Developing Emotional Granularity: A SEL Intervention in High-Poverty Contexts

The challenge:

Research indicates that young women in highly marginalized contexts face severe systemic stressors and cognitive loads (e.g., domestic labor expectations) that compromise emotional well-being. A critical barrier to effective emotional regulation in these demographics is a lack of emotional granularity—the cognitive ability to make high-resolution distinctions between affective states (e.g., distinguishing "frustrated" from "furious").

Without precise linguistic labels, learners rely on a binary valence system ("feeling bad" vs. "feeling good"), depriving them of the diagnostic precision required to select appropriate self-regulation strategies. The challenge was to design a scalable, low-resource intervention that could provide these cognitive scaffolds and measure their acquisition objectively.

What do I do:

As the lead designer for SEL component of the intervention, I structured a 6-week pedagogical sequence and its accompanying evaluation methodology:

• Evidence-Based Pedagogical Design: I designed a three-phase curriculum based on cognitive science principles:

  1. Metacognitive Monitoring: In-person workshops introducing an adapted Mood Meter to transition abstract feelings into concrete cognitive representations.

  2. Spaced Practice & Retrieval: A 5-week remote phase requiring twice-weekly structured logs where students mapped their states. This repeated retrieval practice was designed to automate the cognitive process of emotional labeling.

  3. Social Scaffolding & Embodiment: Final workshops utilizing "emotional clotheslines" to externalize and socially validate internal states, alongside body scan exercises to reinforce embodied cognition.

• Assessment Methodology: To avoid the bias of pre-defined checklists, I implemented a Spontaneous Label Generation Task to measure active vocabulary retrieval.

• Data Processing & Analysis: I digitized the qualitative response data, categorized terms using the "families of emotions" theoretical model, and utilized Python and Tableau to perform paired-samples t-tests and calculate effect sizes.

Results:

The quantitative analysis demonstrated that the pedagogical mechanism successfully altered the cognitive resources available to the students for encoding their emotional experiences:

• Lexical Expansion: The active emotional lexicon increased significantly from a baseline mean of 9.84 words to 15.81 words (t(497)=-25.41, p<0.001). The magnitude of change was substantial, with an effect size of g=1.253.

• Categorical Breadth: The number of distinct "emotion families" accessed by students rose significantly from 6.08 to 7.50 (p<0.001, g=0.82), proving the intervention broadened their emotional spectrum rather than just deepening a single category.

• Contextual Consistency: Subgroup analyses confirmed robust gains across 9 of the 10 distinct communities, regardless of participants' educational levels.

What I learned:

While analyzing the sub-group data, I discovered a crucial anomaly: one specific community (Almoloya de Juárez) did not show a statistically significant increase in categorical breadth. I hypothesized that this cohort's exceptionally high "social density" (close-knit siblings and peer groups) established social scripts that inhibited the risk-taking required to adopt new emotional behaviors. Future iterations must account for social group dynamics, and I plan to employ more rigorous quasi-experimental designs with control groups to isolate these sociolinguistic variables better.

Full report: