News coverage can be dominated by just a few players, even in developed countries with an active free press. This dominance can lead to a reduction in topical and community diversity. Ownership structures might further limit coverage by implicitly or explicitly biasing editorial policies. In this study, we applied ecological diversity measures to quantify the health of the Chilean online news ecology using extensive ownership and social media data. Results indicate that high levels of concentration of ownership and topics coverage characterize the Chilean media landscape. Our methods help to reveal which groups of outlets and owners exert the most significant influence on news coverage. Our work can be generalized to any nation’s news media system.
News is increasingly aggregated by and consumed through social media such as Reddit, Facebook, and Twitter. Due to its reliance on social networking relations for news propagation, social media may be subject to a variety of social issues that can restrict news coverage and topical diversity, for example, as a result of information bubbles and social conformity bias. On the other hand, some might argue that the exponential growth of new communication technologies can solve, or at least alleviate, many diversity problems.
More accessible and cheaper channels of communication should provide new content producers with better opportunities and less friction to compete in a broader media market. However, early indicators point to high levels of bias as well as a lack of diversity in terms of topics covered and communities addressed.
In the past, news ecosystems have been primarily modeled from a political and economic perspective. Theoretical models focusing on the political economy of the mass media show that, for a genuinely democratic society, the more information we can have as voters, the better. Most of these models agree on the negative consequences of the concentration of ownership in the mass media. Having a large share of the media industry in the hands of just a few mega-conglomerates poses the risk of the system not necessarily representing the interest of the common good, the media’s original primary purpose.
In our study, we postulate that the news industry can be modeled as a complex system, an ecosystem that consists of many different interacting components, such as news outlets, their owners, reporters, news consumers, and advertisers, all subject to and responding to a variety of social factors. Through their interactions among themselves and with external drivers, these components collectively shape our news ecosystems.
Given these broad similarities, we hypothesized that we can apply techniques developed to study the health and diversity of natural ecosystems to online news (eco)systems. Previous works that exploit parallels between information systems and ecology have proven to be fruitful. Specifically, we used ecological indicators to analyze the “health” of the “news ecosystem” as viewed from Twitter, an online social network specifically designed for the social propagation of information and increasingly used as a platform for the dissemination of news. Our analysis relied on information about Chilean news outlets since they have established a significant social media presence with a high number of Chilean users. The Chilean media landscape is furthermore well documented due to the availability of detailed, publicly available data for its ownership structure, compiled by Poderopedia.org. Our objective was to measure the diversity of a news ecosystem, taking into account the variety of different news sources, the news content, as well as the news consumers.
We started by considering each news tweet as an entity/individual in our ecosystem and its corresponding news outlet as its type/species. We looked only for tweets corresponding to topics reported from multiple news outlets. With these topics, we can analyze how many outlets cover the same event and/or how many times two outlets coincide in their selection of stories. We then applied well-known ecology indices to quantitatively measure how “healthy”/diverse our system is. We assume that diversity of content is a desired property of any news system. Similarly, it has been proposed that ownership can influence editorial policies and bias content. Thus, we also associated the relationship and type of each entity with the owner of the publisher outlet, rather than with the news outlet itself. This is, potentially, a stronger effect, since several newspapers may publish similar content because they belong to a single ownership group.
We analyzed media ownership using two metrics: numerical diversity and source diversity. Numerical diversity refers to the number of outlets available to the public in a given area; source diversity indicates the number of owners that control those outlets. The rationale for using these indicators is our expectation that having a news industry increasingly dominated by fewer and fewer companies increases the owners’ potential influence on the published content, leading to a higher probability of reallocation of attention to their interests. In this work, we focus on the complexity component of the system by analyzing a variety of quantitative diversity indices, namely the Shannon Diversity and Simpson Diversity indices. Meanwhile, with the Average Taxonomic Distinctness, we get a notion of the similarity that can be expected from the coverage of a story, even in cases where it originates from different outlets.
With the Shannon index, we quantify, for each detected topic, how common it is across the media and whether it was covered disproportionately by just a few newspapers. This measure will give us an indicator to assess whether the topic was generally considered important news and covered accordingly across a wide variety of outlets or pushed as a topic by specific outlets/owners.
While the Shannon index measures the abundance of news sources and diversity of tweets, the Simpson index is considered a dominance index that assigns a higher weight to the most common sources. In other words, the presence of a few tweets of some rare outlets will not have a significant effect on the result. Similar to the ones above, the Average Taxonomic Distinctness takes into account the abundance of the sources, but also includes the taxonomic distance between any two types (out-lets/owners). For us, then, the Average Taxonomic Distinctness is the average editorial “distance” (using an editorial similarity based on overlapping topics selection patterns) between any two news sources in the same topic.
The approach proposed by the Taxonomic Distinctness brings a different dimension to diversity. An ecosystem under environmental disturbance could display not only a reduced number of species (as shown by the Simpson and Shannon indices) but also that the remaining species could be closely related. For a news media ecosystem, this would imply that not only the stories are dominated by a few outlets, but also that the point of view of these outlets could be very similar.
Upon application of the proposed indices, we found low external pluralism or low diversity. The topic selection seems to be driven by individual factors rather than objective criteria, such as the newsworthiness of events. Topics are covered by only a few news outlets and even fewer owners. Furthermore, we found that outlets that cover the same topics exhibit high levels of content similarity, indicating a lack of independent reporting. These findings suggest that many outlets are not only subject to similar editorial policies, but rely on related content. Although our results indicate relatively high numerical diversity (many news outlets), which should in principle contribute to a healthy Chilean news ecology, we observed a significant lack of source-driven diversity.
We estimated the ratio of the actual diversity against its optimal value (based on the number of sources) to see how far the Chilean news ecosystem is from becoming the optimal system: we found that outlets are, on average, 24.7% of their full potential diversity, while owners stand at a deficient 3.1%. We consistently found that the health of the system is considerably more critical when we use owners instead of outlets as types: we saw between five and six times more diversity for outlets according to Average Taxonomic Distinctness and Simpson Index and almost ten times for the Shannon Diversity Index.
The difference in the results between both evaluations (ownership diversity vs. outlet diversity) reveals the false illusion of diversity created by the multiplicity of outlets owned each by dominant companies. This difference is also indicative of high levels of concentration in the Chilean news market: few owners control many outlets and may influence their editorial policies (e.g., topic selection). The fact that outlets owned by the same company systematically share the same topics indicates low internal pluralism. A few mega- conglomerates controlling a large number of outlets presents a clear risk to news diversity, coverage, and representativeness.
All in all, we believe that the methods discussed in this study can help shed light on the health status of the news systems of the world in a convenient way, by neutral parties. More results and in-depth analysis can be found in the full paper.
These findings are described in the article entitled Quantifying the ecological diversity and health of online news, recently published in the Journal of Computational Science.