Abstract
Mass media can set the public’s agenda, particularly during political campaigns. In the social media era, the public can now also set the mass media’s agenda, resulting in intermedia agenda setting. This study’s purpose is to examine the intermedia agenda-setting effects between Instagram posts and mainstream newspapers during the 2016 presidential primary period. To test this relationship, a content analysis was conducted, recording the frequency of political issue mentions in newspaper articles and Instagram posts throughout the presidential primary period. Cross-correlations were then estimated to examine the direction of the influence of the frequency of issue mentions in newspaper articles and Instagram posts. Findings indicate differences between the salient issues in traditional newspapers and Instagram posts during the presidential primary. Additional results suggest a limited intermedia agenda-setting relationship between the issue agendas of mainstream newspapers and Instagram posts.
Introduction
Following the rise of the Internet and then the dawn of the social media era, the mass media landscape has vastly changed. Today, “mass media” comprises a wide variety of user-controlled, online platforms, such as Facebook, YouTube, Twitter, and Instagram, which have altered the mainstream media’s role as the sole gatekeeper of news and information. Citizens themselves can create and share content and have greater control over the content they read and watch. In many ways, citizens embrace participatory practices by writing blogs, producing podcasts, creating memes, sharing images, and shooting live video (see Porlezza, 2019). Therefore, these online platforms often rival traditional broadcast and print mediums, challenging the traditional production, presentation, and consumption of media. This raises the question: who is setting the agenda? Do traditional media sources still set the agenda for the citizenry, influencing the public’s concerns and priorities? Alternatively, in the social media era, do citizens influence the mainstream media’s agenda, encouraging professional journalists to cover specific issues? Or perhaps there is simply a weak tie between the agendas of the citizens and mass media? Indeed, the agenda-setting influence of the mainstream media on the citizenry is well known, but there is limited research on the role of social media in agenda setting. We seek to extend the body of agenda-setting research to newer, unstudied social media sites, mainly the visual-social networking site, Instagram, on the mainstream media and vice versa.
Instagram, launched in 2010, took its first leading role in US political campaigning during the 2016 presidential election, as every major candidate employed an Instagram account (Hootsuite, 2015). In January 2015, The Washington Post wrote an article titled “2016 May Yet be the First ‘Instagram Election,’” discussing Jeb Bush’s use of Instagram rather than a traditional press release, to announce his newly released Super PAC (Schwarz, 2015). Throughout the 2016 campaign, Instagram was dubbed as the new digital battleground (Frumin, 2015; Sanders, 2015), with candidates posting snapshots with family members, informative infographics, graphics of the latest polling, pictures from backstage moments at events and rallies, and videos of debates, events, and interviews (see Munoz & Towner, 2017). Undeniably, voters took notice of the visual, digital war (see Towner, 2017, 2019). During the campaign’s height, Donald Trump and Hillary Clinton’s Instagram accounts boasted about 4.5 million followers and Bernie Sanders with 2 million followers (Hendricks & Schill, 2017). Those on Instagram not only shared content from the candidates and campaigns, but they also posted their images and videos from campaign rallies and events, photographs and digital art expressing political views, and repurposed memes, GIFs, and screenshots. Thus, much of this political content on Instagram is produced and distributed by people outside of the mainstream press, challenging the agenda-setting function of professional journalists and reporters.
To examine the agenda-setting effects between Instagram and the mainstream media, this research focuses on the relationship between Instagram posts and newspaper coverage about the top political issues mentioned during the 2016 presidential primary campaign. It is possible that Instagram captions and hashtags about the candidates and campaign may influence journalists and reporters to cover top trending political topics or issues found on Instagram. The relationship could also be reversed: mainstream news covers emerging and popular campaign topics which users then post about on Instagram. Considering the latter, we argue that there will be a reciprocal relationship between Instagram posts and mainstream media coverage. Hence, this research is looking for evidence of a correlation between the Instagram posters’ issue agenda and the newspaper’s issue agenda. To do this, we coded the number of daily issue mentions in Instagram captions and hashtags as well as in newspaper articles. Then, a series of cross-correlations are estimated of the top 10 issues of the 2016 presidential primary campaign, testing whether Instagram posts are a predictor of newspaper coverage or if the relationship is reversed. Evidence of a reciprocal relationship between Instagram posts and newspaper content would suggest that the traditional agenda-setting process has diminished in the social media era (Williams & Delli Carpini, 2004).
Agenda Setting
Agenda-setting theory is a well-explored topic within media effect theories (McCombs, 2005; Rogers et al., 1993; Wanta & Ghanem, 2000). It acknowledges the important role that the mainstream media (i.e., journalists, editors, and reporters) play in helping to form and craft public agenda (McCombs & Shaw, 1972). First originated by McCombs and Shaw (1972), agenda theory demonstrated that undecided voters’ 1968 presidential election issue responses correlated with issues mentions in traditional mainstream media (McCombs & Shaw, 1972). In the years following, agenda-setting theory has been widely researched with later studies largely confirming the important role that media plays in forming citizen’s agendas (Ghanem, 1997; Kiousis et al., 1999; McCombs, 2004; McCombs et al., 2000; Weaver et al., 2004). It has also evolved to include two different levels; the first level focuses on the media informing citizens on the agenda topics, whereas the second level examines “how” citizens think about issues or candidates by focusing on the agenda’s attributes (i.e., attribute agenda setting) (Balmas & Sheafer, 2010; K. Kim & McCombs, 2007; McCombs et al., 1998). Research has shown the powerful role that mainstream media can play in influencing citizen’s political attitudes and politicians, yet the rise of new media (e.g., social media) is changing the flow of communication and thus, the direction of political communication. Citizens now can influence mainstream media.
