Large language models (LLMs) are popular tools for creative ideation, but have been shown to homogenize outputs across users. We test if approaching an AI tool with high human (vs. low) human agency can mitigate this homogenization effect by encouraging people to use AI to augment their creativity, rather than offload it. Participants were experimentally assigned to one of three conditions (high-agency approach with AI access, low-agency approach with AI access, or a human-only control) and generated creative uses for everyday objects. Contrary to our expectations, ideas did not differ in individual-level quality (overall creativity, originality, and usefulness). However, a preregistered similarity-to-centroid analysis and an exploratory cluster analysis provided convergent evidence of AI-induced homogenization among the low-agency condition. Thus, while AI has enabled greater ideational fluency, our research suggests the degree of agency with which people approach their AI tool has downstream consequences on collective creativity.

AI-based tools that mediate, enhance or generate parts of video communication may interfere with how people evaluate trustworthiness and credibility. In two preregistered online experiments (N = 2,000), we examined whether AI-mediated video retouching, background replacement and avatars affect interpersonal trust, people’s ability to detect lies and confidence in their judgments. Participants watched short videos of speakers making truthful or deceptive statements across three conditions with varying levels of AI mediation. We observed that perceived trust and confidence in judgments declined in AI-mediated videos, particularly in settings in which some participants used avatars while others did not. However, participants’ actual judgment accuracy remained unchanged, and they were no more inclined to suspect those using AI tools of lying. Our findings provide evidence against concerns that AI mediation undermines people’s ability to distinguish truth from lies, and against cue-based accounts of lie detection more generally. They highlight the importance of trustworthy AI mediation tools in contexts where not only truth, but also trust and confidence matter.

Concerns about smartphone dependency have sparked interest in minimal mobile phones: devices supporting basic communication without social apps, web browsing, or games. These design choices are thought to improve well-being, but have not been tested empirically. We conducted a first-of-its-kind longitudinal experiment examining the effects of switching from smartphones to minimal mobile phones on young adults’ psychological well-being over a week (n = 166). To account for individual variation in intrinsic motivation to try a minimal phone, we employed a quasi-experimental design comparing the outcomes of three groups: 1) high-interest volunteers who were asked to use minimal phones or participants that were randomly assigned to either 2) use minimal phones or 3) continue using their own smartphones. Results showed that switching to a minimal phone effectively reduced phone and social media use. However, only high-interest volunteers - those intrinsically motivated to participate - showed significant within-person changes in psychological well-being, reporting reduced stress, increased life satisfaction, and less FoMo. No effects on well-being were observed for those assigned to use the phone. Our results suggest that switching to a minimal mobile phone may support some motivated individuals in improving their sense of agency and well-being in an increasingly connected digital world.

Users trust algorithms more when they can predict the algorithms’ behavior. Simple algorithms trivially yield predictively accurate mental models, but modern AI algorithms have often been assumed too complex for people to build predictive mental models, especially in the social media domain. In this paper, we describe conditions under which even complex algorithms can yield predictive mental models, opening up opportunities for a broader set of human-centered algorithms. We theorize that users will form an accurate predictive mental model of an algorithm’s behavior if and only if the algorithm simultaneously satisfies three criteria: (1) cognitive availability of the underlying concepts being modeled, (2) concept compactness (does it form a single cognitive construct?), and (3) high alignment between the person’s and algorithm’s execution of the concept. We evaluate this theory through a pre-registered experiment ( N = 1250) where users predict behavior of 25 social media feed ranking algorithms that vary on these criteria. We find that even complex (e.g., LLM-based) algorithms enjoy accurate prediction rates when they meet all criteria, and even simple (e.g., basic term count) algorithms fail to be predictable when a single criterion fails. We also find that these criteria determine outcomes beyond prediction accuracy, such as which mental models users deploy to make their predictions.

Personalized algorithms are central to how people discover information and engage with media online. Drawing on interviews and screen-sharing sessions with TikTok users (N=27), we extend the algorithmic crystal framework, which conceptualizes personalized algorithms as reflective surfaces through which users may interpret their experiences with content in relation to their own self-concepts. This research expands the framework to account for the interpersonal dynamics that emerge from user engagement with algorithmic feeds. We found that users who feel ''seen'' by the algorithm use its personalized content recommendations for social signaling: sharing content that represents themselves (''this is me''), acknowledges how they see others (''this is you''), and affirms shared identities (''this is us''). We suggest that these dynamics give rise to a hybrid form of digital selfhood simultaneously shaped by algorithmic profiling and networked social interaction-blurring existing separations in digital identity theory. We also build on the concept of diffracted belonging-the experience of recognizing aspects of oneself in the content of diverse others-to explore how users interpret algorithmically-recommended content as reflective of the self. Our findings suggest that such moments of recognition may contribute to shifts in self-perception and support ongoing processes of identity development. Finally, we illustrate how users engage in the strategic refinement of their feeds to manage how they feel while using the platform. Our findings suggest that this process involves reflective, and sometimes effortful, negotiation with the algorithm, highlighting the co-produced nature of mood management in everyday human-algorithm interactions. Together, these findings underscore the interpersonal and psychological dynamics of interacting with personalized algorithms and provide insights into how social communication and identity work unfold in algorithmically-mediated environments.

