The introduction and widespread use of modern social media (SM), such as Facebook, Twitter, and Instagram social networks (SNs), have changed how politicians interact with the public and conduct political campaigns. Due to the inherent strengths of SM, namely the vast amount of data that can be accessed in real time, a new area of study has formed that focuses on using SM data to forecast election results. Despite the fact that numerous research have been done in the last ten years, the findings are frequently disputed. In this context, this article attempts to analyse and summarise how research on election prediction using SM data has developed from its inception, to lay out the state of the art and the state of the practise, and to pinpoint future research directions in this area. In terms of methodology, we conducted a methodical review of the literature, evaluating the quantity and calibre of publications, the electoral context of studies, the main approaches and traits of successful studies, as well as their main advantages and disadvantages, and compared our findings with those of previous reviews.