The 1990s ushered in a new age of media communication. Emails, websites, and blogs became new communication tools for citizens to disseminate political news and communication (McCombs, 2005). In later years, the era of Web 2.0 brought with it a wave of social media tools (e.g., Facebook, Twitter, YouTube) which easily afforded anyone the ability to create and post content online and communicate electronically. The result is that the mainstream media’s agenda-setting power has potentially diminished (Groshek & Groshek, 2013; Sayre et al., 2010; Williams & Delli Carpini, 2004). The question of “who is influencing who” is not clear as digital media has become part of the media landscape (McCombs, 2005). This research employs an intermedia agenda-setting framework, where news can be now shared between traditional and non-traditional media, is increasingly occurring and will be into the foreseeable future. As McCombs (2005) correctly predicted, “Intermedia agenda setting at both the first and second levels is likely to remain high on the journalism research agenda for a very long time” (p. 549).
Intermedia agenda setting functions within a hybrid media system; a system “built upon interactions among older and new media logics” (Chadwick, 2013, p. 4). Social media, newspapers, TV, and so on are all employed in the delivery of political communication, which “adapt, interact, and coevolve” to advance political issues (Chadwick, 2013, p. 59). However, to what extent, if at all, does one type of media influence another? Research exploring the agenda-setting role of new online media was first conducted on blogs. Scholars overwhelmingly have found a mutual intermedia agenda-setting influence between mainstream media and blogs (Cornfield et al., 2005; Heim, 2013; Meraz, 2009; Sweetser et al., 2008; Wallsten, 2007; Williams & Delli Carpini, 2004). Yet, as other social media platforms rise in popularity, what role do they play in agenda setting?
Social Media as #NewsSource
Social media plays diverse but important roles in the newsroom. The 2017 Global Social Journalism Study found that 90% of journalists surveyed used social media at least once a week, whereas 44% used specific image sharing platforms (e.g., Instagram, Pinterest) (Cision, 2017). Furthermore, 73% of journalists acknowledged that they use social media each work day, and 48% noted that they would be unable to work without it. While the perceived primary role of social media remains to publish and promote a journalist’s work, monitoring other media/what’s going on and sourcing information were other essential uses of social media for journalists (Cision, 2017).
Journalism incorporates social media in a variety of source-related functions: news sourcing, eyewitness contacts, authenticity checks, and profile information (Brandtzaeg et al., 2015). The general public (57%) is the most preferred source on social media, followed by industry/professional contacts (31%), and public relations/related sources (27%) (Cision, 2017). Research suggests that Twitter is the most popular social media tool used by journalists (Brandtzaeg et al., 2015; Nordheim et al., 2018). In particular, older journalists use Twitter, whereas younger journalists use a more diverse range of social media, embracing visual platforms, such as Instagram (Brandtzaeg & Dominguez, 2018). In fact, Instagram was noted as an important and frequently used research tool by younger journalists. Perhaps not surprisingly, younger journalists reported using social media more frequently than older journalists (Cision, 2017).
Journalists have become more strategic in how they monitor and use social media—going beyond the platform’s primary search function and incorporating sophisticated search and publishing tools, such as TweetDeck (Brandtzaeg et al., 2015). Conversations on social media range from the mundane to the profane, but many consumer actions and discussions on social media revolve around political and social activity. For example, a national survey conducted by the Pew Research Center found that 53% of US adults in the past year have been “civically active” on social media, with 14% using hashtags related to a political/social issue. In addition, social media sites can promote political/social issues often overlooked or ignored. Sixty-four percent of Americans reported that social media sites are described well or somewhat well by the statement “social media help give a voice to underrepresented groups” (Anderson et al., 2018). Therefore, journalists can easily search by specific hashtags (#) to find posts on desired topics as well as issues that are underreported. The catalyst behind a hashtag’s creation is often a public relations/marketing function that organically grows. Many hashtags originate from concerned or interested individuals and those that resonant see considerable growth and media exposure.
In exploring the popularity and relationship between specific political/social hashtags and news events on Twitter, a 2018 Pew Research Center study found the #BlackLivesMatter, #MAGA, #Resist, and #MeToo hashtag usage spiked around related news events (e.g., Ferguson, election), but they also remain consistently used after these events. Whereas other hashtags (#JeSuisCharlie) are popular around a news event, but rarely used after. Among a national sample of US adults who have used hashtags related to political/social issues, more are minorities (Blacks 19% and Hispanics 13%), less than 30 years old (25%), and identify largely as Democrats (16%) and Independents (16%) (Anderson et al., 2018). Hashtag use originated out of Twitter but has since expanded to most other social media platforms and Instagram (Anderson, 2018).
Researchers suggest that social media civic and political activity has influenced mainstream media. For instance, evidence was found that newspaper coverage, Google News searches, and YouTube videos of California’s Proposition 8 on same-sex marriage were interrelated (Sayre et al., 2010). Social media (i.e., Twitter, blogs, and forums) and its relationship with traditional media was further examined finding evidence of a reciprocal Granger causality on certain political issues (Neuman et al., 2014). Exploring the power that social media can wield in political protests, Black Lives Matter tweets predicted mainstream media coverage of the movement (Freelon et al., 2016). Further evidence of a reciprocal relationship between presidential candidate and campaign Twitter feeds and mainstream newspapers has also been documented (Conway et al., 2015; Conway-Silva et al., 2017, 2018).