Exposure to misinformation poses significant challenges to democratic processes and public health, particularly during critical events like elections. This study adopts a user-centric approach to analyze the linguistic features of misinformation actually consumed by individuals during web browsing. Using data from a nationally representative panel of 1,240 American adults and their web-browsing data (21M URL visits) during the 2020 U.S. Presidential Election, we examine linguistic and topical differences in the content of 91K unique misinformation and hard news webpages by utilizing natural language processing techniques and Large Language Models. We find that misinformation consumed by users is generally easier to read, exhibits higher negative sentiment, and employs more moral language than hard news. We also find significant linguistic variations across topics--misinformation can be diverse and vary in linguistic features depending on the subject matter. We also identify heterogeneity across key user characteristics: older adults consume more misinformation about COVID-19 and health, with content showing more negative sentiment and fewer moral terms than expected. Republicans engage with misinformation characterized by more negative sentiment and higher moral language, focusing less on health topics and more on social and political issues. These results highlight the importance of a user-centric approach and suggest that interventions to combat misinformation should be tailored to specific topics and user characteristics for greater effectiveness.

Mounting evidence indicates that the artificial intelligence (AI) systems that rank our social media feeds bear nontrivial responsibility for amplifying partisan animosity: negative thoughts, feelings, and behaviors toward political out-groups. Can we design these AIs to consider democratic values such as mitigating partisan animosity as part of their objective functions? We introduce a method for translating established, vetted social scientific constructs into AI objective functions, which we term societal objective functions, and demonstrate the method with application to the political science construct of anti-democratic attitudes. Traditionally, we have lacked observable outcomes to use to train such models-however, the social sciences have developed survey instruments and qualitative codebooks for these constructs, and their precision facilitates translation into detailed prompts for large language models. We apply this method to create a democratic attitude model that estimates the extent to which a social media post promotes anti-democratic attitudes, and test this democratic attitude model across three studies. In Study 1, we first test the attitudinal and behavioral effectiveness of the intervention among US partisans (N=1,380) by manually annotating (alpha=.895) social media posts with anti-democratic attitude scores and testing several feed ranking conditions based on these scores. Removal (d=.20) and downranking feeds (d=.25) reduced participants' partisan animosity without compromising their experience and engagement. In Study 2, we scale up the manual labels by creating the democratic attitude model, finding strong agreement with manual labels (rho=.75). Finally, in Study 3, we replicate Study 1 using the democratic attitude model instead of manual labels to test its attitudinal and behavioral impact (N=558), and again find that the feed downranking using the societal objective function reduced partisan animosity (d=.25). This method presents a novel strategy to draw on social science theory and methods to mitigate societal harms in social media AIs.

The QAnon movement has been credited with spreading disinformation and fueling online radicalization in the United States and around the globe. While some research has documented publicly-visible communications and engagements with the QAnon movement, little work has examined individuals' actual exposure to QAnon content. In this paper, we investigate the extent to which Americans are exposed to QAnon websites, in what contexts, and to what effect. We employ a mixed methods review of 21 million website visits collected from a nationally representative sample of 1,238 American adults across laptops, smartphones, and tablets during the 2020 U.S. presidential election. Quantitative techniques reveal overall levels of exposure to QAnon and who is more likely to be exposed, and qualitative techniques provide rich information about how participants came to be exposed to QAnon and how it fit within their broader media diets. We find that: (1) exposure to QAnon websites is limited and stratified by political ideology and news consumption; (2) exposure tends to occur within right-wing media ecosystems that align with QAnon beliefs; and (3) mixed methods approaches to analyzing digital trace data can provide rich insights that contextualize quantitative techniques. We discuss the implications of our findings for the design of interventions to lessen exposure to problematic material online and future research on the spread of disinformation and extremist content.

Toxic comments are the top form of hate and harassment experienced online. While many studies have investigated the types of toxic comments posted online, the effects that such content has on people, and the impact of potential defenses, no study has captured the behaviors of the accounts that post toxic comments or how such attacks are operationalized. In this paper, we present a measurement study of 929K accounts that post toxic comments on Reddit over an 18 month period. Combined, these accounts posted over 14 million toxic comments that encompass insults, identity attacks, threats of violence, and sexual harassment. We explore the impact that these accounts have on Reddit, the targeting strategies that abusive accounts adopt, and the distinct patterns that distinguish classes of abusive accounts. Our analysis informs the nuanced interventions needed to curb unwanted toxic behaviors online.

