donates $2,800 to Jason Crow’s campaign in 2020” has a dateline of September 10, 2021.) The section, which consists of algorithmically generated articles based on publicly available data slotted into identical design templates, is a mainstay of all Metric Media properties. It was these articles that suggested a far more active publication schedule.
This design proved to be confounding for many of the study’s participants. In the first days of the diary exercise, our inbox was flooded with messages from confused participants unsure of how to review a site with months-old content and a homepage that remained static. We asked them to explore the website and look for updated content beyond the homepage. If they were still unable to find any new content, they were instructed to reflect on the site’s overall utility and design. At least one participant dropped out of the study in frustration at this point. The others continued, but expressed growing disbelief and irritation with the websites and their assigned task.
Brenda, a 34-year-old in Palmdale, California, said during the daily exercise that she was “baffled” by her assigned site, Antelope Valley Today . “The same story has been sitting here [on the homepage] for a week,” she said. “Nothing updates. So I don’t know how you can call yourself a news [site] or Antelope Valley Today. There is nothing from today.”
Tom was similarly dismayed to see the same 2020 political articles day after day on Ames Today . “These are things I would like to read and I would like to discuss,” he said. “I like local news. I love politics. But none of this is current or relevant.” The next day, after seeing the same articles, he confessed, “I’m really frustrated with this. It’s a neat idea for a mission, I was looking forward to it, but I just feel like I got a bum site.”
Besides frustration, other participants noted potential dangers in leaving up outdated content in a quickly changing news cycle. Mary, a 56-year-old in Wilmington, Delaware, was alarmed to see a months-old Delaware Division of Public Health press release about Covid-19 dated April 10, 2020, on her site, the North New Castle News . “So many things are wrong with this story,” she said, “It’s from a press release. It wasn’t written by a journalist … [and it’s] dealing with the coronavirus. Oh, my God. This could be giving wrong information, because pandemic information changes by the second. You can’t have outdated information out there.” As of March 2022, it was still the leading story on the site’s homepage.
Tony, a 24-year-old in Portland, Oregon, also noted a “weirdness” to some of the the articles remaining on prominent display. Referencing an article at the top of the Portland Courant’s homepage about a local resident going to a March for Trump rally in D.C., he said in his video recording, “This was obviously back in January and did not end in the best of situations.” In fact, the article was even older than he realized, dating from November 16, 2020, predating the events of January 6, 2021. But its position on the homepage was even more puzzling and disorienting in light of the later event. As of March 2022, it was still on the Portland Courant ’s homepage.
These observations by our participants can be further contextualized by findings from Asa Royal and Philip M. Napoli’s report , which was published a month after the audience research phase of our study was completed. Royal and Napoli found that over the course of their 78-day observation period, “the median age of front-page stories in the network rose by 70 days.” After the 2020 election “contestation period” ended, the study found that “overall story production by Metric Media outlets declined by an average of >80% in [the three most narrowly won] states. Compared to Metric Media outlets in a control group of states, the amount of fraud coverage and the story production decline among outlets in the battleground states were much more extreme.” The researchers also found that “only 1.31% of all autogenerated stories were placed on the front page of any of the network’s outlets. The rest either sat in outlets’ RSS feeds before being cycled out a few days later or were stowed below the viewport on the front page.”
A deluge of data
A majority of participants said that besides their websites being out of date, they found their content to be mostly irrelevant to their daily lives and communities. They noted the “odd,” “weird,” and “bizarre” content in the Data Points section: listings of small political contributions, expiring local business licenses, FDA inspection dates for local businesses, salaries for low-level bureaucrats, and Mass times for local Catholic churches (which, puzzlingly, appeared under a section titled Ethics and included no similar listings for any other religions, as several participants noted). While some participants had been intrigued by this content at the beginning of the study, after repeated visits, most came to find the content strange, or boring and non-newsworthy. “I just don’t care who gave fifteen dollars to Cory Bush’s campaign in the second quarter of 2020,” said Chiara, a 38-year-old in St. Louis, Missouri. Karen in Plano agreed, asking in her video recording, “Why is it news that someone’s electrical license expired?” Keith, a 36-year-old in Lorain, Ohio, also questioned how listing small political contributions qualified as a news story for Medina Today : “I’m not sure how much this is serving the local community.”
Other participants highlighted the lack of contextual analysis in the data-heavy articles. “There’s a lot of data points on this news website [ The South Bay Leader ], but not a lot of stories,” said Caitlyn, a 51-year-old in Manhattan Beach, California. “Every single person who gave any money to any candidate is not news. … It’s political data points. It’s statistics. It’s not a story.”
