2006 Marketing & Society Dissertation Proposal Award

Kathryn Sharpe, Duke University
 

UNDERSTANDING CONTEXTUAL EFFECTS LEADING TO OVER CONSUMPTION

INTRODUCTION

Policy makers, medical practitioners, and firms have shown increased concern with the steep rise in
obesity rates worldwide. Though many factors influence if someone is obese, weight gain is generally
due to an imbalance between energy expended through physical activity and calories consumed. Despite
the popular belief that exercise and activity levels have declined over the past several decades, at least in
the United States, these levels have remained relatively flat (Cutler, Glaeser, and Shapiro 2003). Thus the
rise in obesity in the U.S. has been primarily driven by the increase in caloric consumption.

As a result, the associated overweight and obesity-related healthcare costs in the United States are
estimated to be well over $100 billion1 annually (DHHS 2001; Finkelstein, Fiebelkorn, and Wang 2003).
Nearly half of this expense is borne by government through Medicaid and Medicare, while the remainder
is primarily absorbed by those paying health insurance premiums (Finkelstein et al. 2003). Though
industry and government have made attempts over the last two decades to reverse the obesity trend with
improved health education and nutritional information, caloric consumption of the population continues
to rise (Russo, Staelin, Nolan, Russell, and Metcalf 1986; Burton, Creyer, Kees, and Huggins 2006). This
dissertation presents a new approach to the problem based upon consumer decision making and economic
modeling. The objective is to control for the economic factor of price and test contextual factors in the
consumer’s decision environment, specifically hypothesized to drive greater consumption. The factors of
interest are extremeness aversion (Essay I), the impact of reference pricing on bundled meals (Essay II),
and the impact of both of these factors on firm actions and government policies aimed at reducing caloric
consumption (Essay III).


DISSERTATION ESSAY I: Estimating the Impact of the Portfolio of Portion Sizes on Choice

The goal of Essay I is to demonstrate the impact of size assortment on caloric consumption. Soft drinks
are used as the case study given the national growth in fast food consumption, and in particular, the
increase in beverage portion sizes (Popkin and Nielsen 2003; Nielsen and Popkin 2003). If consumers
value how much liquid soft drink they receive, their purchase decision is based on a tradeoff between
fluid ounces (quantity) and price (money). In the case of a firm dropping its smallest size, a consumer
who previously chose the smaller size will purchase a larger size, if it is worth paying the additional
expense. Otherwise, he will not choose a drink at all. Based on individual valuations, some customers
will choose the larger size, while others will choose not to purchase a drink. Alternatively, if a firm
removes its largest size, the displaced customers will shift to a smaller size, but all other customers will
choose the same size as before the removal. This is consistent with rational choice theory based on the
tradeoff between quantity and price.

If systematic deviations from this behavior exist when the choice set is changed (either by adding or
deleting a size), then choice is, to some degree, dependant on the choice set. For example, Simonson
(1989) demonstrated that the population tends to avoid products at the extremes in a product portfolio,
termed the “compromise effect.” Thus altering the identity of a product as extreme decreases consumers’
likelihood of purchasing the item. Kivetz, Netzer, and Srinivasan (2004) developed useful models to
parameterize this compromise effect at the population level.

The purpose of this essay is to estimate these individual effects for choice of drink size in a portfolio.
Based on the extremeness aversion literature (Simonson 1989; Simonson and Tversky 1992; Tversky and

1 The estimated cost in 1998 was $78.5B, inflation adjusted, $104.4B in 2005 (Finkelstein et al. 2003). Alternatively, an inflation adjusted estimate of $129 billion for 2004 is based on an estimate of $117 billion in year 2000 dollars (DHHS, 2001) calculated from the 2005 Consumer Price Index - U.S. Bureau of Labor Statistics 2004.  Simonson 1993) and rational utility theory, the following hypotheses regarding overall consumption are
generated:

H1: When the smallest size in a portfolio is eliminated

a. an individual who originally chose the smallest size will choose one of the larger still
available sizes if the person is willing to pay for an available drink, thus increasing caloric
consumption; otherwise, the individual will choose not to purchase a drink at all, thus
decreasing caloric consumption.
b. Additionally, a consumer, who is averse to purchasing the smallest size and finds (after the
elimination of the smallest drink) his or her “preferred” size is now the smallest drink size,
may switch to a larger size. This occurs if the individual’s extremeness aversion is larger
than the disutility of purchasing a larger size.2 Thus in this case, caloric consumption
increases.


