Philadelphia’s news media system: which audiences are underserved?

Abstract

This study builds a multi-dimensional framework for assessing local media systems to identify potential gaps in news provision, especially among socioeconomically marginalized communities. We gather data on income, education, and age of audiences and coverage areas for 38 news outlets in Philadelphia and conduct a content analysis to gauge how these outlets meet critical information needs related to the COVID-19 pandemic. Findings indicate that multiple dimensions of Philadelphia’s news media system—audience size and socioeconomics, staffing levels, forms of ownership, and platform effects—work together to underserve communities with lower levels of income and education and that this structural gap generates a measurable gap in the provision of news content meeting the critical information needs of these communities. Policy interventions such as public funding and subsidies can enhance the capacity for Philadelphia news organizations to meet the critical information needs of marginalized communities.

Local media systems have become critical sites of study as news outlets close and communities become “news deserts” that lack trusted sources of local information (Abernathy, 2016; Ardia et al., 2020; Ferrier et al., 2016; Pickard, 2020). Urban media systems such as Philadelphia, home to one of the largest designated media markets in the United States, also have been affected by declining news resources, which can exacerbate inequalities in news provision (Rafsky, 2020). Gaps in news provision can manifest as absent or inadequate news about issues vital to a community’s interests, which Friedland et al. (2012) refer to as “critical information needs.” From information about public emergencies to coverage of local politics, journalism that fulfills these needs is essential to community cohesion, self-determination and, more broadly, to a functioning democracy.

While Philadelphia appears to have a thriving news market and has experimented with new funding models to sustain news provision (Rieder, 2017; Schaffer, 2010; Wolfson & Funke, 2014), it remains unclear to what degree these recent interventions ensure that the information needs of all residents are equitably met. Philadelphia also has significant levels of socioeconomic and racial inequality (Shields & Siddique, 2020), raising the question of whether the city’s news outlets equitably serve its most marginalized communities. Given the structural nature of both the journalism crisis and the city’s historic inequalities, news gaps may persist, which may call for more significant policy interventions and experimentations with new models of media ownership and governance.

Accordingly, this study builds a multi-dimensional framework for studying the complex structures of local media systems and uses it to assess Philadelphia’s news media landscape with a particular focus on identifying potential news gaps, especially in the city’s most marginalized communities. We begin by reviewing the literature on news gaps, identifying both their structural drivers and social consequences. Next, we outline our methodological approach to assessing multiple dimensions of Philadelphia’s media system: Audience socioeconomics; audience size; news staffing levels; media ownership structures; and news platforms (more recent digital-only entrants in the media system vs. older, legacy outlets that generally combine platforms such as print, broadcast, and digital). Findings indicate that Philadelphia’s news media system underserves communities with lower levels of income and education and that this structural gap generates a measurable gap in the provision of news content meeting the critical information needs of these communities. In our conclusion, we discuss policy implications.

Literature review

Declining advertising revenues associated with digitization have forced many newsrooms to cut staff or close entirely. Such economic pressures on advertiser- and subscriber-funded news in the United States have contributed to the spread of “news deserts”: Pronounced gaps in news provision in which local journalism is almost entirely absent (Abernathy, 2016, 2020; Ardia et al., 2020; Ferrier et al., 2016; Pickard, 2020; Stonbely et al., 2019).¹ Even when a community is served by local news organizations, much of their journalistic output is often not actually local or directly relevant to that community (Napoli et al., 2017). In the absence of quality local news, communities often turn to social media and less-reputable sources for critical local information, enabling misinformation and spectacle to proliferate. Indeed, in Beaver County, Pennsylvania, news-starved residents turned to Facebook alerts for their news needs, which fueled the rapid spread of “fear,” “exaggerations,” and “falsehoods” (Zadrozny, 2021).

Gaps in news coverage, from less than adequate provision of vital information to the virtual absence of such information in news deserts, also raise normative concerns for the pluralistic representation and participation of different publics in democracy (Ferree et al., 2002). Among other functions, media provide a forum for diverse publics to voice their views and serve as a conduit for communication between these publics and their governments (Baker, 2002; Christians et al., 2009, p. 30). When these roles are fulfilled, media can inform different publics that their interests are at stake in political processes and help mobilize them to assert and defend those interests (Baker, 2002). However, news gaps undercut these processes, which lie at the very heart of healthy democracies.

In addition to stoking normative concerns, a systemic lack of quality information can undermine a community’s ability to address concrete, everyday issues vital to its ability to thrive. Friedland et al. (2012) identify eight categories of such “critical information needs”: Emergencies and public safety; health; education; transportation; environment and planning; economic development; civic life; political life. A healthy local news environment is one that meets the diverse critical information needs of different communities within a certain geographic area. Yet reportage that meets critical information needs may not always be favored by journalists and news organizations. Their “news values” constitute a large corpus of research, which often notes that news outlets often favor stories that shock, surprise, or entertain, rather than meet critical needs for information (Galtung & Ruge, 1965; Harcup & O’Neill, 2001, 2017).

