Construction Validation for Algorithmic Content Moderation of Misleading Eco-Claims on Social Media

• Abey Litty

  Texas University

  Author

The proliferation of misleading environmental claims on social media platforms has created an urgent need for automated moderation systems capable of identifying greenwashing at scale. However, existing detection approaches lack validated constructs specifically designed for algorithmic content moderation contexts. This study addresses this gap by developing and validating a multi-dimensional construct framework for identifying misleading eco-claims on social media. Drawing on established greenwashing measurement dimensions including misleading environmental claims, selective disclosure, symbolic environmentalism, and digital greenwashing tactics, we developed a nine-indicator classification instrument and tested it against a stratified sample of 2,847 social media posts from three platforms. Using a hybrid BERT-RoBERTa ensemble architecture, the validated model achieved 89.4% accuracy (F1 = 0.891) in identifying misleading eco-claims, significantly outperforming baseline approaches. Content validity was established through expert panel review (κ = 0.89), while predictive validity was confirmed through structural equation modeling demonstrating significant relationships between detected greenwashing constructs and stakeholder deception perceptions (γ = 0.78, p < 0.001). The study contributes a replicable construct validation framework that bridges greenwashing measurement theory with practical algorithmic moderation implementation.

• pdf

Copyright (c) 2026 Abey Litty (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.