INVESTIGATING THE EFFECT OF COLOR AND ICON ON INFORMATION PROCESSING
BEHAVIORS RELATED TO FRONT-OF-PACKAGE NUTRITION LABELS

By
Raghav Prashant Sundar

A DISSERTATION
Submitted
to Michigan State University
in partial fulfillment of the requirements
for the degree of
Packaging - Doctor of Philosophy
2013

ABSTRACT
INVESTIGATING THE EFFECT OF COLOR AND ICON ON INFORMATION PROCESSING
BEHAVIORS RELATED TO FRONT-OF-PACKAGE NUTRITION LABELS
By
Raghav Prashant Sundar

Obesity is a clear and growing public health crisis in the United States. Providing
consumers with nutrition information has long been at the forefront of obesity prevention
strategies. The
found on US packages
fronts of packages,
have grown increasingly popular as a possible means to increase label effectiveness. It has been
suggested that these FOP labels are more noticeable than the traditional NFPs due to their
position on the front of the package. They are also believed to ease cross-product comparisons
regarding nutritional content. The FDA has featured FOP labels prominently in their strategic
plans and has proposed mandating them in the future. Due to this, there has been an increase
in FOP related research in the US.
This study adds to the existing body of knowledge on FOP labels and establishes and
refines important methodologies for the field of label design research. Most existing FOP
studies rely on qualitative methods like focus groups, surveys and interviews to evaluate
understanding, perceived helpfulness and ease of use of FOPs. This study uses quantitative
methods and represents an important first step in understanding FOP labels in the context of

attention capture and encodation, the first two steps in an established model of information
processing. Additionally, it begins to examine the relationship of the FOP works and the NFP. Six
FOP formats containing color coding, facial icons and checkmarks were studied using four
experiments.
The salient results of Experiment 1 (Change Detection Flicker Task, n=55) indicated that
participants were, in general, more likely to detect changes to the FOP location when compared
to the NFP location (p<0.0001), suggesting that information placed in FOP locations garnered
more attention than traditional strategies. Changes to colored FOPs were detected significantly
faster than their non-color coded counterparts (p<0.0001), suggesting color to be a significant
factor. Experiment 2 (Eye tracking, n=55) suggested th

viewed. There was also evidence to conclude that when the FOP was present, participants were
more likely to fixate on (p=0.0013) and spend longer time viewing (p=0.0032) ANY nutrition
information compared with when the FOP was absent.
Despite the evidence to conclude enhanced noticeability of FOPs when compared with
NFPs in Experiments 1 and 2, there was no indication that FOP presence resulted in higher
encodation of nutrition information in Experiment 3 (Recall Task, n=99). In Experiment 4
(Sorting Tasks, n=93), the presence of FOP labels was beneficial to sort tasks, when they were
used, but a large number of participants did not use them (~75%). Those that did use them had
a higher likelihood of sorting packages correctly (p=0.02).
Across the four Experiments, age, education level, income level, BMI and Children Status
(Yes/No) emerged as demographics of interest.

To Dr.Bix
Thank you for everything

iv

ACKNOWLEDGEMENTS

The four experiments in this study were designed, conducted and analyzed jointly by
Faculty, graduate and undergraduate students from the School of Packaging, the Department of
Psychology and the Department of Food Science and Human Nutrition at the Michigan State
University in collaboration with the Department of Statistics at the Kansas State University. This
study was funded by a National Institute of Health (NIH) R21 Exploratory Grant.
It has been my good fortune that I became a part of such a large and prestigious study.
No PhD student before me ever had the luxury of having an entire team come together for
his/her research project and work tirelessly for the same; this dissertation has my name on the
title page, but it is really the result of the combined efforts of so many professors, students and
staff members. I would like to specially highlight the contributions of all these individuals
below.
I would like to begin by thanking my research advisor and life-long mentor, Associate
Professor Dr. Laura Bix of the School of Packaging, MSU. It has been my privilege and honor to
be a part of her research team since 2007, first as a masters student and then as a PhD student.
There is a reason students who work with her go on to achieve so much in their work and in
their lives. I have grown more, learned more and accomplished more in the 6 years that I have
worked for her than in all the time prior to that. In short, every good thing has ever happened
to me here in the US and will ever happen in the years to come, I owe in part (or maybe
entirely), to her.

v

Dr. Mark Becker, who was the Co-PI along with Dr. Bix for the NIH grant that funded this
study has been an extremely positive influence both on the grant and on my life. His vast
experience and expertise in cognitive psychology were instrumental in getting the grant off the
ground and keeping it there. I believe that his involvement with the grant and the project made
it truly inter-disciplinary and brought it into the realm of translational research. From teaching
me how to design change detection experiments on EPrime® to suggesting publications for my
literature search, his help and guidance have been of incalculable value.
The entire experimental randomization design and statistical analysis of this experiment
were performed by Dr. Nora Bello, Assistant professor at the Department of Statistics at the
Kansas State University. I have been extremely lucky to have had Dr. Bello contribute to the
statistical analyses part of every major project that I have been a part of, right from my
Master’s thesis.
I also wish to extend my gratitude to Dr. Diana Twede and Dr. Rafael Auras who, as a
part of my dissertation committee have been of great support since the day this project began,
offering valuable perspectives on the methods, results and conclusions of the various
experiments performed for every study. An integral part of the grant, Dr. Lorraine
Weatherspoon from the Department of Food Science and Human Nutrition provided important
expertise from a human nutrition standpoint and valuable assistance in recruiting subjects.
The use of actual physical prototypes is one of the main reasons this project stands out
from among the many labeling studies that are published every day. The tutelage and help of
School of Packaging Lecturer, Mr. Dennis Young and the help of my friend and colleague James
Richardson were invaluable during the prototyping process. Together we were able to make
vi

150+ packages, packages that consumers believed were ‘real’. All the paperboard that was used
to make these packages was donated by MeadWestVaco and was obtained thanks to the
efforts of Dr.Bix’s former student, Carly Dehenau. I would also like to thank Professor Robert
Clarke and PhD Candidate Jeff Tazelaar who helped design the innovative tray-slot setup and
RfiD interface for Experiment 4 in this study.
Every experiment in this study needed at least 6 researchers working together to plan,
and administer. I would like to highlight the efforts of graduate students Chad Peltier and Reem
Alzahabi from the Department of Psychology and graduate students Tony Trier, Cory Wilson,
Jiyon Lee and Rita Chen from the School of Packaging who worked tirelessly to test the 200+
human subjects needed for this study. Each of them made this project their own, traveled to
testing locations, scheduled participants, administered tasks and debriefed them. Also involved
in this testing process were a large team of able, undergraduate research assistants from the
School of Psychology. The nutritional component of the four experiments i.e. the
administration of the food questionnaires and taking of weight and height measurements was
entirely handled by a team of research assistants from the Department of Food Science and
Human Nutrition led by graduate students Sumathi Venkatesh, Muna Alchar and Simone
Wilson. We are also deeply indebted to Ms. Lori Strom of the Family Resource Center and Cici
Foster of the MSU Extension office and Tom Little of the Work First Program of the Lansing
Community College for their assistance in helping us to recruit participants for our study. I also
want to thank the MSU Extension office at the Ingham County Health Department, the
Southside community coalition and the offices of the Work First Program at the Lansing
Community College for letting us use their facilities to run participants for several months.
vii

MOST IMPORTANTLY I would like to thank the 200+ participants who gave us their
valuable time and input and helped ensure that this project was completed on time.
I would like to thank my wife Madhumitha, my family back in India and my friends both
here in East Lansing and in India for their unconditional love and unwavering support.

viii

TABLE OF CONTENTS

LIST OF TABLES................................................................................ xiii
LIST OF FIGURES ............................................................................... xv
Chapter 1 Introduction .......................................................................1
Obesity ............................................................................................................ 1
Nutrition Information & Front-of-Pack Nutrition Labels ................................... 2

Chapter 2 Obesity: An overview .........................................................6
Causes ............................................................................................................. 6
Eating Behaviors: ............................................................................................................ 7
Exercise Behaviors: ......................................................................................................... 8
Genetic & Medical: ......................................................................................................... 9

Prevalence: ...................................................................................................... 9
Adults:.......................................................................................................................... 10

Impact ........................................................................................................... 12
Physical ........................................................................................................................ 13
Psychosocial ................................................................................................................. 15
Economic: ..................................................................................................................... 16

Treatment...................................................................................................... 16
Drugs and Surgery: ....................................................................................................... 16
Special diets and Behavioral Modification: .................................................................... 17

Prevention ..................................................................................................... 17
Government: ................................................................................................................ 18
Industry: ....................................................................................................................... 18
Advertising: ........................................................................................................................ 19
Product Reformulation and Repackaging: ......................................................................... 23
Other Initiatives: ................................................................................................................ 25