In order to research patterns and gaps in news provision, scholars have used the concept of a “media system,” which refers to patterned relationships between multiple media outlets and their broader social, economic, and political contexts (Albæk et al., 2014; Aalberg & Curran, 2012; Chadwick, 2017; Hallin & Mancini, 2004). For example, Chadwick’s concept of the “hybrid media system” emphasizes the flow of news across platforms, such as print, broadcast, and the Internet, and different media production logics. Our study focuses on the structural drivers of systemic news gaps, rather than news flows. However, by emphasizing multiple, interacting dimensions of a media system and different production logics, Chadwick’s approach suggests a framework for media system analysis that encompasses both the structural drivers of local news (audience socioeconomics; ownership structures), which influence media production (Benson et al, 2018; Napoli et al., 2015; Picard & van Weezel, 2008; Stonbely et al., 2019), and the different weight that individual outlets carry within a local media system (differences in audience size; digital-only vs. legacy outlets that combine print, broadcast and the Internet to various degrees).

The economic interests generated by different forms of media ownership, such as stock market-traded, private, nonprofit, and public ownership, can produce significant variations in commitments to producing public service-oriented news (Benson et al., 2018; Picard & van Weezel, 2008). For instance, news deserts have been linked to the private equity and hedge fund forms of commercial ownership (Abernathy, 2016; Kuttner & Zenger, 2017). Under such ownership, news properties may be treated like any other asset in large financial portfolios: As targets for extracting maximum profits rather than as civic institutions. News outlets subject to these profit logics often face staff cuts to cost-save and, once they become unprofitable, are either shuttered or sold, often resulting in news deserts.

Profit pressures at stock market-traded and private forms of ownership may encourage certain kinds of substantial news coverage to attract and retain subscribers, but they also incentivize the production of sensationalistic news. Moreover, those that rely on paywalls bar access for those who are unable or unwilling to pay (Knobel, 2018; Pickard, 2020). Nonprofit and publicly funded forms of ownership reduce profit pressures but sustaining these revenue models—particularly in the United States where public funding is relatively low—often requires additional funding streams. The latter may include subscriptions, corporate underwriting, or donor and foundation support, which may encourage news production that does not prioritize a community’s information needs (Benson, 2018; Browne, 2010).

Large, commercial news organizations such as newspapers and TV affiliates continue to be dominant providers of news coverage in U.S. metropolitan areas (Pew Research Center, 2010). However, researchers have found demographic differences among audiences for different news formats: Local television news viewership appears to lean older and lower income and education; Internet use leans somewhat younger and male, as well as higher income and education; newspaper readership leans older and higher income and education (Althaus & Tewksbury, 2007; Althaus et al., 2009; Bucy, 2001; Gaziano & McGrath, 1987; Prior, 2007). Research also suggests that Black audiences consume more television news and less Internet news than white audiences and that Latino audiences are less likely than white audiences to seek out news regardless of media format (Althaus & Tewksbury, 2007; Althaus et al., 2009; Bucy, 2001). Media system gaps in coverage often correlate with socioeconomic indicators of audiences, with lower levels of income and education associated with less news and less coverage fulfilling communities’ critical information needs (Napoli et al., 2015).

Inequities in news provision in Philadelphia’s media system have been noted by prior researchers. Studying the media system in the 1980s and 1990s, Kaniss (1991, 1995) found the city’s news outlets chasing after affluent residents retreating to suburban areas. As digital forms of news distribution emerged in the 2000s, a new wave of research found gaps in public affairs reporting, but also traced networks of emergent digital-only news outlets that potentially could fill some of these gaps (Anderson, 2013; Schaffer, 2010). However, Anderson’s (2013) research shows that even as this digital news sector emerged, longstanding journalistic practices and the large footprint of legacy news institutions such as the Philadelphia Inquirer meant that “stasis” was more often evident than radical change (p. 159). More recent research has identified persistent news gaps that perpetuate negative stereotypes of neighborhoods rather than providing the impetus for positive change, as well as a lack of racial diversity among journalists within local news organizations (Crittenden & Wenzel, 2020; Malo, 2018; Wenzel, 2020a, 2020b; Wenzel et al., 2018). Surveys of Philadelphia residents find that crime news is popular but that what residents desire the most is news and information “of a highly personal nature” that they can put to use in their day-to-day lives (Delli Carpini et al., 2018, p. 8). A 2020 diversity audit found that the Philadelphia Inquirer, the dominant print and online news outlet in the city’s media system, underrepresents people of color in its content and that its staff is overwhelmingly white (74.3% of newsroom staff; Monroe & Wenzel, 2021, p. 22).

Our research builds on this work by assessing Philadelphia’s media system, with a focus on the multi-dimensional influence of different factors—different forms of media ownership, audience size and socioeconomics, and news format—on the production of critical news and information for the city’s marginalized communities. Our research questions are:

• RQ1: To what extent are socioeconomically and demographically differentiated (by income, education, and age) audiences served by Philadelphia’s media system?

• RQ2: To what extent are socioeconomically and demographically differentiated audiences served by digital-only news outlets in Philadelphia’s media system?

• RQ3: How do the socioeconomic statuses and age demographics of different communities correlate with news coverage meeting the critical information needs of these communities?

• RQ4: How do different forms of media ownership affect the provision of news coverage that meets communities’ critical information needs?

Data and methods

For this study, we first sought to identify news media outlets in the greater Philadelphia area, including the city and its suburbs. We reached out to Sarah Stonbely at Montclair State University’s Center for Cooperative Media, who is leading an extensive mapping and content analysis of New Jersey’s media system, which has some overlap with Philadelphia. She generously provided us with data current as of 2019 on outlets in the greater Philadelphia area used in her project. These data sources include Editor & Publisher; BIA/Kelsey; ABYZ News Links; Michele’s List; New Jersey Press Association; the National Newspaper Association; Onlinenewspapers.com; MondoTimes.com; and the Library of Congress.

