Building the Etch-a-Sketch online (ESO) system to broaden and diversify research: Proof of concept for a tool to automate sentiment coding of transcripts from online parent–child observations.

Parent-child relationships are mutually reinforcing dynamics associated with children's psychological adjustment across races and cultures. Observing and rating (coding) naturalistic parent-child interactions is central to understanding psychological mechanisms, targeting interventions, and measuring change, yet pragmatic barriers to in-person assessment (e.g., time, travel, cost) limit their use in large, diverse samples. The online observation tool, Etch-a-Sketch Online (ESO), enables developmentally, racially, and culturally flexible data collection, but human coding of ESO interactions remains resource intensive. This proof-of-concept study introduces SwiftSentiment, an open-access, cost-free, data-secure artificial intelligence-powered tool for sentiment analysis (i.e., identifying emotional tone from text) of ESO-elicited conversations. The sample comprised 62 mothers and fathers recruited in the United Kingdom with their children (46.67% female; Mage = 13.17 years [SD = 1.32], 64.52% White, 20.97% Asian/British Asian, 14.52% mixed/other ethnicity). Associations between human coding of ESO interaction videos (using the established parent-child interaction system) and SwiftSentiment coding of the interactions' audio transcripts were examined, as were human and SwiftSentiment coding associations with self-reported parent-child relationships and child adjustment. Correlations between broad-coded SwiftSentiment emotional positivity and negativity from audio transcripts and human-coded parent-child positivity and negativity from video demonstrated convergent validity of SwiftSentiment coding. SwiftSentiment negativity was associated with self-reported dysfunctional parenting strategies and child externalizing problems, but unlike manual coding, SwiftSentiment positivity was not associated with parent-reported constructs, suggesting that human coding retains advantages in capturing positivity. The study provides initial evidence for partnering ESO and SwiftSentiment as an innovative data collection and automated coding system-the ESO system-to widen and diversify research capacity and reach. (PsycInfo Database Record (c) 2026 APA, all rights reserved).