Search Articles

Measuring Targeting Effectiveness in TV Advertising: Evidence from 313 Brands

We estimate the heterogeneous effects of TV advertising on revenues of physical retail stores and restaurants in the United States for 313 brands using a novel panel of store-level revenue data and a two-way fixed effects design. We find a mean revenue elasticity to TV advertising of 0.094 and a median elasticity of 0.044, along with a significant estimated S-curvature in the marginal effect of advertising. We document significant heterogeneity in estimated effective- ness across store-level characteristics, and in particular find that advertising is more effective for stores in denser areas and for stores in areas with higher numbers of competitor locations. We then use these heterogeneity estimates to construct optimal allocations of ad expenditure across DMAs and project that these counterfactual reallocations would increase returns on advertising by a median of 4.3% and a mean of 22.1%. This study advances recent research demonstrating that TV advertising is less effective than is generally assumed by highlighting the role of suboptimal geographic targeting and by quantifying how much realized effectiveness can understate advertising’s potential effect.

Learn More >>

You want a piece of me: Britney Spears as a case study on the prominence of hegemonic tales and subversive stories in online media

In this work, we seek to understand how hegemonic and subversive (counter-hegemonic) stories about gender and control are constructed across and between media platforms. To do so, we examine how American singer-songwriter Britney Spears is framed in the stories that tabloid journalists, Wikipedia editors, and Twitter users tell about her online. Using Spears’ portrayal as a case study, we hope to better understand how subversive stories come to prominence online, and how platform affordances and incentives can encourage or discourage their emergence. We draw upon previous work on the portrayal of women and mental illness in news and tabloid media, as well as work on narrative formation on Wikipedia. Using computational methods and critical readings of key articles, we find that Twitter, as a source of the #FreeBritney hashtag, continually supports counter-hegemonic narratives during periods of visibility, while both the tabloid publication TMZ and Wikipedia may lag in their adoption of the same.

Learn More >>

Inequalities in Online Representation: Who Follows Their Own Member of Congress on Twitter?

Members of Congress increasingly rely on social media to communicate with their constituents and other members of the public in real time. However, despite their increased use, little is known about the composition of members’ audiences in these online spaces. We address these questions using a panel of Twitter users linked to their congressional district of residence through administrative data. We provide evidence that Twitter users who followed their own representative in the 115th, 116th, and 117th Congresses were generally older and more partisan, and live in wealthier areas of those districts, compared to those who did not. We further find that shared partisanship and shared membership in historically marginalized groups are associated with an increased probability of a constituent following their congressional representative. These results suggest that the efficiency of communication offered by social media reproduces, rather than alters, patterns of political polarization and class inequalities in representation observed offline.

Learn More >>

Inequalities in Online Representation: Who Follows Their Own Member of Congress on Twitter?

Automated tools for parsing and communicating information have increasingly become associated with the production of journalistic content. To study this phenomenon and to explore the development of automated journalism across two locales at the cutting edge of technology, we leverage insights from in-depth interviews with news technologists from pioneering news organizations and Internet companies specialized in the construction of “news bot” technology in the United States and China, including The Associated Press, The New York Times, The Atlanta Journal-Constitution, BuzzFeed, Quartz, Xinhua Zhiyun, Southern Metropolis Daily, Toutiao, and Tencent. Based on these interviews, we document how the creation of automated journalism products is heavily dependent on the successful assembly of actor networks inside and outside organizations. While metrics for measuring the success of automated journalism are applied differently, they often center around the augmentation of existing reportorial activities and focus on replacing rote and mundane (human) work processes. Some of the biggest challenges in automated journalism lie in curating high-quality datasets and managing the associated high stakes of errors in a business defined by trust. Lastly, automated journalism can be seen as a form of experimentation, helping its protagonists to future-proof their respective organizations.

Learn More >>

Inequalities in Online Representation: Who Follows Their Own Member of Congress on Twitter?

Here we represent human lives in a way that shares structural similarity to language, and we exploit this similarity to adapt natural language processing techniques to examine the evolution and predictability of human lives based on detailed event sequences. We do this by drawing on a comprehensive registry dataset, which is available for Denmark across several years, and that includes information about life-events related to health, education, occupation, income, address and working hours, recorded with day-to-day resolution. We create embeddings of life-events in a single vector space, showing that this embedding space is robust and highly structured. Our models allow us to predict diverse outcomes ranging from early mortality to personality nuances, outperforming state-of-the-art models by a wide margin. Using methods for interpreting deep learning models, we probe the algorithm to understand the factors that enable our predictions. Our framework allows researchers to discover potential mechanisms that impact life outcomes as well as the associated possibilities for personalized interventions.

Learn More >>

Democrats are better than Republicans at discerning true and false news but do not have better metacognitive awareness

Insight into one’s own cognitive abilities is one important aspect of metacognition. Whether this insight varies between groups when discerning true and false information has yet to be examined. We investigated whether demographics like political partisanship and age were associated with discernment ability, metacognitive efficiency, and response bias for true and false news. Participants rated the veracity of true and false news headlines and provided confidence ratings for each judgment. We found that Democrats and older adults were better at discerning true and false news than Republicans and younger adults. However, all demographic groups maintained good insight into their discernment ability. Although Republicans were less accurate than Democrats, they slightly outperformed Democrats in metacognitive efficiency when a politically equated item set was used. These results suggest that even when individuals mistake misinformation to be true, they are aware that they might be wrong.