These sources helped us identify 89 print, television, radio, and digital-only outlets, which we cross-checked with additional sources to narrow the scope of our detailed analysis of the city’s media system (FeedSpot, 2021; Matsa & Sumida, 2019; Mercier, 2018). This scope includes mass-oriented outlets prominent across the city and niche-oriented outlets serving specific geographic or social groups. To facilitate a feasible study capable of producing a fine-grained, multi-dimensional analysis of Philadelphia’s media system that includes analysis of news content, we do not include radio stations or low-power TV and smaller affiliate stations. Primarily based on (a) data indicating prominence in terms of circulation, viewership, or web traffic, or (b) frequent mention in published articles as important outlets for news in Philadelphia, our final selection includes 38 outlets in the greater Philadelphia area,² representing a mix of ownership types serving mass audiences and niche audiences.

In addressing our research questions about news provision for socioeconomically and demographically differentiated audiences (RQ1 and RQ2), we gather outlet-level data on audience income (median or average household income), education (percentage of audience with bachelor’s degree or higher), and age (median age or percent below 35); on full-time editorial staffing levels; and on audience size, as measured by daily or weekly print circulation, television households reached by evening news telecasts, and web traffic (unique monthly visitors). These data, which are not available for all of our outlets, derive from the U.S. Census, industry and news sources, and information provided by outlets themselves either through direct correspondence or through their web sites and materials for advertisers and sponsors, often known as “media kits.” Availability limitations require us to draw on data from different years, but most data fall into a narrow time frame (generally 2017 to 2021, though some circulation data is as old as 2009), and therefore they do not preclude comparative analysis. We also contacted each outlet in our sample via email, requesting data on the above variables to ensure we had the most recent and correct information available. We received five responses (13% response rate), four of which provided information that either validated our existing data, corrected that information, or added new information to our analysis.

When audience income, education, and age cannot be determined from an outlet’s media kits, direct communication, or news sources, we use U.S. Census data (American Community Survey) to estimate these data points. First, we determine the geographic coverage area of an outlet, as identified by the outlet in their published materials or emails to us, or as determined by print subscription availability. We then use Census ZIP Code Tabulation Area (ZCTA) data for median household income, education (percentage of people with bachelor’s degrees), and median age for these areas, weighting calculations by population per ZCTA. For outlets with a focus on Hispanic or Black communities, we use 2018 income and education data for Hispanic and Black/African-American majority Census Tracts, as reported by the Economy League of Greater Philadelphia (Shields & Siddique, 2020).

To facilitate comparison between audiences for different outlets, we create easily readable indices for each of these measures of audience demographics. Treating citywide figures as an index of 100, we add or subtract outlet-level audience data to create income, education, and age indices. For example, based on 2019 Census calculations for its coverage area, Delaware County Daily Times’ audience has a (weighted average) median household income (HHI) of $78,106. This is $31,000 (rounded) above Philadelphia’s median HHI of $47,474, creating an income index of 131, as we divide all income data points by 1,000 to assist readability. Education indices above 100 indicate a higher than average number of people with bachelor’s degrees or higher (31% citywide average; adding or subtracting 1 for each percentage point over or under the average). Age indices above 100 indicate a higher percentage of people over 35 or higher median age than found in Philadelphia as a whole (34.7 median age; adding or subtracting 1 for each percentage point over or under citywide percentage, or for each year, in the case of median ages). Lack of available data prevents us from calculating income, education, and age indices for all of our outlets.

To answer our research questions about critical information news provision for socioeconomically and demographically differentiated communities (RQ3 and RQ4), we conduct a content analysis to assess each outlet’s volume of COVID-19 pandemic coverage and compare this to the volume of coverage of individual crimes. We search databases Factiva, NewsBank, and ProQuest to compare the number of news stories published between May 11 and August 18, 2021 that focus on the pandemic vs. the number of stories that focus on crime, attempting to exclude stories primarily focusing on community impacts or structural drivers of crime. We use this time period because several major Philadelphia outlets are archived in Factiva on a rolling, approximately 90-day window. Our searches show that as of August 18, when we began this portion of the data collection, the oldest content included in Factiva for these outlets was published on May 11. We conduct this content analysis for all outlets in our sample that are available in these databases (23 out of 38, or 61%).

To assess the level of critical information needs coverage by outlet, this content analysis operationalizes Friedland et al.’s concept of critical information needs by privileging coverage of the COVID-19 pandemic as potentially fulfilling more critical information needs than coverage of individual-level, breaking-news crime and related legal processes. Both areas of news coverage were highly newsworthy during the time period analyzed: As of early 2022, the pandemic had sickened more than 277,000 people and killed more than 4,000 in Philadelphia County (New York Times, 2022); in 2021, the city of Philadelphia saw a record 557 homicides (Palmer & Orso, 2021).

