Abstract
News outlets are increasingly nationalizing their presentation of news stories, framing and presenting local news in a broad national context. We investigate how supply-side and curation-side factors of the news cycle contribute to the nationalization of news coverage. Through the computational analysis of 1.05 million Google news results on four days in July–August 2020, that corresponded to 1,581 news stories published on the George Floyd protests in Portland, Oregon, and Kenosha, Wisconsin, we examine the relationship between the nationalization of news coverage, stories’ search rank in Google News, and the geographic distance between the news event and the stories’ reading audience. Further, we explore the role of Google News in curating locally focused news. Our findings help to map the media ecosystem in a digital age, highlighting the influence of algorithmic power in politics and showing that excessive circulation of national news may have a profound negative impact on news diversity and social justice.
Politics in America is increasingly nationalized. One often-cited reason for this trend is that readers turn to national outlets, like The New York Times or The Washington Post, when local news outlets shut down. However, we explore another possibility here: that even local news is becoming nationalized.
News nationalization refers to the process by which news outlets frame, present, and discuss news stories within a broader national context rather than focusing on the local aspects of the events. This phenomenon often involves prioritizing national perspectives, political figures, and narratives over the local concerns, histories, or nuances specific to the event’s location (Heider, McCombs, and Poindexter 2005; Moy et al. 2004). Various factors, including the media landscape, audience preferences, and the political or social context, can influence the nationalization of news. For example, the pool of potential readers for an online news source is far broader than that for a print publication circulated within a certain region. To compete for a wider audience, local outlets may be more likely to adopt national frames to appeal to a broader range of readers, even at the expense of the local context (Mullainathan and Shleifer 2005).
In the U.S., local politics tend to be more consensual and less influenced by partisan divides than national politics (Bucchianeri 2020; Holtkamp 2007; Jensen et al. 2021). Therefore, news nationalization helps readers relate local issues to broader political and social trends nationwide (Ferrucci and Perreault 2022; George and Waldfogel 2006). However, news nationalization may also have adverse consequences. For instance, nationalized news reporting may overshadow local perspectives and may potentially diminish the diversity of viewpoints available at the local level (Harte, Howells, and Williams 2018; Janowitz and Burk 1991). Furthermore, as nationalized news frames are typically hyperpartisan (Hopkins 2018), readers may become more affectively polarized (Abramowitz and Webster 2016; Darr, Hitt, and Dunaway 2021; DellaPosta 2020). Consequently, there may be adverse effects on their political attitudes and voting behavior, such as lower rates of voter turnout and of split-ticket voting (Darr, Hitt, and Dunaway 2018; Schulhofer-Wohl and Garrido 2013). In addition, local politics may take on hyperpartisan narratives in keeping with local news trends, as witnessed in how local politicians shift their focus from appealing to local communities to addressing a broader national audience (Moskowitz 2021; Trussler 2021; Warshaw 2019).
Understanding the dynamics of news nationalization is therefore vital for assessing its potential impact on public opinion, civic engagement, and the overall health of the media landscape. Prior research has primarily focused on demand-side factors driving news nationalization. However, a comprehensive understanding of news nationalization requires an examination of both news creation, such as the role of editorial choices, and news curation, such as the role of news aggregators in making news available.
Nationalized local news is a byproduct of the supply-side forces, including the role of media systems and the local environment in shaping news content. Local news outlets play a vital role in giving citizens access to information that enables them to participate in local political and civic affairs (Napoli, this volume). However, in recent years, local news outlets are increasingly pursuing a national agenda in their local news coverage (G. J. Martin and McCrain 2019), potentially impacting the availability of this information. Moreover, according to Hopkins’s (2018) “politicized places” hypothesis, the regional aspects of supply-side factors, such as a community’s demographics or political involvement, might affect outlets’ preference for national versus local news, making some communities more susceptible to the negative impact of news nationalization (Guskin, Moore, and Mitchell 2011; Posey, this volume).