H2: When a larger size is added to a portfolio

a. a consumer who values the increase in quantity more than the combined effects of the
increase in price and the aversion from purchasing the largest size, will switch to the larger
added size, and consequently, increases consumption.
b. A consumer who had previously avoided purchasing the largest size may choose this size
after the addition of an even larger size. Thus is this case, consumption increases.

To generate the data to test these hypotheses, a two part online experiment is employed. The first part is a
within subject design in which participants make eight choices where the portfolio of drink sizes is
manipulated across restaurants. For each respondent, I estimate the value for portion sizes and
extremeness aversion. To measure price sensitivity, the second part of the experiment involves
completing a conjoint task in which participants select between pairs of drinks of different sizes and
prices.

This two-part experiment generates the data to measure (after controlling for price and inherent value for
a drink) an individual’s systemic avoidance of extreme sizes. The overall objective of this essay is
threefold: a) identify and test each of the hypothesized effects, b) develop a structurally flexible model
that allows for each of the effects, and finally c) estimate the model using a Hierarchical Bayes approach.


DISSERTATION ESSAY II: The Promotional and Contextual Aspect of Bundling

The purpose of Essay II is to examine the effect of bundling on caloric consumption. Currently in the
marketplace, most national fast food firms offer bundled meals, more commonly known as “combo
meals” or “value meals.” Nearly all of the fast food firms bundle together an entrée, a medium side (most
often fries), and a 21oz drink. The goal of this essay is to determine the effects of this practice on a) an
individual’s tendency to purchase more food because he or she is attracted to other perceived aspects of
the bundle and b) how the decisions of consumers who decide not to purchase the bundle are impacted
due to the presence of bundled meals on the menu.

In addition to testing the effect of offering bundled meals on consumption, this essay also addresses
different bundled menu formats. In most fast food establishments, bundled meals occupy a separate
section of the menu display. Recently, McDonald’s has incorporated the ala carte entrée price on the
value meal menu. Thus when consumers search for the ala carte price for an entrée they observe both the

2 This assumes larger sizes are equal or greater in price

ala carte price and the value meal price. Alternatively, most other fast food firms (e.g. Wendy’s or Burger
King) keep the prices separate; thus, when consumers search for the price of a bundled meal and the
respective ala carte entrée, the prices are found in two different places. As examples, Wendy’s (separate)
and McDonald’s (combined) menus are provided below:

Separate

Menu

Combined

Menu

Preliminary results of a study conducted suggest that consumers who choose a combo meal in both
formats do not significantly differ in choice behavior; thus, consumption is not significantly impacted for
these individuals. However, for consumers who prefer to purchase food items ala carte, there exists a
very subtle context effect, namely the saliency of the “value” (reference price) of the smaller bundle
composed of the side order and the drink. In other words, consumers are more likely to compare the price
of the entrée with the combo price and then determine the “value” of the remaining items, i.e., the side
order and the drink. This value is hypothesized to be less when consumers do the comparison than when
they do not do the comparison. This intuition is based on the assumption that mental accounting (Thaler
1985) is more likely to occur in the combined menu format than in the separate menu format.

A within subject experiment is designed to generate the data to test these hypotheses and estimate the
magnitude of the effects. In this study, each of the choice occasions is a restaurant menu where menu
format is manipulated. Additionally, the portfolio of drink sizes varies; however, unlike the previous
study, prices for all selections vary. This allows for the estimation of both the context effect of the
portfolio of drink size and price sensitivity, in addition to the effects of bundles and the bundle menu
format. Furthermore a conjoint task follows such that participants are asked to perform a series of choice
tasks to allow for a more accurate estimation of their price sensitivity and valuation of different menu
items such as sizes of drinks, fries, and meals.

To my knowledge, there are no publicly available studies that measure the impact of bundled meals. The
study uses a national sample and the results will establish a baseline for the positive or negative effects on
caloric consumption. These effects will depend on the magnitude of the increases and decreases in
consumption. The second major objective is to determine the effects of two different value meal menu
formats. Lastly, the study will test the effects of drink size, this time in an environment with value meals.
Analysis of these data will be done at the individual level, and Hierarchical Bayes will be used for the
estimation.
 

