18 January 2022 in Journal of Medical Internet Research
Background: Communication campaigns using social media can raise public awareness; however, they are difficult to sustain. A barrier is the need to generate and constantly post novel but on-topic messages, which creates a resource-intensive bottleneck. Objective: In this study, we aim to harness the latest advances in artificial intelligence (AI) to build a pilot system that can generate many candidate messages, which could be used for a campaign to suggest novel, on-topic candidate messages. The issue of folic acid, a B-vitamin that helps prevent major birth defects, serves as an example; however, the system can work with other issues that could benefit from higher levels of public awareness. Methods: We used the Generative Pretrained Transformer-2 architecture, a machine learning model trained on a large natural language corpus, and fine-tuned it using a data set of autodownloaded tweets about #folicacid. The fine-tuned model was then used as a message engine, that is, to create new messages about this topic. We conducted a web-based study to gauge how human raters evaluate AI-generated tweet messages compared with original, human-crafted messages. Results: We found that the Folic Acid Message Engine can easily create several hundreds of new messages that appear natural to humans. Web-based raters evaluated the clarity and quality of a human-curated sample of AI-generated messages as on par with human-generated ones. Overall, these results showed that it is feasible to use such a message engine to suggest messages for web-based campaigns that focus on promoting awareness. Conclusions: The message engine can serve as a starting point for more sophisticated AI-guided message creation systems for health communication. Beyond the practical potential of such systems for campaigns in the age of social media, they also hold great scientific potential for the quantitative analysis of message characteristics that promote successful communication. We discuss future developments and obvious ethical challenges that need to be addressed as AI technologies for health persuasion enter the stage.
1 June 2020 in Journal of Communication
In this special issue devoted to speaking across communication subfields, we introduce a domain general explanatory framework that integrates biological explanation with communication science and organizes our field around a shared explanatory empirical model. Specifically, we draw on David Marr’s classical framework, which subdivides the explanation of human behavior into three levels: computation (why), algorithm (what), and implementation (how). Prior theorizing and research in communication has primarily addressed Marr’s computational level (why), but has less frequently investigated algorithmic (what) or implementation (how all communication phenomena emerge from and rely on biological processes) explanations. Here, we introduce Marr’s framework and apply it to three research domains in communication science—audience research, persuasion, and social comparisons—to demonstrate what a unifying framework for explaining communication across the levels of why, what, and how can look like, and how Marr’s framework speaks to and receives input from all subfields of communication inquiry.
1 October 2020 in Annals of the International Communication Association
The Limited Capacity Model of Motivated Mediated Message Processing (LC4MP) aims to understand message processing dynamics. Despite 20 years of research, no meta-analysis has assessed LC4MP effects. We conducted a meta-analysis of the model to examine three theoretical research domains in the LC4MP: cognitive load, motivation, and memory. Results from 142 articles and 683 effects demonstrate that pooled effect sizes for research domain range from r = .314–.398. Effect sizes vary by measurement modality with self-report resulting in the largest pooled effect size, followed by behavioral, and finally psychophysiological measures. We did not detect evidence of publication bias. These findings offer meta-analytic support for LC4MP research domains and are discussed in terms of falsifiability, predictive and explanatory power.
11 November 2021 in Journal of Communication
Flow is thought to occur when both task difficulty and individual ability are high. Flow experiences are highly rewarding and are associated with well-being. Importantly, media use can be a source of flow. Communication scholars have a long history of theoretical inquiry into how flow biases media selection, how different media content results in flow, and how flow influences media processing and effects. However, the neurobiological basis of flow during media use is not well understood, limiting our explanatory capacity to specify how media contribute to flow or well-being. Here, we show that flow is associated with a flexible and modular brain-network topology, which may offer an explanation for why flow is simultaneously perceived as high-control and effortless, even when the task difficulty is high. Our study tests core predictions derived from synchronization theory, and our results provide qualified support for the theory while also suggesting important theoretical updates.
1 July 2022 in Journal of Media Psychology
Online contexts are becoming a widely available space to disseminate health information and target specific populations for health campaigns. Limited evidence for health message engagement in these contexts exists. This study draws on the elaboration likelihood model and construal-level theory to predict processing time and recall when individuals are presented with messages for or against electronic cigarette use from socially close or distant sources. Participants (N = 159) were shown messages about electronic cigarettes, designed to look like tweets, from socially close and socially distant message senders. Processing times were highest for pro-attitudinal messages while messages from socially close sources were more likely to be recalled, and furthering social distance increased the difference in processing times for pro- and counter-attitudinal messages. We demonstrate the applicability of behavioral measures in online studies, while finding that attitudes, social distance, and their interaction affect measures of message processing. These findings suggest further exploration may be needed to differentiate between processing time and counterarguing. From our findings, we offer applied practitioners guidance on how to develop messages that target audiences will spend more time considering and are more likely to remember.
13 December 2021 in Health Communication
Most adults in the United States (U.S.) fail to meet guidelines for physical activity (PA) and nutrition outlined by the Centers for Disease Control. One important predictor of engagement in healthy behavior is support from one’s romantic partner. However, messages from partners may fail to motivate healthy behavior if they threaten recipients’ face and cause reactance. The present study examines face-threatening acts (FTAs) and face management strategies (FMSs) in conversations between romantic couples and their associations with reactance, healthy eating, and PA behaviors. Cohabitating couples (N = 70) were recruited, and one member of the couple completed a 10-day diary survey in which they reported on daily memorable conversations they had with their partner about PA and/or healthy eating. Participants completed measures of positive and negative face threat, as well as the extent to which they engaged in healthy eating and PA that day. Trained raters assessed reported conversations for positive and negative face threat as well as positive and negative FMSs. Results indicate that both participant and trained raters’ reports of positive face threat were associated with increased reactance, whereas only participants’ reports of negative face threat were associated with increased reactance. Both positive and negative FMSs were associated with reduced reactance. Reactance was negatively associated with healthy eating and PA. Results are discussed in terms of implications for reactance and politeness theories, as well as pragmatic implications for the millions of partnered individuals in the U.S. seeking to improve their health.