Evidence of social media’s influence on mainstream media’s agenda is mixed, suggesting only a moderating influence (Groshek & Groshek, 2013; Kwak et al., 2010; Vargo et al., 2015). To illustrate, Kwak et al. (2010) found that CNN was ahead of Twitter in reporting a majority of the time. Whereas Groshek and Groshek (2013) found evidence that traditional mainstream media influenced social network sites, but social networking site’s ability to impact traditional media was restricted to specific issues (i.e., cultural coverage). There also simply may be a weak tie between the citizens’ agenda and the mainstream media’s agenda. Recent research examining the agenda-setting ability of dual-screeners reveals an inverse relationship between the agenda of Twitter users and the agenda of second screeners’ Twitter users (i.e., citizens who tweet while watching television); however, second screeners react to Twitter trends in the opposite direction of the general Twitter user (Ceron & Splendore, 2019). The latter implies that second screeners do not have a second-level agenda-setting influence on the Twittersphere. Overall, this research illustrates that intermedia agenda setting is complex and not fully understood. Thus, these inconsistent results suggest that more research is needed. In particular, we need to understand the role of citizen accounts on Instagram’s ability to influence agenda setting.
Instagram as #PoliticalPlatform
In 2016, 32% of US online adults (28% of all Americans) reported using Instagram (Greenwood et al., 2016). At this time, Instagram claimed over 700 million monthly active users, growth outpacing Facebook and Twitter (Instagram, 2017). During the 2016 US presidential election primaries, Instagram users were mostly young adults (59% being 18–29 years old in the United States), more often women (38% vs. 26%), completed some college (37%), and made less than $30,000 a year (38%) (Greenwood et al., 2016). Instagram relies prominently on visual interactions. Users communicate with images and a short video that is accompanied by a caption, text, or tags to other users. Uniquely, Instagram has a high character limit of 2,200 and 30 hashtags. Compared to other social networks, hashtags are used more extensively within Instagram (Anderson, 2018). Many hashtags used by users relate to civic movements. To illustrate, in 2018, #metoo was used 1.5 million times, #timesup 597K and #marchforourlives 562K (Instagram, 2018b). Recently, Instagram sought to further strengthen their political involvement by promoting a voter registration website in advance of the 2018 midterm elections and providing an “I Voted” selfie sticker. Tapping the sticker led users to a polling location look-up website (Instagram, 2018a). These recent efforts in driving political action signal the potential political influence that Instagram can wield.
Few political science, marketing, or communication scholars have sought to study Instagram’s use in politics. Research conducted has primarily explored how US, Austrian, Canadian, and Swedish political candidates use the platform as a communication and image management tool (Eldin, 2016; Filimonov et al., 2016; Hendricks & Schill, 2017; Lalancette & Raynauld, 2019; Liebhart & Bernhardt, 2017; Ludwig, 2017; Munoz & Towner, 2017; Towner & Munoz, 2017). In relationship to the United States, scholars have noted that 2016 presidential candidates applied different communication strategies on Instagram (Hendricks & Schill, 2017; Ludwig, 2017; Munoz & Towner, 2017). During the 2016 presidential primary campaign, Trump commonly posted attack ads, whereas Clinton posted images of campaign photos and videos (Hendricks & Schill, 2017). Employing a content analysis of presidential candidate Instagram posts, Towner and Munoz (2017) and Ludwig (2017) found that candidates primarily used Instagram to cover campaign events and address policy issues. Despite the different issues and image content communicated by the candidates, Munoz and Towner (2017) research illustrates that candidates sought to present themselves as the “ideal candidate” to Instagram users. Yet, there may be little effort by the parties or candidates to actually interact with Instagram users on the platform, as Russmann and Svensson’s (2017) research on the 2014 Swedish national elections suggests.
While some research has explored political candidates’ Instagram use, less attention has been allocated to how citizens are politically using Instagram to discuss civic and political issues. As mentioned, Instagram posters have utilized political issue–related hashtags (e.g., #metoo) (Instagram, 2018b). Anecdotally, the large follower counts of the presidential political candidates, coupled with active consumer engagement within these accounts suggest that these political conversations are also occurring within Instagram. To date, however, none of this work has examined citizen’s conversations as they relate to US politics. Instead, it focuses on Instagram usage around the 2014 Scottish independence referendum by Scottish citizens and the 2015 UK General Election (Feltwell et al., 2015; Mahoney et al., 2016). Feltwell et al. (2015) followed relevant hashtags (e.g., #IndyRef) and conducted a thematic image analysis. Their findings revealed a variety of visual communication strategies that users employed: referendum-related leaflets and signs, selfies and photos that display their issue position, and images of referendum-related symbolic flags and bagpipes. In a similar study about the Scottish Independent Referendum (#IndyRef) and the 2015 UK general election (#GE2015), Mahoney et al. (2016) examine the type of content Scottish Instagram users post. Instagram images were coded for themes, such as political expression, symbolism, process, and egocentrism. Findings reveal that political conversions furthering the democratic process were occurring on Instagram. These results, in the international context and consistent with previous research conducted on Twitter (Freelon et al., 2016), suggest that political conversions facilitated by political and issue-based hashtags do occur within Instagram and have the potential to influence and be influenced by mainstream media.