According to Jamelia, a 32-year-old in San Francisco, the main weakness of her site, the San Francisco Sun , was that “they rely too much on giving their audience these numbers without any context. Why is this data important and has that data changed from when it was last published?” Sam, a 29-year-old in Gunter, Texas, was worried about how the North Texas News ’ reporting on an expiring cosmetology license, for example, might impact local members of the community. “Many of the articles seem to paint a bad light on people,” he said. “Statistics need context around them.”
While most participants were confused or annoyed by automated content, a small number of participants did find utility in the collected data around actionable information like low gas prices, store holiday closings, Catholic Mass listings, and new regulations. “I just thought it was nice … that kind of a personal touch,” Donald, a 61-year-old in Salt Lake City, said of the Mass listings in the SLC Times . “It was something that you might not normally see. … It was kind of a little bit refreshing.”
Alongside automated, data-filled stories, the assigned websites feature a smaller number of articles with reporter bylines. In many cases, participants found these stories to be irrelevant as well, disconnected from themselves and their communities in subject matter or geography.
“They don’t have a lot of stories that are based in San Francisco,” said Jamelia of the San Francisco Sun. Reading a headline about vaccination policies in the cities of Santa Monica and Burbank, nearly 400 miles south of San Francisco, she said, “That’s great. But I don’t live anywhere near Santa Monica or Burbank.”
Rahim, a 35-year-old in Laguna Beach, California, was initially intrigued by his assigned site, the Southern California Record , due to its focus on legal issues. But after reading a headline about a case in the Central California city of Fresno, more than 250 miles north, he said, “These aren’t really local. You know, to me, something that happens in Fresno isn’t really a local news story.” In his final reflections, Rahim wrote that his site “essentially lumped in stories from all over the state. If the focus was strictly on legal proceedings and government rulings within So Cal [Southern California], that could have been a promising premise, but that simply was not the case.”
Alice in Houston said she was “surprised” to see an article in a Texas paper about Rudy Giuliani losing his license to practice law. “There’s no Texas link here,” she said. “And from what I read in the About section of the website, it seemed like this was supposed to be super focused on Houston and Texas.” At the same time, she noticed, another major national story with a clear local angle was missing: “I was really surprised not to see anything about Vice President Harris’s visit to the state” the previous week to speak to the immigration crisis on the border.
Alice was one of many participants who were disappointed by the lack of coverage of other issues on their sites, both specific stories as well as broader topics. On the day former Minneapolis Police Officer Derek Chauvin was sentenced for the killing of George Floyd, two participants were shocked to find no mention of the news on their assigned sites. “It’s called MinnesotaStateWire.com and they don’t even have the biggest story from the state of Minnesota today on their Web page about the sentencing of Derek Chauvin,” said Donna, a 42-year-old in New Germany, Minnesota. Mona, a 38-year-old in Apopka, Florida, was similarly surprised to find nothing about the collapse of the condominium that killed nearly 100 people in Surfside, Florida, on her assigned site, the North Lake Times . Abigail, a 37-year-old from Homer Glen, Illinois, was concerned that her assigned site, The Prairie State Wire , wasn’t covering a salmonella outbreak in shrimp sold in the state. Frank, a 36-year-old in River Forest, Illinois, said that it was “very serious” that his assigned site, West Cook News , had no new content despite a tornado warning issued in his area. “This isn’t a news site. I don’t even know what this is,” he said.
Several participants noticed the heavy focus on local politics and government at the expense of other beats. “Everything on this page is politics!” said Chiara of the St. Louis Reporter . Of the eight critical information needs defined in a study for the Federal Communications Commission, the only topic that the majority of participants identified as having been valuably covered by their outlet was political information. Meanwhile, despite the pandemic and other ongoing crises, a majority of participants said the outlets provided no valuable information related to emergencies and public safety, and half said they provided no valuable information related to health. Several participants observed a lack of information about Covid restrictions, which, as Dory, a 25-year-old in Miami, noted while reviewing the Miami Courant, is “something really important to put on a news site.” Others noted the absence of other staples of local news coverage, such as local events and holiday celebrations. “I don’t see anything about sports” in the North Birmingham Times , said Ken, a 43-year-old in Arab, Alabama, “which is kind of weird, because Alabama loves its sports.”