Chapter 3 Nutrition Labeling: An Overview ....................................... 28
Nutrition Information & the Nutrition Labeling Education Act (NLEA): ........... 29
The Nutrition Facts Panel (NFP):.................................................................................... 32

The Impact of the NLEA.................................................................................. 34
Product Sales and Formulation: .................................................................................... 35
Information Search and Label Use: ................................................................................ 37
ix

Dietary Choices and Behavior: ...................................................................................... 39

Consumer Characteristics and Nutrition Label Use ......................................... 42
Age: .............................................................................................................................. 42
Education: .................................................................................................................... 45
Gender: ........................................................................................................................ 46
Time Pressure: .............................................................................................................. 48
Income: ........................................................................................................................ 48
Working Status: ............................................................................................................ 50
Household Label Usage: ................................................................................................ 51
Location: ...................................................................................................................... 51
Special diets & Medical Conditions: .............................................................................. 52
Prior Knowledge: .......................................................................................................... 53

Discussion ...................................................................................................... 54

Chapter 4 Front-of-Pack Nutrition Labeling ....................................... 56
Introduction & Classification .......................................................................... 56
Based on Information Contained:.................................................................................. 57
Based on Format: ......................................................................................................... 59

Usage and Regulatory History ........................................................................ 59
United Kingdom:........................................................................................................... 59
The European Union: .................................................................................................... 64
United States: ............................................................................................................... 66

Peer Reviewed Literature............................................................................... 73
The United Kingdom (UK): ............................................................................................ 74
The European Union (EU): ............................................................................................. 85
Australia: ...................................................................................................................... 91
United States: ............................................................................................................... 94

Chapter 5 Background ................................................................... 100
Background.................................................................................................. 102
Information Processing Models:.................................................................................. 102
Change and Change Blindness: .................................................................................... 103
Eye Tracking: .............................................................................................................. 106
Design Features in FOP labels: .................................................................................... 107

Chapter 6 Experiments .................................................................. 109
Materials ..................................................................................................... 109
Front of Pack labels: ................................................................................................... 109
Product Types and Health levels: ................................................................................ 112
Product Types: .................................................................................................................. 112
Health Levels: ................................................................................................................... 113

Experiments ................................................................................................. 115
x

Experiment 1 – Using Change detection to evaluate Attentional Priority: .................... 117
Overview: ......................................................................................................................... 117
Stimuli: ............................................................................................................................. 117
Recruiting: ........................................................................................................................ 121
Procedure: ........................................................................................................................ 121
Experiment 2 – Using eye tracking to evaluate attentional priority and label use: ........ 122
Overview: ......................................................................................................................... 122
Recruiting: ........................................................................................................................ 126
Procedure: ........................................................................................................................ 126
Experiment 3 – Using recall and recognition tasks to evaluate incidental encoding: ..... 128
Overview: ......................................................................................................................... 128
Stimuli: ............................................................................................................................. 129
Recruiting: ........................................................................................................................ 130
Procedure: ........................................................................................................................ 131
Experiment 4 – Using sort tasks to evaluate FOP ease of use: ...................................... 131
Overview: ......................................................................................................................... 131
Stimuli: ............................................................................................................................. 132
Procedure: ........................................................................................................................ 133
Subject Characterization & Demographics: .................................................................. 136

Chapter 7 Results .......................................................................... 138
Experiment 1 ............................................................................................... 138
Subjects: ..................................................................................................................... 138
Data Processing: ......................................................................................................... 140
Results: ...................................................................................................................... 140
Binary Variable – Change Detected (Yes/No): ................................................................. 140
Continuous Variable – Time to detect change (milliseconds): ......................................... 142
Discussion: ................................................................................................................. 147

Experiment 2 ............................................................................................... 149
Subjects: ..................................................................................................................... 149
Data Processing: ......................................................................................................... 152
Results: ...................................................................................................................... 153
Response to ANY nutritional information: ....................................................................... 155
Response to NFP information: ......................................................................................... 165
Discussion: ................................................................................................................. 173

Experiments 3 & 4 ........................................................................................ 177
Subjects – Experiments 3: ........................................................................................... 177
Data Processing – Experiment 3: ................................................................................. 179
Results – Experiment 3: .............................................................................................. 179
Binary Variable – Correct Package Chosen (Yes/No): ...................................................... 179
Discussion – Experiment 3: ......................................................................................... 181
Subjects - Experiment 4: ............................................................................................. 181
Data Processing – Experiment 4: ................................................................................. 184
xi

Results – Experiment 4 ............................................................................................... 184
Binary Variable - Sorted Correctly (Yes/No): .................................................................... 184
Continuous Variable - Time to sort (seconds): ................................................................. 187
Binary Variable – Self Report of FOP label use (Yes/No) ................................................. 197
Discussion – Experiment 4: ......................................................................................... 199

Chapter 8 Conclusions, Limitations & Future work .......................... 202
Conclusions .................................................................................................. 202
Limitations ................................................................................................... 203
Future Work ................................................................................................ 204
Exploring Color-Coding further .................................................................................... 204
Exploring Product Type Effects .................................................................................... 205
Location, FOP Size, Package Form and other design effects ......................................... 205
Working with Children and speakers of other languages ............................................. 206

APPENDICES................................................................................... 207
Appendix 1 - Consent Forms: ....................................................................... 208
Appendix 2 - Recruitment advertisements: .................................................. 216
Appendix 3-Products Designed for Study: .................................................... 220

REFERENCES ................................................................................... 231

xii

LIST OF TABLES

Table 1 - Optional and Required nutrients on Nutrient Fact Panels (NFP) .................................. 34
Table 2 - Research investigating the effect of Age on Nutrition Label use .................................. 44
Table 3 - Research investigating the effect of Education on Nutrition Label use ........................ 46
Table 4 - Research investigating the effect of Gender on Nutrition Label use ............................ 47
Table 5 - Research investigating the effect of Subject Income on Nutrition Label use................ 49
Table 6 - Research investigating the effect of Work Status on Nutrition Label use ..................... 50
Table 7 - Research investigating the effect of Household type on Nutrition Label use ............... 51
Table 8 - Research investigating the effect of Geographical location on Nutrition Label use ..... 52
Table 9 - Research investigating the effect of Special Diets and Medical Conditions on Nutrition
Label use ....................................................................................................................................... 53
Table 10 - Research investigating the effect of Prior Knowledge on Nutrition Label use ............ 54
Table 11 – Classification of FOP label schemes by information contained. Classification scheme
sourced from Wartella et al.[30]. ................................................................................................. 58
Table 12 - Food Standards Agency (FSA, UK) consumer research evaluating Front of Pack Label
Formats (see Figure 14 a,b,c,d)..................................................................................................... 79
Table 13 – Description of FSA's comprehensive FOP label evaluation performed in 2009 ......... 81
Table 14 - Advantages and Disadvantages of two major FOP label formats. Reproduced from
Louie et al.[233] ............................................................................................................................ 92
Table 15 - Categories for nutrients per 100 grams of product [321]. See Figure 13 for graphical
representation. ........................................................................................................................... 111
Table 16 - Summary of experimental procedures ...................................................................... 116
Table 17 – Labeling scheme generated for Experiment 4. For the four FOP nutrients, each
brand’s FOP label has a ‘high’ amount of one nutrient, ‘low’ amount of one nutrient and
‘medium’ amounts of two nutrients, i.e. resulting in one red panel, one green panel and two
xiii

yellow panels in the color coded FOP label types. This was also true for each nutrient across the
four brands.................................................................................................................................. 133
Table 18 - Breakdown of participants who fixated upon ANY nutritional information, NFPs and
FOPs on Packages........................................................................................................................ 154