Digital news aggregators hold considerable power in prioritizing national news content for consumers. For example, when they curate local events, they may favor nationalized coverage and deprioritize stories that originate from local news outlets (Fischer, Jaidka, and Lelkes 2020; Nechushtai and Lewis 2019). Because news aggregators exercise some control over how people access news online, it is crucial to understand how this news curation affects representation of communities and perception of political issues.
Building on these research gaps, we offer a methodological and theoretical contribution to how news nationalization affects news coverage. We propose a new method for measuring the extent of nationalized reporting through the local and national entities mentioned. Our theory extends the politicized places hypothesis (Hopkins 2018) to consider the role of geographic and structural factors in determining whether local preferences can influence news coverage. We focused on the nationalization in news coverage of the 2020 protests in Portland, Oregon, and Kenosha, Wisconsin, to examine how local conditions and national politics might shape news coverage of an event. In this context, our analysis highlights the role of news creators and curators in the media ecosystem and how they affect the complex relationship between the nationalization of news coverage, search rank, and the geographic distance between the news event and the audience. Further, our study of news coverage of the Portland and Kenosha protests gives a better understanding of the implications of news nationalization for individuals, communities, and the broader American society.
Our analysis of a data set of Google News search results sampled over one month suggests that coverage of the Portland and Kenosha protests were heavily nationalized. However, news outlets outside of urban areas were less likely to publish nationalized coverage. Furthermore, an analysis of Google News rankings suggested that its search curation may amplify the effect that distance and political ideology have on the availability of locally focused news. The findings have implications for how news nationalization affects local news reporting. They also highlight how algorithms can ameliorate or amplify news echo chambers by limiting access to diverse perspectives.
Literature Review
Nationalizing local news could have individual, county, and regional consequences. At the individual level, if local sociopolitical issues are increasingly interpreted through a hyperpartisan, nationalized lens (DellaPosta 2020), then local readers are less aware of their civic and economic ties to rural areas, leaving them with fewer opportunities to find common ground with others who seem physically and ideologically distant. A nationalized discourse promotes an understanding of political events as partisan narratives devoid of local context. Consequently, the nationalized framing of local news could increase affective polarization (Abramowitz and Webster 2016; Levendusky 2010; Prior 2010, 2013), as readers focus on partisan cues to process information rather than local perspectives (Darr, Hitt, and Dunaway 2018, 2021). For instance, Hopkins (2018) reported that native-born white and Black Americans are more likely to take on anti-immigrant views in response to a local influx of immigrants if the news reports immigration as a salient national issue.
At the county level, readers might change their political behavior because they do not have complete information on how issues affect their local communities or livelihoods (Darr, Hitt, and Dunaway 2018; Hopkins 2018). They may even drop out of political participation altogether (Hayes and Lawless 2018). At the regional level, the lack of local reporting could exacerbate social and economic inequality in the U.S. by expanding its news deserts, which are the vast swaths of the nation with no local reportage (Abernathy 2018; Fischer, Jaidka, and Lelkes 2022).
The nationalization of news is occurring amid major changes in American politics and society. With the proliferation of digital media in the first two decades of the 21st century, Americans experienced a dramatic reduction in the cost of accessing information and media. With these reduced costs, it would have been plausible for Americans to develop a preference for hyperlocal news outlets (Hopkins 2018). Instead, Americans focused chiefly on a few large, nationally focused outlets (Hindman 2009, 2011). The following section discusses our study of the news creation and curation factors affecting nationalization in the coverage of news events.
Factors affecting local news production
A news reporting hierarchy refers to the taxonomy of media outlets that cater to different audiences and cover issues at various geographical scales, from local newspapers to national TV networks. Thinking of news reporting as a hierarchy helps us understand how media coverage can include local, regional, and national interests, ensuring that these interests are cross-cutting and safeguarding against the formation of polarized factions—monoliths that share attitudes and identities (Kernell 1986). However, polarized factions may still emerge because the inequity in resources and attention favors national news outlets. Because they have more resources, they may be better able to provide high-quality coverage that includes diverse perspectives and in-depth analysis. However, in doing so, they inadvertently shift the public’s focus to national issues and potentially diminish local media’s role as a source of information and accountability for local communities (Min et al. 2022; Moy et al. 2004).