Mary in Delaware, assessing the North New Castle News, asked, “Where is my weather report? My traffic report? What’s happening in my community today? Activities I can take my family to this weekend? None of that is here.” In her final reflections, she expanded upon these critiques. “It was not what I consider a news website. It was simply a website mocked up to look like a news website. … This is a website with an ulterior motive that is using the words ‘news website.’”
4b. Formulating perceptions of trustworthiness and bias
Mary was not alone in doubting the legitimacy of her site—and a significant number of participants also questioned whether their outlets had, to reuse her phrase, ulterior motives. Yet despite their overwhelmingly negative final impressions, perceptions around the sites’ trustworthiness and potential bias in their coverage were much more mixed, and in some cases, almost evenly divided. While a majority of participants found the outlets to be untrustworthy (though by a much smaller margin than with overall assessments on quality), a very narrow majority nonetheless assessed their coverage to be fair and balanced. The split between these related but subtly distinct questions is revealing in terms of some of the wider takeaways about how these sites are interpreted. Specifically, the distinct ways in which participants responded to the algorithmically generated articles that are core components of pink slime journalism.
Making sense of algorithmically generated news
When assessing the trustworthiness of their assigned websites, nearly all participants mentioned the Data Points articles that are prevalent among Metric Media sites and, as low-cost, algorithmically generated output, typify quintessential pink slime journalism. Participants’ reactions, however, differed significantly in ways that have implications for the broader effect of the outlets on local markets.
Sam in Texas observed at the beginning of the study that the section appeared to consist of “A.I. that has grabbed a bunch of statistic pages and is just regurgitating statistics … it just comes across as so fake and it’s also surface level,” he said in a video recording. “It just definitely does not seem like a real person is putting this together, but rather a computer program.” For the few bylined articles, he noted that he could not find any information about the author.
Miranda in Virginia had a similar reaction, saying that the Shenandoah Valley News site “doesn’t really feel real … the content really looks to me like it’s all generated by AI or something like that … it just feels off.” She also noted the lack of bylines on most articles and a resulting failure to “establish credibility.” She was one of a few participants who took an extra step to review her site’s social media pages and was surprised to find that its Facebook page had zero followers and no content. In her final reflection, she wrote, “After several days I still don’t really know anything about the website or its purpose.”
Ellen, a 40-year-old in Potsdam, New York, concluded that her assigned site, Empire State Today , did not seem trustworthy because the “authors [often] aren’t clearly identified, the source/company responsible doesn’t seem to be authoritative or genuine, the site gives off a spammy feel to it and makes me think that some of the content is auto-generated.” Cora in Michigan categorized her site as an “auto-generated robot website.”
Martin, a 37-year-old in Indianapolis, Indiana, also concluded in his first day reviewing his site that the Indy Standard “is very much driven by probably just pulling from a database or some spreadsheet somewhere and taking some numbers from it and basically generating articles based off of it.” After noting that the bylines didn’t appear to belong to “an actual reporter,” he said, “The content doesn’t really feel like news at all. It feels more so like, ‘Here are just some numbers that we can pull about your general area and we’re going to build a website that sort of tries to represent it as news, even if it doesn’t really function in the traditional sense of news.’”
Despite these observations, in his final reflections at the end of the study Martin wrote, “Because almost all the stories were number-focused, they felt fairly trustworthy … very few of the pieces I read included anything involving opinion.” While he wrote that he would like to know more concerning the source of the data for these articles, he concluded that “as a result [of the focus on numbers and lack of opinions], they seem to be free of bias.” Thus, despite reservations about a lack of transparency around authorship and sourcing, the reliance on data proved effective in creating a perception of neutrality and, to some extent, trustworthiness. And while the automated articles and lack of new content left participants such as Martin feeling the sites were “sketchy” and not entirely normal, that didn’t mean they thought the coverage was politically skewed one way or another.
The automated articles’ use of unvarnished numbers and a straightforward editorial tone were repeatedly cited among the roughly half of participants who found their website to be trustworthy, fair and balanced, or both. Rahim in California, who had been disappointed by his site, the Southern California Record, nonetheless wrote in his final reflections:
I didn’t find any of the content to be strongly skewed toward either left or right ideology and viewpoints. I would say my initial impression of the website was that it was cautiously worthy of my trust. But I think some of that has to do with the fact that the stories themselves were rather bland and apropos of a legal journal, so perhaps I was bored into trusting the content. … [I] didn’t run into any strong opinion pieces (and it definitely wasn’t slanted in the vein of something like Fox News). But again, when the majority of your articles are simply providing banal reporting on legal matters and regulatory issues without any form of opinion interjected, there’s little room for impropriety.
