Over the past decades, the behavior of people who do not maximize their payoff but instead seem to be other-regarding has received much attention in the (behavioral) economics literature. Many different social preference models that ideally explain such other-regarding behavior across a large span of contexts have been proposed and tested. Building on this literature, this dissertation studies social preferences in different contexts and expressions across three manuscripts. The first manuscript examines the behavior of people who avoid a situation that allows them to express social preferences. Drawing on psychological game theory, we tested whether guilt-aversion or self-image concerns could better explain this behavior. It was found that guilt-aversion, but not self-image concerns, can explain the behavior of these people. The second manuscript made use of crowdfunding donations data and showed that the reversal of the compassion fade effect when going from a separate to a joint evaluation condition extends from the lab to the field. Social preferences can also manifest themselves through people donating their time. The third manuscript examines how the severity of a catastrophe (i.e., the COVID-19 pandemic) affects the provision of catastrophe-related voluntary labor. We found a concave relationship between the weekly COVID-19-related death numbers and the amount of voluntary work provided by individuals. Thus, by drawing on prosocial behavior expressed in three different environments, this dissertation extends the current literature by studying how social preferences are influenced by the context in which they are carried out.