Interpersonal Communication and Maternal Behavioral Practice: A Case Study of Explanatory Causal Machine Learning in...

Area of Science:

• Public Health
• Nutrition Science
• Health Communication

Background:

• Understanding the mechanisms of nutrition interventions is crucial for program effectiveness.
• Few studies have explored the pathways through which nutrition education impacts behavior.

Purpose of the Study:

• To assess the total effect of maternal interpersonal communication on nutrition practices.
• To determine the extent to which this effect operates through improved awareness and knowledge.
• To compare machine learning (ML) and structural equation models (SEMs) in analyzing these pathways.

Main Methods:

• Utilized data from the Suaahara study (n=2040 mothers with children under 2).
• Estimated total and indirect effects of interpersonal communication on nutrition practices using ML and SEMs.
• Adjusted for household, child, and maternal covariates.

Main Results:

• Maternal interpersonal communication positively impacted nutrition-related practices.
• Both ML and SEMs showed similar total effects (ML: 0.58, SEM: 0.60).
• ML models attributed a larger proportion of the effect to indirect pathways via awareness and knowledge (ML: 0.24 vs. SEM: 0.16).

Conclusions:

• Interpersonal communication is vital for improving nutrition practices, partly by boosting awareness and knowledge.
• Disparities between ML and SEMs highlight the complex, nonlinear nature of intervention effects.
• Simple linear models may not fully capture the intricate relationships between awareness, knowledge, and outcomes.