Research Programming Solutions

Did you know there is a team of developers who can assist with your research? Wharton Computing’s Research Programming team is a centralized resource available to PhD and faculty members. In this session, you will learn how the team works, and we will highlight some of the shared solutions we offer.

Is It Better To Pursue Goals In Sequence Or In Parallel?

When people have several important, discrete goals, is it better to work on them all at once or tackle them one at a time, in order? At first glance, parallel pursuit seems superior: it allows people to shift effort across goals as circumstances change, and it preserves the option to focus on a single goal whenever desired. Yet we propose that adding the constraint of sequential pursuit – that is, starting with one goal and not proceeding to the next goal until the first is complete – can improve performance. In practice, unconstrained parallel pursuit may enable procrastination, increase task-switching, and make difficult goals easier to avoid. Across two longitudinal experiments, participants set personal goals for a week and were randomly assigned to pursue them either sequentially or in parallel. In both studies, sequential pursuit significantly increased completion of the highest-priority goal, with no meaningful reduction in total goals achieved. The benefits were strongest when the top-priority goal was more difficult, consistent with the idea that sequential pursuit helps people stay focused on hard but important work. Overall, the findings reveal a counterintuitive insight: although parallel pursuit feels more flexible and efficient, the added constraint of working on goals one at a time can meaningfully improve real-world goal achievement.

Al-Assisted Adversarial Collaboration Accelerates Resolution of Debate on Minority Salience

The advancement of knowledge depends on scientific debates and rigorous tests of competing hypotheses. Yet this process is often slow and acrimonious. Artificial intelligence can facilitate the process by summarizing arguments on both sides as propositions, isolating conflicting propositions and proposing empirical tests. We applied an Al-assisted adversarial collaboration to a debate in PNAS on minority salience: the overestimation of minorities in visual displays of faces. Kardosh et al. claimed that overestimation increased when minorities in lab experiments were also minorities in participants’ environment. White participants from the USA overestimated black minorities in facial displays more than white minorities. Kardosh et al. argued the boost occurred because blacks were minorities in the USA. Gayet et al. viewed the evidence as weak and open to alternative explanations. We attempted an Al-assisted adversarial collaboration to resolve the debate. Al identified key points in the dispute and helped design tests of hypotheses. We conducted two experiments that manipulated the ethnicity of minority faces and the ethnicity of participants. Participants from different countries viewed 1Qx1Q facial displays (minority proportion ==25%) and estimated percentages of minorities. Minority percentages were systematically overestimated, and overestimation increased when minorities in the display were also minorities in participants’ countries. Finally, we measured beliefs about hypotheses on both sides before and after seeing the experimental results. Confidence in key hypotheses converged. The exercise is a proof of concept that Al-assisted adversarial collaboration provides valuable support, focuses the debate on key issues and can produce results that ultimately change minds.

From Stigma to Support: How Others’ Income Volatility Elicits Prosocial Attitudes and Behaviors

Increasing prosocial attitudes and behaviors toward lower-income people is an important
goal for many policymakers and organizations. Yet, influencing prosociality toward lower-income people is notoriously difficult. This work introduces a novel approach to increase
prosociality: highlighting income volatility. Income volatility is an important facet of consumers’ income, affecting over 40% of US consumers. Seven preregistered experiments (N = 4,703) demonstrate that consumers donate more money, express less stigma, and report greater support toward others with higher (vs. lower) income volatility, even when total income is held constant. Evidence from a series of mediation and moderation experiments shows that this effect is driven, at least in part, by differences in the dispositional and situational attributions that consumers make about others’ financial circumstances (e.g., low work ethic vs. a weak economy). Specifically, greater income volatility shifts consumers’ attributions away from dispositional explanations and toward situational ones, resulting in increased prosociality. These findings have important implications for policymakers and organizations working to increase support and reduce stigma toward lower-income populations, including government benefit applicants.

A Reference-Point Theory of Reflection Effects: Predicting Individual Choice under Uncertainty

Understanding how people make choices under uncertainty is a central goal in the social sciences. Across three preregistered online experiments (Ns = 490, 395, and 290), we test a reference-point theory of individual preferences under risk (known odds), ambiguity (unknown odds) and direct comparisons between these options using incentivized choices between small-stakes gambles. Participants exhibit reflection effects—systematic preference reversals around zero—across all three types of choices. Our framework accounts for these patterns by distinguishing between belief and hedonic motives estimated from the ratings of anticipated pleasure for options and outcomes. When uncertainties are small, individuals tend to overestimate the likelihood of winning (optimism) and losing (pessimism). When uncertainties are larger, they often underestimate the likelihood of winning (pessimism) and losing (optimism). Reference-point theory clarifies identical choices can arise from different psychological motives and how similar motives can result in different choices, offering a unified account of preferences under risk and ambiguity.

Speaking Up with Negative Emotion: How Nonverbal Expression and Affect Labeling Influence Voice Endorsement

The Dynamics of Confidence: How Task Structure and Difficulty Shape Confidence Over Time

Across six studies, we investigated how confidence evolves over the course of repeated decisions. Whereas prior work has documented general declines in confidence across multi-item tasks, we sought to explore when and why such erosion occurs and when it reverses. We explore the role of task difficulty and decision ease in shaping confidence judgments. Results revealed that confidence declined more steeply on objectively harder tasks, and was sensitive to subtle manipulations of subjective difficulty. Together, these findings suggest that confidence change is not solely a consequence of performance monitoring but is shaped by the ongoing experience of the task itself.