Expectations
As previously discussed, the rise of social media sources has not debunked the conventional understanding of agenda setting. In other words, there is no evidence that the mainstream media no longer influence the citizenry in the Internet era (Johnson, 2014; Tran, 2014). In fact, there is an acknowledgment that political communication exists in an interrelated, interdependent hybrid media system (Chadwick, 2013). Research on citizen-generated websites, blogs, and social media sites indicates extending the traditional theory of agenda setting into intermedia agenda setting. Intermedia agenda setting asserts that citizen-generated content overlaps or intersects with traditional media content (McCombs, 2005). That is, one medium is influencing the other and vice versa. Building on relevant intermedia agenda-setting research (Conway et al., 2015), we examine whether mainstream newspapers influenced citizen’s issue agenda on Instagram or whether the citizen’s issue agenda, measured by their Instagram posts, predicted the issue agenda of newspapers. We expect an intermedia agenda-setting relationship between newspaper issue content and Instagram posts about issues. Specifically, we anticipate that the traditional understanding of agenda setting will hold—that newspaper’s issue agenda during the campaign period will predict citizen’s issue agenda on Instagram (Hypothesis 1). To find evidence that the newspaper agenda does not influence Instagram posters would be surprising, especially as prior work finds that the mainstream press can predict the agenda among social media users (Groshek & Groshek, 2013; Neuman et al., 2014; Sayre et al., 2010; Vargo et al., 2015; Wallsten, 2007).
In addition, we expect that Instagram posts about political issues will predict the newspaper issue agenda (Hypothesis 2). The latter expectation is consistent with evidence suggesting that social media content can significantly predict the mainstream media agenda (Conway et al., 2015; Conway-Silva et al., 2017, 2018; Freelon et al., 2016; Sayre et al., 2010; Towner & Munoz, 2017). As Instagram posters engage with the campaign and candidates, they react to and create digital, visual content that they share on Instagram. Campaign supporters (and non-supporters), as well as candidate surrogates, can craft sensational captions, memorable memes, and eye-catching images that gain momentum on Instagram, capturing the attention of a keen journalist. In this bi-directional media environment, professional journalists and editors now monitor their audience, relying on social media and websites for leads and trending topics. As prior research illustrates, we expect that professional journalists and reporters will likely use Instagram as a news and information source, particularly during campaign periods (Cision, 2017; Parmalee, 2013, 2014). Indeed, several studies conclude that traditional journalists rely on blogs for sources and topics, often referencing some of the top blogs in their reports (Cornfield et al., 2005; Meraz, 2009; Messner & Garrison, 2011; Sweetser et al., 2008; Wallsten, 2007).
Data Collection
Instagram Posts
This study examined the Instagram posts about the Republican, Democratic, and Independent candidates during the 2016 presidential primary. To gather Instagram posts about the presidential primary candidates, a novel, visual intelligence platform, Beautifeye, (www.Beautifeye) was contracted to scan and collect images, videos, captions, and tags of Instagram posts made by regular citizens during the primary period: 1 January 2016 to 30 June 2016. The logic for the starting point was that 1 January 2016 marked 1 month before the Iowa Caucuses held on 1 February 2016. Since the last presidential primary was held on 14 June 2016, in the District of Columbia, the 30 June 2016 endpoint was appropriate. Beautifeye crawled worldwide Instagram posts, searching and collecting posts with the following hashtags and tags: #hillaryclinton, #hillary2016, #imwithher, #clinton2016, #hillary, #bernie, #berniesanders, #feelthebern, #bernie2016, #berniesanders2016, #donaldtrump, #trump, #makeamericagreatagain, #trump2016, #trumptrain, #donaldtrump2016, #tedcruz, #cruz2016, #cruzcrew, #marcorubio, #rubio, #rubio2016, #kasich, #johnkasich, #garyjohnson, and #garyjohnson2016. Specifically, Beautifeye used these hashtags as entry points in their Instagram application program interface (API) to gather and download relevant posts. Any hashtags not listed above were not included in the sample. This search resulted in a sample of N = 638,977 Instagram posts.
Beautifeye’s Instagram API carefully crawled username type and username data for each post. Therefore, the resulting 638,977 Instagram posts in this sample include posts from regular citizen posters and did not include posts from professional journalists, media outlets (e.g., @nytimes, @washingtonpost, @foxnews), professional pundits, the presidential candidates, the political parties, official interest groups, universities, companies, and businesses. During the content analysis, the first author examined all usernames in the final sample, confirming that these posts did not include journalists, media outlets, political candidates, and so forth. As a result, this sample represents all Instagram posts from the average citizen poster. We acknowledge that this sample may include posts from bots or inauthentic Instagram accounts, as it was not possible to manually identify spam accounts. Based on location id and location tagging information provided by Beautifeye, the authors removed all Instagram posts that originated from outside of the United States (except for Puerto Rico), leaving N = 636,100 Instagram posts. We do acknowledge, however, that this sample may include posts from non-US citizens. From the 636,100 posts, about 1% of the Instagram posts were randomly selected and extracted for content analysis. Therefore, the final sample includes N = 5,573 Instagram posts. 1 All Instagram posts are examined as an “Instagram Index” to measure the overall Instagram issues mentioned.
Mainstream Newspapers
To compare Instagram posts to mainstream media content, we examined four of the highest circulating national newspapers in the United States (MediaMiser, 2016). The newspaper sample included The New York Times, The Washington Post, USA Today, and The Los Angeles Times. The New York Times and The Washington Post were selected because they are considered the national newspapers of record, read by political elites and highly engaged citizens. The USA Today was included because it targets the more general reader who has some interest in politics. To include West Coast political perspectives, we also included the Los Angeles Times. Lexis-Nexis and Pro Quest were employed to gather all newspapers articles published in these newspapers during the primary period: 1 January 2016 to 30 June 2016. To search for relevant articles and editorials in each database, we used the key term “U.S. Presidential elections,” in combination with “primaries” or “caucuses.” The final sample includes 724 relevant articles. In our analysis, the four newspapers are examined as a “Newspaper Index” to measure the overall issue agenda in the mainstream press.
Method
Content Analysis
For Instagram, the unit of observation was the daily post, including the caption, hashtags, and tags. For newspapers, the unit of observation was daily content about the presidential primary election published in a newspaper. The lead author conducted the coding of the newspaper articles and Instagram captions, hashtags, and tags. To develop an initial coding sheet, we drew upon previous studies that content analyze issue coverage during campaign periods (Conway et al., 2015; Conway-Silva et al., 2017, 2018; Hansen & Benoit, 2001; K. Kim & McCombs, 2007; Tedesco, 2001, 2005). Based on this prior work as well as preliminary coding, we then randomly selected 20 newspaper articles and 20 Instagram posts for preliminary coding.
The initial coding produced a more detailed coding sheet, including the major issues expected in an election campaign, such as economy, environment, and health care. But also, many issues, and some sub-issues of the major issues, unique to the 2016 election cycle. Indeed, every election cycle is different, with candidates and parties emphasizing some policy issues while downplaying others. In the 2016 primaries, some of these more uniquely emphasized issues include #blacklivesmatter, climate change, email, and Muslim. For instance, “climate change” was a highly partisan issue during the primary season, with Democratic candidates Hillary Clinton and Bernie Sanders arguing about who is doing more to fight climate change, and Republican candidates Marco Rubio and Ted Cruz signing a “no climate change tax” pledge. As a result of the FBI’s investigation into Hillary Clinton’s use of her personal email for State Department official business, another unique issue, “email,” was noted in the preliminary coding. The issue “Muslim” was also included, as Donald Trump’s campaign proposed to create a policy banning Muslims from entering the United States as well as proposing to build a wall to slow the number of immigrants. Many of these unique issues are consistent with recent research focusing on the prevalent policy issues in the 2016 primary period (Ludwig, 2017; Nawara & Bailey, 2017).
Therefore, our final coding sheet includes a comprehensive list of 33 main issues and unique issues in the 2016 presidential primary: abortion, affirmative action, banking, #blacklivesmatter, budget, campaign finance/donate, civil rights, climate change, corruption/ethics, crime, deport, drugs, economy, education, email, employ/job, environment, equality, foreign policy/trade, gay/LGBTQ, gun, health care/Affordable Care Act, immigration, income equality, military, minimum wage, Muslim, racial, religion, social security, tax, weed, and welfare.
With the final coding sheet, the lead author coded the newspaper articles and Instagram captions, hashtags, and tags for issue frequencies. 2 The newspaper article and Instagram post were coded for how many times an issue—or issues—was mentioned, meaning that more than one issue could be coded per article or post. For example, one newspaper article may mention banking five times, the economy twice, and immigration 10 times. A minimum threshold of five issue mentions for at least one newspaper was employed to identify the primary campaign’s important issues. Similarly, the same minimum was applied to Instagram captions and hashtags—a minimum threshold of five issue mentions for each issue in total. Instagram issue coding did not include coding of images or video. Considering the set threshold, three issue categories were removed from the analysis—religion, affirmative action, and weed—as they did not meet the threshold.
To calculate intercoder agreement, a random subsample of 10% of all newspaper articles and Instagram captions and hashtags were coded by a second coder. This second coder was a political science student trained in coding and content analysis. We calculated the percent agreement, relying on the proportion of agreement of coded units between the two coders. The intercoder agreement ranged from 82% to 100% agreement across the 30 issues coded.
Findings
Table 1 reports the frequency of issue mentions for the Newspaper Index as well as the Instagram Index. Instagram issue mentions (N = 5,248) exceeded the number of issue mentions in newspapers (N = 4,433) (See Table 1). The latter finding is not surprising, as Instagram has more carrying capacity for issue mentions than newspapers. For instance, Instagram posters have sufficient space to post about an issue with a generous character limit (2,200) and hashtag limit (30). Moreover, unlike mainstream journalists and reporters, Instagram posters have no limit on their number of posts per day.
Issue Frequencies and Rankings for Newspaper Index and Instagram Accounts for 30 Issues.
The mainstream press and Instagram posts contrasted on the issues at the very top of their agenda, but overall there were some commonalities of the issues in the top 10 rankings. To more easily examine the issue agendas, the issue frequencies were ranked from 1 = the top issue mentioned to 30 = the least issue mentioned in Table 1. For example, the Newspaper Index illustrates that the top five issues in the mainstream press were immigration (N = 578), followed by the economy (N = 339), campaign finance (N = 324), banking (N = 321), and health care/ACA (N = 300). On Instagram, the top five issue were civil rights (N = 582), followed by corrupt/ethics (N = 535), climate change (N = 424), gay/LGBTQ (N = 415), and abortion (N = 317). There are no common issues in the top five rankings for the Newspaper Index and Instagram Index, suggesting a divergence in issue emphasis during the presidential primary. Among the top 10 issue rankings, Table 1 shows that the mainstream press and Instagram both place the issues of immigration and the military prominently on their news agenda. The mention of “military” shared the most commonality with 258 mentions in newspapers and 279 mentions on Instagram. Yet, newspapers placed “immigration” and “military” much higher on their agenda than Instagram.
