Communicating data-driven insights typically through narrative visualizations is gaining increasing popularity in both practice and academia. Data videos identified as one of the main genres of narrative visualization are short motion graphics that incorporate visualizations about facts. Their unique characteristics make them a great candidate for telling compelling data stories to a broad audience. However, very little is systematically recorded about what elements are featured in data videos, the processes used to create them and what features make data videos effective. As a result, the solutions available to facilitate crafting these videos and taking advantage of their storytelling power are scarce and demand much needed attention from the research community. To this aim, in this thesis work, I present a series of exploratory studies to shed light on data videos, their constituent components, and creation process. Based on the lessons learned from these studies, I have designed and developed, DataClips, a web-based authoring tool to consolidate the creation of data videos by lowering the skill level required to create data videos using common data visualizations and animations. To apply the resulting data video authoring solution, I demonstrate use cases in which effective communication of the data insights to a broad audience is of significant importance. Through a large-scale online study, I have tested different design features of data videos to find answers to basic questions regarding their effectiveness. In particular, I have assessed the effects of animation, pictographs and icon-based visualizations on viewer engagement and preference in comprehending the communicated information. The results provide design implications for authoring effective data videos by maximizing viewer engagement and comprehension.