Rethinking wellbeing measurement: learning from a scale development study with adolescents living with HIV in Zimbabwe

Technical Summary: Rethinking Wellbeing Measurement in Adolescents Living with HIV in Zimbabwe

Problem Statement
Adolescents living with HIV (ALHIV) in Eastern and Southern Africa face high rates of treatment attrition, morbidity, and mortality, significantly exacerbated by common mental health conditions. While interventions like the Zvandiri model aim to improve wellbeing to enhance treatment adherence, current measurement tools are insufficient. Existing instruments predominantly focus on negative outcomes (e.g., depression, anxiety) or rely on positive psychology scales developed in high-income, individualistic settings. These tools often fail to capture the culturally embedded, relational, and fluctuating dimensions of wellbeing specific to ALHIV in sub-Saharan Africa, creating a gap in the ability to quantify protective attributes and intervention mechanisms.

Methodology
The study employed a multi-stage, mixed-methods scale development process (2023–2025) to create the Zvandiri Character Strength (ZCS) scale. The process followed a rigorous, iterative pathway:

1. Scoping and Systematic Review: Identified 15 positive psychological constructs used in ALHIV research, resulting in an initial item bank of 263 items.
2. Content Validity Assessment: Three reviewers rated items for relevance, reducing the bank based on Item-level Content Validity Index (I-CVI) scores.
3. Qualitative Exploration: Conducted interviews, focus group discussions (n=73), and expert panel meetings to identify key constructs (connectedness, happiness, hope, motivation, optimism, perseverance) and contextual nuances.
4. Co-Design and Prioritization: ALHIV participants (n=24) prioritized constructs, leading to a 53-item candidate scale spanning 14 constructs.
5. Cognitive Interviews: Assessed comprehension of items and response options.
6. Quantitative Testing and Reduction:
   • Administered the 53-item scale to 530 ALHIV.
   • Applied Exploratory Factor Analysis (EFA) and Rasch analysis to assess dimensionality, item fit, and threshold ordering.
   • Reduced the scale to 38 items (ZCS 38) and subsequently to 25 items (ZCS 25) based on statistical performance and participant feedback regarding ceiling effects and social desirability.
7. Retrospective Pretest/Post-test: Administered the ZCS 25 to a small pilot group (n=8) to test feasibility and sensitivity to change, including retrospective ratings of past wellbeing.

Key Results

• Construct Heterogeneity: Qualitative data revealed that wellbeing is relational, contextually situated, and often co-occurring with distress. Constructs like "happiness" were described as "happiness with some tears," challenging the notion of a singular, positive latent trait.
• Psychometric Challenges:
  • Ceiling Effects: The initial 53-item and subsequent 38-item scales showed extremely high positive responses (81–98% agreement), indicating restricted variance.
  • Disordered Thresholds: Rasch analysis revealed disordered response thresholds, suggesting the Likert scale options were not functioning as intended.
  • Weak Internal Structure: Factor analysis showed low inter-item correlations and cross-loadings, failing to support a single interpretable total score.
  • Response Bias: Participants reported that items felt like "positive affirmations," leading to acquiescence and social desirability bias, particularly in interviewer-administered settings.
• Retrospective Instability: In the pilot phase, retrospective pretest scores fluctuated unpredictably compared to post-test scores, and qualitative accounts of improvement often contradicted quantitative score changes.
• Outcome: The ZCS 25 scale did not demonstrate the reliability or validity required for use as a standalone outcome measure. The authors did not proceed to large-scale validation.

Key Contributions and Significance
The paper's primary contribution is methodological rather than the validation of a new tool. It serves as a case study illustrating the practical consequences of applying standard psychometric approaches to relational, context-specific interventions in low-resource settings.

• Critique of Imported Measures: The study highlights the limitations of adapting high-income country instruments without deep cultural grounding, noting that concepts like "self-acceptance" in Zimbabwe are relational (embedded in community) rather than individualistic.
• The "Jangle" Problem: It underscores the conceptual fragmentation in positive psychology literature, where overlapping labels (e.g., resilience, self-efficacy, hope) are often treated as distinct without clear differentiation.
• Response-Process Validity: The research emphasizes that response patterns (e.g., ceiling effects) were driven by the interaction between item wording (affirmations), administration mode, and social desirability, not just the underlying construct.
• Measurement Philosophy: The authors argue that future measurement must explicitly decide whether the goal is to capture "lived wellbeing" (which is fluctuating and relational) or "intervention-relevant change" (specific domain shifts). They suggest that if the former is the goal, methods may need to shift toward qualitative, visual, or ecological momentary assessment (EMA) approaches rather than static total scores.

Conclusion
The ZCS scale development process did not yield a validated instrument for a single total score of positive wellbeing. However, it provides critical lessons for future research: measurement must begin with clear construct definitions grounded in local lived experience, prioritize response-process validation, and align the measurement strategy with the specific theory of change of the intervention. The authors conclude that rigorous psychometric methods are essential but must be preceded by construct pruning and domain-level testing to ensure cultural and conceptual coherence.