While Table 1 shows that the newspaper issue agenda and Instagram posters’ issue agenda contrasted during the presidential primary campaign, it remains to be tested if there is an empirical relationship between their overall issue agendas. The relationship between the issue rankings (all 30 issues) in the Newspaper Index and Instagram Index were first compared by employing Spearman’s p correlation. A Spearman’s p correlation reveals that there is no significant correlation between the Newspaper Index and the Instagram posts (rho = –.074, p = .697). The latter suggests there is no relationship between the overall issue agendas of the mainstream media and Instagram.
To examine the most salient issues covered in the presidential primary campaign, the top 10 issues were identified by summing issue frequencies across the Newspaper Index and Instagram Index. The top 10 issues in rank order were immigration (N = 767), civil rights (N = 627), corruption/ethics (N = 626), military (N = 537), campaign finance/donate (N = 483), health care/ACA (N = 463), gay/LGBTQ (N = 460), climate change (N = 453), abortion (N = 417), and banking (N = 417). The top 10 issue rankings for the Newspaper Index and Instagram posts were compared by employing Spearman’s p correlation. This empirical test indicates a significant and negative relationship between the top 10 issue mentions in the Newspaper Index and Instagram Index (rho = –.855, p = .002). The results imply that the top issue agendas of the mainstream press and Instagram posts are related.
Next, we seek to examine further the relationship between the issues mentioned in the mainstream media agenda and Instagram agenda. To test this link, cross-correlations were estimated. In agenda-setting research, cross-correlations empirically test the association between two agendas over time (McCombs et al., 2013; Shaw & McCombs, 1977). We want to note, however, that cross-correlations do not prove real-world causality and only offer evidence for possible intermedia agenda-setting effects. If Instagram posts influence newspaper content, then the Instagram issue mentions will be significantly correlated with newspaper issue mentions. If newspaper coverage influences Instagram posts, then newspaper issue mentions will be significantly associated with Instagram posts.
Before conducting the cross-correlations, the top 10 issues in the Newspaper Index and Instagram Index were examined for non-stationarity, specifically testing each issue series for linear and quadratic trends. The latter was a preliminary pre-processing step conducted to uncover any auto-correlation or non-stationarity in each issue series. If a time-series is non-stationary, the cross-correlation will not reflect the true relationship between the two series. This pre-processing step uncovered trends in the Instagram issue series for immigration (linear function with R2 = .04, p < .05) and gay/LGBTQ (quadratic function R2 = .04, p < .01). To remove linear and quadratic trends from the aforementioned series, detrending was employed to achieve stationarity. There were no linear or quadric trends uncovered for issues in the Newspaper Index.
To test Hypotheses 1 and 2, cross-correlations were estimated to examine the direction of the frequency of issue mentions between the Newspaper Index and Instagram Index for the top 10 issues. See Table 2. When examining the daily number of issue mentions in media content, it is important to consider the optimal time frame to study. That is, what is the gap between the mention of an issue in one media agenda and the mention of the issue in another media agenda. There is no consensus in the agenda-setting literature on the optimal time lag. Agenda-setting research in the Internet era has employed a variety of time frames, including a 2-month period (Luo, 2014), a 7-day period (Kushin, 2010), and a 1-day period (Vonbon et al., 2016). Considering the online environment and an accelerated news cycle during an election campaign, a shorter time frame is ideal. Since we are examining daily newspaper content, it is essential to consider the lagged effects or to control for the level of content 7 days before the present value (seven lags back). A 7-day time lag is also consistent with previous research examining the intermedia agenda-setting relationship between Twitter and newspaper content (Conway et al., 2015; Conway-Silva et al., 2017, 2018; Kushin, 2010; Luo & Harrison, 2019). A reported lag suggests that Instagram posts significantly predicted newspaper issue mentions 1–7 days before the issue mentions in the newspaper indices. A lead indicates that issue mentions in the newspaper indices significantly predicted issue mentions in the Instagram posts 1–7 days prior to issue mentions in the Instagram posts. A reported lag of 0 indicates that any influence occurred on the same day.
Daily Lag and Lead Cross-Correlations for Instagram Posts and Newspaper Index.
Note: A reported lag suggests that Instagram posts significantly predicted newspaper issue mentions 1–7 days before the contemporary frequencies. A lead indicates that issue mentions in the newspaper indices significantly predicted issue mentions in the Instagram posts 1–7 days prior to the contemporary frequencies. A reported lag of 0 indicates that any influence occurred on the same day.
In general, the results in Table 2 demonstrate very little evidence of an agenda-setting relationship between the mainstream media and Instagram posts during the 2016 primary period. From the top 10 issues tested, only six of the issues—immigration, civil rights, military, abortion, health, and bank—contained significant lags and/or leads. There were no significant lags or leads between newspapers and Instagram posts for four issues: corruption, campaign finance, climate change, and gay/LGBTQ. Table 2 reports the cross-correlations results for the six issues containing significant lags and/or leads. Considering Hypothesis 1, there is very little evidence that the newspaper’s issue agenda predicted Instagram posters’ issue agenda between January and June 2016. As Table 2 shows, there are three significant leads for two issues: civil rights and banking. There is no evidence that newspaper paper coverage of immigration, military, abortion, and health care predicted Instagram posts. Looking at the length of the leads, newspaper issue content predicted Instagram posts more than 4 days before the issues of civil rights. But on the issue of banking, the newspaper’s agenda swiftly predicted—within 2 days—Instagram posts. In addition, newspaper content on banking had a rather high impact (r = .37) on Instagram posts.
