Serving Up
Algorithmic Transparency with Strategic Users
Qiaochu Wang et al.
Management Science, forthcoming
Abstract:
Should firms that apply machine learning algorithms in their decision making make their algorithms transparent to the users they affect? Despite the growing calls for algorithmic transparency, most firms keep their algorithms opaque, citing potential gaming by users that may negatively affect the algorithm’s predictive power. In this paper, we develop an analytical model to compare firm and user surplus with and without algorithmic transparency in the presence of strategic users and present novel insights. We identify a broad set of conditions under which making the algorithm transparent actually benefits the firm. We show that, in some cases, even the predictive power of the algorithm can increase if the firm makes the algorithm transparent. By contrast, users may not always be better off under algorithmic transparency. These results hold even when the predictive power of the opaque algorithm comes largely from correlational features and the cost for users to improve them is minimal. We show that these insights are robust under several extensions of the main model. Overall, our results show that firms should not always view manipulation by users as bad. Rather, they should use algorithmic transparency as a lever to motivate users to invest in more desirable features.
What’s Next? Artists’ Music after Grammy Awards
Giacomo Negro, Balázs Kovács & Glenn Carroll
American Sociological Review, forthcoming
Abstract:
Do the cultural works artists produce after receiving major awards change in character? As awards lessen the constraints artists typically face, we argue that award winners receive more opportunities, gain more autonomy, and are more likely to pursue unique creative paths. Empirically, we analyze the consequences of winning a major Grammy award, a high-profile (often status-shifting) honor in the popular music industry. Using a neural learning approach, we examine the subsequent artistic differentiation of albums of award winners from albums of other artists. We analyze whether the music styles and sonic content of post-Grammy albums of winners change, and whether they become more or less similar to the combined corpus of albums of other artists. In panel regression estimates, we find that after winning a Grammy, artists tend to release albums that stand out more stylistically from other artists. Surprisingly, artists who were nominated but did not win a Grammy became more similar to other artists than they were before the nomination. The findings suggest symbolic awards can regularly induce change and affect the heterogeneity of cultural products.
Political Costs and Strategic Corporate Communication
Christine Cuny, Jungbae Kim & Mihir Mehta
NYU Working Paper, May 2022
Abstract:
Do firms strategically use advertising campaigns when subject to the threat of political costs? Communication via advertising can assuage public concerns, which, in turn, reduces the pressure on elected officials to impose political costs on the industry. We identify expected political costs using cases of repeated industry testimony at congressional hearings. To disentangle strategic advertising in response to the threat of political costs from advertising for other reasons (e.g., reputation building or to generate sales), we exploit the fact that only politicians overseeing industry-relevant hearings can impose costs on a given industry. We find that subsequent to these hearings, affected industries increase their advertising by 132% more in the electorates of the politicians overseeing the hearings, relative to the increase in the electorates of other politicians. The effects are magnified in the electorates served by the most senior politicians on the committees, in election years, and when the hearings are longer or include a higher proportion of legislation-related words. Moreover, our results are not driven by politicians' decisions to serve on committees relevant to their local-area firms. In sum, our findings provide novel evidence about corporate communication with non-investor stakeholders.
Do employees' tattoos leave a mark on customers' reactions to products and organizations?
Enrica Ruggs & Mikki Hebl
Journal of Organizational Behavior, July 2022, Pages 965-982
Abstract:
Previous research has shown negative evaluations of tattooed employees in the workplace, particularly in white-collar jobs and by hiring managers (e.g., Henle et al., 2021, as they are perceived to possibly damage an organization's image. Drawing on the stereotype and stigma literatures (e.g., Kunda & Spencer, 2003, Zhang et al., 2021), we examined how customers evaluate tattooed employees and the organizations for which they work. We also explored the role of tattoo-related stereotypes as a mechanism to explain the influence of employee tattoos on customers' reactions. Across two studies, we found that customers held both negative and positive stereotypes about tattooed employees but they did not display more negative attitudes or behaviors toward tattooed (vs. non-tattooed) employees. Further, in white-collar jobs that involve artistic skills, tattooed employees were viewed more positively, which in turn was related to greater hiring intentions for these employees compared to non-tattooed employees. We discuss implications of our findings with respect to the shifting nature of tattoos as stigma, the role of stereotype application in understanding tattoo stigma, and the value of considering greater contextual factors in the evaluation of how tattooed employees affect organizations.
Creative artificial intelligence and narrative transportation
Tanja Messingschlager & Markus Appel
Psychology of Aesthetics, Creativity, and the Arts, forthcoming
Abstract:
Artificial intelligence (AI) is increasingly used to accomplish complex tasks, including the creation of artworks and entertainment products. Our focus here is on user responses to AI systems as authors of fictional stories. Across two experiments, we examined how the information that a story was written by AI influences narrative transportation and related experiences. In Experiment 1 (N = 325) the information that an AI had created a short story (contemporary fiction) reduced narrative transportation into this story. Experiment 2 (N = 489) was an extended replication in which genre differences (contemporary fiction vs. science fiction) were addressed. As expected, ostensible AI authorship reduced transportation, but this effect was qualified by genre: Whereas the AI-authorship effect was replicated for contemporary fiction stories, transportation did not differ between human and AI authorship when participants read science fiction stories. Across both experiments, individual differences (openness, affinity for technology and attitude toward AI) did not moderate the effect of AI authorship on any of the dependent variables.
The speed of stories: Semantic progression and narrative success
Henrique Laurino Dos Santos & Jonah Berger
Journal of Experimental Psychology: General, forthcoming
Abstract:
Why are some narratives more successful? Although this question has ancient roots, studying it empirically has been challenging. We suggest that semantic progression (i.e., semantic similarity between adjoining portions of a narrative) might shape audience responses but that this role changes over the course of a narrative. Specifically, although slower semantic progression (i.e., greater semantic similarity between adjoining portions) is beneficial at the beginning of narratives, faster semantic progression is beneficial toward the end. To test this possibility, we used natural language processing and machine learning to analyze over 40,000 movie scripts. Consistent with our theorizing, deep-learning-based embeddings find that movies with slower semantic progression early and faster semantic progression later are evaluated more positively. Analysis of over 10,000 TV episodes finds similar results. Overall, these findings shed light on what makes narratives engaging, deepen understanding of what drives cultural success, and underscore the value of emerging computational approaches to understand human behavior.
Social learning and local consumption amenities: Evidence from Yelp
Elliot Anenberg, Chun Kuang & Edward Kung
Journal of Industrial Economics, June 2022, Pages 294-322
Abstract:
We estimate the effect of social learning through Yelp on average restaurant quality across different types of markets. We use a regression discontinuity design to show that restaurants are more likely to exit receiving low ratings on Yelp. The effects of ratings on exit are especially strong for restaurants in zipcodes with high Yelp usage (e.g. more urban neighborhoods with higher income and education). Simulations show that in the long-run, the selective restaurant exit caused by Yelp increases average restaurant quality by 0.098 Yelp stars in the average zipcode, and by at least 0.238 stars in high usage markets.