In the summer of 2021, users on the livestreaming platform Twitch were targeted by a wave of "hate raids," a form of attack that overwhelms a streamer's chatroom with hateful messages, often through the use of bots and automation. Using a mixed-methods approach, we combine a quantitative measurement of attacks across the platform with interviews of streamers and third-party bot developers. We present evidence that confirms that some hate raids were highly-targeted, hate-driven attacks, but we also observe another mode of hate raid similar to networked harassment and specific forms of subcultural trolling. We show that the streamers who self-identify as LGBTQ+ and/or Black were disproportionately targeted and that hate raid messages were most commonly rooted in anti-Black racism and antisemitism. We also document how these attacks elicited rapid community responses in both bolstering reactive moderation and developing proactive mitigations for future attacks. We conclude by discussing how platforms can better prepare for attacks and protect at-risk communities while considering the division of labor between community moderators, tool-builders, and platforms.

Artificial Intelligence (AI) is a transformative force in communication and messaging strategy, with potential to disrupt traditional approaches. Large language models (LLMs), a form of AI, are capable of generating high-quality, humanlike text. We investigate the persuasive quality of AI-generated messages to understand how AI could impact public health messaging. Specifically, through a series of studies designed to characterize and evaluate generative AI in developing public health messages, we analyze COVID-19 pro-vaccination messages generated by GPT-3, a state-of-the-art instantiation of a large language model. Study 1 is a systematic evaluation of GPT-3's ability to generate pro-vaccination messages. Study 2 then observed peoples' perceptions of curated GPT-3-generated messages compared to human-authored messages released by the CDC (Centers for Disease Control and Prevention), finding that GPT-3 messages were perceived as more effective, stronger arguments, and evoked more positive attitudes than CDC messages. Finally, Study 3 assessed the role of source labels on perceived quality, finding that while participants preferred AI-generated messages, they expressed dispreference for messages that were labeled as AI-generated. The results suggest that, with human supervision, AI can be used to create effective public health messages, but that individuals prefer their public health messages to come from human institutions rather than AI sources. We propose best practices for assessing generative outputs of large language models in future social science research and ways health professionals can use AI systems to augment public health messaging.

Whose labels should a machine learning (ML) algorithm learn to emulate? For ML tasks ranging from online comment toxicity to misinformation detection to medical diagnosis, different groups in society may have irreconcilable disagreements about ground truth labels. Supervised ML today resolves these label disagreements implicitly using majority vote, which overrides minority groups’ labels. We introduce jury learning, a supervised ML approach that resolves these disagreements explicitly through the metaphor of a jury: defining which people or groups, in what proportion, determine the classifier’s prediction. For example, a jury learning model for online toxicity might centrally feature women and Black jurors, who are commonly targets of online harassment. To enable jury learning, we contribute a deep learning architecture that models every annotator in a dataset, samples from annotators’ models to populate the jury, then runs inference to classify. Our architecture enables juries that dynamically adapt their composition, explore counterfactuals, and visualize dissent. A field evaluation finds that practitioners construct diverse juries that alter 14% of classification outcomes.

Because algorithm audits are conducted by technical experts, audits are necessarily limited to the hypotheses that experts think to test. End users hold the promise to expand this purview, as they inhabit spaces and witness algorithmic impacts that auditors do not. In pursuit of this goal, we propose end-user audits-system-scale audits led by non-technical users-and present an approach that scaffolds end users in hypothesis generation, evidence identification, and results communication. Today, performing a system-scale audit requires substantial user effort to label thousands of system outputs, so we introduce a collaborative filtering technique that leverages the algorithmic system's own disaggregated training data to project from a small number of end user labels onto the full test set. Our end-user auditing tool, IndieLabel, employs these predicted labels so that users can rapidly explore where their opinions diverge from the algorithmic system's outputs. By highlighting topic areas where the system is under-performing for the user and surfacing sets of likely error cases, the tool guides the user in authoring an audit report. In an evaluation of end-user audits on a popular comment toxicity model with 17 non-technical participants, participants both replicated issues that formal audits had previously identified and also raised previously underreported issues such as under-flagging on veiled forms of hate that perpetuate stigma and over-flagging of slurs that have been reclaimed by marginalized communities.

This research examines how TikTok users conceptualize and engage with personalized algorithms on the TikTok platform. Using qualitative methods, we analyzed 24 interviews with TikTok users to explore how algorithmic personalization processes inform people's understanding of their identities as well as shape their orientation to others. Building on insights from our qualitative data and previous scholarship on algorithms and identity, we propose a novel conceptual model to understand how people think about and interact with personalized algorithmic systems. Drawing on the metaphor of crystals and their properties, the algorithmic crystal framework is an analytic frame that captures user understandings of how personalized algorithms (1) interact with user identity by reflecting user self-concepts that are both multifaceted and dynamic and (2) shape perspectives on others encountered through the algorithm, by orienting users to recognize parts of themselves refracted in other users and to experience ephemeral, diffracted connections with groups of similar others. We describe how the algorithmic crystal framework can extend theory and inform new lines of research around the implications of algorithms in self-concept development and social life.

