Healthy Alternatives and Consumer Choice

Impact of Healthy Alternatives on Consumer Choice: A Balancing Act

Trivedi, Sridhar, & Kumar (2016) — Journal of Retailing

Motivation and Research Questions

  • Health consciousness is rising — 32% of consumers switching to healthier options (FMI 2012)
  • But obesity rates exceed 25% in most US states — a disconnect between concern and behavior
  • Prior research: mostly experiments with stated data, single-category focus
  • Core puzzle: Do consumers “balance” healthy and indulgent choices across categories?
RQ Question
RQ1 Do consumers balance consumption of a common health element (fat, salt, sugar) across categories?
RQ2 Is this balancing behavior consistent across different health elements?
RQ3 Are stated health orientations consistent with actual purchase behavior?

Conceptual Framework: Compensatory Balancing

  • Consumers treat negatively-perceived attributes (fat, salt, sugar) as a “budget” to manage across categories
  • A healthy choice in one category compensates for an indulgent choice in another
  • Example: Buy low-fat milk but regular cheese → total fat intake is “balanced”
  • Unlike prior work (single-category, stated preferences), this study:
    • Models actual cross-category purchase behavior
    • Links stated survey attitudes with revealed scanner data
    • Captures household-level heterogeneity

Theoretical foundation: compensatory decision-making (Fishbein & Ajzen 1975; Chernev & Carpenter 2007) extended to negatively-perceived attributes across categories.

Data and Methodology

Dual-source data: - Scanner panel: actual purchases from 400 households - Survey: psychographics + demographics from the same households

Three health elements, 10 categories:

Element Categories
Fat Milk, Yogurt, Creamer, Cheese
Salt Canned Soup, Crackers
Sugar Cereal, Ice Cream, Cookies, Peanut Butter

Method: Augmented Latent Class Model (ALCM) - Multi-category choice model with cross-category interactions - Allows households to be a weighted mixture of segments (Dirichlet) - Captures true heterogeneity beyond discrete membership

Three Consumer Segments (Consistent Across All Elements)

Segment Description Behavior
Health Driven Strong health orientation Seeks healthy versions; low price/promo sensitivity
Balancing Moderate approach Mixes healthy + regular; compensates across categories
Hedonic Indifferent to health Prefers regular products; high price/promo sensitivity
  • BIC selects 3 segments for each health element
  • Segment proportions vary dramatically by element
  • r parameter (proportion of “pure” segment members) supports ALCM over standard LCM

RQ1: Evidence of Balancing Behavior

Significant negative cross-category interactions indicate balancing:

Element Health Driven Balancing Hedonic
Fat (31%) 1 neg / 12 sig 7 neg / 12 sig 9 neg / 12 sig
Salt (41%) Minimal balancing Substantial Substantial
Sugar (23%) Minimal balancing Substantial Substantial

Key insight: ~69% (fat), ~59% (salt), ~77% (sugar) of households exhibit balancing behavior

The Health Driven segment shows almost no balancing — they consistently buy healthy. But they are the minority for fat (31%) and sugar (23%).

RQ2: Balancing is NOT Consistent Across Elements

Segment membership migration (Fat → Salt):

Fat Segment → Healthy (Salt) → Balancing (Salt) → Hedonic (Salt)
Healthy 38% 19% 33%
Balancing 25% 17% 57%
Hedonic 52% 30% 17%

Core finding: Consumers who balance on one health element typically become Hedonic on others. - 57% of Fat-Balancing → Salt-Hedonic - 61% of Fat-Balancing → Sugar-Hedonic - Only 17% stay in Balancing segment for another element

RQ3: Stated vs. Revealed Health Orientation

HOS (Health Orientation Score) comparison:

Segment Revealed (HOS param) Stated (Factor score)
Health Driven 0.49 +0.315
Balancing 0.06 +0.052
Hedonic 0 (base) −0.622

Critical mismatch: The Balancing segment’s revealed behavior is close to Hedonic (HOS = 0.06 vs. 0), but their stated perception aligns with Health Driven.

Balancing consumers believe they eat healthy, but their actual purchase patterns show otherwise. This explains why public health messaging often fails — consumers already think they’re doing enough.

Promotional Strategy Insights

Cross-category promotional elasticities (Fat categories):

Promotion on → Healthy Milk Regular Milk Healthy Cheese Regular Cheese
Healthy Milk 3.11 −0.75 0.13 0.09
Regular Milk −0.98 3.01 0.07 0.19
Healthy Cheese 0.04 0.14 1.21 −0.25
Regular Cheese 0.02 0.11 −0.27 1.82

Insight: Promoting healthy milk depresses regular milk sales more effectively than promoting healthy cheese depresses regular cheese. Strategic bundling (healthy milk + healthy cheese) can amplify healthy consumption.

Managerial and Public Policy Implications

  1. One-size-fits-all messaging doesn’t work — three distinct segments need tailored strategies
  2. The Balancing segment is the key target — largest group for fat (36%) and sugar (37%), but they overestimate their healthfulness
  3. Cross-category bundling — promote healthy versions of categories with strong negative cross-effects (milk suppresses cheese)
  4. Choose which product to promote — promoting healthy milk has greater spillover benefits than promoting healthy cheese
  5. Element-specific strategies — fat/salt: more focused segments; sugar: significant mixing requires broader approaches
  6. Public policy must be more specific — “eat healthy” is ineffective when consumers believe their balancing behavior already achieves this

Limitations and Future Research

  • Symmetric interaction effects assumed (Russell & Petersen 2000 framework) — could be relaxed
  • Computational constraints limit number of categories per health element
  • No unified framework across all three elements simultaneously
  • Cannot distinguish “naturally low” consumption from “deliberately restricted” consumption
  • Quantity purchased not modeled
  • Spatial/geographic peer influences not explored

Future directions: - Unified cross-element framework - MCMC methods for larger category sets - GIS-based spatial analysis of healthy consumption patterns - Accommodate purchase quantity in balancing models

Citation: Trivedi, M., Sridhar, K., & Kumar, A. (2016). Journal of Retailing, 92(1), 65–82. https://doi.org/10.1016/j.jretai.2015.05.003