The shift toward nationalization could also be occurring because of a decline in local news (Wahl-Jorgensen 2019), including the closure (Abernathy 2018) or hollowing out of existing local news outlets (Peterson and Dunaway, this volume), which is occurring amid an increase in national news production and availability (Darr, Hitt, and Dunaway 2018; Moskowitz 2021). Even the local outlets that remain may have adopted a nationalized focus in order to compete online (Djourelova, Durante, and Martin 2021) or favor investors who focus more on profit than on the news (Peterson and Dunaway, this volume). Furthermore, as readership moves primarily online, local news is no longer tethered by geographic market boundaries and demographics. Therefore, outlets may see the benefit of framing local stories from a hyperpartisan perspective that would be intriguing to a national readership. The change in reporting style may also help them to compete for advertisers.
In summary, the changes in local news outlets and the shifts toward nationalized coverage more broadly have fomented a drop in the quantity and quality of local news available, meaning that communities lack the types of information essential to local information needs (Napoli et al. 2017). However, even with the growing understanding of what local news and, inversely, its loss means for communities in terms of civic and political health, we have limited empirical evidence regarding news nationalization. We, therefore, pose the following research questions:
Research Question 1 (RQ1): What is the prevalence of national versus local news entities in all the local news coverage available through Google News?
RQ2: What county-level features of a news outlet predict higher nationalization in its news coverage?
Another factor in the news cycle is curation by a third-party news aggregator such as Google News, Apple News, or Facebook. News aggregators act as a central pathway for information access, providing access to news articles within online services and platforms that readers are already familiar with (Guess 2021). However, when aggregators rank national outlets over local ones, it becomes more challenging for consumers to find and engage with content from local outlets.
In prior work, we have reported on how search algorithms in news aggregators like Google News may widen the gap in the reach of national versus local news outlets. Through an analysis of search results from different counties, we reported that Google News prioritizes the coverage of national outlets over that of local outlets in general and locally oriented search queries (Fischer, Jaidka, and Lelkes 2020, 2022). Google News often did not feature local news in prominent positions when returning search results, a finding that reiterates prior audits of news search results (Bandy and Diakopoulos 2020). However, we needed to examine whether simply increasing the diversity of outlets in the search results is sufficient to address the representation gap and whether the news provided by local outlets is relevant to local communities. To accurately assess the role played by third-party digital news platforms in news nationalization, it is necessary to audit the semantic characteristics of local news reports across different news outlets. Therefore, we posit the following research questions:
RQ3: To what extent are search algorithms in Google News privileging national news entities over local news entities?
RQ4: What county-level features predict Google News serving consumers nationalized coverage in its search results?
By answering these questions, we hope to better understand the different facets of news nationalization, such as the supply-side factors affecting local news production and the role of third-party digital news platforms in exacerbating this trend.
Protests, near and far
To identify the correlates of news nationalization, we examined the coverage of the same events by different outlets, holding the topic constant but observing variation in how outlets reported on the event in local or national terms. We focused on the news coverage of two major protests in Portland, Oregon, and Kenosha, Wisconsin, in July and August 2020, which occurred during a time of racial reckoning in the U.S. following the murder of George Floyd in Minneapolis. While the Portland and Kenosha protests had substantial backstories rooted in local history, they garnered national attention as the events grew in size and scale. They also gained international attention as issues of racial justice were addressed by political parties and voters across the political spectrum. This made the events an ideal testbed to understand the competing interests in the news hierarchy because local outlets competed for a wider audience when national audiences paid attention to the protests.
These events also allow us to validate the nationalization hypothesis, which would anticipate that even local-to-the-protest news outlets may privilege national frames over local frames. Even so, we expect that they may be less nationalized than other outlets.
Methodology
Our goal is to examine the factors predicting the choices made by local news outlets in covering local events of national importance, as well as whether these factors affect the consumption of news by readers who use news aggregators. First, we calculated the nationalization of news based on the prevalence of national over local entities mentioned in individual headlines. Next, we explored the nationalization of news coverage of the Portland and Kenosha protests across different types of news outlets. Finally, we considered the impact, if any, that news aggregators have on exposure to national and local news through curation of search results relating to the protests.
