@inproceedings{01JNNMJHBMVDEABPN2BH7FXEKN,
  abstract     = {{Consumers face over more than 1,000 food-related decisions annually (Food Insight, 2018). As a result, 
consumers rely (heavily) on cognitive shortcuts to simplify decision making, particularly when selecting and 
buying low-involvement food products (Del Campo, Pauser, Steiner, & Vetschera, 2016). Amongst these 
shortcuts, we find lay beliefs and intuitions, such as the "unhealthy = tasty" and "healthy = expensive" 
intuitions, which have been shown to influence consumer purchasing behavior (Haws et al., 2017; Li, 
Heuvinck, & Pandelaere, 2022; Raghunathan, Naylor, & Hoyer, 2006).

The existence of these intuitions is well established, in both descriptive and experimental research (Haws et 
al., 2017; Jo & Lusk, 2018; Raghunathan et al., 2006). Prior research also revealed that these intuitions can vary 
as a function of the context (Werle, Trendel, & Ardito, 2013). It remains unclear, however, where these intuitions 
originate from and whether they hold true in the real world. As an example, consider the 'healthy = expensive' 
intuition central to the present study, which refers to the belief that healthier food is more expensive than 
unhealthy food. The present study seeks to investigate if this (well-established) intuition holds true in a real-world retail setting. We therefore collected food products’ prices from two major Western European retailers 
and used the Nutri-score as an indicator of these products’ actual healthiness.

These retailers were selected because (a) a (very) large portion of their food products are assigned a Nutri-score, and (b) they have a well-established online presence, thereby allowing for the use of web scraping 
techniques to acquire the information needed for our research purposes. In total, we collected data of 4,070 
products from Retailer 1 (2,085 of which lacked a Nutri-score, i.e., 51.23%) and 3,068 products from Retailer 2 
(857 of which lacked a Nutri-score, i.e., 27.93%). Sophisticated machine learning techniques were used to 
assign Nutri-scores to food products that initially lacked one. It may be noted that the retailers selected for this 
study do not follow an "Everyday Low Prices" strategy but instead occupy the middle to higher end of the price 
spectrum, providing a more accurate reflection of standard market prices.

The price of each product was determined as the price per kilogram or liter, expressed relative to the median 
price within each product category. To control for brand-specific price variations, we applied the TrueSkill 
algorithm, an advanced method for accounting for complex multilevel data structures. Promotional prices 
were not considered in this analysis. 

Table 1 summarizes our main findings. Notably, the “brand” effect was highly significant in both datasets, 
showing that the brand of a product had a strong, positive impact on product pricing. In contrast, the presence 
vs. absence of a Nutri-score had no impact whatsoever on product pricing. Most importantly, the healthiness 
of a product (i.e., the NS-category) also had a significant, positive impact on product pricing: Food products 
with less favorable Nutri-Score values were more expensive than healthier options. This suggests that, across 
all food categories, less healthy products were consistently priced higher than healthier ones. Crucially, this 
observation holds true irrespective of the product category, given that random effects controlled for variations 
at the category level.

In sum, while consumers tend to believe that healthy food is more expensive than unhealthy food, we find no 
evidence for this intuition in real-world data. In fact, our findings suggest that the opposite is true. Given that 
lay beliefs and intuitions are known to influence consumer behavior (Haws et al. 2017; Li et al., 2022; 
Raghunathan et al., 2006), this observation constitutes an important contribution to the literature on lay 
intuitions. In addition, this paper taps into the potential of using web data and scraping for academic research. 
Although this approach to date remains underutilized in marketing and retailing research, it offers great 
potential to advance and provide more comprehensive insights in real-world retail settings (Guyt, Datta, & 
Boegershausen, 2024). The present study contributes to this growing body of work and indicates the potential 
of leveraging web data and using complex mathematical methods. 

Despite this study’s contribution, future studies could test if the same data pattern emerges in (a) retailers 
that generally charge lower prices (e.g., with an everyday-low-prices approach) and (b) in other parts of the 
world (see Jo & Lusk, 2018; Teele 2014). Future research could also explore the factors contributing to these 
price differences. This exploration could include both top-down as well as bottom-up supply and demand 
dynamics. Finally, it seems particularly interesting to engage in an empirical analysis of potential 
interventions that may help alleviate the influence of lay intuitions which, so our data show, have little to do 
with reality. To the extent that these lay beliefs contribute to suboptimal behavior (e.g., preferring unhealthy 
foods over healthy foods), such a research agenda will ultimately help to develop interventions that can 
nudge consumers toward optimal purchasing behavior.}},
  author       = {{Rumes, Daan and Janssens, Bram and Spruyt, Adriaan and Verstraeten, Julie}},
  booktitle    = {{2025 AMA Winter Academic Conference Marketing in Service of Nature and Humanity, Proceedings}},
  isbn         = {{9780877570226}},
  keywords     = {{Consumer Behaviour,Marketing Analytics,Retailing}},
  language     = {{eng}},
  location     = {{Phoenix}},
  pages        = {{1}},
  publisher    = {{American Marketing Association}},
  title        = {{Healthy choices, lower prices?}},
  volume       = {{36}},
  year         = {{2025}},
}