When Metrics Mislead: Parents’ Lived Realities in the Public Safety Net

Paper Title

When Metrics Mislead: Parents’ Lived Realities in the Public Safety Net

Publication Info

• Topic area: Human-computer interaction in public sector data systems and child welfare.
• Keywords: Child welfare system, public safety net, sociocentric metrics, egocentric perspectives, interpretive computational analysis, informal support networks, information gaps, data-driven systems, service delivery, narrative-based data.

Background and Problem

• Problem / challenge: Metrics used by public agencies to measure support often misrepresent the lived experiences of families, failing to capture the relational and contextual dynamics of care and neglect.
• Significance: Misaligned metrics can obscure systemic gaps, reinforce inequities, and hinder effective service delivery, impacting vulnerable families navigating the child welfare system.
• Motivation and related work: Prior HCI research highlights the tension between sociocentric metrics and egocentric lived realities, calling for systems that integrate qualitative narratives to better represent complex care dynamics. This paper builds on these calls by centering parents’ experiences to interrogate existing metrics.

Solution

• Proposed approach: An interpretive computational narrative analysis workflow combining thematic coding, a small language model (Phi-4), and multidimensional scaling (MDS) to analyze parents’ accounts of support and unmet needs.
• Novelty:
  1. Introduces a proof-of-concept for integrating egocentric narratives into sociocentric data systems.
  2. Maps relational structures of support and neglect using computational techniques.
  3. Highlights information gaps in standardized metrics through parents’ lived experiences.
  4. Proposes design implications for collaborative, interpretive systems in mission-driven organizations.
• Procedure and key techniques:
  • Interviews with 75 parents involved in the child welfare system.
  • Thematic coding and computational modeling to identify patterns in narratives.
  • MDS visualization to map semantic relationships between themes of support and unmet needs.
  • Validation of computational outputs through manual coding and exemplar transcript analysis.

Results

• Concrete findings:
  • Parents with robust informal networks (e.g., family, friends, employers) received more flexible and effective support from caseworkers, while isolated parents faced procedural overload and neglect.
  • Sociocentric metrics like referrals and completions fail to capture timeliness, accessibility, and relational quality, masking systemic gaps.
  • Informal supports are critical but invisible in administrative records, creating information gaps that misrepresent compliance and progress.
• Advantage over baselines: Demonstrates how egocentric perspectives can expose misalignments in sociocentric metrics, offering richer insights into service delivery and structural inequities.
• Experiments / evaluation:
  • Qualitative interviews with 75 parents, analyzed using thematic coding and Phi-4.
  • MDS visualization to reveal clustering of support and neglect themes.
  • Validation through manual coding and exemplar transcript analysis.
• Limitations and future work:
  • Limited generalizability beyond the child welfare context.
  • Calls for further development of interpretive, collaborative systems to integrate egocentric narratives into public service metrics.

Summary

This paper investigates how parents involved in the child welfare system experience support and neglect, revealing critical misalignments between sociocentric metrics and egocentric realities. Using an interpretive computational workflow, the study highlights the fragmented and conditional nature of the ad hoc safety net parents rely on, shaped by informal networks and systemic gaps. Findings show that standardized metrics often obscure timeliness, accessibility, and relational quality, reinforcing inequities for isolated families. The paper proposes design implications for collaborative systems that integrate narrative-rich qualitative data to better align evaluation practices with the lived experiences of families.