Data collection
We followed an established approach for auditing Google News results that has been used in prior studies of search results (Fischer, Jaidka, and Lelkes 2020; Robertson, Lazer, and Wilson 2018). For one day each week in July and August 2022, we repeatedly queried Google News for news on “protest near.” We scraped the top 100 results returned each time for 3,142 county or county equivalents in the U.S. To mitigate biases created through cookies and possible ordering effects, we collected the search results 12 times—three times each on July 22, August 7, August 14, and August 28. We also used the headless Selenium browser to “browse” Google News through an incognito Google Chrome window each time so that no user cookies would be stored that would bias the results. Finally, we randomized the query order each time to avoid any ordering effects repeated querying may introduce.
In this manner, we queried Google News a total of 37,704 times. We collected 1.36 million results comprising the top 100 ranked results from 396 news outlets, as reported in Table 1. Table 2 reports the most and least frequently occurring news outlets in the data set. Next, we filtered the entire data set to retain only the news headlines referencing the protests, leaving us with 1.05 million results. Finally, deduplicating news headlines left us with the headlines data set comprising 1,581 unique headlines.
Descriptive Statistics of the Protest Data Set
Entity Statistics Identifying (a) the Types and Most Frequently Encountered Entity Types and (b) the Most Frequently Occurring Local and National Entities
Content analysis
To study the nationalization of news, we used a combination of automatic and manual content analysis. We first extracted mentions of all people and organizations in news headlines (Moskowitz 2021). Next, we separated these mentions into local and nonlocal entities. Then, to understand how the news media prioritize one kind of news coverage over another (Wells et al. 2020), we calculated news nationalization scores for each headline as the relative advantage of national entities over local entities. Finally, we calculated each outlet’s average news nationalization score based on all its headlines. The entity extraction and classification approach details are reported in the online appendix.
Analytical approach
We conducted a pair of regression analyses. First, to examine the effect of supply-side forces on news nationalization, we regressed a news headline’s nationalization score based on the distance from the county publishing the news to either Portland, Oregon, or Kenosha, Wisconsin, and a set of county-level covariates capturing the socioeconomic and ideological characteristics of the county. Second, to examine curation effects, we regressed the headline’s nationalization score on its position in the search results for a county and the distance from the searching county to either Portland, Oregon, or Kenosha, Wisconsin, once again including county-level covariates about the searching county. Finally, we examined the interaction between the rank and the distance after including county-level random intercepts to account for unobserved heterogeneity. We anticipated that collecting data on different days would introduce systematic bias; therefore, in all models, we included random effects for the day of data collection.
Findings
Entity distribution
Table 2 reports entity statistics at the word and headline levels. Table 2a shows that the most frequently encountered entities refer to common nouns, followed by organizations, and then named individuals. The headlines most frequently referenced protesters and the police. Terms also included references to other significant figures, such as the mayor of Portland. References were also made to a veteran who was attacked by the police during the protests and a teen who shot and killed a protester. When these entities were aggregated into headline-level statistics, we observed that they had more national entities on average, as reported in panel b of Table 2. Figure 1a and 1b visualize the most frequent entities labeled as local or national that occurred at least twice in the data set after dropping common words and stemming similar entities (e.g., protester and protesters). In Figure 1, the font size of the word is proportional to its frequency in the data set. Examples of how the nationalization ratio was calculated are provided in the online appendix.
Word Clouds Depicting the Most Frequently Occurring (a) Local and (b) National Entities in the Collected News Headlines Data Set
Figure 2 explores news nationalization at the supply level in terms of the socioeconomic environment of the reporting outlets. Information about the publishing county was obtained by crowdsourcing annotations about the headquarters of all the news outlets in the data set and cross-referencing this information with socioeconomic indicators and voter demographics from the U.S. Census Bureau. We observe that the distance from the protest did not significantly predict nationalization. News outlets from counties that lean to the right (b = 19.21, standard error = 7.7, p < .05) had a higher nationalization in protest coverage. On the other hand, the news coverage from outlets in counties with a lower median household income (b = −0.37, standard error = 0.126, p < .01) was less nationalized. At the same time, the effect of the population was marginally significant (b = −6.33, standard error = 3.08, p < .10). Taken together, these results suggest that news outlets from urban counties were less nationalized in their coverage of protests.
