Constructing causal loop diagrams from large interview data sets.

Area of Science:

• Urban planning and development
• Systems thinking and modeling
• Qualitative research methodologies

Background:

• Urban development decision-making is complex and often opaque.
• Understanding stakeholder mental models is crucial for effective interventions.
• Causal loop diagrams (CLDs) can visualize these complex systems.

Purpose of the Study:

• To compare manual and semi-automated methods for constructing CLDs from qualitative data.
• To illuminate mental models and collective decision-making processes in urban development.
• To assess the efficiency and accuracy of different CLD construction approaches.

Main Methods:

• Application and comparison of four variations of CLD construction methods.
• Utilized 123 semi-structured interviews from the 'Tackling the Root Causes Upstream of Unhealth Urban Development' project.
• Employed both manual and semi-automated processes on interview transcripts and thematic analysis datasets.

Main Results:

• Semi-automated CLD construction offers time savings for large qualitative datasets compared to manual methods.
• Careful interpretation is needed for peripheral variables at the boundaries of thematic analysis.
• The choice between manual and automated approaches depends on the specific modeling objectives.

Conclusions:

• Both manual and semi-automated methods can effectively visualize mental models for urban development decision-making.
• Recommendations include recording quantitative descriptors for CLD construction processes from large qualitative datasets.
• Future research should further refine automated methods for complex systems analysis.