However, research on “news values” indicates that journalists may not deem pandemics and crimes as equally newsworthy (Galtung & Ruge, 1965; Harcup & O’Neill, 2001, 2017). For instance, in terms of what Galtung and Ruge call “frequency,” individual crimes occur often, though they typically are of a short term nature, while the pandemic continues to unfold over a long time period, though with frequent events relevant to most, if not all of Friedland et al.’s critical information needs. In terms of “unambiguity,” the level of information required to clearly understand COVID-19 may be higher, especially as official advice for staying well and reducing viral transmission develops and changes. In terms of “unexpectedness,” pandemics and crimes are both surprising, though particularly violent or unusual crimes may have more shock value. Additionally, Harcup and O’Neill (2001) identify the “magnitude” of an event as a key news value, as with a pandemic that affects very many people in many ways. Harcup and O’Neill (2017) also have noted that the rise of social media has made “shareability” a news value, as in news likely to provoke a response on Facebook or Twitter.

Though it may lack the pandemic’s magnitude, crime news is popular with audiences (Delli Carpini et al., 2018). However, it is less clear whether such information wants can be easily equated with information needs that, if fulfilled, benefit rather than harm a community. While Friedland et al. include “emergencies and public safety” among their critical information needs, the authors note the negative impacts of reportage on crime, which can generate “fear of crime and minorities in local communities,” and they emphasize a community’s interest in “fair and accurate representation” (p. 11). Indeed, a large corpus of research has noted the sensational aspects of crime coverage (Adams, 1978; Ehrlich, 1996; Knight, 1989; Wang, 2012), primarily intending to capture audience attention rather than to assess the impact on a given community. Crime coverage is subject to further criticisms: It can permanently cast a negative light on those accused of or convicted of minor crimes (Benton, 2021; Boston Globe, 2021; Wemple, 2021); overreporting crime can overrepresent the problem, creating false public perceptions of supposed “crime waves” and increasing public support for punitive measures such as longer prison sentences (Beale, 2006; Sharkey, 2018); the way crime news is covered can reflect and reinforce racial biases, such as more frequent reporting on criminal histories when Black victims and suspects are involved than when white victims and suspects are involved (Monroe & Wenzel, 2021); episodic news accounts of crime, often primarily using police and prosecutors as sources, can distract public attention from structural causes of crime, such as systemic drivers of poverty and racism and inadequate mental health and substance abuse services (Hechinger, 2021).

By contrast, the COVID-19 pandemic, which has impacted virtually all aspects of daily life, is a critical concern for all communities and is relevant to all eight of Friedland et al.’s critical information needs. Because of this broad impact across all dimensions of a community’s well-being, our content analysis seeks to assess the degree to which news outlets focus on COVID-19 coverage versus coverage of crime incidents, taking a higher ratio of COVID-19 to crime coverage as an indicator of a stronger focus on an unarguably critical information need for Philadelphia’s communities, including socioeconomically marginalized communities.

Pandemic coverage need not always be spectacular to meet a critical information need, especially given audience surveys that find a clear desire for usable information that can be put to use in the day-to-day life of a community (Delli Carpini et al., 2018). Examples of COVID-19 coverage in our analysis include stories with headlines such as “COVID cases increasing, vaccine lowers severity” (Chestnut Hill Local); “Chester County distributing $5.9M in state pandemic relief to local hospitality businesses” (Philadelphia Business Journal); “Philly area restaurants requiring vaccination: a running list” (Billy Penn). Examples of crime coverage in our analysis typically have a spectacular dimension and are less clearly critical for the daily functioning of a community, as in the case of stories with headlines such as “West Philadelphia double shooting under investigation” (Philadelphia Patch); “Woman accused of using GoFundMe money raised for husband’s cremation, funeral to pay living expenses” (WTXF/Fox29); “Cobbs Creek man cleared of 1984 murder by evidence that came from police file” (Philadelphia Daily News). For crime stories, our search strings are designed to exclude references to police killings that have sparked Black Lives Matter protests, as these raise community-wide concerns for pervasive, systemic racial bias, and to exclude references to the January 6, 2021, riot at the U.S. Capitol, which occurred beyond the Philadelphia area.

To conduct qualitative assessments of coverage, we download for each outlet random (every Nth item, chronologically ordered), 100-story samples surfaced by our searches for COVID-19 and crime stories; we download the entirety of stories if searches for an outlet surfaced fewer than 100 stories. Search strings were developed through a rigorous testing process (for details of methodology, see appendix), but these subsamples enable us to conduct additional assessments of the relevance of items returned by our search results. For most outlets, these subsamples indicate that our search strings achieve the high levels of precision indicated by search string testing, meaning that relevant items comprise roughly 86% of COVID items and 91% of crime items.³ Since the pandemic has had very wide-ranging impacts on a community’s life—such as social distancing at church events, concert cancellations, vaccination clinics, and travel disruptions—our search string terms for COVID items produce similarly wide-ranging results with a high degree of relevance. Additionally, we assess the degree to which COVID items in our subsamples are either locally produced or locally focused and find that such items comprise the majority of items for most outlets, with the exception of WCAU/NBC10 (12% of COVID items being locally focused or produced); Philadelphia Tribune (22%); WPVI/ABC6 (31%); Metro (48%). However, nationally and internationally produced or focused items can include critical information about the pandemic, as in the case of national recommendations for wearing masks or receive vaccinations.

Among subsamples for crime stories, we find many instances of nonlocal coverage, including news accounts of celebrity trials. Such nonlocal crime items, regardless of subject matter, are suitable for inclusion in our counts of crime stories because they clearly are less relevant to a broad set of critical information needs in Philadelphia than information about the pandemic.

