Leveraging computational methods for nonprofit social media research: a systematic review and methodological framework

While social media platforms are valuable for examining the online engagement of nonprofit and philanthropic organizations, the research considerations underlying social media data remain opaque to most. Through a systematic review of nonprofit studies that analyze social media data, I propose a methodological framework incorporating three common data types: text, engagement and network data. The review reveals that most existing studies rely heavily on manual coding to analyze relatively small datasets of social media messages, thereby missing out on the automation and scalability offered by advanced computational methods. To address this gap, I demonstrate the application of supervised machine learning to train, predict, and analyze a substantial dataset consisting of 66,749 social media messages posted by community foundations on Twitter/X. This study underscores the benefits of combining manual content analysis with automated approaches and calls for future research to explore the potential of generative AI in advancing nonprofit social media research.