Shonda, a 23-year-old in Heath, Texas, who had found the NE Dallas News ’s content unnewsworthy and irrelevant, also concluded that the site “did seem fair. There was nothing on the site that was condescending to a specific community or one sided. All of the articles presented a factual statement and point of view that was elaborated on with research and data. There were no opinion pieces or anything I would have considered to be slander.”
Jordan, a 42-year-old in Midlothian, Illinois, agreed that the South Cook News felt trustworthy because he didn’t think the articles “felt like they were approached with any sense of bias, or had intent to distract/cause chaos. … It didn’t feel politically divisive. The writing and the tone of the writing feel honest. … I didn’t read anything or see any headlines that made me question fairness in any way.”
Ludvig in California, one of the few participants with an overall positive final impression of his website, the Central OC Times , wrote that the site seemed trustworthy and without bias because “the nature of reporting concerning campaign donations and statistics concerning schools were not motivated by a point of view. I did not find any sensationalistic or partisan reports in this paper.” Moreover, the site “seems more fair and balanced than some of my local papers such as the Voice of OC, which can skew at times to the far left. The LA Times skews to the far left on certain occasions as well.”
Julie, a 36-year-old in Houston, wrote that “only the facts were addressed or quoted” on her site and the articles seemed as if they were written from the perspective of a “neutral person.” Edward, a 31-year-old in Richmond, Kentucky, found his site, The Bluegrass Times , untrustworthy. Yet he still wrote in his final reflections that the content “seemed balanced because the articles were short and simple without a lot of commentary. The articles did not seem to have a political bias one way or the other. The authors did not express their opinions.”
Caroline in Chicago wrote that she found her site, the Cook County Record, to be “factually trustworthy.” She had recorded positive impressions of the website’s design and layout, and how its focus on legal issues exposed her to topics she normally would not see. However, she noted, “its bias is evident in the types of information it presents—like the way it highlights the conservative perspective on partisan legal issues, or only includes statements from the conservative side” on articles about a Supreme Court decision and a local environmental lawsuit. The site, she concluded, had a “pretty conservative slant,” and many of the articles “made me kind of angry.”
“Not a legit news site”
Caroline was among the approximately one third of participants who perceived a specifically conservative slant to their assigned websites’ coverage. Erica, a 44-year-old in Collingswood, New Jersey, noted that while the data-heavy articles at Gloucester Today had no overt “statement of bias … [the content] was arbitrary and looked like right-wing selective data. Selective reporting of facts has its own implicit biases.” In her final reflections, she concluded that the site “seemed like a fake news churning site and/or just something thrown together by some random people using Gloucester County and other counties as a way to get agendas across. But it absolutely did not have actual relevant local news.”
Jocelyn, a 38-year-old in Florence, South Carolina, was incensed while reading articles in the Palmetto State News about Republican Senator Lindsay Graham, critical race theory, and conservative activist organizations. “This paper is so right-wing biased that it is, like, insane,” she said in a diary entry. “It just feels like this is where fake news comes from. As a social studies teacher who taught my kids about analyzing sources, this kind of thing makes me really angry and really quite sad for the country.”
Several participants accused their sites of using the pretext of local news to push a conservative or Republican political agenda. Morris, a 39-year-old in Chicago, wrote that the Chicago City Wire “seemed to be heavily focused on a conservative viewpoint. All of the stories seem closer to conservative talking points or propaganda rather than fair and balanced stories. … There wasn’t really much news content, just writing to steer the reader towards the conservative perspective.”
When asked in the final reflection section if her assigned site had deepened her understanding of her local area, Judy, a 59-year-old in Doylestown, Pennsylvania, answered, “Sadly it deepened my understanding of the extremely Republican biased websites that are out there … my website was written only from the perspective of a rabid Republican and seems to be obsessed with showing [political] donations.”
Karen in Texas noted in her initial response that the Dallas City Wire “does not look like a legit news site. My gut says something is off.” In one of her later diary entries, Karen questioned a June 18, 2021, article titled “Bennett: the media is chock full of bullies.” Its author, Monty J. Bennett, accused the local Dallas magazine, D Magazine, of presenting sponsored content as journalism. “This is really interesting to me. It feels like some content I might want to share,” she said. But she was dismayed by the lack of information on the site about its author. “Who is Monty J. Bennett? I have no idea. Is this someone who works for Dallas City Wire? Is he a guest [writer]? What’s his expertise? That’s something I would like to know.”