Findings

Income, education, and age indices

Based on available data for news outlets included in this study, Philadelphia’s media system underserves socioeconomically marginalized communities. Table 1 shows that the city’s media system on average leans toward producing news for audiences with income and education levels higher than city medians and averages (RQ1), with an average income index of 122 and an average education index of 111. The system on average also produces news for older audiences, with an age index of 109. However, a multi-dimensional analysis of the media system taking account of the relative weight of different outlets (staffing levels and audience size) and different formats (digital-only vs. outlets combining print, broadcast, and digital formats) raises further concerns about the degree to which Philadelphia audiences are equitably served.

Based on staff levels and audience size, the Philadelphia Inquirer arguably occupies a dominant position in the Philadelphia media system, though an online paywall at Inquirer.com limits access for non-subscribers. Since email exchanges with the Inquirer indicate all its journalists are included under the same Inquirer brand regardless of which outlet their content appears, we count Inquirer and Daily News (both under the same ownership) editorial staff together as Inquirer staffers. However, based on the Inquirer’s reported data for advertisers (Philadelphia Inquirer, 2018), income and education indices are higher for Inquirer audiences than for Daily News audiences. Daily News print circulation also trails Inquirer circulation (11,813 average daily copies vs. 80,940 average daily copies, respectively).

Major television affiliates for ABC, CBS, NBC, and Fox also wield significant journalistic resources, but data on the composition of their audiences are unavailable. Public media organization WHYY has a television audience education index (102) close to the Philadelphia average, based on a media kit citing 2013–2014 figures (WHYY, n.d.).

Though Philadelphia is home to many smaller publications that serve various racial, ethnic, religious, and other social groups (e.g., Al Día, El Hispano, Impacto, Catholic Philly, Jewish Exponent, Philadelphia Gay News), as well as hyperlocal sites focused on certain neighborhoods (e.g., Passyunk Post, Northeast Times, West Philly Local), journalistic resources at these outlets pale in comparison to the city’s dominant news outlets. These socially and geographically oriented publications have small newsrooms, typically fewer than a dozen people, and in some cases only two or three people. Few of these smaller publications have established paywalls requiring people to pay for access, which are more likely to be found at larger outlets serving audiences, who often are relatively affluent, regardless of location, such as the Philadelphia Business Journal, the Inquirer/Daily News, and Philadelphia Magazine.

Because Philadelphia news outlets have different weights in terms of audience size (using print circulation and TV viewership numbers if available and web traffic if not; weekly and monthly figures adjusted to daily estimates by dividing by 7 or 30, respectively), we conduct an additional analysis using weighted average indices. This analysis produces average income and age indices for the media system that are slightly lower than raw averages: 117 for income; 111 for education; 108 for age.

Assessing differences in formats, we find that digital-only outlets, several of which are newer entrants in the city’s media system (e.g., Billy Penn, Passyunk Post, and West Philly Local), do not appear to be significantly altering the system’s socioeconomic dynamics (RQ2). In Table 2, we aggregate circulation, web traffic, and staffing figures by high-/low-income indices, as well as by digital-only sites. Digital-only outlets do not have the same level of journalistic staffing as large, legacy organizations, and they tend to serve audiences that have above average income and education and that are older. These more affluent, educated, and older demographics are found among outlets that have the largest audience size in terms of web traffic (unique monthly visitors and page views) and the largest contingents of journalists.

Content analysis: meeting critical information needs

Our analysis of news content, reported in Table 3, shows that nearly all outlets give more attention to the COVID-19 pandemic than to breaking news of individual crimes (RQ3). Excluding outliers Jewish Exponent and Philadelphia Magazine (ratios more than one standard deviation above the overall average), the average ratio is 5.5,⁴ meaning that on average outlets produce 5.5 times as many COVID-19 stories as they do crime stories.

When ratios are weighted by audience size, the systemwide average drops to 2.9 (excluding Philadelphia Patch, for which audience size data are not available, and Roxborough Review, which has an artificially low ratio due to a low volume of content; 2.5 excluding Jewish Exponent and Philadelphia Magazine, as well). Unlike our assessment of average income, education, and age indices, ratio averages include major commercial TV affiliates, which generally have large audiences but low COVID-to-crime coverage ratios and thus have a systemwide suppression effect on average ratios. Excluding their ratios, Philadelphia’s media system has a weighted average ratio of 4.8 (3.9 excluding Jewish Exponent and Philadelphia Magazine, as well).

However, news outlets supply news that fulfills critical information needs in ways that are related to audience demographics: Among outlets included in our content analysis, wealthier, more educated, and older audiences are associated with a stronger news focus on COVID-19 pandemic versus crime. Higher levels of attention to COVID-19 coverage are positively correlated with income, education, and age indices, with bivariate correlations of 0.545 (p = .07), 0.21 (p = .54), and 0.854 (p = .003), respectively for these indices and COVID-to-crime coverage ratios.⁵ Correlations for income and age are considered moderate to strong, and the correlation with education is weak.⁶

Most television stations have far lower than average ratios of COVID to crime coverage, though WPVI/ABC6’s ratio of 3.3 exceeds ratios at most of Philadelphia’s larger outlets. WTXF/Fox29 is one of two outlets in our study that give more attention to crime than to the pandemic (−3.2). The only other outlet to show a similar negative ratio is South Philly Review (−2.9), though this ratio is based on a low overall volume of stories.