The nine significant lags shown in Table 2 indicate that Instagram posts predicted newspaper coverage on six issues during the 2016 presidential primary period: immigration, civil rights, military, abortion, health care, and banking. The latter offers only moderate support for Hypothesis 2, which asserts that Instagram posts predict the mainstream newspaper’s issue agenda. Interestingly, Instagram posts led newspaper coverage 5–7 days prior on the issues of immigration, military, and banking. That is, there is a longer time span (more days) between Instagram mentions of immigration, military, and banking issues and the newspaper’s coverage of these issues. Looking at Column 1 in Table 2, the cross-correlation indicates that Instagram posters’ mentions of immigration predicted newspaper coverage 5–7 days prior with no reverse effects: newspaper coverage of the immigration did not predict Instagram posts. There were also no reverse effects for military, abortion, and health care. Instagram posts predicted newspaper coverage more quickly—1–2 days prior—on the issues of military and health care. The latter results suggest that Instagram posts about these issues were rather successful in predicting issues emphasis on abortion and health care in newspapers. It is also worth highlighting the somewhat reciprocal relationship between Instagram posts and newspaper content on the issue of banking. The impact of newspaper coverage on banking, particularly at Lead 2, had a higher magnitude than all other lags and leads in Table 2. The latter findings suggest that newspaper content was a stronger predictor of the Instagram posters’ agenda on the banking issue than the reverse.
Overall, the cross-correlations of these issue series in Table 2 reveal that there are more lags (12) and leads (3), which implies that Instagram posts have more predictive power over newspaper’s issue agenda. The evidence for intermedia agenda setting is not overwhelming; however, as only 15 lags/leads out of the possible 84 lags/leads were significant (Considering that a total of 10 issues were empirically examined, technically 15 lags/leads out of 140 lags/leads tested were significant). It is also notable that there are no same-day correlations (Lag 0) for any of the issues tested. That is, newspaper issue content and Instagram issue content does not predict “day of” coverage or posts.
Discussion
This research strived for a deeper understanding of how the mainstream media’s issue agenda and the agenda of the citizen’s accounts on the visual-social networking site, Instagram, intersected during the 2016 presidential primary. It further sought to understand whether the conventional agenda-setting relationship between traditional media and audience held, or whether the issue agenda on Instagram was linked to the mainstream media’s agenda. The results in Table 1 indicate that issue emphasis in Instagram posts differs substantially from top issues in the mainstream press. During the 2016 primary period, Instagram posts and newspapers shared only two issues among the top 10 issues. Instagram posters cover more about civil rights, corruption, climate change, gay rights/LGBTQ issues, and abortion rights, whereas newspapers cover immigration, the economy, campaign finance reform, banking, and health care. Indeed, Instagram is more popular among young adults and minorities than it is among adults overall (Pew Research Center, 2018); therefore, Instagram’s issue content predominantly includes political issues more important to these individuals (Ember, 2018). The traditional press continues to cover traditional, political issues—foreign affairs, economic concerns, and health and medicine (Pew Research Center, 2010). Instagram posters discuss issues that tend to focus on matters that garner much less attention in the mainstream press, suggesting that this platform is a supplementary place for political issue discussion.
Findings suggest minimal evidence for an intermedia agenda-setting relationship between newspapers and Instagram posters’ issue agendas. There is unconvincing evidence that the newspaper issue agenda influenced the Instagram posts during the primary period, indicating that the agenda-setting effect of the mainstream media is not strong among Instagram posters. Presently, daily newspaper circulation (print and digital) is at an all-time low in the United States (e.g., 52,329,000 in 2006 vs. 34, 657,199 in 2016) (Barthel, 2018). Digital newspapers have increased the number of younger newspaper readers; however, newspaper readers are disproportionately college-educated and affluent (annual household incomes over US$100,000) (Neilsen Insights, 2016). These demographics are not consistent with Instagram users (Pew Research Center, 2018). Research from Pew Research Center also supports Instagram users turning away from traditional print newspapers (12% access) to access news more readily from local TV (33%), cable TV (25%), and news websites and apps (42%) (Gottfried & Shearer, 2016). One could argue that Instagram users, who are mostly young, non-Whites (Gottfried & Shearer, 2016), perceive today’s mainstream media to be “out of touch,” unrepresentative, and unresponsive to issues important to their demographic profile.
There is more convincing evidence that Instagram posts influence the newspaper’s issue agenda. However, the correlations between the newspaper agenda and Instagram agenda on the six issues reported are not strong and are somewhat inconsistent. Presently, the median age of journalists is 47 (Thompson, 2014), which is considerably older than the vast majority of Instagram users. Willnat and Weaver (2018) report that about half of journalists (53%) working for newspapers incorporate social media into their reporting routine, with most journalists using micro-blogs—Twitter—as a news source. Therefore, it is likely that Instagram is not frequently used in the newsrooms of the top US newspapers. Also, journalists would also be unlikely to pick up information gleaned from other individuals indirectly. Research from Pew Research Center indicates that platforms such as Reddit, Facebook, and Twitter were used more often as news sources for US adults compared to Instagram and those that did use Instagram as a new source did not align demographically with newspaper readers (e.g., Instagram users are mostly young 18–29 years [58%], no college degree [69%], and non-White [57%]) (Gottfried & Shearer, 2016).