Algorithmically-mediated content is both a product and producer of dominant social narratives, and it has the potential to impact users' beliefs and behaviors. We present two studies on the content and impact of gender and racial representation in image search results for common occupations. In Study 1, we compare 2020 workforce gender and racial composition to that reflected in image search. We find evidence of underrepresentation on both dimensions: women are underrepresented in search at a rate of 42% women for a field with 50% women; people of color are underrepresented with 16% in search compared to an occupation with 22% people of color (the latter being proportional to the U.S. workforce). We also compare our gender representation data with that collected in 2015 by Kay et al., finding little improvement in the last half-decade. In Study 2, we study people's impressions of occupations and sense of belonging in a given field when shown search results with different proportions of women and people of color. We find that both axes of representation as well as people's own racial and gender identities impact their experience of image search results. We conclude by emphasizing the need for designers and auditors of algorithms to consider the disparate impacts of algorithmic content on users of marginalized identities.

AI-Mediated Communication (AI-MC) is interpersonal communication that involves an artificially intelligent system that can modify, augment, or even generate content to achieve communicative and relational goals. AI-MC is increasingly involved in human communication and has the potential to impact core aspects of human communication, such as language production, interpersonal perception and task performance. Through a between-subjects experimental design we examine how these processes are influenced when integrating AI-generated language in the form of suggested text responses (Google's smart replies) into a text-based referential communication task. Our study replicates and extends the impacts of a positivity bias in AI-generated language and introduces the adjacency pair framework into the study of AI-MC. We also find preliminary yet mixed evidence to suggest that AI-generated language has the potential to undermine some dimensions of interpersonal perception, such as social attraction. This study contributes important concepts for future work in AI-MC and offers findings with implications for the design of AI systems in human-to-human communication.

With the emergence of conversational artificial intelligence (AI) agents, it is important to understand the mechanisms that influence users' experiences of these agents. In this paper, we study one of the most common tools in the designer's toolkit: conceptual metaphors. Metaphors can present an agent as akin to a wry teenager, a toddler, or an experienced butler. How might a choice of metaphor influence our experience of the AI agent? Sampling a set of metaphors along the dimensions of warmth and competence---defined by psychological theories as the primary axes of variation for human social perception---we perform a study $(N=260)$ where we manipulate the metaphor, but not the behavior, of a Wizard-of-Oz conversational agent. Following the experience, participants are surveyed about their intention to use the agent, their desire to cooperate with the agent, and the agent's usability. Contrary to the current tendency of designers to use high competence metaphors to describe AI products, we find that metaphors that signal low competence lead to better evaluations of the agent than metaphors that signal high competence. This effect persists despite both high and low competence agents featuring identical, human-level performance and the wizards being blind to condition. A second study confirms that intention to adopt decreases rapidly as competence projected by the metaphor increases. In a third study, we assess effects of metaphor choices on potential users' desire to try out the system and find that users are drawn to systems that project higher competence and warmth. These results suggest that projecting competence may help attract new users, but those users may discard the agent unless it can quickly correct with a lower competence metaphor. We close with a retrospective analysis that finds similar patterns between metaphors and user attitudes towards past conversational agents such as Xiaoice, Replika, Woebot, Mitsuku, and Tay.

Live streaming services allow people to concurrently consume and comment on media events with other people in real time. Durkheim's theory of "collective effervescence" suggests that face-to-face encounters in ritual events conjure emotional arousal, so people often feel happier and more excited while watching events like the Super Bowl with family and friends through the television than if they were alone. Does a stronger emotional intensity also occur in live streaming? Using a large-scale dataset of comments posted to news and media events on YouTube, we address this question by examining emotional intensity in live comments versus those produced retrospectively. Results reveal that live comments are overall more emotionally intense than retrospective comments across all temporal periods and all event types examined. Findings support the emotional amplification hypothesis and provide preliminary evidence for shared attention theory in explaining the amplification effect. These findings have important implications for live streaming platforms to optimize resources for content moderation and to improve psychological well-being for content moderators, and more broadly as society grapples with using technology to stay connected during social distancing required by the COVID-19 pandemic.

We are entering an era of AI-Mediated Communication (AI-MC) where interpersonal communication is not only mediated by technology, but is optimized, augmented, or generated by artificial intelligence. Our study takes a first look at the potential impact of AI-MC on online self-presentation. In three experiments we test whether people find Airbnb hosts less trustworthy if they believe their profiles have been written by AI. We observe a new phenomenon that we term the Replicant Effect: Only when participants thought they saw a mixed set of AI- and human-written profiles, they mistrusted hosts whose profiles were labeled as or suspected to be written by AI. Our findings have implications for the design of systems that involve AI technologies in online self-presentation and chart a direction for future work that may upend or augment key aspects of Computer-Mediated Communication theory.

This paper examines relationships between perceptions of warmth and competence, emotional responses, and behavioral tendencies in the context of social robots. Participants answered questions about these three aspects of impression formation after viewing an image of one of 342 social robots in the Stanford Social Robots Database. Results suggest that people have similar emotional and behavioral reactions to robots as they have to humans; impressions of the robots' warmth and competence predicted specific emotional responses (admiration, envy, contempt, pity) and those emotional responses predicted distinct behavioral tendencies (active facilitation, active harm, passive facilitation, passive harm). However, the predicted relationships between impressions and harmful behavioral tendencies were absent. This novel asymmetry for perceptions and intentions towards robots is deliberated in the context of the computers as social actors framework and opportunities for further research are discussed.