Assessing coverage in relationship to forms of ownership (RQ4), we find that commercial outlets are associated with lower ratios of critical versus noncritical information needs coverage than noncommercial outlets, with an average ratio of 6.8 for commercial (stock market-traded and privately held) outlets versus an average ratio of 15.9 for noncommercial (public and nonprofit) outlets (4 vs. 8.9, excluding outliers Philadelphia Magazine and Jewish Exponent; p < .001, chi square tests using COVID and crime item totals, either including or excluding outliers). Public media organization WHYY and its digital-only publication Billy Penn show either higher than average or nearly average ratios of pandemic to crime coverage (ratios of 4.7 and 20.4, respectively). Most nonprofits give proportionally low attention to crime, though the Inquirer, which has a hybrid ownership structure as a public-benefit corporation owned by a nonprofit, is near the weighted system average (ratio of 2.8) and its Daily News edition is below the average (2.4). The stock market-traded form of ownership is largely represented in our study by the affiliates of CBS, ABC, NBC, and Fox, which have a low average ratio of COVID-19 to crime coverage (average ratio of 0.9 for outlets owned by stock-market traded companies, vs. 11 for all other outlets; p < .001, chi square tests, using COVID and crime item totals).

With the exception of magazine-style outlets attracting affluent audiences (e.g., Philadelphia Business Journal, Philadelphia Magazine, and Philadelphia Weekly), commercial, privately held outlets tend to show low ratios of COVID-19 to crime coverage. This includes the Delaware County Times, Metro (Philadelphia), and digital-only Philadelphia Patch, but not Philadelphia Tribune, which largely serves Philadelphia’s Black community. Analysis of random subsamples of stories shows that COVID and crime items in the Tribune often are wire service stories that raise racial justice concerns, and the Tribune’s COVID-to-crime coverage ratio may be higher than indicated in Table 3.

The Inquirer often covers crime against the backdrop of concerns over high homicide rates in Philadelphia. For example, a short item (292 words) published August 18, 2021, reports on a shooting death and notes in the final paragraph that Philadelphia and other cities have seen “spikes in homicides and shootings” (McCarthy, 2021). Reporting on rising homicide rates arguably meets a community’s critical information needs, but in this case and in many others the focus remains on short, breaking news-style coverage of crime incidents rather than on larger structural issues of inequality and justice, even as the Inquirer has attempted to move away from brief, “ledger” style crime coverage (Monroe & Wenzel, 2021).

Discussion

The patterns that emerge from our data paint a multi-dimensional portrait of Philadelphia’s media system that strongly suggests that differences in audience size and staffing levels, ownership structures (commercial vs. noncommercial), and format (print vs. television vs. digital-only) together work to underserve socioeconomically marginalized populations. Though our findings are limited to a subset of news outlets that serve the greater Philadelphia area,⁷ and we cannot calculate income, education, and age indices for the city’s major television affiliates due to lack of available data, these media system patterns support our finding, in answer to RQ1 concerning the relationship between news provision and audience socioeconomic status, that socioeconomic divides are potential sources of systemic news gaps.

The system on average leans toward serving audiences with higher income and education and disproportionately older than 34. Local television networks are positioned to provide a counterweight to this overall pattern, as prior research has found that television draws audiences with lower income and education levels (Althaus & Tewksbury, 2007; Althaus et al., 2009; Bucy, 2001). However, even if such a counterweight exists, it raises questions about divides related to the different ways that television, print, and digital media cover the news, and future research should assess the degree to which local TV stations provide in-depth public affairs coverage and analysis compared to an outlet such as the Inquirer with an editorial staff of more than 200 people. Perhaps offering an additional counterweight, many niche news and information sources in Philadelphia specifically serve marginalized communities, such as Impacto, Al Día, Philadelphia Tribune, and Telemundo and Univision affiliates. However, these outlets generally do not have the same journalistic resources in terms of staffing levels as Philadelphia’s dominant news sources.

In answer to RQ2 concerning news provision by digital-only outlets, these more recent entrants to the Philadelphia media system appear to largely mirror the longstanding socioeconomic divides of traditional news media, as they reach audiences that, in the aggregate, have income, education, and age indices that exceed Philadelphia medians and averages. Additionally, the small staffing levels at digital-only outlets with lower income and education indices represent an additional inequality in Philadelphia’s media system (Table 2). Though these findings are limited to a small number of digital-only outlets due to data availability issues, they raise a concern, deserving of further research, for what might be termed “digital news deserts” in which online-only outlets devote their already small staffs to reaching the same higher socioeconomic audiences served by large, legacy outlets and online-only outlets reaching lower socioeconomic audiences lack the capacity to significantly change this broader trend. Consequently, lower socioeconomic audiences may be further underserved by online news, especially considering a long-standing digital divide that disproportionately excludes disadvantaged communities.