In fact, Bennett’s ties to Metric Media have come up multiple times. The Texan CEO of a hospitality and real estate company, Ashford, Inc., Bennett is also the publisher of The Dallas Express, a news website that he launched in early 2021.(A press release from late 2021 stated the ownership of Dallas Express had changed, but an April 2022 op-ed was attributed to “Monty Bennett | Publisher”) Prior to founding The Dallas Express, Bennet also ordered the Metro Business Network, another part of the Metric Media-related networks, to publish articles that would support a stimulus bill to help his company and to attack his creditors, according to a New York Times investigation . (Bennett has strenuously denied the allegations, but a Tow Center investigation also found that Ashford, Inc. is a client of Metric Media’s parent company .) According to D Magazine , Bennett’s Dallas City Wire piece attacking the magazine as “bullies” came after it published an article discussing the New York Times’ investigation into his alleged practices. Several weeks after Karen’s observations, the link to the article no longer worked and the story had seemingly vanished from the Dallas City Wire website, according to our review of the site. As of June 2022 it still appears to be missing. In her final reflections she reiterated her impression that her site “was not trustworthy … my gut from the start was that this site is not legitimate.”
Of the roughly one third of participants who perceived conservative bias in their website’s content, approximately half suspected an actual partisan organization of being behind the site. Sandra, a 58-year-old in Winterville, Georgia, said reading The Athens Reporter ’s coverage of voting issues made her “feel like there are people within my community who are not happy with the previous presidential election … I think that it is a politically motivated news publication put out by the Republican Party.” Noah, a 37-year-old in Kansas City, Missouri, said the KC Reporter ’s “lack of original local content made me suspect this publication as a front for some sort of ideologically motivated group.”
Miles, a 30-year-old in Washington, D.C., was suspicious of a December 2020 article in the D.C. Business Daily about the dire economic situation for the hotel industry. “I think this really represents the talking points of the AHLA [the American Hotel and Lodging Association, a lobbying organization for the industry],” he said. “I’m beginning to feel like it’s a little bit more propaganda. That there is definitely someone that is funding this, that’s pushing their agenda.” (As investigations by both the New York Times and the Tow Center have found, the Metric Media network had ties to the hotel industry, and the publications reflected the talking points of the hospitality industry through the pandemic lockdowns.)
Several of the participants cried foul when reading their sites’ boilerplate About pages and their stated mission, in the case of Metric Media, to “provide objective, data-driven information without political bias.” Judy in Pennsylvania responded with “Ha ha ha,” when she read the Lower Bucks Today About page. “Total garbage … its strengths are to push the Republican agenda,” she said. Denver City Wire’s About page “says nothing about being a conservative organization,” said Lara, a 36-year-old in Denver, Colorado. “They make it sound like they’re unbiased and neutral, which is completely false.”
Russ, a 57-year-old from Doylestown, Pennsylvania, said he was “quickly getting a sense” from the Lower Bucks Today “that this whole thing is pretty biased in a right wing Republican way. … It looks like it would like to be a source of local information but my sense is that it’s a political outlook is slanted and misleading.” He later said, “I don’t have much experience with local news websites other than my own newspaper, but I believe there is a trend of politically motivated sites pretending to be local news.” In his final reflections he wrote, “This website did not deepen my understanding of my local news, but did make me think more about political polarization, and a concern about biased sites pretending to be neutral news.” Despite his suspicions, Russ had not researched the ownership of his assigned site before being prompted to in the final reflections section. This decision, or lack thereof, was not uncommon, even among the dozens of participants who were suspicious of their assigned sites’ intentions.
Transparency and Ownership
A key area of interest in this study was to examine the extent to which participants actively sought to establish the provenance of their assigned websites, and the process through which they sought this information. The New York Times observed that Metric Media sites “typically lack mastheads, local addresses and clear disclosure of their ownership or revenue sources,” adding that “Voters could easily become confused about the origins of information from these seemingly innocuous local-sounding outlets.” Newsguard assigned a red “nutrition label” to sites in the Metric Media network—meaning they “generally fail … to meet basic standards of credibility and transparency”—and determined the network’s sites “do not meet the standard for disclosing information about ownership and financing.”
None of our participants had encountered their assigned websites before. Yet despite many raising concerns or highlighting issues over the course of the study, only a small number looked into the ownership of their assigned sites of their own volition. Of those who did, comparatively few engaged in what education researchers have referred to as “ lateral reading ”—the practice of seeking out other sources to evaluate a site’s credibility.
