Using Machine Learning Technology to Measure Changes in Emotional State During Parenting Helpline Calls

“Emotional support helplines are a longstanding strategy for supporting parents, but few studies have measured how much these programs work to address negative feelings in parents.

Drawing on a sample of 281 helpline calls from parents to the California Parent and Youth Helpline^®, the authors analyzed changes in emotions expressed through language, while parents spoke to a counselor. The authors used a sequential explanatory mixed methods design. They worked with a Natural Language Inference (NLI) model, BART, to measure and quantitatively display changes in feelings parents express. They found that 85% of the parents made more positive statements as the call proceeded. The authors tested BART by measuring its rate of agreement with qualitative coding by the authors and against a codex constructed by service users. The authors discuss the implications of this study for addressing parental isolation and how the research methods could be adopted by other helplines.”