Project Overview

• Whether student grit differs between individualistic (United States) and collectivist (Argentina) cultures
• Whether student grit can change after sustained interaction with MathSpring, an intelligent tutoring system that integrates growth-mindset messaging and affective learning companions

Research Questions

1. How do students display grit, and how can grit be effectively assessed within MathSpring?
2. Does grit differ between students in individualistic and collectivist cultures (United States vs. Argentina)?
3. Can grit change through exposure to growth-mindset messaging embedded within MathSpring?

My Role

• Designed and executed an independent, mixed-methods research study
• Managed and analyzed large, multi-site datasets across two countries
• Conducted statistical analyses including:
  • Correlation analyses
  • Independent-samples and paired-samples t-tests
  • Regression modeling
• Interpreted behavioral log data from an intelligent tutoring system
• Authored a full undergraduate thesis and presented findings publicly

Study Context: MathSpring

• Personalizes math problem difficulty using cognitive and effort-based models
• Collects detailed log data (e.g., attempts, hints, skips, response timing)
• Uses animated learning companions that provide encouragement and growth-mindset messaging

Participants

United States Dataset

• N = 324 students (primarily grades 6–7)
• Data collected in Massachusetts (2017–2018)
• Post-test grit data only

Argentina Dataset

• N = 179 sixth-grade students
• Three middle schools across multiple sections
• Pre-test and post-test grit data collected
• Study conducted over 5 weeks with weekly MathSpring sessions

Measures

Grit Assessment

• Duckworth’s 8-item Grit Scale (Likert-type)
• Translated and culturally adapted for Spanish-speaking participants

Behavioral Indicators (from MathSpring logs)

• Problem-solving behaviors (e.g., attempts, guesses, skips, hints)
• Indicators of disengagement (e.g., answering too quickly without reading)
• Performance measures (accuracy, mistakes, mastery patterns)

Analytical Approach

• Descriptive statistics to examine grit distributions by country and school
• Independent-samples t-tests to compare U.S. and Argentina post-test grit
• Paired-samples t-tests to examine pre-/post-changes in Argentina
• Correlation and regression analyses to explore relationships between grit and learning behaviors

Key Findings

Cross-Cultural Differences

• Argentina students demonstrated significantly higher post-test grit than U.S. students (p = .02)

Change Over Time

• Argentina students showed a significant increase in grit from pre-test to post-test after using MathSpring (p < .01)

Behavioral Predictors

• Behavioral log data showed weak predictive power for grit
• Certain disengagement behaviors were negatively associated with grit, but accounted for less than 10% of variance, highlighting the complexity of measuring affect through behavior alone

Interpretation

• Grit may be influenced by cultural context, consistent with theories of collectivism
• Grit appears malleable, showing modest improvement after exposure to growth-mindset-aligned learning environments
• Behavioral data alone may be insufficient to fully capture affective constructs without richer contextual signals

Why This Matters

• Learning analytics and affective computing
• Cross-cultural psychology
• Educational technology and intelligent tutoring systems

Skills Demonstrated

• Cross-cultural research design
• Learning analytics and log-file analysis
• Statistical modeling and hypothesis testing
• Survey adaptation and translation
• Independent research and academic writing
• Data-driven evaluation of educational technology