Understanding the Features of Text-Image Posts and Their Received Social Support in Online Grief Support Communities

Abstract

People in grief can create posts with text and images to disclose themselves and seek social support in online grief support communities. Existing work largely focuses on understanding the received social support of a post in pure text but often overlooks the post that attaches an image in grief communities. In this paper, we first computationally characterize the textual (e.g., theme), visual (e.g., color), and text-image coherence (i.e., semantic and sentiment coherence) features of text-image posts in a grief support community. Then, we conduct regression analyses to systematically examine the effects of these features on their received informational, emotional, esteem, and network support. We find that attaching a selfie image in the post positively predicts received informational and emotional support, while the social image of a post is a positive predictor of network and esteem support. A post is also likely to get more social support if its text is describing the visible content or telling a story depicted in the image or the perceived emotions in the text and image are not conflict. These results supplement existing research on mental health communities and provide actionable insights into assisting grief people to seek social support online.