Consumer Behavior Dataset - Theory of Consumption Value (TCV) in Indian Staple Food Delivery (Linear and Non-Linear Analysis)
Description
This dataset contains 761 valid survey responses from Indian Gen-Z consumers (aged 18-28) regarding their consumption behaviour of staple foods via Food Delivery Apps (FDAs) like Zomato and Swiggy. The data was collected between July and September 2025 using a judgemental sampling method. The research framework is grounded in the Theory of Consumption Value (TCV), measuring five key dimensions: functional, emotional, social, conditional, and epistemic values. The dataset includes Likert-scale responses for these values, as well as measures for consumer attitude, behavioural intention, and actual buying behaviour. Demographic variables such as sex, education level, family size, work setting, and ordering frequency are also included. This data was utilised to perform a dual-stage analysis: Structural Equation Modelling (SEM) for linear relationships and Multivariate Adaptive Regression Splines (MARS) for non-linear threshold effects.
Files
Steps to reproduce
1. Survey Instrument: Utilise the survey items mapped to the TCV framework as described in the accompanying manuscript. Responses were recorded on a 5-point Likert scale. 2. Data Cleaning: The raw data was screened for completeness. Responses with missing values or failing the two filter questions (active FDA user and staple food ordering) were removed (N=39 removed, N=761 retained). 3. Linear Analysis (SEM): Load the .xlsx file into Smart-PLS (v4.1.1.5) or similar SEM software. Perform Confirmatory Factor Analysis (CFA) to check Factor Loadings (threshold > 0.7), CR (> 0.7), and AVE (> 0.5). Run the Bootstrapping algorithm (5,000 subsamples) to test the structural paths (H1 through H7). 4. Non-Linear Analysis (MARS): Export the latent variable scores from the SEM model. Import data into RStudio (v4.5.1). Use the earth package to execute the Multivariate Adaptive Regression Splines (MARS) model. Set attitude as the dependent variable and the five consumption values as predictors to identify knots and hinge functions as specified in the research paper. 5. Robustness Check: Apply the Ramsey RESET test in R to confirm the presence of non-linearity before interpreting MARS results.
Institutions
- National Institute of Technology KurukshetraHaryana, Kurukshetra