xiv

LIST OF FIGURES

Figure 1 - Front and Side views of Kellogg Cocoa Krispies package showing a Front of Pack (FOP)
label and a Nutrition Facts Panel (NFP. For interpretation of the references to color in this and
all other figures, the reader is referred to the electronic version of this dissertation. Text inside
the figure is not meant to be readable, but is for visual reference only........................................ 4
Figure 2 - Product Attributes ........................................................................................................ 30
Figure 3 - A typical Nutrition Facts Panel (NFP). Image retreived from
http://www.fda.gov/ucm/groups/fdagov-public/documents/image/ucm153391.png .............. 33
Figure 4 - Examples of Multiple Traffic Light (MTL) FOP labels in different configurations.
Retrieved from http://www.eatwell.gov.uk/foodlabels/trafficlights/ ......................................... 60
Figure 5 – A typical %GDA label with absolute and percentage amounts of nutrients but no
traffic light colors. ......................................................................................................................... 61
Figure 6 - An artist's rendering of the Hybrid FOP system being proposed by the Food Standards
Agency in the UK. .......................................................................................................................... 64
Figure 7 – Proposed Model IOM 3-point FOP label format. The product on the far-left does not
qualify for points while the product on the far right qualified for points and meets predetermined criteria for all three nutrients - saturated fat+trans fat, sodium and added sugar.
Reproduced from the IOM Phase 2 report on Front of Pack Nutrition Rating Systems and
Symbols [29].................................................................................................................................. 70
Figure 8 - The GMA's Facts-Up-Front (FUF) FOP label scheme [269]. .......................................... 71
Figure 9 - FOP label concepts tested in preliminary FSA focus groups circa 2004. a. %GDA Label
Format b. Key Nutrients Format (an early version of the Multiple Traffic Lights Format) c.
Simple Traffic Light d. Healthy Icon Format e. Extended Traffic Light Format. Formats b and c
were noted by the FSA to have significantly more promise than competing designs ................. 76
Figure 10 - Four FOP label designs quantitatively tested by the FSA following concept testing of
6 potential designs (See Figure 24) a) Multiple Traffic Light (MTL) b) Simple Traffic Lights c)
Color-Coded %GDA d) Monochrome %GDA ................................................................................. 77
Figure 11 - The Flicker task paradigm, a loop of 4 images - an original and a changed image with
an interleaved grey screen. On looping, the change itself will appear to flicker, giving this

xv

paradigm its name. Text within image is not meant to be readable and is for visual reference
only.............................................................................................................................................. 106
Figure 12 - Traffic Light colors & facial icons were manipulated to create 4 FOP label designs 110
Figure 13 - Graphical representation of data in Table 15; nutrient categories per 100 grams of
product........................................................................................................................................ 112
Figure 14 - BranBlast, a brand of breakfast cereal created for this study represented with a
Color + Facial Icon FOP in both health levels - healthy and unhealthy. Text within package is not
meant to be readable and is for visual reference only ............................................................... 114
Figure 15 - The IOM's 3-point labels added to Experiment 1 ..................................................... 118
Figure 16 – a. The Three Brands used in this Change Detection Experiment b. Example Change
detection image divided into 10 sectors. Per brand, The FOP and NFP sectors featured 12 critical
changes each. The filler sectors featured 12 changes each to non-critical aspects of packaging
(brand names, graphics etc). Spread over the three brands this resulted in a total of 72 * 3 =
216 trials per subject. Text within package images is not meant to be readable and is for visual
reference only ............................................................................................................................. 120
Figure 17 – An example 8 package set that a participant viewed in Experiment 2 (Eye Tracking)
..................................................................................................................................................... 125
Figure 18 – The custom built plywood and acrylic pane used for running Experiment 2 .......... 127
Figure 19 - RfID sorting setup designed for Experiment 4. Text within image is not mean to be
readable and is for visual reference only.................................................................................... 134
Figure 20 - Demographic Information for 55 usable participants of Experiment 1 ................... 139
Figure 21 - Effect of Location of Change on Probability of Change detection. Participants were
more likely to detect changes to the FOP than changes to the NFP (p<0.0001) ....................... 142
Figure 22 - The effect of Location of Change (p<0.0001) on Time to detect change. It is
estimated that participants took 61% longer time to detect changes to the NFP Location when
compared to the FOP Location ................................................................................................... 144
Figure 23 - The effect of Color (p<0.0001) on the estimated time to detect change. There was
evidence at the 5% statistical significance level to conclude that the non-color coded FOPs took
longer to detect than their color coded counterparts. P-values for pair-wise comparisons
between color coded and non-color coded FOP Type pairs are shown above each pair of bars.
Alphabets indicate statistically significant pair-wise comaparisons between all bars at the 5%
significance level ......................................................................................................................... 146

xvi

Figure 24 - Experiment 2 Subject Demographics for 55 usable subjects. Note that the Education
demographic is interpreted differently for this experiment. ..................................................... 151
Figure 25 - Schematic representation of the two eye tracking Lookzones on a flattened package.
Note that the red borders around the zones are for visual reference and were not actually
present on the packages that were used in the Experiment. Text within the package is not
meant to be readable and is for visual reference only ............................................................... 152
Figure 26 - Effect of Label type, FOP + NFP vs. NFP only and Probability of fixation on nutrition
information (p<0.0001)............................................................................................................... 156
Figure 27 - The effect of Label type (p=0.0032), Product Type (p=0.0070) and Health Level
(p=0.0096) on the total time spent on nutrition information .................................................... 158
Figure 28 - The effect of Subject BMI category on the time spent on nutrition information
(p=0.03) Obese subjects (with a BMI of 30+) spent an estimated 56% more time on nutrition
information than Overweight subjects (with BMI of 25 to 29.9). Alphabets represent statistically
significant pair-wise comparisons at α=0.05 .............................................................................. 160
Figure 29 - The Effect of Children Status (p=0.0008) and Income Range (p=0.0042) on estimated
total time spent on nutrition information .................................................................................. 162
Figure 30 - The effect of subject visual acuity on time spent on nutrition information (p=0.003).
Bars and alphabet are indicative of statistically significant differences at α =0.05 ................... 163
Figure 31 - Effect of Label type, FOP + NFP vs. NFP only on Time to first fixation of nutrition
information (p=0.0013)............................................................................................................... 165
Figure 32 - Estimated time spent in the NFP by product type by label type. There is a statistically
significant effect of Label Type for the Cereal Product Type (p=0.0085) but not for the Crackers
Product Type (p=0.95). ............................................................................................................... 167
Figure 33 - The effect of children status (p=0.0005) and income per year (p=0.0094) on the
estimated time spent on the NFP zone (seconds) ...................................................................... 169
Figure 34 - The effect of Visual Acuity on the time spent on the NFP zone (p=0.0027). Alphabets
indicate statistically significant differences at α=0.05................................................................ 170
Figure 35 - The effect of label type on number of visual hits to the NFP by product type. As
shown, there is significantly (p=0.0002) higher number of the NFP zone visual hits for cereal
packages that have no FOP when compared to cereal packages that have the Color + Facial Icon
FOP label. There is no such statistically significant difference for crackers (p=0.4) .................. 172
Figure 36 - Demographic information for Exp 3 Participants ..................................................... 178

xvii

Figure 37 - The effect of run order on the probability of selecting the correct/healthier package
in the incidental encoding task (p=0.016). Lines and alphabets of indicative of statistically
significant differences at α=0.05 ................................................................................................ 181
Figure 38 - Demographic information for Experiment 4 Subjects .............................................. 183
Figure 39 - Estimated Probability of correctly sorting as determined by self reported use of the
FOP (p=0.02) ............................................................................................................................... 186
Figure 40 - The effect of nutrient on the probability of correctly sorting (p<0.0001). There was
an estimated lower likelihood of sorting packages by sat-fat when compared to the other
nutrients. Alphabets are indicative of statistically significant differences at α=0.05 ................ 187
Figure 41 - The effect of subject visual acuity on time to correct sort (p=0.0055) .................... 189
Figure 42- The effect of the Education Range demographic factor on time to correct sort
(p=0.0091) ................................................................................................................................... 190
Figure 43 – The effect of FOP Type on the estimated time to sort (seconds). Effects are
interpreted for each level of the Used FOP (Yes/No) factor for the a) Fat and b) Saturated Fat
nutrients because of the significant 3-way interaction between Nutrient, Self Reported FOP Use
and FOP Type (p=0.0329). Alphabets are indicative of statistically significant pair-wise
comparisons within each chart at α=0.05. Y-axes represent ‘Estimated time to sort (seconds)’
..................................................................................................................................................... 192
Figure 44 – The effect of FOP Type on the estimated time to sort (seconds). These effects are
interpreted for each level of the Used FOP (Yes/No) factor for a) Salt/Sodium and b) Sugars
nutrients. because of the significant 3-way interaction between Nutrient, Self Reported FOP
Use and FOP Type (p=0.0329). Alphabets are indicative of statistically significant pair-wise
comparisons within each chart at α=0.05. Y-axes represent ‘Estimated time to sort (seconds)’
..................................................................................................................................................... 193
Figure 45 - The effect of self reported FOP use on time to sort (seconds) for each nutrient type
for the two FOP Types that had facial icons. Significant differences in time to sort at α=0.05 for
each nutrient within each FOP Type between FOP Used (Yes) and FOP Used (No) cases are
shown in bold. ............................................................................................................................. 195
Figure 46 - The effect of self reported FOP use on time to sort (seconds) for each nutrient type
for the two FOP types that did not have facial icons. Significant differences in time to sort at
α=0.05 for each nutrient within each FOP Type between FOP Used (Yes) and FOP Used (No)
cases are shown in bold. ............................................................................................................. 196
Figure 47 - Impact of education range on FOP label use (p=0.0009). Participants who had more
than a high school education were more likely to self-report use of the FOP then participants
who had less than a high school education. ............................................................................... 198
xviii

