American Sociological Review, Forthcoming
Although conspiracies are consequential and social, online conspiracy groups that consist of individuals (and bots) seeking to explain events or a system have been neglected in sociology. We extract conspiracy talk about the COVID-19 pandemic on Twitter and employ the biterm topic model (BTM) to provide a descriptive baseline about the discursive and social structure of online conspiracy groups. We find that individuals enter these communities through a gateway conspiracy theory before proceeding to extreme theories and adopt more diverse conspiracy theories than bots. Event-history analyses show that individuals tweet new conspiracy theories, and tweet inconsistent theories simultaneously, when they face a threat proxied by a rising COVID-19 case rate and receive attention from others via retweets. By contrast, bots are less responsive to rising case rates, but are more consistent, as they mainly tweet about how COVID-19 was deliberately created by sinister agents. These findings suggest that human beings are bricoleurs who use conspiracy theories to make sense of COVID-19 while bots seek to create moral panic. Taken together, our findings suggest that conspiracy talk by individuals is defensive in nature whereas bots engage in offense.
Henrich Greve, Hayagreeva Rao, Paul Vicinanza, and Echo Zhou
Relative frequencies of Covid-19 Conspiracies on Twitter
Quarterly Earnings Calls as an Interrogation Game: Implications for Analyst Stock Recommendations
Job Market Paper
Many forms of social interaction in market economies, from job interviews to entrepreneurial pitches, are a type of “interrogation game.” In these interactions, the interrogator seeks accurate and replete information from a subject who aims to appear forthcoming while simultaneously concealing sensitive information. Drawing up sociological conversation analysis, I argue the unique question-answer structure of the interrogation game imbues the interrogator with discursive control. A core question for the subject, therefore, is whether they should try to control the conversation or assume the subordinate interactive role. To answer this question empirically, I develop an unsupervised word-embedding-based method to computationally quantify conversational control as deferential acts of alignment by one’s conversation partners. I validate this measure against hand-annotated data and apply it to study the conversations between sell-side analysts and executives in over 50,000 quarterly earnings calls (QECs). Contrary to theories of agenda setting in conversation, I find that executives achieve the greatest success, operationalized as more post-call stock upgrades and fewer post-call stock downgrades by analysts, when analysts lead the conversation. I discuss the implications of these results for the sociological study of social interaction, computational conversation analysis, and executive communication in QECs.
Example computations of conversational alignment using word embeddings
A. Likelihood of firm founded by founder wrongdoing history. B. Rates of wrongdoing by new ventures by firms with and without a founder with a prior history of wrongdoing
The Dark Side of Entrepreneurship: Wrongdoers and the Formation of Fraud Trait Ventures in the Financial Advisory Industry
In preparation for resubmission, Management Science
The bright side of entrepreneurship has been emphasized at the expense of the dark side. We study the formation of fraud-trait organizations: new ventures created with the purpose of committing more wrongdoing. We predict that prior offenders–those with a prior conviction of wrongdoing–are more likely than clean employees to create new firms because they face labor market constraints and self-employment enables the evasion of scrutiny and punishment. We hypothesize that their firms commit significantly more wrongdoing because wrongdoing founders a) hire other wrongdoers and b) imprint the new venture with routines that encourage misconduct. We test our model using the complete population of all 1.2 million registered financial advisers in the United States between 2008 and 2018. Consistent with our theory, wrongdoing quadruples the probability a financial advisor starts a new firm in the next year. We find that ventures started by prior offenders are 100 times more likely to hire other wrongdoers and commit 21 times more wrongdoing than comparable ventures opened by clean founders. Thus, our findings indicate that new ventures are progenitors of wrongdoing rather than sources of redemption for founders.
Paul Vicinanza, Mark Egan, Gregor Matvos, Hayagreeva Rao, and Amit Seru
R&R, PNAS Nexus
Where do prescient ideas—those that initially challenge conventional assumptions but later achieve widespread acceptance—come from? Although their outcomes in the form of technical innovation are readily observed, the underlying ideas that eventually change the world are often obscured. Here we develop a novel method that uses deep learning to unearth the markers of prescient ideas from the language used by individuals and groups. Our language-based measure identifies prescient actors and documents that prevailing methods would fail to detect. Applying our model to corpora spanning the disparate worlds of politics, law, and business, we demonstrate that it reliably detects prescient ideas in each domain. Moreover, counter to many prevailing intuitions, prescient ideas emanate from each domain's periphery rather than its core. These findings suggest that the propensity to generate far-sighted ideas may be as much a property of contexts as of individuals.
Paul Vicinanza, Amir Goldberg, and Sameer B. Srivastava
Illustration of how prescience is computed using Bidirectional Encoder Representations from Transformers (BERT).
Network of firm mobility, colored by percentage of employees with a history of prior misconduct from blue (low) to red (high)
Sorting Amplifies Peer Effects and Recidivism: Evidence from the Financial Advisory Industry
The literature on organizational wrongdoing has been fragmented between firm-centric accounts of corrupt organizational cultures or peer effects and broader industry-level perspectives. We integrate these literatures, documenting how assortative matching in the labor market exacerbates the peer effect of employee misconduct. We study employee wrongdoing using the complete population of 1.2 million registered financial advisers in the United States from 2008 to 2018. Using firm mergers as an exogenous source of coworkers, we estimate the treatment effect of coworker misconduct and find that prior offenders are six times more susceptible to negative peer effects than their clean coworkers. Whereas prior work treats selection as a confounder to peer effects estimation, we explicitly model the allocation of employees into firms and find that prior offenders are placed together, amplifying peer effects and recidivism. To isolate peer effects from labor market sorting, we simulate the labor market as a random walk process. Using these simulated data, we find that peer effects increase employee wrongdoing by only 18 percent. By placing wrongdoers together in the same firms, however, labor market sorting increases employee wrongdoing by 80 percent, a result driven by the recidivism of prior offenders. In doing so, we demonstrate how labor market sorting and peer effects work in unison to magnify recidivist behavior.
Paul Vicinanza, Mark Egan, Gregor Matvos, Hayagreeva Rao, and Amit Seru
How Organizational Cultures Shifted During the COVID-19 Pandemic... And What Might Need to Change Back
California Management Review Insights, 2021
Derek N. Brown, Yixi Chen, Hope Harrington, Paul Vicinanza, Jennifer A. Chatman, Amir Goldberg, and Sameer B. Srivastava