Social computing platforms are often celebrated for their capacity to connect; yet expressing one's identity, and seeking as well as providing support on these platforms can be difficult when people experience distress and stigma. In this dissertation, I theorize social media behaviors such as sensitive disclosures and interactions around them. I address how we can design social computing systems that facilitate disclosures of difficult and stigmatized human experiences and enable supportive interactions to form around them. I focus on pregnancy loss, as a common reproductive health experience that is often traumatizing and not disclosed in spite of potential disclosure benefits such as access to social support.

First, I contribute a framework to explain direct pregnancy loss disclosures and non-disclosures on social media. I conducted in-depth phenomenological interviews with women in the U.S. who are social media users and had experienced pregnancy loss within the past two years. This framework includes six types of decision factors related to the self, audience, network, society, affordances and platform, and time. I also introduce the novel concept of network-level reciprocal disclosure to describe broadcast disclosures that happen in response to and in anticipation of others' disclosures and a perceived reduction in stigma.

Second, I contribute a typology of indirect disclosure strategies and an understanding of decision factors informing indirect disclosures of sensitive experiences across social media platforms. Indirect disclosure strategies include hinting about an experience or an identity facet in such a way that the audience must infer the full message, or explicitly sharing information through another person. Factors related to the self, audience and affordances, and time motivate individuals to disclose pregnancy loss in an indirect manner, although in distinct ways compared to direct disclosures.

Finally, responses to sensitive disclosures are important to understand if we aim to design social computing systems that encourage supportive interactions. I used interviews and designed vignette instruments to provide a response decision making framework (RDM) that explains factors informing whether and how individuals respond to sensitive disclosures from their social media connections. This framework includes factors related to the self, poster, and disclosure context. I identify empirically grounded design and research implications to support potential disclosers and responders in sensitive or stigmatized contexts on social media.