A Typology of Social Media Use by Human Service Nonprofits: Mixed Methods Study

Area of Science:

• Computational Social Science
• Nonprofit Management
• Digital Communication

Background:

• Human service nonprofits face challenges in fully leveraging social media for communication.
• Limited understanding exists regarding the diverse applications of social media beyond self-promotion.
• A need for conceptual clarity is identified to enhance training for nonprofit social media utilization.

Purpose of the Study:

• To examine the online communication content of nonprofit sexual assault (SA) centers in Canada using Twitter (X).
• To develop a typology for classifying tweets from SA centers through qualitative and machine learning methods.
• To provide a framework for understanding nonprofit social media objectives and outcomes.

Main Methods:

• A mixed-methods approach combining qualitative analysis and supervised machine learning.
• Manual coding of 10,809 tweets from 133 SA centers (March 2009-March 2023) for training data.
• Classification of 286,551 organizational tweets using machine learning, followed by unsupervised topic modeling and sentiment analysis.

Main Results:

• Seven thematic categories identified: community engagement, organization administration, public awareness, political advocacy, support for others, partnerships, and appreciation.
• Organization administration was the most frequent category; political advocacy and partnerships were least frequent.
• Machine learning model achieved 63.4% accuracy; sentiment analysis showed neutral prevalence, with fear dominant and joy linked to partnerships/appreciation.

Conclusions:

• An original theoretical model elucidates how human service nonprofits utilize social media across seven thematic categories.
• The study offers a comprehensive typology of nonprofit communication objectives and content on social media.
• Findings support enhanced training for nonprofit leaders to improve public, policy, and stakeholder engagement.