Figure 48 - Effect of Age group on the probability of FOP label use. Alphabets are indicative of
statistically significant pair-wise comparisons at α=0.05 ........................................................... 199
Figure 49 - Bran Blast (Breakfast Cereal). Text within image is not meant to be readable and is
for design reference only ............................................................................................................ 220
Figure 50 - Golden Harvest (Breakfast Cereal). Text within image is not meant to be readable
and is for design reference only ................................................................................................. 221
Figure 51 - Spyros (Breakfast Cereal). Text within image is not meant to be readable and is for
design reference only ................................................................................................................. 222
Figure 52 - Sunrise (Breakfast Cereal). Text within image is not meant to be readable and is for
design reference only ................................................................................................................. 223
Figure 53 - Earls (Crackers) Text within image is not meant to be readable and is for design
reference only ............................................................................................................................. 224
Figure 54 - Fish Eye (Crackers) Text within image is not meant to be readable and is for design
reference only ............................................................................................................................. 225
Figure 55 - Marvel (Crackers). Text within image is not meant to be readable and is for design
reference only ............................................................................................................................. 226
Figure 56 - Snappers (Crackers). Text within image is not meant to be readable and is for design
reference only ............................................................................................................................. 227
Figure 57 - Best Cuisine (Prepared Meals) Text within image is not meant to be readable and is
for design reference only ............................................................................................................ 228
Figure 58 - Read Meals (Prepared Meals) Text within image is not meant to be readable and is
for design reference only ............................................................................................................ 229

xix

Chapter 1
Introduction
Obesity
In its simplest sense, obesity is the presence of excess fat (adipose tissue), more than
what is required for normal biological functioning of the human body. This increase in ‘body
mass’ may be attributable to varied reasons, both medical and psycho-social [1]. Obesity leads
to several complications, ranging from a general reduction in mobility to more serious ailments
like heart disease and stroke [1-3]. It is known to affect all individuals irrespective of age,
gender, race and geographical location, and is widely considered one of the most significant
public health problems of the developed world today [4-7].
Individuals are classified as being overweight or obese mainly based on their Body Mass
Index (BMI). The BMI of an individual is calculated by dividing their weight (in kilograms) by the
square of their height (in meters). BMI is thus expressed in kilogram per meter square (kg/m2).
The CDC classifies adults based on BMI into four categories – underweight (<18.5),
normal (18.5 to 24.9), overweight (25 to 29.9) and obese (>30) [8]. BMI’s of 45-50 not
uncommon among individuals with severe obesity [9].
In the US, a child is said to be obese when his or her BMI is above the 95th percentile and
is overweight when the BMI is above the 85th percentile of the general population of children
[10]. These percentiles were determined by the National Health and Nutritional Examination
Survey II (1963-1965) and the National Health and Nutritional Examination III (1966-1970) in the
USA.

1

Nutrition Information & Front-of-Pack Nutrition Labels
Providing consumers with nutrition information and empowering them to make healthy
dietary choices has long been has been at the forefront of obesity prevention strategies [11].
Almost all packaged food products sold in the US must contain nutrition information in the
form of a Nutrition Fact Panel (NFP) [12]. The NFP is a tabular representation of nutrition
information required in a standard format since the enactment of the Nutrition Labeling &
Education Act of 1990 [5, 13] (See Figure 1). These NFPs are intended to help consumers make
objective comparisons between products in order to make healthy dietary choices. Despite
NFPs being provided to consumers for the past 20 years, the very high rates of obesity in the US
suggest that they may not be as impactful as initially predicted [14-17] (Obesity prevalence in
the USA will be examined in detail in Chapter 2). This led the FDA’s August 2003 obesity
working group to conclude that “it is clear that consumers would benefit if they were to pay
more attention to and make better use of information… on food labels. Providing
encouragement and making it as easy as possible for consumers to do so are worthy public
health objectives. [18]”
More recently, the Federal Government has become more explicit in its
recommendations for labeling. In 2010, the Whitehouse’s Task Force on Childhood Obesity
recommended the development and implementation of a “standard system of nutrition
labeling for the front-of-packages” based on scientific research [19]. This recommendation was
largely catalyzed by a proliferation of simplified tabular and iconic representations of nutrition
information on the fronts-of-packages that has taken place over the last ten years. Commonly
referred to as Front-of-pack or Front-of-package labels and abbreviated ‘FOPs’, they frequently
2

feature graphics, symbols and color-coding that are intended to facilitate quick and easy
comparisons between products (See Figure 1 for an example). Due to their positioning (front of
the pack), it has been suggested by authors in the UK [20], EU [21] and here in the US [22] [23]
that they may be more noticeable than the traditional nutrition information which is present on
the side or back the packages. It has also been suggested by academic researchers in both the
US [23] [24] and abroad [21] that FOP labels may induce manufacturers to reformulate their
offerings.
Inconsistent formatting and information within FOP formats and symbologies has led to
consumer confusion and frustration [23] [24, 25] [26]. Recognizing the merits of FOPs and the
consumer frustration generated from non-standardized, unregulated systems, the US Food and
Drug Administration (FDA) has recently identified “the exploration of FOP nutrition labeling
opportunities” as a key initiative in its 2012-2016 Strategic Plan [27] [28].
To this end, in 2010, the Institute of Medicine was commissioned by the Centers for
Disease Control and Prevention (CDC) working under direction from Congress to “examine and
provide recommendations regarding FOP nutrition systems and symbols [29]” The IOM has
released two reports thus far: a Phase 1 report released in 2010 that focused on reviewing the
benefits of FOPs and evaluating existing systems [30] and a Phase 2 report released in mid 2012
that, based on the IOM committee’s deliberations, proposed a model FOP system that
contained several desired characteristics (See Figure 7) [29].

3

Figure 1 - Front and Side views of Kellogg Cocoa Krispies package showing a Front of Pack
(FOP) label and a Nutrition Facts Panel (NFP. For interpretation of the references to color in this
and all other figures, the reader is referred to the electronic version of this dissertation. Text
inside the figure is not meant to be readable, but is for visual reference only.

Due to the rise in their popularity and in anticipation of potential regulatory
developments, there has recently been a sharp rise in peer-reviewed publications concerning
FOP labels in the US [22, 24, 25]. Almost all studies released in the US thus far focus on
evaluating consumer understanding [24, 25] [31] [32], product perceptions [33] and purchase
choices [34] (Readers are directed to Chapter 4 of this report and to Hawley, Roberto et al. [23]
for reviews on the subject).
4

Both in the US, and around the world, most of the work that has been done on FOPs is
qualitative in nature, comprised of: focus groups (EU - [35] [36] , US - [37]), assessment of in
store behaviors (UK- [38] [39]), surveys (US - [32] [33] ) and depth interviews (UK - [38]).
Considering the potential cost and reach that a mandated FOP system will eventually
have, “it is critical that an informative, easily understood, science-based FOP labeling system be
implemented [23].”
This study adds to the existing body of knowledge on FOP labels and establishes and
refines important methodologies for the field of label design research. While most previous
studies cited above concentrate on understanding and use, this study represents an important
first step towards the understanding of FOP labels in the context of attention capture and
encodation, the first two steps in an established model of information processing [40]. Six FOP
label formats are studied across four experiments.
The study was conceived, designed and conducted jointly by faculty, graduate and
undergraduate students from the School of Packaging, Department of Psychology and the
Department of Food Science and Human Nutrition at the Michigan State University, East
Lansing, MI in collaboration with the department of Statistics at the Kansas State University,
Manhattan, KS. All testing was approved by the Institutional Review board at the Michigan
State University under IRB #10-459.