Examining News Nationalization at the Supply Level in Terms of the Socioeconomic Environment of the Reporting Outlets
News nationalization in news curation
Next, we examined the extent of news nationalization in Google News search results. Accordingly, we calculated the median nationalization score in every county’s top 10 Google News results. Figure 3 depicts these scores in a choropleth visualization. The deeper shades across most U.S. counties show that in most counties, the top search results about the Portland and Kenosha protests contained more nationalized coverage than local coverage.
The Median News Nationalization of the Top 10 Search Results Returned by Google News
Figure 3 obscures the dynamism of news results and news events. For instance, more local news coverage could be deeper into the search results. Furthermore, as the protests continued over weeks and months, news outlets may have changed their coverage style to reflect the evolving on-ground realities. Therefore, in Figure 4, we have plotted the average ratio of national references to local ones over all the search results in each of the four weeks. The four plots show whether news nationalization across search ranks varied over time. For example, in our first capture of content related to the Portland protests on July 22, the most prominent results (i.e., those with low search ranks) were markedly nationalized in their coverage. By contrast, the overall levels of nationalization were lower in the three subsequent capture windows. Even so, in the data captured on August 7 and August 14, the most prominent results with the lowest search rank still had higher levels of nationalization than the results that followed. In the final capture window on August 28, when we collected stories related to the protests in Kenosha, the pattern reversed, with the most prominent stories returned being more localized in their coverage than those following them.
Average Nationalization of Stories Returned by Google News Based on Their Rank in the Search Results
To calculate the effects of rank and distance on the nationalization ratio, we rescaled the dependent and independent variables as percentages of the maximum value. For instance, the nationalization ratio was transformed into a percentage of its maximum value, that is, when national entities were represented in the headline precisely four times more than local entities. The results of the regression analysis are reported in Figure 5. From this plot of the model coefficients, we can see that our variables of interest, search rank, and the county’s distance to the event are significantly associated with the nationalization of the returned stories. For example, stories with a higher search rank—those that appear lower in the results produced by Google News—have lower nationalization scores (b = −0.12, standard error = −0.77e−3, p < .001). This implies that a 1 percent increase in rank reflects a 0.12 percent decrease in the nationalization ratio.
Regression Coefficients for Predicting the Nationalization of Local News in News Curation by Google News as a Function of the Search Rank and the Distance to the Protest Event
On the other hand, the Google News results farther from Portland or Kenosha had higher nationalization scores than those for counties closer to those communities (b = 0.89, standard error = 1.91e−3, p < .001). This implies that a 1 percent increase in the distance to the county reflected a 0.89 percent increase in the nationalization ratio. The model included many sociodemographic variables and suggested that a higher nationalization ratio was significantly positively correlated with many of the markers of rural counties, for example, a higher rural population (b = 0.26, standard error = 1.00e−2, p < .001) and a lower population density (b = 0.03, standard error = 1.00e−2, p < .05). In addition, counties with a higher voter turnout percentage (b = 0.03, standard error = 3.80e−3, p < .001) and counties with a vote margin percentage in favor of the Republican Party (b = 0.03, standard error = 3.80e−3, p < .001) were more likely to see more nationalized coverage in their search results. However, these effects may result from other endogenous variables and should be interpreted cautiously.
Table 3 reports the interaction of these measures of interest after controlling for random effects due to the day of data collection and the within-county differences. We find that the interaction effect is also significantly associated with the nationalization of returned stories (β = −0.45e−3, standard error = 0.001e−3, p < 2e−16); that is, in counties farther from the protest, higher-ranked news headlines were more nationalized than in closer counties, while a focus on local entities was buried deeper in the results. Figure 6 breaks down the marginal effect of rank on the nationalization ratio across different distances between the county and the news event. Distinguishing these interactions for within-state and within-region counties, we observe that while the trends are similar across all the counties, in the search results about the protests in the first 50 ranks, the nationalization ratio was lower in the search results for counties within the same state. Moreover, this ratio fell more gradually compared to counties within the same region.
