Application of Choice-Based Conjoint Analysis

This repository contains the implementation of a Choice-Based Conjoint Analysis to study user preferences regarding AI assistant features (e.g., ChatGPT).

Project Description

The project uses the Choice-Based Conjoint Analysis methodology to analyze user preferences across five main attributes of an AI assistant:

Attribute	Levels
Specialization	Assistant, Code, Content
Speed	Slow, Fast
Quality	Sufficient, Optimal
Privacy	Low, High
Cost	15€, 20€, 25€

Repository Structure

├── Analisi.R                           # Main analysis script
├── Design_Project.R                    # Experimental design generation
├── Convert_survey_to_choice_format.R   # Google Form data conversion
├── Extract_info_per_respondent.R       # Demographic info extraction
├── data/
│   ├── Choice_Data_Converted.csv       # Converted data for analysis
│   └── demographic_information.csv     # Respondent demographic info
└── raw_data/
    └── Questionario Lab (Risposte).csv # Raw responses from Google Form

R Scripts

Script	Description
Analisi.R	Main data analysis script using Multinomial Logit and Mixed Multinomial Logit models
Design_Project.R	Optimal experimental design generation
Convert_survey_to_choice_format.R	Conversion of Google Form data into the format required for analysis
Extract_info_per_respondent.R	Extraction of respondent demographic information

Data

• raw_data/: Raw data collected from the Google Form
• data/: Processed data ready for analysis
  • Choice_Data_Converted.csv: Responses converted into the format required for Conjoint Analysis
  • demographic_information.csv: Demographic information (age, gender, education level)

Contributors

• @filippostanghellini
• @MolteniF
• @01andreabattaglia
• @matteo-massari
• @TommasoCestari