5

Chapter 2
Obesity: An overview

Causes
While an in depth study of energy metabolism is beyond the scope of this report, it is
essential to understand that obesity occurs as a result of energy ‘imbalance’. There are two
phases of energy utilization in the human body – an absorptive phase and a post-absorptive
phase. When food is consumed, some of the energy is used immediately and the rest is stored.
This is the absorptive phase.
A few hours after a meal, the body enters a post-absorptive phase, where it needs to
draw energy from stored sources within the body.
Human Energy Balance is commonly expressed as indicated in Eq 1.
Energy Intake = RMR + TEF + TEE

(Eq1)

Where, the RMR (Resting Metabolic Rate) is the amount of energy that the human body
needs to function normally while at rest, at room temperature. This is the energy used for vital
body processes involving the heart, lungs, brain, kidneys and nervous system. RMR is measured
only in the post absorptive state which means that digestion and other processes involved in
the break down and absorption of food are not taking place. This RMR is fairly constant for a
given individual but changes with age, gender and special needs. The TEF (Thermic Effect of
Feeding) is the energy expended by the body in digestion and absorption of food. It is generally
measured as the energy needed for the breakdown and ultimate transformation of complex

6

raw food to simple usable forms [41]. The TEE (Thermic Effect of Exercise) is the energy used
to perform external work using the muscular and skeletal system [42].
Energy Intake from food must equal energy expenditure and storage (a consequence of
the first law of Thermodynamics) [6]. When there is an imbalance in Eq 1 i.e. an excess of
energy intake, the excess is compensated by storage in the body, eventually resulting in excess
mass [43].
The aforementioned energy imbalance may occur as a result of a confluence of reasons,
both external and internal. These are discussed below under three sub divisions – eating
behaviors, exercise behaviors and medical reasons; the first two being external and the third
being internal.
Eating Behaviors:
As explained in the discussion about energy balance, any food constitutes an intake in
energy, and if not offset by energy expenditure, the differential results in storage, leading to
excess weight and obesity [6]. There are vast number of factors that determine what individuals
eat, when and how much.
Among children in an American society, where it is very likely that both parents are
working and children are under the supervision of a secondary caregiver, excessive pampering
of the child resulting in consumption of calorie laden treats is common [6], additionally many
parents do not have the necessary time to create a nutritious meal plan for their children.
Childhood diet can also suffer in divorced families; the efforts by one parent to create a good
diet may be obviated by the other parent. The dietary habits of parents and immediate family
have a profound influence on eating behaviors. The children in households where at least one
7

parent consumes fatty foods regularly are twice as likely as their peers to consume high fat
foods. Those children who are from households where both parents consume high fat foods are
three to six times more likely to consume high fat foods [44].
Society also significantly influences consumption, energy expenditure, and therefore,
obesity prevalence. Families in Europe and France eat their heaviest meal during the afternoon
and have a light dinner. The opposite happens in USA, where the heaviest meal is eaten before
sleep [6]. In some countries, heavy eating and large size are regarded as symbols of wealth and
prosperity. This is common in areas of Asia and the pacific islands like Nauru, which has one of
the highest rates of obesity in the world with 90% of adults having a BMI above the 90th
percentile [3].
It has been suggested that in the US, we live in an ‘obesogenic environment (conducive
to obesity prevalence)’; Ard suggests this environment manifests in “the availability of high
energy dense, palatable, inexpensive food [which] is only surpassed by the mechanized, labor
saving and entertainment devices designed to keep us from moving too much[45].”
Exercise Behaviors:
After the industrial revolution, the level of physical work performed by human beings
reduced drastically resulting in “fewer natural opportunities for vigorous physical activity [6]”.
Research suggests that, as with eating preferences, exercise behaviors are also largely
determined in early childhood, with children of inactive parents six times less likely than those
of active parents to indulge in physical activity and outdoor play [46]. Other reasons for
decreased physical activity include living in unsafe neighborhoods where outdoor play is not
feasible and greater emphasis on academic achievement in schools than physical activity. [6].
8

Researchers have often tried to validate/reject the popular ‘couch potato’ hypothesis
where it is proposed that increases in weight are often due to some form of sedentary
entertainment like television and video games that result in lower time spent in physical
activity. However, there is as of yet no empirical evidence to support/debunk this. Studies that
have tried to link number of hours spent viewing television or playing video games with BMI
have yielded contradictory results [47-50].
Genetic & Medical:
While the link between genetics and obesity is not fully understood, researchers have
used methods like comparing obesity prevalence between identical twins (two individuals with
identical genetic material) and twins who have grown up separately with mixed conclusions
[51-56].
One contention is that certain individuals have a greater tendency to become obese
because of an inherited lower Resting Metabolic Rate (RMR), thereby resulting in higher
amounts of fat storage. The issue is not completely understood; available studies have
produced varied conclusions [57] [58] [59] [60, 61].

Prevalence:
Obesity was once thought to be a concern only of developed countries with high food
availabilities. It wasn’t until 1997 that the World Health Organization (WHO) recognized obesity
as a global concern and called for immediate action [62]. The WHO estimates that, as of 2008,
1.4 billion adults are obese worldwide and this number is almost quadruple the 2005 estimate
of 400 million adults [63]. It is well known that the problem of obesity is not limited to adults;
the World Health Organization (WHO) estimates that in 2010, nearly 42 million children under
9

the age of five were overweight worldwide. Approximately 35 million of those children
currently live in developing countries [64].
US prevalence and trends for adults and children are discussed below.
Adults:
A major source of obesity and health related data in the United States is the NHANES
(National Health and Nutrition Educational Survey) database; a database updated and
maintained by the National Center for Health Statistics (NCHS). A 2010 paper in the Journal of
the American Medical Association (JAMA) analyzed data sets from NHANES, estimating that
nearly 35% of American men and women were obese in 2008 [65]. This figure rises to 65%
when overweight adults (greater than 20 years of age with a BMI over 25) were also
considered. The authors mention that the likelihood of being obese was highest if an individual
was between 40 and 59 years of age. The obesity rate is highest for non-Hispanic Black women
(51%) between the ages of 40 and 59 [65].
On the subject of trends and increases the authors mention that,
“The prevalence of obesity for adults aged 20 to 74 years increased by 7.9 percentage
points for men and by 8.9 percentage points for women between 1976-1980 and 19881994, and subsequently by 7.1 percentage points for men and by 8.1 percentage points
for women between 1988-1994 and 1999-2000 [65].”

However, the study concludes that although obesity levels are very high and there has
been a dramatic increase in obesity rates between 1976 and 2000, there have been no
statistically significant increases in obesity prevalence between 2003-04 and 2004-05 for either
men or women [65].

10

Updated numbers released in January 2012 by the NCHS mention that this ‘leveling off’
effect is continuing and there is no evidence of increasing obesity prevalence (α=0.05) among
adults when 2007-2008 and 2009-2010 are compared [66]. The 2012 report from the NCHS also
mentions that there is no evidence of significant difference between prevalence rates in men
and women in the US, but it is worth mentioning here that globally, more women are obese
than men [63]. The NCHS indicates that obesity likelihood was greatest for those who were 60
years and over as opposed to the 2010 numbers that concluded that those between 40 and 59
years of age were most likely to be obese [66].
In addition to age and gender, ethnicity is also worth mentioning in the context of
obesity prevalence [1, 6, 65], with the non Hispanic Black and Hispanic sections of the
population in the US reporting a higher proportion of obese individuals than Caucasians [65].
Behavioral Risk Factor Surveillance System (BRFSS) data analyzed by the Centers for Disease
Control and Prevention (CDC) suggest that “Blacks had 51 percent higher prevalence of obesity,
and Hispanics had 21 percent higher obesity prevalence compared with whites [67].”
Location and ethnicity have been found to interact. BRFSS data suggests that the
“greater prevalence of obesity for blacks and whites were found in the South and Midwest than
in the West and Northeast. Hispanics in the Northeast had lower obesity prevalence than
Hispanics in the Midwest, South or West [68].”
Children and Adolescents:
Data from the most recent NHANES surveys analyzed and compiled by Ogden et al.
reveal that 16.9% of US children are obese as of 2010. The authors indicate that there was no
evidence of a significant increase when compared to the 2007-2008 percentages, suggesting a
11

‘leveling off’ effect similar to that discussed in the section on obesity prevalence in adults. The
prevalence is highest among non-Hispanic black children with an estimated 28.6% obesity rate
for children ages 6-11 years [16].
A letter published in early 2013 in the Journal of the American Medical Association
analyzed data from the Pediatric Nutrition Surveillance System (PedNSS) containing health and
nutrition data for almost 28 million pre-schoolers (2-4 years). The authors mentioned that
statistical analyses revealed a drop in extreme obesity (i.e. BMI more than 120% of the 95th
percentile) – 2.22% to 2.07% from 2003 to 2010 as opposed to an increase from 1.75% to 2.22
percent from 1998 to 2003 in the age group studied (2-4 years) [69].

Impact
In a statement declaring September 2012 National Childhood Obesity Awareness
month, US President Barack Obama stated,
“Over the past several decades, childhood obesity has become a serious public
health issue that puts millions of our sons and daughters at risk. The stakes are
high: if we do not solve this problem, many among America's next generation
will face diabetes, heart disease, cancer, and other health problems associated
with obesity… [70]”

Across the Atlantic, Alan Johnson, then Secretary of State for Health in the UK, likened
obesity to other grave problems, like climate change and pollution [71]. Other officials have put
the obesity epidemic on par with counterterrorism, with obesity related deaths far
outnumbering those caused by terrorists and wars [1].