Main Effects and the Interaction Effect of Rank and Distance from the Event on the Nationalization of News Headlines
*p < .05. **p < .01. ***p < .001.
The Conditional Effect of the Average Rank of a Headline in a County on the Nationalization of Local News in News Curation by Google News, as a Function of the County’s Distance to the Protest Event, after Including Random Date and County Effects
Discussion
Our study examines the influence of news creation and curation practices on the nationalization of local news. Notwithstanding the decline in local news outlets, the number of outlets in an area may be an imperfect indicator of whether a community has access to the types of local information that experts deem essential for civil health. Furthermore, it is increasingly important to understand the pathways through which readers are exposed to different news frames.
We propose expanding the definition of “news deserts” to consider whether communities receive quality localized news coverage, rather than merely measuring the quantity of local news outlets. Our analysis of over 1,500 news headlines about two protests in the U.S. found that all headlines, including those local to the protests, were more likely to be nationalized than localized. Our findings corroborate research on an increasingly important element of the study of news production and its effects: place (Usher 2019, 2021). News outlets published in urban areas were more likely to focus on local entities in their protest coverage, possibly because they had more resources than outlets published in rural areas. Journalistic resources are increasingly concentrated in major cities due to economic forces and the increasingly digital nature of editorial work. The onus of providing local perspective falls on outlets with more resources, which are more likely to be national in scope and focus.
While localized coverage is inequitably produced, it is also inequitably served to its consumers. Our findings showed that news curation by Google News exacerbated the effect of distance as urban counties (Branton et al. 2007; S. R. Martin 1988) and places farther from Portland or Kenosha received less localized and more nationalized frames in the top-ranked search results for queries about the protests. On the other hand, the localized coverage of the protests surfaced higher in the returned results in counties closer to the protests.
These patterns of search results may match the expectations of relevance and the preferences and demands of consumers in different geographies and points in time. However, the discrepancy between the availability of localized coverage and the delivery of localized coverage via a news aggregator like Google News demonstrates a critical avenue that can reinforce the polarizing effects of nationalized news coverage.
The attention economy, as applied to news media, refers to the competition among news outlets to capture and retain the attention of audiences in an era where information is abundant but attention is scarce (Myllylahti 2018). In this landscape, news content is often tailored to be more sensational, clickable, or shareable to attract viewership and engagement. The insights from our analysis here complement prior work that examines how algorithms managing the attention economy (i.e., commodifying people’s attention) may reinforce offline realities. Our prior research found that Google News prioritizes national over local outlets (Fischer, Jaidka, and Lelkes 2020, 2022). In this study, we show that they may also need to consider the importance of local information and context provided alongside national coverage. National debates are often intertwined with local issues. For example, the catalysts for the Black Lives Matter protests were localized events, particularly the murder of George Floyd by the Minneapolis Police Department, an event steeped in the local history of Minneapolis (Sood, Speagle, and Ehrman-Solberg 2019). Even policies under national debate, such as gun regulation or tax policies, have primarily localized effects across the country.
Nationalized framing of events, such as the Portland and Kenosha protests, can lead to biased perspectives that lack local, racial, and social context (Dixon 2008). For instance, the danger in covering the protests from a top-down view, as The New York Times did (Mansoor 2020), is that it could promote the claims of elite actors or institutions (Michel 2020), even when their claims are inaccurate and used to justify social code violations. Furthermore, a characterization of the Portland and Kenosha protests as a riot (Shaw 2020) or a conflict with federal officers (Barker, Baker, and Watkins 2021; Owen, Barr, and Date 2020) advocates a prejudicial perspective devoid of local, racial, and social context (Hoffman 2022). Instead, reporting that is sensitive to local events could clarify how the escalation of conflict in Portland was directly related to the local history of racism and police brutality (Associated Press 2020; Levinson, Wilson, and Haas 2020).