Concern about algorithmically-curated content and its impact on democracy is reaching a fever pitch worldwide. But relative to the role of social media in electoral processes, the role of search results has received less public attention. We develop a theoretical conceptualization of search results as a form of media-search media-and analyze search media in the context of political partisanship in the six months leading up to the 2018 U.S. midterm elections. Our empirical analyses use a total of over 4 million URLs, scraped daily from Google search queries for all candidates running for federal office in the United States in 2018. In our first set of analyses we characterize the nature of search media from the data collected in terms of the types of URLs present and the stability of search results over time. In our second, we annotate URLs' top-level domains with existing measures of political partisanship, examining trends by incumbency, election outcome, and other election characteristics. Among other findings, we note that partisanship trends in search media are largely similar for content about candidates from the two major political parties, whereas there are substantial differences in search media for incumbent versus challenger candidates. This work suggests that longitudinal, systematic audits of search media can reflect real-world political trends. We conclude with implications for web search designers and consumers of political content online.

Self-presentation is a process that is significantly complicated by the rise of algorithmic social media feeds, which obscure information about one's audience and environment. User understandings of these systems, and therefore user ability to adapt to them, are limited, and have recently been explored through the lens of folk theories. To date, little is understood of how these theories are formed, and how they tie to the self-presentation process in social media. This paper presents an exploratory look at the folk theory formation process and the interplay between folk theories and self-presentation via a 28-participant interview study. Results suggest that people draw from diverse sources of information when forming folk theories, and that folk theories are more complex, multifaceted and malleable than previously assumed. This highlights the need to integrate folk theories into both social media systems and theories of self-presentation.

We interact with dozens of web interfaces on a daily basis, making inclusive web design practices more important than ever. This paper investigates the impacts of web interface design on ambient belonging, or the sense of belonging to a community or culture. Our experiment deployed two content-identical webpages for an introductory computer science course, differing only in aesthetic features such that one was perceived as masculine while the other was gender-neutral. Our results confirm that young women exposed to the masculine page are negatively affected, reporting significantly less ambient belonging, interest in the course and in studying computer science broadly. They also experience significantly more concern about others' perception of their gender relative to young women exposed to the neutral page, while no similar effect is seen in young men. These results suggest that gender biases can be triggered by web design, highlighting the need for inclusive user interface design for the web.

The popularity of social media platforms today makes them an important venue for self-presentation, but the unique affordances of these platforms challenge our existing models for understanding self-presentation behavior. In particular, social media provide multiple platforms on which the self may be presented, expand the role other individuals can play in one's own self-presentation, and expand the audience while often simultaneously providing less information about who is in that audience. This paper presents an affordance-based approach to self-presentation on social media platforms rooted in these three challenges and presents a systematic taxonomy for considering aspects of platforms that affect self-presentation. Results from an exploratory study of 193 users suggest significant variation in user perception of our proposed affordances across social media platforms, participant experience levels, and participant personality traits.

Online peer-to-peer platforms like Airbnb allow hosts to list a property (e.g. a house, or a room) for short-term rentals. In this work, we examine how hosts describe themselves on their Airbnb profile pages. We use a mixed-methods study to develop a categorization of the topics that hosts self-disclose in their profile descriptions, and show that these topics differ depending on the type of guest engagement expected. We also examine the perceived trustworthiness of profiles using topic-coded profiles from 1,200 hosts, showing that longer self-descriptions are perceived to be more trustworthy. Further, we show that there are common strategies (a mix of topics) hosts use in self-disclosure, and that these strategies cause differences in perceived trustworthiness scores. Finally, we show that the perceived trustworthiness score is a significant predictor of host choice--especially for shorter profiles that show more variation. The results are consistent with uncertainty reduction theory, reflect on the assertions of signaling theory, and have important design implications for sharing economy platforms, especially those facilitating online-to-offline social exchange.

Words symbolically represent communicative and behavioral intent, and can provide clues to a communicator's future actions in online communication. This paper describes a sociotechnical study conducted from 2008 through 2015 to identify deceptive communicative intent within group context as manifested in language‐action cues. Specifically, this study used an online team‐based game that simulates real‐world deceptive insider scenarios to examine several dimensions of group communication. First, we studied how language‐action cues differ between groups with and groups without a compromised actor. We also examine how these cues differ within groups in terms of the group members' individual and collective interactions with the compromised actor. Finally, we look at how the cues of compromised actors differ from those of noncompromised actors, and how communication behavior changes after an actor is presented with an ethical dilemma. The results of the study further our understanding of language‐action cues as indicators for unmasking a potential deceptive insider.

Self-disclosure is rewarding and provides significant benefits for individuals, but it also involves risks, especially in social media settings. We conducted an online experiment to study the relationship between content intimacy and willingness to self-disclose in social media, and how identification (real name vs. anonymous) and audience type (social ties vs. people nearby) moderate that relationship. Content intimacy is known to regulate self-disclosure in face-to-face communication: people self-disclose less as content intimacy increases. We show that such regulation persists in online social media settings. Further, although anonymity and an audience of social ties are both known to increase self-disclosure, it is unclear whether they (1) increase self-disclosure baseline for content of all intimacy levels, or (2) weaken intimacy's regulation effect, making people more willing to disclose intimate content. We show that intimacy always regulates self-disclosure, regardless of settings. We also show that anonymity mainly increases self-disclosure baseline and (sometimes) weakens the regulation. On the other hand, an audience of social ties increases the baseline but strengthens the regulation. Finally, we demonstrate that anonymity has a more salient effect on content of negative valence.The results are critical to understanding the dynamics and opportunities of self-disclosure in social media services that vary levels of identification and types of audience.

The ubiquity of mobile devices has resulted in more opportunities to interact with more people than ever before. Given a finite capacity for interaction with others, people commonly manage their availability by limiting others' access to them. Prior work has demonstrated the importance of doing so in a relationally sensitive way and identified the butler lie, in which deception is used to manage availability, as a common linguistic strategy. Two key limitations of existing exploratory work, however, are limited samples of primarily students and a focus on media properties in understanding ambiguity that enables butler lies to be plausible. This paper aims to address these issues via a broad field study of deception and butler lies using a novel message-sampling method employed via a custom mobile app. Results show clear evidence of butler lies occurring in a broader population, with some gender differences; and urge adoption of a multi-level framework for understanding ambiguity that also includes private information and infrastructure-level attributes of interaction media.

Big social data, such as that produced by Facebook and Twitter, have the potential to transform the social sciences and lead to advances in understanding human behavior. At the same time, novel large-scale methods and forms of collaboration between academia and industry raise new and important ethical questions.

With an increasing number of online users, the potential danger of online deception grows accordingly - as does the importance of better understanding human behavior online to mitigate these risks. One critical element to address such online threat is to identify intentional deception in spontaneous online communication. For this study, we designed an interactive online game that creates player scenarios to encourage deception. Data was collected and analyzed in October 2014 to identify certain deceptive cues. Players' interactive dialogue was analyzed using linear regression analysis. The results reveal that certain language features are highly significant predictors of deception in synchronous, spontaneous online communication.

While we tend to think of self-presentation as a process executed by the self, reputation management on social network sites, like Facebook, is increasingly viewed as a collective endeavor. The information users share about one another can have significant impacts on impression formation, and at times this other-generated content may be face threatening, or challenging to one's desired self-presentation. However, we know little about the nature of these other-generated face threats and the ways that people perceive them. Using an online survey of 150 Facebook users, we report on what these users consider to be other-generated face threats and how they feel after experiencing them. Results suggest that many face threats result from other Facebook users neglecting or misunderstanding a target's audience and/or self-presentation goals, as well as a target's fear of creating an unwanted association with another Facebook user. Experience of these threats is affected by both individual and situational factors. We also report on a new unique measure capturing Facebook skills.

Motivating students to be active learners is a perennial problem in education, and is particularly challenging in lectures where instructors typically prepare content in ad-vance with little direct student participation. We describe our experience using Twitter as a tool for student "co-construction" of lecture materials. Students were required to post a tweet prior to each lecture related to that day's topic, and these tweets -- consisting of questions, examples and reflections -- were incorporated into the lecture slides and notes. Students reported that they found lectures including their tweets in the class slides to be engaging, interactive and relevant, and nearly 90% of them recommended we use our co-construction approach again.

In an always-connected world, managing one's unavailability for interaction with others can be as important and difficult as coordinating mutual availability. Prior studies have identified the butler lie, a linguistic strategy commonly used to manage unavailability, and examined message-level data to examine how message senders' use of butler lies varies across media and situations. This study is the first to examine how butler lies are perceived by those who receive them. Pairs of student participants provided messages sent to each other in real conversations and indicated whether these messages were deceptive or not. These messages were then passed to the partner, who indicated perceived deception and provided an explanation. Results suggest that participants expect butler lies regularly although not as often as they are actually produced, and participants are not very accurate in identifying butler lies. Moreover, detailed analysis of messages and explanations suggests that butler lies play a relational role that is expected by both parties in a dialog.

Consumers' purchase decisions are increasingly influenced by user-generated online reviews. Accordingly, there has been growing concern about the potential for posting deceptive opinion spam---fictitious reviews that have been deliberately written to sound authentic, to deceive the reader. But while this practice has received considerable public attention and concern, relatively little is known about the actual prevalence, or rate, of deception in online review communities, and less still about the factors that influence it.

In response to the death of a close friend or relative, bereaved individuals can use technology as part of the grieving process. We present a study that analyzes the messages of the friends and family of the deceased to their Facebook profile before and after their passing. Our analysis reveals that mourners use profiles as a way to maintain a continuing bond with the deceased, as well as a way to accomplish specific front stage bereavement communication, such as sharing memories, expressing sorrow and providing social support. These observations may improve the design of social networking technologies so that they remain useful, sensitive tools for the bereaved.