12

In a series on childhood obesity published in the Huffington Post, Dr. Robert Murray,
chair of the ‘American Academy of Pediatrics Council on School Health’ and advisor to the
national ‘Action for Healthy Kids’ initiative, mentions that, in the context of healthcare, “no
issue is more important than the future implications of obesity among children [4]”
The impacts of obesity are discussed below under three subheadings – physical,
psychosocial and economic.
Physical:
Obesity carries a host of complications and directly causes gallstones, osteoarthritis and
various gynecological problems [1]. It is also an independent risk factor for several diseases and
debilitating medical conditions like heart disease, stroke and cancer [1, 5, 6, 13].
It is estimated that the majority of the 500,000 Americans who die every year as a
result of heart problems, die from Coronary Heart Disease (CHD). This number comprises
approximately one fourth of all deaths that occur annually in the USA. It is also well established
that adults who are obese or overweight and engage in little or no physical activity are the most
at risk for this disease [72]. Apart from the 500,000 deaths that are caused by CHD, nearly 1.4
million non-fatal heart attacks occur as well. Apart from obesity’s effects on the arteries
supplying blood to the heart (a.k.a. CHD), it can also affect the heart muscles directly, in a
condition known as Cardiomyopathy. In this condition, the heart is forced to increase its power
to pump blood to all the excess fat tissue in the body. Excessive stress on the heart as a result
of this may cause heart failure. It is believed that increased levels of lipids in the blood that
result from obesity represent an additional risk factor for CHD as well [6]. Although CHD is not

13

commonly found in children even if they are extremely obese, it is a proven medical fact that
CHD is a long degenerative process that starts in early childhood [6].
Despite the lack of a complete understanding regarding the link between obesity and
hypertension, people who are obese are three times more likely to develop high blood pressure
than non-obese people, irrespective of their age [73]. Although many obese children have high
blood pressure, it does not frequently result in any debilitating conditions in them. However, if
neglected, high blood pressure may persist into adulthood and cause serious problems.
Diabetes, the inability of the blood to regulate glucose levels is another leading cause of
death in the US. It is a well-established paradigm that the risk of contracting this condition is
increased when the person’s body weight is high. People who are overweight are five times
more likely to contract diabetes than people who are not [74] and a study published in 1996 by
PinhasHameil et al. [75] suggested that the onset of adult diabetes may be accelerated in
adolescents who are obese.
Several studies have established that an increase in weight increases the risk of several
types of cancer (uterine [76, 77], colon [78-80], kidney [81, 82], esophageal [83, 84], prostate
[85, 86]). There is also medical evidence of a complex, but direct, link between breast cancer
and obesity [87].
Several other health complications accompany obesity. The most important of these
involve the inability of the lungs to function properly because of the great amount of fat that is
stored in the abdomen and stomach. This is called Obesity Hypoventilation Syndrome. Other
lung-related problems include sleep apnea, where the airways that provide the lungs with air

14

are blocked during sleep because of the relaxation of the muscles around the throat. Obese
individuals frequently have sleeping disorders as a result [6].
Obesity is also known to trigger osteoarthritis and other diseases of the skeletal system,
notably Blount’s disease or bowing of the legs. 75% of the children in the USA with this
condition are obese [6]. In females, obesity results in fertility problems and irregular menstrual
periods. It may also result in early and irregular development of sexual characteristics [6]. Skin
conditions like psoriasis are also known to be aggravated by obesity [1].
Psychosocial
Apart from the impact on the physical health of the individual, obesity also affects a
person’s mental and social well-being.
Maguire and Haslam discuss the controversy surrounding the “chicken-egg” question of
obesity and depression; whether depression leads to binge eating and obesity or vice versa is
debated. There is, presently, no conclusive medical evidence to prove that obese people are
more prone to mental illness, although self reports from overweight American women have
revealed that 37% of them suffer from low self esteem and self confidence [88].
Obese children may be teased and bullied frequently and may suffer from a poor self
image, not being able to generally indulge in play like their peers. Recent surveys have revealed
that children and young adults rate obese children as ‘less likable’ than those with missing limbs
and facial disfigurement [6].
Deep rooted mental conditions like anorexia (forced starvation) and bulimia (excessive
consumption of food followed by regurgitation) are prevalent among adults and children with
obesity [6] and must be identified and treated early.
15

Economic:
The now oft-cited 2001 release, ‘Surgeon General’s Call to Action,’ indicates obesity is a
major public health concern and suggests that excess weight results in the death of at least
300,000 people every year in the US [89]. As of 2008, it is estimated that obesity related
healthcare spending was more than $145 billion, with the medical cost for obese individuals
$1,429 higher, on average, than the cost of those who were not obese [90]. Koplan estimates
that “half of obesity-related health care costs are paid for with public funds [91].”
The Institute of Medicine (IOM) suggests that by 2015, the healthcare expenditures of
the US will reach $4 trillion (roughly around 20 % of the country’s entire GDP) as opposed to the
$1.6 trillion presently spent. The same IOM report indicates that 12% of the increase will come
from an increase in obesity-related spending [13].
In the US, the national costs related to childhood obesity are approximately $11 billion
for children that are privately insured and $3 billion for those who participate in Medicaid and
Medicare programs [13].

Treatment
Obesity, being a disease with no single cause, has treatment methods that take on
various forms (e.g. simple changes in the individual’s lifestyle, increased exercise, careful
management of diet and medical interventions, drugs and surgery etc). In most leading obesity
treatment centers, each patient is carefully studied to evaluate the unique individual factors
that led to the onset of his or her condition. Once this is done, an individualized set of
treatment methods is formulated [92-94].
Drugs and Surgery:
16

Most anti-obesity drugs work by affecting the complex chemical and nervous processes
that affect appetite and satiety, usually acting as inhibitors for the two chemical appetite
inducers in the human body – catecholamine and serotonin [95, 96]. Surgical obesity treatment
methods like dental or gastric wiring are used only when education, exercise and residential
programs to encourage weight loss have not worked and patients have developed many lifethreatening complications [6].
Special diets and Behavioral Modification:
Poor diet being one of the major causes of obesity, special dietary plans are almost
always recommended for obese individuals; these may feature a wide range of dietary
modifications from general reductions in fat and sugar content to specialized diets.
Counseling and behavioral modification usually center around three approaches:
stimulus control (removing everything from the individual’s immediate environment that makes
him/her think of eating), cognitive restructuring (attitude change brought about by
individualized counseling) and self monitoring (maintaining detailed records of physical activity
and food consumed) [6]. Behavioral modification usually brings with it recommendations for
increased physical activity and individualized exercise regimens.

Prevention
It is generally believed that from a national perspective, the prevention of diseases like
obesity, which have both medical and psychosocial causes, should involve large and small scale
initiatives by two major stakeholders – the government (federal government, state
government, local communities/municipalities, schools) and the industry (food and beverage

17

manufacturers, retailers and, to a certain extent, the Non-Governmental Organizations (NGOs)
that collaborate with them).
Government:
In the past decade, the US Institute of Medicine (IOM) and the Centers for Disease
Control and Prevention (CDC) have recognized the significance of obesity as a public health
problem and focused greatly on prevention and education. In the US, the recommended dietary
intake standards for the nation are set by the United States Department of Agriculture (USDA)
every five years in the form of the ‘Dietary Guidelines for Americans’ document. The latest set
of recommendations were published in 2010 [97] and drew attention to physical exercise and
obesity prevention. The federal government, through the Department of Health and Human
Services (DHHS), is charged with the responsibility of funding and organizing initiatives that help
in nutrition education and obesity prevention at the state and federal level. The DHHS also
frequently releases guidance documents that help state, community and school level bodies
formulate their own nutrition education and awareness programs. Readers are directed to
Coplan et al.[13] and two workshop reports published by the National Academics and the IOM
[14] [98] and for a detailed examination and summary of obesity prevention measures initiated
by various government bodies at the state and federal level and their outcomes.
Industry:
Koplan and Brownell, in a commentary piece in the Journal of the American Medical
Association (JAMA) examine industry responses to the obesity threat. Koplan mentions that
while the industry has many ‘pluses’ in this regard: nutrition labeling, portion control packaging
and endorsement of public health campaigns to name a few, it still indulges in several
18

counterproductive practices; practices like advertising calorie-dense products to children, and
deceptive product reformulations [91]. While it is not possible to examine all industry’s
responses to the obesity epidemic in a few paragraphs, the following section provides an
overview of some of the high-budget, long-term initiatives. This information is divided under
three subheadings: Advertising, Product Reformulation/ReBranding and, Alternative Media.
Advertising:
A report released by the USDA in November of 2012 estimated that in 2010 US
consumers spent approximately $1.169 trillion on food (both consumed at home and away
from home) [99]. It is estimated that children and adolescents spend around $165 billion a year
of their own money and influence the spending of another $200 billion every year [100, 101].
As a result, children have frequently been targeted by aggressive marketing practices.
A report published in 2008 by the Federal Trade Commission (FTC) calculates that in
2006, food companies spent $9.6 billion in advertising, out of which $1.6 billion was spent
targeting children between 2 and 17 years of age. $ 1 billion was spent on ads targeted at
adolescents [102]. It has been estimated that, as of 2004,children between the ages of 2-11 are
exposed to approximately 5,500 ads per year, teens (aged 13-19) approximately 8,000 ads and
adults 11,000 ads per year [103]. Note that these are just television ads related to food
products.
Research by the Federal Trade Commission (FTC) concludes that when 1977 was
compared with 2004, there was no increase in the television ads that children viewed. There is
presently no clear academic consensus on whether high exposure to food related advertising is