Metaphors in Negotiation

Negotiation outcomes depend not only on the offers exchanged but also on the mental models of the negotiating parties and the interpretations they form of one another’s motives. The language a negotiator uses to communicate can greatly shape the way they are understood and the way they understand others. One ubiquitous feature of the language surrounding negotiations is metaphor. Research across health, climate, communication, vaccination, and finance indicates that metaphor use can alter judgment and behavior. Building on this body of work, our research develops a collection of fundamental negotiation metaphors and outlines how different metaphors can cue different approaches to common negotiation goals such as value creation and value claiming. Our central claim is that negotiation communication through metaphor shapes shared mental models, relational outcomes, and perspective taking.

A New Look at Homophily: People Underestimate the Extent to Which Dissimilar Others Are Interested in Talking to Them

Do people systematically underestimate the extent to which dissimilar others are interested in talking to them? Results from six experiments, in the lab online, and in the field, document people’s overly pessimistic beliefs about others’ interest in talking to them, across several types of dissimilarity, including age, race, personality, job type, and sociocultural background. Our work makes three key contributions. First, we extend psychological phenomena first documented in the intergroup literature, showing that they apply more broadly to many types of dissimilarity. Second, we provide a new perspective on
“homophily,” people’s tendency to initiate and pursue deeper relationships with those who are similar to themselves. We argue that homophily is not always the result of people’s direct preference for similar interaction partners and their limited opportunities to talk to dissimilar people-the primary explanations offered in the literature-but instead may often reflect people’s pessimistic beliefs about what dissimilar others think of them. Third, we draw on metaperception research, showing how uncertainty and salient negative thoughts can bias people’s expectations of conversational success, making interactions with dissimilar others feel unnecessarily challenging. Finally, we explore people’s pessimistic beliefs in a field experiment in collaboration with Mystery Minds, an organization that connects employees across group divides through conversation. These findings contribute, both theoretically and practically, to our understanding of homophily, metaperception, intergroup interaction, networking, and the social cognition of conversation.

Do We Write What AI Tells Us To?

Consumers are increasingly turning to large language models (LLMs) to enhance their written communications. Yet this growing reliance raises a critical question: when consumers adopt AI-generated writing suggestions, do their words still reflect their own sentiments and intentions, or those of the algorithm? In this paper, we explore this issue by examining consumers’ willingness to adopt LLM-generated writing suggestions that would alter the valence of an original communication. Drawing on evidence from a large-scale field study of how AI use affects the valence of consumer complaints (CFPB; N = 458,316), two controlled experiments, and a replication (N = 7,413), we find that consumers readily adopt valence-changing suggestions, even when unsolicited. Moreover, these choices have significant downstream consequences: we find that adopting AI writing suggestions carries over to how messages are subsequently written without AI assistance and when AI is used to edit complaint letters, it makes them less persuasive to third-party readers. The mechanisms underlying decisions to adopt AI writing suggestions, as well as the factors that explain individual differences in writing advice adoption, are explored.

What People Say to ChatGPT and What That Says About Them

Millions of people now regularly interact with chatbots for a variety of purposes, including work, advice, and companionship. Yet little research has examined how the actual content of people’s conversations with Al relates to their psychology. We collected ChatGPT conversation histories from hundreds of participants and had them complete a survey assessing well-being (e.g., loneliness, depression, life satisfaction) and attitudes toward Al (e.g., beliefs about whether using Al makes people happier and more productive). Using computational text analysis, we explore how the content and characteristics of people’s actual conversations-what they discuss and how often-relate to their well-being and attitudes. By examining actual conversation data alongside self-report measures, this research provides insight into the psychological correlates of real-world human-Al interaction.

A Balancing Act: Why Efforts to Improve Work-Life Balance Backfire for East Asian Employees

An increasing number of organizations encourage employees to strike a balance between work and life. Scholars have assumed that the growing prioritization of work-life balance is a trend that improves work engagement and job satisfaction. However, a complication surfaces when integrating theory on ethnic identity with theory on organizational culture: organizations that place a premium on work-life balance may not benefit – and may even harm – employees who, as a function of their ethnic identities, prioritize working hard to serve their families (e.g., employees with East Asian ethnic
identities). We predict that, when working in a culture that promotes work-life balance, individuals with an East Asian ethnic identity are likely to experience value incongruence. Accordingly, they identify less with their organizations than (1) employees with non-East Asian ethnic identities who work in organizations with strong work-life balance cultures and (2) employees with East Asian ethnic identities who work in organizations with weak work-life balance cultures. This, in turn, causes
individuals with East Asian ethnic identities to be less engaged and satisfied with their work. We ested our predictions with a survey of full-time employees (N = 263), archival data from the General Social Survey (N = 9,079), and a pre-registered quasi-experiment (N = 596). We also conducted three supplementary studies featuring data from the World Values Survey (N = 181,911), Glassdoor (N = 17,456,881), and another quasi-experiment (N = 362). The results, which largely support our hypotheses, redirect the literature on work-life balance, well-being, and culture.