The importance of emotion to group outcomes in FtF highlights the need to understand emotion contagion in distributed groups. The present study examines the transfer of negative emotion in online groups. Negative emotion was induced in one of three group members completing a task in CMC. The data suggest that emotion contagion took place at the group level, with partners experiencing more negative emotion, more disagreement, higher verbosity, and use of more complex language in induced groups compared to control groups. Induced groups also performed better on the group task, raising questions about the effects of negative emotion contagion in online groups.

The proliferation of communication technology has led to potential stratification of contacts across different media, which has important implications for interpersonal dynamics, such as deception. The present study examines how two text-based communication media, BBM and SMS, involve different kinds of social contact networks, and how these differences lead to changes in the frequency and nature of lies. The results reveal that BBM social contacts are relationally closer and include more friends but fewer family and acquaintances than SMS. More deception was also observed in BBM, which included more lies about managing social interactions. The results have important implications for the impact of design features, such as PIN exchange, in text messaging.

Group cohesiveness is a vital social dynamic that is difficult to achieve in virtual teams, but leadership can help groups move past these challenges. We used the Language Style Matching metric to measure group cohesiveness over the course of interaction while groups with either assigned or emerging leaders worked via online chat to complete a collaborative task. We find that overall, successful groups are more cohesive than unsuccessful groups at all times. For groups with assigned leaders, we find this same pattern of cohesiveness. For groups with emerging leaders we find that successful groups and unsuccessful groups are similar in group cohesiveness during the first two-thirds of interaction, but during the final third successful groups are more cohesive than unsuccessful groups.

In this paper, we analyze the way in which international Facebook users who had recently moved to the United States used different languages to selectively self-disclose to their old (native-language) and new (English-speaking) social circles. We found significantly more intimate self-disclosure, covering a broader range of cognitive and emotional topics, in native-language status updates compared to updates in English. Self-disclosure was also more positive in English. These patterns support our hypotheses that users exploit language barriers to serve different self-presentational goals for different social circles and generate implications for SNS privacy control.

Managing one's availability for interaction with others is an increasingly complex act, involving multiple media and the sharing of many types of information. In this paper we draw on a field study of 183 SMS users to introduce the idea of the "interpersonal awareness narrative" -- the coherent, plausible and sometimes deceptive stories that people tell each other about their availability and activities. We examine participants' use of deception in these accounts, and focus in particular on "butler lies," those lies told to enter or exit conversations or to arrange other interactions. We argue that participants use this type of deception in SMS strategically, drawing on the inherent ambiguities of SMS while maintaining plausible narratives.

This study investigates whether deception in online dating profiles is detectable through a linguistic approach, which assumes that liars nonconsciously produce different word patterns than truth-tellers. We objectively measure deception in online dating profiles and analyze the linguistic composition of the open-ended component of the profile (i.e., "about me" section) using computerized text analysis. Results show that profile deceptions correlate with fewer self-references, increased negations, fewer negative emotion words and fewer overall words used in the textual self-description. Results are discussed in terms of (1) practical implications for detecting deception in online profiles; and (2) theoretical implications regarding the impact of media affordances (i.e., asynchronicity and editability) on the occurrence of linguistic cues to deception.

This article explores the operation of warrants, connections between online and real-world identities, on deceptive behavior in computer-mediated communication. A survey of 132 participants assessed three types of warrants (the use of a real name, a photo, and the presence of real-world acquaintances) in five different media: IM, Forums, Chat, Social Networking Sites (SNS) and Email. The effect of warrants on lies about demographic information (e.g., age, gender, education, etc.), one's interests (e.g., religion, music preferences, etc.), and the seriousness of lies was assessed. Overall, deception was observed most frequently in Chat and least often in SNS and Email. The relationship between warrants and deception was negative and linear, with warrants suppressing the frequency and seriousness of deception regardless of medium, although real-world acquaintances were especially powerful in constraining deception in SNS and emails.

Instant messaging (IM) is a common and popular way for co-workers, friends, and family to stay in touch, but its"always-on properties can sometimes lead people to feel overexposed or too readily available to others for conversation. This, in turn, may lead people to deceive others about their actual status or availability. In this paper, we introduce the notion of the "butler lie to describe lies that allow for polite initiation and termination of conversations. We present results from a field study of 50 IM users, in which participants rated each of their messages at the time of sending to indicate whether or not it was deceptive. About one tenth of all IM messages were rated as lies and, of these, about one fifth were butler lies. These results suggest that butler lies are an important social practice in IM, and that existing approaches to interpersonal awareness, which focus on accurate assessment of availability, may need to take deception and other social practices into account.

While most collaboration technologies are concerned with supporting particular tasks such as workflows or meetings, many work groups do not have the teamwork skills essential to effective collaboration. One way to improve teamwork is to provide dynamic feedback generated by automated analyses of behavior, such as language use. Such feedback can lead members to reflect on and subsequently improve their collaborative behavior, but might also distract from the task at hand. We have experimented with GroupMeter - a chat-based system that presents visual feedback on team members' language use. Feedback on proportion of agreement words and overall word count was presented using two different designs. When receiving feedback, teams in our study expressed more agreement in their conversations and reported greater focus on language use as compared to when not receiving feedback. This suggests that automated, real-time linguistic feedback can elicit behavioral changes, offering opportunities for future research.