19

one of the causes for the rise in obesity, especially the alarming increase in childhood obesity
[103].
Nevertheless, it is known that advertising and marketing methods have a strong
influence on the food choices of children, adolescents and adults alike. [104]. Children as young
as 2 or 3 years develop an affinity for certain brands [13], decreasing their ability to
differentiate between healthful and non-healthful products. In a study published in the Journal
of the American Medical Association (JAMA), researchers reported that brand logos and spokes
characters registered a high degree of recognition among preschoolers, with those characters
featured on the Disney Channel being the most easily recognized (n=229, 3 to 6 years of age).
Recognition levels increased as age increased [105]. The brand preferences of children are
known to influence the dietary choices of the entire family, not only the child in question [106].
Readers are directed to Mangleberg [107] and Jenkins [108] for reviews on this topic.
The FTC, the government body that regulates advertising practices and monitors media
related to unfair ads, currently has no authority when it comes to ads for food products
directed at children. However, a 2010 report indicates that the FTC is considering an overhaul
of its approach to this subject [109].
The industry has responded to food marketing criticism and the possibility of FTC
regulation with self-regulation. The Children’s Advertising Review Unit (CARU), set up by a
coalition of industry partners in 1974, has established a voluntary set of guidelines for child
related advertising [110]. In 2006, the Children’s Food and Beverage Advertising Initiative
(CFBAI) was set up by 10 food and beverage companies (the CFBAI has since grown to include
17 companies representing a total of 60 to 90% of product sales in their food category) [111].
20

The companies that are a part of the CFBAI agree to “devote at least 50% of their advertising
directed to children under 12 to include messages that encourage good nutrition and healthy
lifestyles. [102]” Response to both CARU and CFBAI initiatives over the years has been mixed.
Some have praised the industry for taking the positive steps of self-regulation. Others have
leveled a variety of criticisms against them [102, 106, 109, 112]. Both the CARU and CFBAI are
primarily funded by the very entities it seeks to regulate (large food manufacturers) [112]. It
has also been mentioned that while the CARU sets forth guidelines that companies should
follow in their advertising practices, it does little to enforce them [113]. The CARU, while
keeping a close watch on advertising methods, has been accused of not paying attention to
products themselves [114]. The CFBAI guidance documents do not specify what exactly
constitutes “advertising directed at children under 12” and do not specify what are messages
that encourage good nutrition [102]. CFBAI member companies have complete independence
to decide which products they choose to self regulate and how. The member companies merely
pledge to apply “established government and scientific standards” and the CFBAI does not
specify a blanket standard for all companies and products [111]. Another noteworthy fact is
that both the CARU and the CFBAI do not include packaging (graphics or structure) as part of
their self-regulation initiatives, despite the widely accepted convention that food packages are
part of the promotional marketing mix and very frequently have structural and graphical
features to entice children (Spiderman Macaroni and Cheese, Crush Cups®, GoGurt ®, Fruit Rollup tongue tattoos, and Milk Chugs® etc). It has also been suggested that these initiatives
merely serve to avoid eventual government regulation of food marketing practices [115].

21

Parent and consumer groups frequently criticize the aforementioned industry ad
practices. The Campaign for a Commercial Free Childhood (CCFC) receives frequent media
coverage for their efforts to create awareness about industry practices relating to child
targeted advertising [116] [98]. In 2006, the CCFC partnered with another consumer giant, The
Center for Science in the Public Interest (CSPI), to target the Kellogg Company. The two groups
drew attention to the fact that the packages of cereal that Kellogg was sending to pre-schools
to use in art projects were their high-sugar Froot Loops and this constituted a case of ‘under the
table marketing’ [98].
Advertising in the form of Public Service Announcements (PSAs) on healthy eating are
frequently broadcast on major networks and radio stations. These are generally produced both
by large food corporations and non-government agencies, or both working together. One
example is The Ad Council’s Coalition for Healthy Children (CHC) initiative which produces and
releases research based PSA’s on nutrition to increase awareness [117]. Their most recent
campaign to generate awareness on obesity prevention began in mid-2012; it was broadcast to
237 million listeners over radio stations and cost $30 million [118].
At present, it is extremely difficult to study the effects of PSA’s and nutrition education
campaigns. Gantz et al. indicate that, in 2005, adolescents viewed an average of approximately
25 minutes of PSAs relating to nutrition and physical activity in contrast to 40 hours of food and
beverage advertising. These numbers were different for children aged 8 to 12, who were
exposed to 1 hour and 15 minutes of PSAs and over 50 hours of advertising for food and
beverages. Gantz and Rideout state that while the amount and content of PSAs and advertising
is known, quantifying precise effects is difficult [15, 119].
22

Canadian data collected by the provincial government of Quebec suggests a link
between advertising and obesity. In 1980 Quebec passed a law banning all fast-food
advertisements that targeted children. Researchers report that over the past 32 years this has
resulted in an $88 million (13%) reduction in fast-food spending and a calorie consumption
reduction of almost 4 billion calories. Authors report that early childhood influence may have
translated into purchasing behaviors in adulthood. Young adults in Ontario (Ontario does not
have a similar advertising ban) are 38% more likely to purchase fast food than Quebec young
adults [120-122].
Product Reformulation and Repackaging:
The IOM stresses that the industry has an important part to play in the prevention of US
obesity by ‘developing and promoting products, opportunities, and information that encourage
healthful eating behaviors and regular physical activity [13]’
The Grocery Manufacturer’s Association (GMA) has stated that almost 4,496
reformulated products were introduced to the US market between 2002 and 2005. Most of
these (67 percent) were lower in saturated and trans-fat and approximately 20 percent had less
added sugar and carbohydrate than previous formulations [123]. IOM reports suggest that large
food and beverage manufacturers like General Mills, PepsiCo and Quaker Oats are spending
significant amounts of money in reformulating products to create healthier nutrition profiles
[13]. In 2011, Wal-mart, as a part of a 5-point obesity prevention agenda, announced a
commitment to reformulate “thousands of packaged foods” that bear its own Great Value®
brand, reducing sodium by 25% and added sugar by 10% by 2016 [98] [124].

23

Along with reformulation, repackaging and rebranding of existing products is another
approach to assist people in making healthy dietary choices. Several studies have conclusively
indicated that people have a “tendency to eat more when offered more [125]”. One study
provided adult men and women (n=60) with bags of potato chips with similar graphics but in
two different sizes (170g and 85g). The subjects who were given the larger portion consumed
more than subjects who were given the smaller portion (p<0.001). Researchers also reported
that when a full meal was served a few hours later, there was no compensation for the amount
of snack consumed [126].
The effect of taste on consumption quantity has been investigated as well. A similar
study performed by Wansink and Park provided people in a movie theatre (n=161) with tubs of
popcorn in two different sizes in two different taste levels. Researchers reported that subjects
who were given the larger portion, consumed more, despite their popcorn being less tasty
(p<0.01) [127].
Children as young as preschoolers have also been observed to consume more when
offered more. Researchers observed that when the amount of an entrée served during lunch
was doubled, children of the same age group consumed more (n=30, p<0.05). This effect was
observed irrespective of the age of the subject [128]. A similar study which created a multiple
regression model to compute energy intake in children found portion size to be the greatest
predictor. Data for this study was obtained from the US government’s continuing ‘Survey of
Food Intakes by Individuals’ (CSFII, n=1045, p<0.001) [129].
In terms of commercially available products, an often cited example of a portion
controlled package is the Nabisco’s 100 calorie cookie pack by Kraft®, which was first launched
24