Regarding RQ3 concerning how critical information needs are met by Philadelphia news outlets, our study shows that in the aggregate, COVID coverage eclipses crime coverage. However, when assessed at the outlet level, we find pronounced inequity in the provision of such coverage along socioeconomic markers. Among outlets included in our content analysis, audiences that are more affluent, more educated (though correlations with education indices are borderline weak/negligible), and older receive proportionally more COVID-19 pandemic coverage while audiences that are less wealthy, less educated, and younger receive news coverage with a stronger focus on often individualized, breaking-news crime. Local TV news, where we would expect to find lower income and education indices (though we lack the data to produce these indices), does not appear to help close these gaps, as most COVID-to-crime news ratios at local TV affiliates are low, occasionally even negative. These findings are limited to database search results and to a subset of outlets in our study, most of them with education, income, and age indices above 100. Income and education correlations do not achieve statistical significance at p < .05, which should be taken into account when extrapolating findings beyond included outlets. However, correlations align with prior research finding associations between socioeconomics and local news provision (e.g., Napoli et al., 2015; Stonbely et al., 2019). Future research should include manual coding of content and additional outlets, if data are available. Future research also should assess how journalists’ access to data and the availability of news sources impact these news production levels.

Additionally, though we argue that during the pandemic a broader range of critical information needs are met by COVID coverage than by coverage of individual crime events, we acknowledge that the importance of COVID news versus crime news is an open question. If we set aside a dichotomous relationship implied by COVID-to-crime ratios and simply assess raw percentages of COVID and crime items (i.e., percentages of COVID and crime items out of all items produced by each outlet during our content analysis time period), correlations with audience socioeconomics show similar results: More COVID coverage is associated with higher income levels (income index and COVID correlation of 0.33, p = .27) and more crime coverage is associated with lower income and education levels (negative correlation of −0.489, p = .09, for crime and income indices; −0.301, p = .34, for crime and education indices).⁸ Thus, our study indicates that socioeconomic statuses of audiences are associated with inequities in fulfillment of potentially critical information needs. This is a cause for concern and further research regardless of any assessment of how different news topics represent critical information needs.

In answering RQ4 concerning ownership impacts on critical information needs, we find that public-funded and nonprofit media produce more COVID-19 coverage relative to crime stories than their commercial counterparts. Hedge fund ownership, often criticized for treating news holdings as primarily vehicles for profit rather than public service (Abernathy 2016; Kuttner & Zenger 2017), and other private investment entities are associated with low ratios of COVID-19 to crime coverage: Alden Global-owned Delaware County News is among outlets giving far lower than average attention to the COVID-19 pandemic relative to crime (1.6); Patch, owned by Hale Global, a self-described investment holding company specializing in turning around troubled companies (Bercovici, 2014), gives nearly equal attention to crime as it does to COVID-19 coverage (1.2). Stock market-traded ownership, a commercial form of ownership prevalent among major TV affiliates, also is associated with far lower than average ratios of COVID-19 to crime coverage and, at times, levels of crime coverage that are greater than COVID-19 coverage (WTXF/Fox 29, ratio of −3.2).

Conclusion

Despite what might appear to be a robust and thriving media system, Philadelphia’s patterns of news provision indicate informational inequalities along socioeconomic lines. Our study, which builds and applies a multi-dimensional framework for analyzing local media systems, shows that news gaps can arise when some news outlets produce substantial coverage for some audiences while neglecting others and when they provide overall inadequate coverage of vitally important issues related to public health and other critical information needs. These gaps seem particularly pronounced when commercial outlets attempt to draw audiences with crime coverage that focuses on individual incidents rather than on the broader social and economic factors that fuel crime.

When it comes to fulfilling communities’ critical information needs—a fundamental prerequisite for pluralistic democratic participation—our findings show that less wealthy and younger audiences are underserved compared to wealthier residents. This analysis also demonstrates that public media offer proportionally more critical needs coverage, while commercial news outlets, particularly those owned by private equity/hedge fund investors and stock market-traded companies, compound these patterns. Although hyperlocal, niche outlets have the potential to address some of the ensuing gaps, these are often relatively under-resourced news organizations, raising questions about their long-term sustainability.

Policy interventions with the potential to close news gaps for underrepresented communities should take account of systemic market failures that tend to under-produce certain kinds of news and information that are less likely to generate revenues but that the public still needs (Picard & Pickard, 2017). Encouraging diverse forms of ownership—especially public and nonprofit media institutions—has the potential to shield news organizations from the economic incentives to bait their audiences with sensationalistic coverage of crime and other topics. Various kinds of media subsidies could increase staffing levels at outlets serving marginalized communities, both across the city and at the hyperlocal, neighborhood level, and increase the racial and gender diversity of staffing at dominant news organizations.

Calling for more public funding to reduce commercial pressures on struggling news organizations and expanding the budget for public media is a challenging reformist project. The United States stands out globally for how little it invests in its public media, amounting to merely $1.40 of federal spending per person, which is a fraction of what most other developed democracies around the world spend on public media (Neff & Pickard, 2021). As economic stability and outlook worsen for local news outlets for the foreseeable future, nonmarket means of support will likely become even more crucial in the years ahead.

Data Availability

The data underlying this article are available in the article.

Conflict of interest: The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Timothy Neff’s and Pawel Popiel's postdoctoral research fellowships were supported, in part, by the Independence Public Media Foundation.

Acknowledgments

The authors thank Sarah Stonbely at Montclair State University’s Center for Cooperative Media for sharing data central to this research. We also thank our reviewers for their insightful comments and criticisms, which were invaluable in revisions to this article.

