Social Media Integration and Adoption Antecedents in Government and Non-Government Organizations of Coastal Bangladesh: A TOE Framework Perspective

Abstract

Aims: To examine how social media is integrated into the communication strategy and structures of government and non-government organizations in coastal Bangladesh and to identify factors that influenced this adoption for disaster management within the Technology, Organization, and Environment (TOE) framework. This will contribute to disaster communication literature by empirically testing the applicability of the TOE framework and assist practitioners of disaster management in designing more targeted interventions in disaster-prone coastal districts of Bangladesh.  Place and Duration of Study: Paikgasa Upazila under Khulna district and Assasuni Upazila, Satkhira district, Bangladesh. Data were collected in March 2025. Methodology: A mixed-method approach was adopted in this study. Quantitative data were collected from a total of 30 organizations, which include nine government, seven local government, and 14 NGOs, using a bounded near-census. To improve and validate quantitative results, qualitative data were collected through 15 Key Informant Interviews (KIIs) from five government organizations, four local governments, and six NGOs. The respondents were selected purposively. Quantitative Data was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) bootstrapping, and qualitative data was analyzed using thematic analysis. A mixed-method research approach could quantify the relationship between the TOE factors and social adoption, while qualitative data could give a deeper explanation of contextual factors. PLS-SEM is suitable for theory building in small, medium, and non-normal data, while qualitative analysis could answer why and how the organizations adopt social media.   Results: The study found 97% of organizations used Facebook for disaster communication, particularly during preparedness and response phases. However, only 8% have a separate social media communication strategy, and 25% use project-based approaches. Insufficient technological access, misinformation, and limited staff expertise are the barriers to fully institutionalized social media. According to the TOE framework, only the technological factors like access to technological infrastructure (smartphone) and skilled staff showed a statistically significant impact on social media adoption (β = 0.495, p = .012) by the organizations. Environmental factors showed a positive but non-significant impact (β = 0.265, p = .201) while organizational factors had no meaningful impact (β = 0.015, p = .941). Qualitative data ensured that technological readiness, youth engagement, and supportive policies influenced social media adoption, while organizational structure, e.g., bureaucratic inertia, hierarchical decision-making, and resource constraints, acted as challenges regarding this adoption. The findings extend the TOE framework by revealing that in the setting where resource constraints existed, technology-led diffusion can precede and later shape organizational and environmental adaptation, which indicates a sequential rather than balanced influence among the three dimensions. Conclusion: This study extended the TOE frameworks by identifying that in the resource-constrained disaster context, instead of balanced influence, the TOE factors are sequential, where technological readiness is the primary driver. In this context, organizational and environmental factors become later. In coastal Bangladesh, social media adoption by governmental and non-governmental organizations remains technology-driven; however, it’s not fully institutionalized yet. To make the use of social media effective in the long run, organizations should stop using it in a random way. Instead, they should make it part of their official communication plans, train their staff to use it well, and create systems to fight misinformation. However, this study was conducted only in two coastal upazilas and sample was small near-census, future research should test the model with larger samples across diverse hazard settings.