1 of 12

ESTEEM: Measuring Socio-emotional Skills

Clara Delavallade

www.poverty-action.org/socio-emotional-skills

2 of 12

Which skills matter for economic empowerment?�TOC: Skills key to goal achievement, but they depend on the goal

Developing Preferences,

Setting “Achievable” Internalized Goals

Emotional awareness

Self awareness

Listening

Empathy

Action Planning

Personal Initiative

Problem Solving

Taking Action

Self Control

Emotional Regulation

Perseverance

Expressiveness

Negotiation

Interpersonal Relatedness

Influence

Collaboration

Key Behaviors for Economic Empowerment

Developing Networks/ Obtaining support

Overcoming obstacles, binding constraints, social norms

Finding and utilizing opportunities/solutions: jobs/business/programs

3 of 12

Limitations of Existing Measures

How ESTEEM Responds

LMIC-Specific Challenges

    • Literacy, translation, scale usage, offline formats
    • Western bias in objective tasks (Laajaj et al., 2019; Dinarte-Diaz & Egana-DelSol, 2022)

Cultural Sensitivity

    • Designed for low-literacy contexts.
    • Adaptable; piloted and validated in Nigeria, Tanzania, Côte d’Ivoire, Rwanda, DRC

Scalability

    • Modular, flexible for multi-topic surveys or standalone use

Bias in Self-Reports (cf. STEP):

    • Acquiescence, reference effects, recall bias, social desirability
    • SRs may reflect perceptions more than skill levels
    • Can measure traits rather than skills
    • Beyond SRs: behavioral measures
    • Evidence-Based: Developed through psychometric review, FGDs, and interdisciplinary research
    • Policy Relevance: Provides disaggregated data to inform targeting, program design, and cross-country benchmarking

Why ESTEEM? Addressing Measurement Challenges in LMICs

4 of 12

Countries -

Introducing ESTEEM

Intrapersonal

Interpersonal

Awareness

Emotional Awareness

Self Awareness

Listening

Empathy

Management

Emotional Regulation

Self Control

Personal Initiative

Perseverance

Problem Solving

Expressiveness

Interpersonal Relatedness

Influence

Negotiation

Collaboration

Development

  • Lit on Employment & Gender Differences
  • Map to other Frameworks
  • Focus Group Discussions
  • Spanning Range of SES

5 of 12

Self Report Scale (Empathy: Cognitive & Affective)

Example Measures

Situational Judgement Tests (Emotional Regulation)

Tasks: Counting triangles, CPTX, Listening comprehension/behaviors, SJT-embedded tasks

6 of 12

Measure Development�Iterative Psychometric Analysis

Psychometrics in each context

Content Validity

Focus Groups/ Interviews w. Entrepreneurs

    • Which skills
    • Which behaviors
    • Critical incident sourcing

Cognitive Interviews

    • Assess understanding

Intensive Translations

    • Translation ratings, Linguistic Feedback, Glossary

Review from Subject Matter Experts

    • Psychologists, SES Curricula, Education Assessment, Field Coordinators

Key Learnings

  • Importance of each content validity step
  • Tasks often time consuming and focus on sub-skill
  • 2 skills had 2 dimensions: Listening & Relatedness
  • Behaviors depend on spheres -> scenario-development was key
  • Method differentiates factors more than skill
  • Reversed items could not be included on the same scale

7 of 12

Key take-aways from validation papers

  • Which measure:
    • Behavioral measures outperform self-reports in predicting labor market outcomes
    • Self-reported measures better predict mental health outcomes

  • Which skills? Relationship management is only consistent predictor of income: Collaboration, negotiation, influence and empathy are the strongest predictors of income across multiple countries.

  • Mixed results on gender: For women, SES are a better predictor of income if norms are less restrictive. Seems like a promising avenue for further research.

  • Implications
    • Curricula and measures need to consider relationship management skills, which is often not included
    • In development process, we learned about how skill “levels” depend on the “sphere” of usage
    • Is it levels of skills that matter, how much you value them, or whether others value you having them?

8 of 12

Pillar 3: RCTs testing various SES trainings�1. BRAC Tanzania IE�Sample: 4800 NEET Youth

40 urban & peri-urban communities in Tanzania (Dar es Salam, Dodoma & Iringa)

2400 men & 2400 women (60 men & 60 women per street) aged 16-27

Not in full-time salaried employment, education or training

T1: Awareness training

27 hours

T2: Management training

27 hours

T3: Awareness and management training

54 hours

Control

Baseline, Follow-up 3 and 12 months after intervention: May 2021-Oct 2022

9 of 12

Baseline Results

  • Female disadvantage in widely used self-reported SES measures, which disappears with behavioral measures.
  • Gender differences in self-reports are strongly correlated with social desirability and internalized gender norms.
  • There is a larger gap between self-reported and behavioral measures among men, also more strongly correlated with SD and regressive gender beliefs on women’s ability to problem-solve among men.
  • Suggests that men over-report their skills more than women underestimate theirs.

Preliminary IE Results

  • Limited impacts on behavioral measures of SES but significantly positive impacts on most self-reported measures at ML, which are not sustained after one year.
  • Significantly positive impact on employment and income only for men seeking jobs at baseline.
  • No significant difference between the types of SES training (awareness, management, or both)
  • These results suggest that providing SES training alone may not be sufficient to improve young women’s employability and income.

BRAC Tanzania IE�Sample: 4800 NEET Youth

10 of 12

2. Nigeria APPEALS IE: Intrapersonal v. Interpersonal skills

Sample: APPEALS (Agri-business training + grant) beneficiaries

Design: 4 days interpersonal v. 4 days intra+inter v. Control

Preliminary Results from follow-up 2.5 years after intervention:

    • Different impacts of SES trainings for men and women
      • Women specialize in one business, hire more female workers, and increase profits by 50%+, but see no change in decision-making or empowerment.
      • Men diversify income sources, increase household income, and reduce spending on temptation goods.

      • Male employers hire fewer women, female employers hire more, redistributing female workers but reinforcing occupational segregation.

    • Gendered skill development reinforces social norms
      • Women gain interpersonal (communal) skills (e.g., collaboration, empathy).
      • Men develop intrapersonal (agentic) skills (e.g., decision-making, problem-solving).
      • SRs increase for women, only where gender norms less restrictive

    • Policy implications
      • Tailored training: Interpersonal skills benefit women most; combined training is more effective for men.
      • SES trainings improve skills and business outcomes, especially for women, but do not help them deviate from norms.

11 of 12

Lessons and next steps

Lessons learnt:

  • Importance of interpersonal skills (relationship management) for women’s labor market outcomes
  • Importance of going beyond self-reports 🡪 behavioral measures less subject to biases
  • Self-reports matter too
      • if interested in wellbeing/mental health outcomes
      • Proxy for perception of skills/self-confidence, a predictor of econ outcomes too

ESTEEM measure applications:

  • Measuring impact of SES or other interventions on SES
  • Initial assessment of SES, to inform intervention scoping:
    • To highlight skills to focus on
    • To capture potential heterogeneity along dimensions such as gender, age, initial goals, etc…

Next steps:

    • Dissemination: GIL research uptake (incl. webinars) + JPAL-IPA research agenda
    • Digitizing SES measures and incorporating them into SkillCraft platform
    • Designing short version
    • Adapting and testing in other contexts
    • Which skills to signal to employers?
    • Examining returns to SES when relaxing gender norms (Nigeria for Women)

12 of 12

Website: www.poverty-action.org/ses

Thank you!