Footnotes

References

Appendix—Content analysis method

For our content analysis generating ratios of COVID-19 coverage to crime coverage, we follow the rigorous method outlined by Stryker et al. (2006) to develop and validate search term strings that demonstrate high levels of precision in surfacing this coverage. First, we developed decision trees to guide our discussion of the types of stories to include and exclude from our database searches. Next, we conducted intercoder reliability tests to ensure that we agreed on rules for inclusion and exclusion. A final reliability test on 300 stories published by our outlets from the Factiva and NewsBank databases and including two coders (co-authors) resulted in strong intercoder agreement scores for COVID-19 stories (0.907 Holsti’s method; 0.812 Krippendorff’s alpha) and for crime stories (0.947 Holsti’s; 0.837 Krippendorff’s).

We then developed “open” and “closed” search strings to calculate precision and recall rates. As explained by Stryker et al. (2006), recall measures the degree to which a search string includes all relevant stories; precision measures the degree to which a search string includes only relevant stories. An open string attempts to gather all relevant items by opening the search as widely as possible, which also will gather many irrelevant items (high recall, low precision). A closed string, which is the string used in our final database searches, will filter out most irrelevant items (higher precision than the open string, though often at the expense of slightly lower recall).

As our content analysis does not involve human coding, precision is the most important measurement for assessing our database search strings (Stryker et al. 2006). Therefore, we sought to achieve at least 90% precision and 80% recall for our search strings (with a 5% margin of error). Search string testing involved a full year’s worth of Philadelphia Inquirer crime stories (4,060 stories gathered by our open search string) and a random sample of Inquirer COVID-19 stories (710 stories; random sample used in accordance with Stryker et al.’s method due to the large number—more than 10,000—of potentially relevant items found by our open search string).

Our final, closed search string for crime stories achieved 91% precision and 87% recall. Our final, closed search string for COVID-19 stories achieved 86% precision and 80% recall. COVID’s relevance for a broad swath of news topics creates many borderline cases of relevance that reduce search precision. Both search strings exclude sports-related items; the crime search string excludes stories related to Black Lives Matter and related protests over incidents of police killings, as well as stories related to the January 6, 2021, riot at the U.S. Capitol.

To adjust for less than perfect precision and recall, we follow Stryker et al.’s method of using recall and precision rates to produce a coefficient to apply to the number of COVID-19 (multiplied by coefficient of 1.07) and crime stories (multiplied by coefficient of 1.05) surfaced by our searches to produce an estimated total number of stories in each category of coverage for each outlet during our search period. Dividing adjusted numbers of COVID-19 stories by adjusted numbers of crime stories generates a ratio of COVID-19 to crime coverage for each outlet. A higher ratio indicates more coverage of COVID-19 vs. coverage of crime.

Final search string for COVID-19 coverage:

(headline=(covid* OR coronavirus* OR pandemic* OR vaccin* OR virus* OR Wuhan OR Hubei OR CDC* OR quarantin*) OR lead=(covid* OR coronavirus* OR pandemic* OR vaccin* OR virus* OR Wuhan OR Hubei OR CDC* OR quarantin*)) NOT (headline=(Eagles OR Flyers OR Phillies OR Sixers OR 76ers OR NCAA OR MLB OR NHL OR MLS OR “NFL” OR “soccer federation” OR “minor league” OR “major league” OR “happy valley” OR PSU OR golf OR soccer OR football OR “Penn Relays” OR Owls OR “auto racing” OR NASCAR) OR lead=(Eagles OR Flyers OR Phillies OR Sixers OR 76ers OR NCAA OR MLB OR NHL OR MLS OR “NFL” OR “soccer federation” OR “minor league” OR “major league” OR “happy valley” OR PSU OR golf OR soccer OR football OR “Penn Relays” OR Owls OR “auto racing” OR NASCAR) OR obituaries OR “sound off” OR “letters to the editor” OR “clarifications” OR “clearing the record” OR “Dear Abby”)

Final search string for crime coverage:

((“police said” OR “police spokesperson” OR “detectives said” OR “investigators said” OR “authorities said” OR “according to investigators” OR “according to police” OR “according to the police” OR “prosecutors said” OR “was charged with” OR “is charged with” OR “arrested and charged” or “was accused of” OR “is accused of” or “was arrested for” or “on charges of” or “cops id” or “police id”) OR headline=(stabbing* OR stabbed OR “aggravated assault” OR assault* OR murder* OR manslaughter OR rape* OR robbery OR robbed OR killer OR attacker OR shooting* OR gunshot OR “gun shot” OR knife or burglar* OR vandal* OR detectives OR embezzle* OR theft OR stole* OR slain OR homicide* OR felon* OR misdemeanor*)) NOT ((Eagles OR Flyers OR Phillies OR Sixers OR 76ers OR NCAA OR MLB OR NHL OR MLS OR “NFL” OR “soccer federation” OR drown* OR thunderstorm OR tornado OR “house fire” OR “sound off” OR “college basketball” OR “letters to the editor” OR “police brutality” OR “Black Lives Matter” OR “George Floyd” OR “Breonna Taylor” OR “Daunte Wright” OR “Rayshard Brooks” OR “Daniel Prude” OR “Walter Wallace” OR clarifications OR legislat* OR editorial OR “live life love” OR politics OR sports OR “Dear Abby” OR capitol OR “tear gas” OR “*plane crash” OR “President Trump” OR “Donald Trump”) OR headline=(coronavirus OR covid OR virus OR vaccin*))