In line with prior research (e.g., Groshek & Groshek, 2013), these findings continue to suggest that the agenda-setting relationship between traditional media and social media is complicated and likely issue specific. For example, both outlets mentioned military issues with about the same frequency, but there is no evidence that Instagram users followed military issues covered by newspapers during the 2016 primaries. Perhaps, Instagram posters do not turn to mainstream newspapers for military information. However, newspaper journalists and Instagram posters may diverge on what aspects of the military in which to focus. For instance, newspapers tend to analyze how presidential candidates will lead the military and act as military commander as well as how candidates feel about funding the military budget, whereas those on Instagram post about family and friends who are military veterans, the veteran’s health care program, general feelings about patriotism and protecting the flag, and supporting our troops. For example, one Instagram poster captioned, “Support the Troops #military #usmilitary #navy #veterans #usveterans #Bernie #berniesanders #bernie2016 #feelthebern #berniesandersforpresident #clinton #history #Hilary #clinton #militaryfamily #war #troops #bringourtroopshome” along with an image of Bernie Sanders (Instagram, 24 February 2016). This may explain why Instagram posts influenced newspaper’s military content, but there was no reciprocal effect from newspaper coverage.
The cross-correlations offer some evidence of a bi-directional or reciprocal relationship between newspaper content and Instagram posts on the issues of civil rights and banking. It is worth noting that Instagram posts on banking took 7 days to permeate the newspaper agenda, whereas newspaper coverage took fewer days (2 days) to influence Instagram posts on the topic. For example, on the issue of banking, on 17 February 2016, the New York Times covered the primary candidates’ plans to create jobs among poor and minority communities, noting that Clinton’s job plan “. . . would be paid for by a ‘risk fee’ imposed on Wall Street banks and changes in the tax code” (Chozick & Alcindor, 2016). On Instagram, someone posted: “#HillaryClinton represents #Banks, #Corporations and #Media. #BernieSanders represents the people. Do you think corporations give $ for no reason? No matter what party you’re for . . . the people of the #usa should NOT stand for it!” (Instagram, 21 February 2016). The aforementioned post on Instagram was not uncommon, as many posted about corruption on Wall Street, financial reform, Hillary Clinton’s paid speeches to Wall Street, taxes on the middle class, and more.
Our findings reinforce the speculation that the news media is not actively monitoring Instagram as the same level of intensity compared to other new media platforms such as blogs, Twitter, and Facebook. With only half of newspaper journalists incorporating social media into their reporting (Willnat & Weaver, 2018), it would be wise for all journalists (young and old) to regularly monitor Instagram, as it is vital to reach the next generation of readers, watchers, and viewers. On the issue of civil rights, both outlets influenced each other’s agenda within 4 days. One may only speculate this quick, robust reciprocal effect is the result of campaign events focused on these issues. Civil rights, particularly women’s issues, human rights, and racial equality were important issues echoed throughout the presidential primaries, covered during debates, rallies, and press releases. For instance, an Instagram user rallying for Bernie Sanders posted, . . . #berniesanders marching with #martinlutherkingjr #civilrights #peacefulprotests #mlk #blackhistorymonth #feelthebern #rockthevote #ihaveadream Bernie Sanders has been #forthepeople and a public servant his whole career. He cares about ALL people and not special interests. #realdeal He is consistent with his beliefs and doesn’t flip flop. We are lucky he is running and will likely not get a chance like this again. (Instagram, 7 February 2016)
Similarly, in February 2016, the USA Today reported that Bernie Sanders campaigned to Black voters by emphasizing that he was active in the 1960s civil rights movement (Page, 2016). Both professional journalists and Instagram posters covered these issues on almost the same days, likely responding to campaign events.
This analysis also shows no contemporaneous relationships on any issues, suggesting that newspapers and Instagram posters are not covering the same issues on the same day. This is surprising because mainstream media and social media often cover the same “hot” issues and significant events on the day of occurrence (Conway et al., 2015; Conway-Silva et al., 2017, 2018; Towner & Munoz, 2017). This lag time implies that issues covered by newspapers are not transferred to Instagram at a fast rate and vice versa: issues covered by Instagram posters are not quickly transferred to newspapers. On the one hand, this is likely because professional journalists and reporters place a priority on monitoring other online outlets for political topics, mainly blogs and Twitter, and then turn to Instagram. The same thinking could also be applied to Instagram—Instagram users may be relying on candidate and campaign content first and then mainstream media coverage when posting about political issues. Instagram is still a relatively new social media platform in the political-digital sphere; therefore, this may be why there are no immediate “day-of” effects.
This is the first empirical examination of the intersection between the top circulating US newspapers and Instagram posts. Although the mainstream media includes television, radio, newspapers, and magazines, this study was limited to examining issue mentions in hard-copy newspapers. We acknowledge that this is an incomplete look at the mainstream media issue agenda. Broadening to television—cable and network coverage—would give more insight into intermedia agenda setting. Television coverage may do a better job of covering often-overlooked issues pertinent to young, non-Whites. Given Instagram’s focus on visual communication, a reciprocal relationship may be more evident in television coverage. The largest limitation of this study is that we cannot suggest causality. That is, Instagram posts do not cause newspaper content and vice versa. We can only suggest that there is a correlation or link between the two online platforms regarding political issues. Consistent with the hybrid media system, we also acknowledge that other media sources (television, other forms of social media) may have influenced our findings. This research is also limited in scope, as we only examine media content during the 2016 presidential primary period. The media landscape may be very different in future special elections, midterm election periods, and presidential primary campaigns. As the digital landscape is ever changing, scholars should continue to examine intermedia agenda-setting effects.
Increasingly more Americans prefer to watch their news, as opposed to reading it (Mitchell, 2018). In the future, as Instagram’s user adoption rate continues to climb, and as its core audience becomes older and more diverse, we may expect to see stronger evidence for intermedia influence between newspapers and Instagram. However, for now, Instagram scholars should turn their attention to television to explore potential intermedia agenda-setting effects.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
