BoPMLPIP: Application of Classification Techniques to Explore the Impact of PIP among BoPs

Debadrita Panda, Sabyasachi Mukhopadhyay, Rajarshi SahaInternational Journal of Intelligent Systems and Applications 2022 · DOI

Abstract

This study tries to gain insight into the effect of demographic and psychological variables on the Bottom of the Pyramid (BoP) consumers for making Packaging Influenced Purchase (PIP) decisions by focusing on two specific consumer behaviour theories - compensatory consumption and consumers’ resistance. Being the product's face, packaging contributes heavily to the above mentioned two streams of consumption behaviour. A collection of ten demographic variables and four psychological variables have been administered on a sample of 1400 BoP consumers to explore their effect behind making PIP of selected FMCG products. Various classification techniques have been deployed to capture the impact of these variables. This experimental research design revealed that both demographic and psychological variables affect the PIP. The comparison between urban and rural BoPs potentially comes with the guidelines for practical marketing implications.

Introduction

The "Bottom of the Pyramid" (BoP) represents a massive, often overlooked global market with over $5 trillion in purchasing power. However, research into BoP consumer behavior is polarized: one side sees an untapped opportunity for "inclusive capitalism," while the other warns of potential exploitation where attractive marketing lures the poor into unnecessary spending. This paper addresses a critical gap in understanding how Packaging Influenced Purchase (PIP) decisions function within this demographic.

Central to this investigation are two competing behavioral theories: Consumer Resistance, where packaging serves as a "silent salesman" providing vital functional information, and Compensatory Consumption, where packaging triggers impulsive buying to compensate for socio-economic status. Furthermore, the study investigates whether these drivers differ between urban and rural BoP populations, who navigate distinct infrastructure and employment realities.

Methodology

The researchers employed an experimental design combining traditional behavioral surveys with modern machine learning classifiers. The study utilized a primary dataset collected via structured questionnaires from 1,400 BoP consumers in two districts of West Bengal, India (Paschim Medinipur and North-24-Pgs). The sample was split into 800 urban consumers (often unskilled or semi-skilled laborers) and 600 rural consumers (primarily dependent on agriculture), all with an average monthly income $\le$ Rs. 10,000 (~$120 USD).

Research Methodology Flowchart

As illustrated in the methodology flowchart (Figure 2), the researchers tracked 10 demographic variables (e.g., age, income, education, TV habits) and 4 psychological variables (Customer Expectation, Perceived Opportunity, Perceived Self-Efficacy, and Political Interest/Ethical Concerns). Four classification algorithms—Random Forest (RF), Logistic Regression (LR), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM)—were deployed to predict PIP behavior.

Results and Discussion

The machine learning models achieved high predictive accuracy, proving that PIP behavior is not random but tied to specific socio-psychological markers. As shown in the ROC curve comparison (Figure 3), the Random Forest and Logistic Regression models emerged as the most effective classifiers.

ROC Curve Comparison for Classification Models

Key Findings:

  • Model Performance: The Random Forest model was the top performer for demographic analysis, achieving 82% accuracy for urban BoPs and 75% accuracy for rural BoPs. For psychological variables, Logistic Regression led with 85% accuracy for urban and 78% for rural segments.
  • The Urban-Rural Divide: Rural BoP consumers are more influenced by the visual elements of packaging. In contrast, urban BoP consumers respond more to information and innovation, suggesting that urbanization increases "consumer resistance" and a focus on functional value.
  • Social Capital and Psychological Drivers: In rural areas, "Political Interest" (a proxy for social capital) was a significant predictor, increasing the chance of a PIP by 3.54 times. Universally, Customer Expectation and Perceived Self-Efficacy were critical, indicating that BoP consumers use packaging to judge quality and status rather than being easily misled.
  • The "TV Effect": Duration spent watching TV was a significant demographic factor for both groups, acting as a catalyst for brand awareness that eventually influences purchase decisions.

Conclusion

This study demonstrates that PIP behavior among BoP consumers is a structured phenomenon influenced by both demographic and psychological factors. While the quantitative machine learning approach is robust, the authors acknowledge limitations in geographic scope (limited to West Bengal) and product category (focused on FMCG). Future research should incorporate qualitative deep-dives and a broader range of psychological variables to further refine these predictive models and explore the impact of packaging on durable goods.

Summary updated

Abstract

This study tries to gain insight into the effect of demographic and psychological variables on the Bottom of the Pyramid (BoP) consumers for making Packaging Influenced Purchase (PIP) decisions by focusing on two specific consumer behaviour theories - compensatory consumption and consumers’ resistance. Being the product's face, packaging contributes heavily to the above mentioned two streams of consumption behaviour. A collection of ten demographic variables and four psychological variables have been administered on a sample of 1400 BoP consumers to explore their effect behind making PIP of selected FMCG products. Various classification techniques have been deployed to capture the impact of these variables. This experimental research design revealed that both demographic and psychological variables affect the PIP. The comparison between urban and rural BoPs potentially comes with the guidelines for practical marketing implications.
BibTeX
@article{panda2022bopmlpip,
  author = {Debadrita Panda and Sabyasachi Mukhopadhyay and Rajarshi Saha and Rajarshi Saha},
  title = {{BoPMLPIP: Application of Classification Techniques to Explore the Impact of PIP among BoPs}},
  journal = {International Journal of Intelligent Systems and Applications},
  year = {2022},
  doi = {10.5815/ijisa.2022.06.02},
}
Figure