Contemporary work increasingly involves interacting with strangers in technology-mediated environments. In this context, we come to rely on digital artifacts to infer characteristics of other people. This paper reports the results of a study conducted in a global company that used expertise search as a vehicle for exploring how people interpret a range of information available in online profiles in evaluating whom to interact with for expertise. Using signaling theory as a conceptual framework, we describe how certain 'signals' in various social software are hard to fake, and are thus more reliable indicators of expertise. Multi-level regression analysis revealed that participation in social software, social connection information, and self-described expertise in the corporate directory were significantly helpful in the decision to contact someone for expertise. Qualitative analysis provided further insights regarding the interpretations people form of others' expertise from digital artifacts. We conclude with suggestions on differentiating various types of information available within online profiles and implications for the design of expertise locator/recommender systems.

An enduring assumption about computer-mediated communication is that it undermines emotional understanding. The present study examined emotional communication in CMC by inducing negative affect in one condition and neutral affect in another. The results revealed that 1) participants experiencing negative affect produced fewer words, used more sad terms, and exchanged messages at a slower rate, 2) their partners were able to detect their partners emotional state, and 3) emotional contagion took place, in which partners interacting with participants in the negative affect condition had significantly less positive affect than partners in the control condition. These data support a relational view of CMC.

With the widespread use of social networking sites, it is easy to acquire a great deal of personal information about someone before meeting them. How do people use this information when initiating relationships? In the present study, participants either had access to an unknown partner's Facebook profile or did not, and were instructed to get their partners to like them in a short instant messaging conversation. Participants used social network and profile information in two ways: probes, asking questions whose answer they already knew, and implicit mentions, referencing information that made them seem more similar to their partner. These strategies successfully increased interpersonal attraction. Participants, however, frequently rated these strategies as deceptive, raising important concerns about the use of asymmetrical personal information for interpersonal gain.

Our ability to express and accurately assess emotional states is central to human life. The present study examines how people express and detect emotions during text-based communication, an environment that eliminates the nonverbal cues typically associated with emotion. The results from 40 dyadic interactions suggest that users relied on four strategies to express happiness versus sadness, including disagreement, negative affect terms, punctuation, and verbosity. Contrary to conventional wisdom, communication partners readily distinguished between positive and negative valence emotional communicators in this text-based context. The results are discussed with respect to the Social Information Processing model of strategic relational adaptation in mediated communication.

Online dating is a popular new tool for initiating romantic relationships, although recent research and media reports suggest that it may also be fertile ground for deception. Unlike previous studies that rely solely on self-report data, the present study establishes ground truth for 80 online daters' height, weight and age, and compares ground truth data to the information provided in online dating profiles. The results suggest that deception is indeed frequently observed, but that the magnitude of the deceptions is usually small. As expected, deceptions differ by gender. Results are discussed in light of the Hyperpersonal model and the self-presentational tensions experienced by online dating participants.

Effective communication in project teams is important, but not often taught. We explore how feedback might improve teamwork in a controlled experiment where groups interact through chat rooms. Collaborators who receive high feedback ratings use different language than poor collaborators (e.g. more words, fewer assents, and less affect-laden language). Further, feedback affects language use. This suggests that a system could use linguistic analysis to automatically provide and visualize feedback to teach teamwork. To this end, we present GroupMeter, a system that applies principles discovered in the experiment to provide feedback both from peers and from automated linguistic analysis.

Ambiguity is an important concept for HCI because of its pervasiveness in everyday life, yet its emergent nature challenges the role of design. We examine these difficulties with regards to Aoki and Woodruff's [1] proposal to use ambiguity as a resource for designing space for stories in personal communication systems. We challenge certain assumptions about ambiguity and propose a set of design and evaluation guidelines that flow from this re-conceptualization of ambiguity and design.

Social psychology has demonstrated that lying is an important, and frequent, part of everyday social interactions. As communication technologies become more ubiquitous in our daily interactions, an important question for developers is to determine how the design of these technologies affects lying behavior. The present research reports the results of a diary study, in which participants recorded all of their social interactions and lies for seven days. The data reveal that participants lied most on the telephone and least in email, and that lying rates in face-to-face and instant messaging interactions were approximately equal. This pattern of results suggests that the design features of communication technologies (e.g., synchronicity, recordability, and copresence) affect lying behavior in important ways, and that these features must be considered by both designers and users when issues of deception and trust arise. The implications for designing applications that increase, decrease or detect deception are discussed.

In this paper we examine the relationship between emergent social network characteristics in a computer-supported collaborative learning course and locus of control. An emergent communication network of engineering students that took place in a distributed distance learning environment was examined. Three measures of an actor's social network, density, brokerage, and reach, and participants' locus of control, internal vs. external, were assessed. The data suggest that, relative to participants with external locus of control, participants with internal locus of control decreased their network density over time but increased their brokerage and reach. The results are discussed in the context of instrumental action, through which participants are assumed to develop personal networks in pursuit of maximizing potential social capital.