Connecting with others enhances one’s health and happiness, yet people routinely forego opportunities to socially engage because they expect these interactions to unfold less positively than they do. In this dissertation I propose (Chapter 1), and test (Chapters 2-5), an expectancy-value theory to understand why people are overly avoidant in their social interactions. The theory predicts that a person’s interest in engaging is guided by the perceived likelihood that engaging would lead to specific outcomes (expectancies) and by how much value they expect to attach to these outcomes (values). People mismanage their social relationships because they underestimate how social others will be during an interaction, expect others to derive less value from the warmth (vs. competence) of one’s actions than others do, and overlook the influence of the communication medium on the outcomes of an interaction. People’s miscalibrated expectancies may lead them to be less social than would be ideal for their well-being. In Chapter 2, I examine how wisely people choose whether to engage. Participants expected strangers to be less social during conversation than close others or oneself, but later reported that the strangers they spoke with were more social than expected (Experiment 1). These miscalibrated beliefs may create a barrier to engaging: Participants reported being especially likely to engage when they expected others to be highly social and expected to attach high value to others’ responses (Experiments 2a-5). In contrast, I found mixed evidence that uncertainty about others’ interest creates a barrier to engaging independent of people’s expectations (Experiments 3a-5). In Chapter 3, I examine how wisely people choose how to engage. Interactive communication media, such as back-and-forth conversations, should foster stronger connections than non-interactive media such as voice messages. However, people may not focus on the medium when assessing the consequences of engaging. Consistent with these hypotheses, participants established stronger connections through interactive than non-interactive media (Experiments 6-8b), especially when discussing areas of disagreement (Experiment 7). However, participants did not anticipate differences between these media (Experiments 6-7, 8b). People may undervalue interactive media when choosing how to engage. In Chapter 4, I examine how wisely people choose what to talk about. People may withhold personal details about themselves because they expect others to react negatively to this content. However, participants underestimated how social others would be during conversation, and they consequently overestimated the awkwardness of deep conversation (Experiments 9-10). They also expected others to care less about the warmth of one’s self-disclosures than others did, causing them to overestimate the reputational costs of revealing negative secrets (Experiments 11-12). People may be overly reluctant to open up to others. In Chapter 5, I test whether calibrating people’s expectancies removes a barrier to engaging. Participants preferred deeper conversations (Experiments 13a-b) and were more likely to reveal negative secrets (Experiment 14) when they expected others to respond favorably. Calibrating people’s expectancies may lead them to engage in ways that strengthen their connections and enhance their well-being. Finally, in Chapter 6, I discuss open questions and propose directions for future research.