DISSERTATION ESSAY III: Application of Behavioral and Estimation Research on Policy and Firms

The models developed in Essay I and Essay II are applied in Essay III to examine the effects of potential
firm and government actions to reduce obesity. The individual utility functions developed in Essays I and
II are used to generate demand functions for entrees, sides, and drinks. Using these demand functions, I
assume firms are profit maximizing entities which set prices and assortment sizes. This combined
consumer and firm action model allows for the examination of how firm profits and caloric consumption
are impacted through consumption reduction policies such as a fixed tax, a graduated tax based on
consumption, an average size requirement, or fixed limitations on the assortments available. Preliminary
analyses have indicated large differences in both the impact on consumption and profitability from
different regulatory regimes.

This research can have consequential effects on the current obesity problem, because changing one aspect
of a consumer’s behavior can have measurable impact in the long run. For example, an individual who
adds (reduces) intake by 120 calories a day (this is the difference between a 21oz and a 32oz non-diet soft
drink3) will gain (shed) eight pounds a year. Moreover, a seemingly harmless extra 120 calories a day
seems to go unrecognized. It is just a single size increase. Yet a 10 to 20 cent decision, e.g., the choice
between purchasing a 21oz drink and 32oz drink, can greatly affect weight. By understanding how
individuals make tradeoffs, this research not only contributes to the understanding of the behavior of
consumers, the actions of firms, and policy decision making related to obesity, but also to consumers’
wallets and overall health.
 

DISSERTATION PROGRESS

This completed dissertation expands upon our understanding of the contextual factors affecting consumer
choice. Together the essays involve modeling and estimating the magnitude of these effects and how
these factors impact firms, policies, and consumers as related to the obesity problem. The methods
involved to complete this proposed dissertation include experimental methods, Hierarchical Bayes
estimation, and non-linear optimization. The thesis contributions are substantive and theoretical but also
conceptual in that it utilizes a portfolio of sophisticated analyses to obtain more valid conceptual insights.
The timetable summarizing the three parts of the dissertation (to be completed before the Fall 2007 job
market) is described below:


REFERENCES

Burton, Scot, Elizabeth Creyer, Jeremy Kees, and Kyle Huggins (2006), “Attacking the Obesity
Epidemic: An Examination of the Potential Health Benefits of Nutrition Information Provision in
Restaurants,” American Journal of Public Health, 96 (9) 1-6.

Cutler, David M., Edward L. Glaeser, and Jesse M. Shapiro (2003), “Why Have Americans Become More
Obese?” Working Paper National Bureau of Economic Research: Massachusetts.

DHHS (2001), The Surgeon General’s Call to Action to Prevent and Decrease Overweight and Obesity.
Rockville, MD: U.S. Department of Health and Human Services, Office of the Surgeon General.

Finkelstein Eric A., I. C. Fiebelkorn, G. Wang (2003), National Medical Spending Attributable to
Overweight and Obesity: How Much, and Who’s Paying? Health Aff (Millwood) 22:219–226.

Kivetz, Ran, Oded Netzer, and V. Srinivasan (2004), “Alternative Models for Capturing the Compromise
Effect,” Journal of Marketing Research, XLI (August) 237-257.

Nielsen, Samara J., and B.M. Popkin (2003), “Patterns and Trends in Portion Sizes, 1977-1998,” Journal
of the American Medical Association, 89(4), 450-453.

Popkin, B.M. and Samara J. Nielsen (2003), “The Sweetening of the World’s Diet” Obesity Research,
11(11) 1325-1332.

Russo J.E., Staelin R., Nolan C.A, Russell G.J. and Metcalf B.L. (1986), “Nutrition Information in the
Supermarket,” Journal of Consumer Research, 13, 48-70.

Simonson, Itamar (1989), “Choice Based on Reasons: The Case of Attraction and Compromise Effects,”
Journal of Consumer Research, 16 (September), 158-174.

Simonson, Itamar and Amos Tversky (1992), “Choice in Context: Tradeoff Contrast and Extremeness
Aversion,” Journal of Marketing Research, 29(August), 281-295.

Thaler, Richard (1985), “Mental Accounting and Consumer Choice,” Marketing Science 4 (Summer),
199-214.

Tversky, Amos and Itamar Simonson (1993), “Context-dependent Preferences,” Management Science,
39(10) 1179- 1189.