Our findings raise concerns that a national focus in the news coverage, heightened by search-engine algorithms, make it difficult for Americans to learn how the highest levels of government decision-making affect them. Citizens in news deserts face social isolation (Eubanks 2018), and there is a high variance in how and whether these scarcity effects are perceived nationwide. Limited exposure to local issues may also reinforce elite advantage at the expense of public well-being and knowledge.
A concerted effort to curate localized news coverage could help reduce electoral and affective polarization in the U.S. Localized frames can decrease the influence of partisan cues on voting behaviors and encourage split-ticket voting (Darr, Hitt, and Dunaway 2018; Hayes and Lawless 2018; Peterson 2017). Furthermore, local news coverage can reorient citizens to local events and people as well as promote engagement in local affairs and political participation. In the case of protests, localized coverage can focus on protesters’ experiences and local authorities’ appeals, helping citizens connect events to familiar contexts (Ruigrok and van Atteveldt 2007).
We offer recommendations corresponding to our findings regarding local news creation and curation. First, the problem of the scarcity of high-quality local news should be included as a news quality indicator by independent news watchdogs such as Reporters without Borders. This would create greater awareness and spur news outlets to make editorial changes to reflect the local perspective in their news coverage. Second, news aggregators may want to manage their news curation to offer more local perspectives. Finally, because of the significant interaction effect of rank and distance (possibly a proxy variable for news relevance), a delicate balance is still possible by identifying multiple dimensions of news quality and then curating news to achieve those ends. A case in point is Apple News, which combines algorithmic and editorial logic to produce a human-curated Top Stories section (Bandy and Diakopoulos 2020), which performs better on source diversity, concentration, and evenness than an algorithmic approach alone.
A limitation of our work is that while our context allows us to understand how local events can affect national news coverage, it is unclear whether our findings would generalize to other local news events. Furthermore, focusing only on four days of coverage would mean we have ignored many incidents during the protests, which could have affected our data and findings. Google News algorithms are black-boxed, meaning we do not know whether the aggregator is dynamically curating search results in response to repeated queries. Nevertheless, using incognito browser windows without cookies, repeatedly sampling the data, and spacing out our repeated data samples across four weeks could address these limitations. Furthermore, our analyses suggest that place and rank effects on nationalized news curation are stable despite these known unknowns. Although Figure 4 evidences a change in the position of local news frames in the results, our results cannot disentangle whether these were caused by a shift in news creation or Google News’ curation practices. In future work, we plan to develop our approach with more data to explore temporal changes in news narratives and news curation practices in the coverage of news events for further insight into how media outlets set the public agenda (Palmgreen and Clarke 1977).
Supplemental Material
sj-docx-1-ann-10.1177_00027162231217873 – Supplemental material for News Nationalization in a Digital Age: An Examination of How Local Protests Are Covered and Curated Online
Supplemental material, sj-docx-1-ann-10.1177_00027162231217873 for News Nationalization in a Digital Age: An Examination of How Local Protests Are Covered and Curated Online by Kokil Jaidka, Sean Fischer, Yphtach Lelkes and Yifei Wang in The ANNALS of the American Academy of Political and Social Science
Footnotes
Supplemental Material
Supplemental material for this article is available online.
Kokil Jaidka is an assistant professor at the National University of Singapore’s Department of Communications and New Media. Her research involves using big data to analyze human behavior and societal trends, with a focus on online communication and social media.
Sean Fischer is an independent researcher who earned his PhD from the Annenberg School for Communication at the University of Pennsylvania. His research has explored how cultural information contributes to cycles of political polarization. He also studies local news economies and the threats they face from digital platforms.
Yphtach Lelkes is an associate professor at the University of Pennsylvania’s Annenberg School for Communication. His research primarily revolves around partisanship, polarization, and media’s role in shaping public opinion.
Yifei Wang is a PhD student in the Department of Communication at the University of California, Santa Barbara. His research interests center on group dynamics, information behavior, civic engagement, and computational social science.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