Lack of research around ownership
A sizable majority of participants—roughly two thirds—did not attempt to research who owned or operated their assigned site before they were prompted to do so in the final reflection section. Most of these participants said it had never occurred to them. Some participants, however, indicated they were embarrassed by their lack of initiative, suggesting at least some familiarity with this aspect of media literacy.
Ruth, a 56-year-old in Richmond, Texas, was apologetic and surprised when asked whether she had previously researched who owned and operated her assigned site, the SE Houston News . “Oh, my gosh, so sorry! I never looked or thought about who ran the website at all. … I wonder where would I find that?” Caroline in Chicago said she was “kind of surprised” that she hadn’t thought to investigate who was behind her site. Of the roughly one third who had attempted to identify the site’s ownership before being prompted to do so, only a small number researched the outlet beyond the scant information available on the website.
A significant portion of participants—both those who had conducted research of their own accord and those who waited until they were prompted—were unable to find any relevant information on their outlets’ websites. Many found the parent company’s name (e.g. Metric Media or LGIS), but no other information. The About pages of at least five sites, all of which belonged to the Record network, displayed error messages that read “This page isn’t working.” (The link remained broken for several months after the completion of the study, but has since been fixed.) These findings were indicative of the networks’ overarching lack of transparency.
Participants’ reactions to the general lack of information around ownership, however, varied significantly, from mild annoyance to severe alarm.
At one end of the spectrum, participants commented on the lack of transparency, but did not express strong opinions or indicate that this affected their view of the site’s integrity. Speaking of his attempts to find who owned the Cook County Record, Ivan, a 43-year-old in Chicago, said he “had so much trouble.” After discovering the About page didn’t work, he “tried all these different options to find it. I looked on Facebook, Twitter, still couldn’t find anything. So, yeah, this site doesn’t make it too easy to find out who owns the site.”
Julie in Houston also found the ownership information available on her site to be lacking. She read aloud from its About page, which describes the site as a “daily online news source … founded to change the local news focus from celebrities, gossip, and partisan political issues to matters of greater interest to all Houstonians. … We focus on fact-based reporting and do not allow opinionated journalism in our news stories.” Responding to this description she said, “I mean, I see that. But it still doesn’t tell me who actually is writing this.”
Kate, a 50-year-old in Dallas, Texas, also said she could not find anything on her assigned site “that shows who runs the site.” “I’m not seeing anything other than them talking about their principles and stuff,” she explained. “I don’t see who owns it or anything. I went to the contact info. Didn’t really seem to be anything there. So I have no idea who owns it.”
Notably, all of these participants had previously said that they found their sites to be trustworthy, fair, and without bias. None of them indicated in their reflections that this lack of transparency had altered that assessment. In further indicators of possible media literacy and trust vulnerabilities, these participants uncritically accepted the claims of objectivity in these sites’ nebulous About pages or interpreted nonprofit status as further confirmation of virtuous intentions. Adrianne, a 27-year-old in Dallas, said, “I didn’t realize this until just now, but they don’t have a side to their stories. They just provide facts,” after reading the outlet’s stated mission to report “dispassionately [and] place emphasis on facts over all else. News reports … should never be ‘spun’ or carry editorial opinions to suit the political persuasions of our reporters or editors.” Adrianne said she believed this principle was “really important, because sometimes you look at certain news websites and they are very one-sided. So it’s good that it’s just reporting the facts, which not a lot of people do recently.”
Suman in Florida noted that Metric Media was a 501(c)(3) nonprofit news content provider, “so it tells me that it doesn’t have a political agenda, or it’s not backed by any biased thing, which is great.” However, this assumption is incorrect. While there are IRS codes that don’t allow 501(c)(3) organizations from partaking in political campaigns on behalf or against a candidate (directly or indirectly), nonprofit organizations can partake in issue advocacy. Suman was among the few participants who had a final positive impression of his assigned site, the NC Florida News . In his final reflections he praised the site’s trustworthiness: “It definitely feels like a more balanced and unbiased news outlet to report relevant community related news.”
Vivek, a 33-year-old from Rolling Meadows, Illinois, was another of the few participants who had recorded a final positive impression of his site, the Cook County Record. However, he was the lone participant who seemed to question that perception after attempting to research its ownership. “I was a little worried when I clicked on ‘About’ and it didn’t take me anywhere but here,” he said, indicating the error message on its website. “Maybe it’s perfectly innocent,” he said, but “maybe not.”
Prior research, lack of transparency, an