WITHDRAWN: Social Media As a Source of Data - Lessons Learned and Implications for Utilizing Computational Methods in Social Work Research

The 2020 SSWR Annual Conference featured a Workshop entitled Social work research and technology: leveraging AI, topic modeling and community-based methods for research on human services, violence, and grief. This 2021 roundtable continues and deepens a segment of conversation around social media data and computational modeling - the use of algorithms to explore, predict, and find language patterns that offer new interpretations and theories (Blei & Smyth, 2017). One methodology, unsupervised machine learning, is a data-driven approach likened to grounded theory. Comparatively, supervised machine learning can label a subset of a larger sample to learn an algorithm and generate predictive claims. Yet these methodologies are critiqued for lacking reflexivity and exacerbating racist inequities. As aligned with The Grand Challenge, social workers can harness the power of these technologies to better understand and respond to social problems (Berzin et al., 2018). But it requires innovation, adaptation, and capacity building among social work scientists.

Social media (e.g. Twitter, Reddit) is an enticing data collection source for several reasons: space for sharing narratives; less stigma and low barrier for participation; access to populations less engaged with research studies; low cost for researchers; more accessible for doctoral students; and potential to induce theory building. Computational modeling and machine learning have been successfully employed to intervene on social media within a behavioral health context (e.g. depression, substance abuse). With the excitement around these models and the possibility to expand our existing knowledge, this area is relatively under-examined in social work research.

Four scholars with experience conducting social media studies will provide lessons learned for the social work field. In brief, the first scholar will bring an international perspective on how to combine social media with satellite images to predict poverty and shape aid distribution in sub-Saharan Africa and offer expertise in obtaining grants within this area. The second scholar will discuss accessing Instagram and Twitter and coordinating data collection and cleaning efforts. The third scholar will bring experience within both social media (e.g., Reddit) and computational methods to discuss scenarios of utilizing different modeling analyses. The fourth scholar will share expertise from published work on understanding social media-based movements and ethical algorithmic decision making, as well as offer their experience as a current member of the Twitter Academic Research Advisory Board.

The authors will facilitate discussion about: How can social media and computational modeling be helpful for social work researchers and what RQs can be addressed; What results can computational models generate when employed with social media data within the social work scholarship context; What implications does this have for intervention development, implementation and evaluation across micro-meso-macro levels; How can researchers work towards social justice using these methods and what ethical debates are present within this area; What suggestions can be offered regarding publishing this work in SW journals, as well as getting this work funded? This roundtable will encourage participants to discuss the application of social media and computational modeling in social welfare research and ponder several concerns within this trend.