This paper presents a review of empirical methods used to assess the behavioral, economic, and political outcomes of Internet and social media usage. Instead of merely surveying the various impacts of the Internet, we examine the methods adopted to identify these impacts. We describe two main approaches for establishing causal effects, each with strengths and limitations. The first approach involves searching for exogenous sources of variation in the access to fast Internet or specific content. The second approach takes the form of field or laboratory experiments. In this paper, we focus on the first approach, delving into the methodological threats, empirical design, and main findings of the most prominent studies that exploit natural or quasi-experiments for identifying the causal impact of high-speed Internet or specific social media. This undertaking allows us to highlight the key empirical challenges in the field of Internet and social media economics while summarizing the causal relationships that the literature has uncovered so far.