Learn More >>

Data Representation as Epistemological Resistance

Over the last two decades quantitative data representation has moved from a specialization of the sciences, economics, and statistics, to becoming commonplace in settings of democratic governance and community decision making. The dominant norms of those fields of origin are not connected to the governance and activism settings data is now used in, where practices emphasize empowerment, efficacy, and engagement. This has created ongoing harms and exclusion in a variety of well-documented settings. In this paper I critique the singular way of knowing embodied and charts and graphs, and apply the theories of epistemological pluralism and extended epistemology to argue for a larger toolbox of data representation. Through three concrete case studies of data representations created by activists I argue that social justice movements can embrace a broader set of approaches, practicing creative data representation as epistemological resistance. Through learning from these ongoing examples the fields of data literacy, open data, and data visualization can help create a broader toolbox for data representation. This is necessary to create a pluralistic practice of bringing people together around data in social justice settings.

Learn More >>

Stitching Politics and Identity on TikTok

Parker Bach, Adina Gitomer, Melody Devries, Christina Walker, Deen Freelon, Julia Atienza-Barthelemy, Brooke Foucault Welles, Diana Deyoe, and Diana Zulli

Though a relative newcomer among social media platforms, social video-sharing platform TikTok is one of the largest social media platforms in the world, boasting over one billion monthly active users, which it garnered in just five years (Dellatto, 2021). While much of the early attention to the platform focused on more frivolous elements, such as its dances and challenges, the political weight of TikTok has become ever clearer. In the 2020 US election, Donald Trump’s plan to fill the 19,000-seat BOK Center in Tulsa was stymied by young activists who reserved tickets with no intention of attending, organized largely on TikTok (Bandy & Diakopoulos, 2020). In the years since, political discourse on TikTok has continued to emerge from everyday users and political campaigns alike (see Newman, 2022), even as TikTok itself has become an object of political contention: calls for banning the app in the United States–citing security concerns influenced by xenophobia, given the app’s Chinese ownership–began in the Trump presidency (Allyn, 2020) and have recently culminated in state- and federal-level bans on the app for government-owned devices in the U.S. (Berman, 2023). While some studies have navigated limited data access and the platform’s relative infancy to offer an examination of political TikTok (see Literat & Kligler-Vilenchik, 2019; Medina Serrano et al., 2020; Vijay & Gekker, 2021; Guinaudeau et al., 2022), there remains a significant need for more analysis and theorization of how TikTok can become both a site for political discourse and a feature caught up within political mobilization. This panel seeks to bring together emerging work that deals with political participation on TikTok, in order to share current wisdom and forge future research directions. The presented works specifically focus on the relationship between political participation on TikTok and political identity for three primary reasons. First, as a video-based and thus embodied platform (Raun, 2012), creator identity is more prominent and easily perceptible in the visual and auditory elements of TikTok videos than in the primarily text-based posts on platforms like Twitter and Facebook. Second, TikTok relies more heavily on its recommendation algorithm for content distribution than its competitors traditionally have (Kaye et al., 2022; Cotter et al., 2022; Zeng & Kaye, 2022; Zhang & Liu, 2021), leading to the creation of “refracted publics” (Abidin, 2021) or Gemeinschaft-style communities (Kaye et al., 2022) around users’ common interests, which may include and/or be heavily informed by identity. Third, TikTok has long prioritized and found success with Generation Z and younger users more broadly (Zeng et al., 2021; Vogels et al., 2022; Stahl & Literat, 2022), which has made generational identity extremely salient on the app, while also implicating political identity, as young people tend to hold political beliefs more cognizant and accepting of diverse identities than older generations (Parker et al., 2019). The papers in this panel consider a wide range of identity characteristics of TikTok users and how these identities shape and are shaped by political discourse on TikTok. Paper 1 builds on TikTok’s targeting of Gen Z, considering the identities of age and generation through a content analysis of political remix on TikTok to uncover how younger users use TikTok for political activism as compared to their older counterparts, and finding evidence that TikTok is a powerful site of collective action. Also building from TikTok’s appeal to GenZ, Paper 2 presents a digital ethnographic analysis of the Trad-Wife phenomena on TikTok, offering that TikTok quietly (and thus insidiously) offers space for the cultivation of Christian Nationalist, ‘gentle fascisms’ within GenZ women, often without mention of ‘politics’ at all. Paper 3 offers a computational content analysis of political posts on TikTok with a focus on the interactions between identity and partisanship, and particularly the ways in which creators of marginalized identities on the right act as identity entrepreneurs, offering conservative critiques of their identity groups in ways which find popularity among conservative audiences of hegemonic identities. Finally, Paper 4 looks at differences in how TikTok users respond to male and female politicians’ TikTok videos using a combination of computational and qualitative methods, with exploratory analysis suggesting that male politicians receive more neutral and positive comments than female politicians. By focusing on identity and political discourse on TikTok, we recognize the wide range of political activity occurring on a platform often denigrated as frivolous, and foreground the importance of identity characteristics to the technological and social shaping of these dialogues.

Learn More >>

Generative AI and User-Generated Content: Evidence from Online Reviews

Samsun Knight and Yakov Bart

How has generative artificial intelligence affected the quality and quantity of user-generated content? Our analysis of restaurant reviews on Yelp.com and product reviews on Amazon.com shows that, contrary to prior lab evidence, use of generative AI in text generation is associated with significant declines in online content quality. These results are similar both in OLS and in two- period differences-in-differences estimation based on within-reviewer changes in AI use. We also find that use of generative AI is associated with increases in per-reviewer quantity of content, and document heterogeneity in the observed effect between expert and non-expert reviewers, with the strongest declines in quality associated with use by non-experts.

Learn More >>

11 – 20