in 2004. The original package for this product was a thermoformed tray with 30-36 cookies. The
pre-portioned packs had 5-8 individual pouches each containing 100 calories worth of cookies
inside a folding carton [130]. One source mentions that this brand was well received by
consumers and reached sales of more than $100M in the first year of its introduction [131]. A
report released by the Grocery Manufacturers Association in 2004 discusses packaging changes
to encourage healthy eating in detail and mentions that a majority of US companies are
contemplating, or have already made significant changes, to their packaging. [131].
Rebranding of existing healthy products to enhance appeal is also common. A recent
effort by Bolthouse Farms and several other carrot growers to rebrand baby carrots as ‘Junk
food’ received significant coverage in popular media. Three ounce packages of baby carrots
featuring trendy graphics were released in several parts of the US and were sold both in retail
stores and in school vending machines [132]. An FTC report draws attention to several brands
of frozen vegetables and fresh fruit that make use of licensed characters like Curious George®
and Spongebob Squarepants® and recommends that “companies should continue to expand
efforts to package more nutritious products in ways that are more appealing to children [102]”
A large proportion of the total food (slated to be almost 53 percent in 2010) that
children and adolescents eat is consumed at restaurants and food service establishments [133].
However, only 7 percent of total fruits and vegetables are consumed at these places and it has
been suggested that improving presentation of fruits and vegetables served at restaurants will
increase this percentage [13].
Other Initiatives:

25

Large retailers, like Wal-Mart (USA) and Tesco (UK), run promotional campaigns inside
their stores to encourage healthy eating [134]. Food industry giants like Dole, Chiquita and
Sunkist have increased promotional spending to create awareness of the merits of consuming
fruits and vegetables [13]. Food companies also contribute to obesity prevention efforts by
altering their own product portfolios and product priorities. IOM reports feature references to
press releases from the senior managements of companies like PepsiCo and Kraft Foods,
committing themselves to ensuring that a large part of their revenue comes from healthful
product lines [13]. Several large companies and foundations are also making improvements to
their online content to serve the goal of nutrition education and obesity prevention.
Innovative video games, many of which are NGO backed, such as ‘The escape from
Obeez city’ and ‘Squire’s Quest’ have in-game situations that create awareness about physical
activity and the consumption of fruits and vegetables. IOM reports discuss how studies have
determined that children who play these games have increased their consumption of fruits and
vegetables by at least one serving each day [135, 136].
With the increased use of the internet by individuals of all ages, there are also several
online resources that aim to educate users about healthy eating and exercise. Several of these
resources target children and hope to encourage healthy eating from an early age. For instance,
Kraft Foods announced in 2005 that its constituent websites intended for the viewing by
children would feature only healthful products [137]. Most ad campaigns and in-store
promotions are also accompanied by micro websites that reinforce major messages of the
campaign and feature additional information. A noteworthy example is the ‘Eat Smart, Grow
Strong’ campaign, formed as a collaboration of consumers and companies including Anne’s
26

Home-Grown, Newman’s Own and Whole Foods to encourage consumption of vegetables and
exercise through an interactive website targeted at children which featured both a website and
a series of TV spots [138].

27

Chapter 3
Nutrition Labeling: An Overview

Labeling on food packages in the US is largely regulated by five major laws passed over a
span of 84 years (1906-2004).
The Pure Food and Drug Act of 1906 – Required that food and drug labels not be
misleading in any way and carry information on all ingredients [139]. The Federal Food, Drug
and Cosmetic Act (FFDCA) of 1938 – established the agency that became the FDA and gave it
authority over most packaged foods in general [140]. The Fair Packaging and Labeling Act
(FPLA) of 1966 –authorized specific labeling requirements intended to facilitate value
comparisons for consumers. Misleading statements on package fronts were no longer allowed
(eg. Jumbo Quart, Giant Liter); this law also required that the brand name, product name and
net weight information be present on the front of the package, also known as the Principle
Display Panel (PDP) [141].The Nutrition Labeling and Education Act (NLEA) of 1990 – required
that almost all packaged food products sold in the US contain nutrient composition, in standard
content and format, in order to facilitate nutritional comparisons [142]. The most recent
legislation that regulated food labeling was the Food Allergen Labeling and Consumer
Protection Act (FALCPA) that was signed into law in 2004. The FALCPA regulated labeling of
allergens on food packages and made it compulsory for manufacturers and retail
establishments to list all of the main 8 food allergens (milk, egg, fish, crustacean shell fish, tree
nuts, wheat, peanuts and soybeans) that are contained in any product in a prescribed manner
[143, 144].
28

Of the five acts mentioned above, the NLEA of 1990 is the most relevant to this study.
The passage and postulates of this act are reviewed briefly below.

Nutrition Information & the Nutrition Labeling Education Act (NLEA):
The publication of the ‘Product Characteristics Theory’ by Lancaster in 1966 is seen as a
milestone in the field of consumer research. Lancaster proposed that consumers did not simply
look at food products as commodities meant for consumption, but as bundles of attributes
[145]. Lancaster’s contemporary, Stigler, famously proposed that when information is made
more accessible to the purchaser, the costs of information search goes down, increasing the
probability of search, leading to an increase in the amount of information found on packages
[146].
By 1973, Lancaster’s ‘Product attributes’ were expanded and researchers divided the
concept into three categories of attributes: experience, search and credence attributes [12]
(See Figure 2)

29

Experience
attributes
Those that
consumers
assign to
products after
experience
(taste, texture)

Credence
attributes
Those that
consumers
cannot assign
to products
without
information
obtained from
an external
source
(nutrition
content)

Search
Attributes
Those that
consumers can
readily link to a
product during
product search
(color, weight,
category)

Figure 2 - Product Attributes
In the context of the three attributes, a nutrition label is something that converts the
is
‘credence’ attribute of nutrition content into a ‘search’ attribute by providing pre
pre-purchase
information, thereby creating the opportunity for consumers to consciously make healthier
dietary decisions. It is for this reason that nutrition labeling has long been seen as a means to
that
enable consumers to lead healthier lives [11, 147].
During the World Wars, an era of deficiency diseases, nutrition labels in the US were still
voluntary. Where nutrition information was present on packaging, it generally focused on
ere
vitamins and minerals (i.e. those nutrients that people weren’t getting enough of) Analyses of
of).
packages in 1977 indicated that 40 percent of all food products in the US had some form of
f
nutritional labeling [148]; this had risen to approximately 70 percent in 1991 [149].
;
[149]
According to the IOM, the 1980’s were a period of “growing consensus of the link
between diet and health [150].” Experts cite the publication of two reports, one titled, “The
30

Surgeon General’s Report on Nutrition and Health [151]”; and the other titled, “Diet and Health:
Implications for reducing chronic disease risk [152]” as significant contributors to this shift,
receiving widespread coverage in national media. Overall, this resulted in a radical shift in the
information presented on package labels, with emphasis shifting to nutrients that people were
known to consume in excess (sodium, fats and sugar).
Until mid 80’s the provision of nutrition information on food packages was still
voluntary. Manufacturers were free to decide what information went on their products and
‘health claims’ were completely unregulated [12]. While writing about the situation that
preceded the NLEA, authors frequently refer to outrageous claims made by manufacturers in an
attempt to cash in on the growing public awareness of the science linking diet and health. For
instance, Kellogg’s featured a ‘cures cancer’ claim on their high fiber cereals in 1984 [153].
Furthermore, these claims were backed by the National Cancer Institute (NCI) [154, 155]. The
FDA immediately released public statements that such health claims on any product would
automatically invoke the idea of a product intended for the diagnosis or cure of a disease (i.e. a
drug), forcing Kellogg’s to remove the label in question. At the same time, the FDA was urged
by consumer advocates to consider the possibility of allowing health claims on food products
under special circumstances [153].
This increasing need for uniform nutrition labeling and greater regulation of health
claims on packages prompted the signing of the 1990 Nutrition Labeling and Education Act
(NLEA) by President George HW Bush [12]. This law, and its accompanying regulations, brought
about significant changes in the way nutrition information was displayed on packages and
initiated the regulation of health claims [156, 157].
31

The Nutrition Facts Panel (NFP):
The NLEA required a standardized format, placement and content of nutritional
information for almost all packaged foods sold in US commerce, in the form of a table called the
‘Nutrition Facts Panel,’ or NFP [156, 157].
The Nutrition Facts Panel (NFP) (depicted in Figure 3) contains the absolute amounts of
several nutrients contained in a single serving of the food. Along with these absolute amounts,
the NFP also contains the percentage of the nutrient in a serving of the food with reference to
the Recommended Daily Intakes (RDI) for micronutrients, like vitamins and minerals, and Daily
Reference Values (DRV) for macronutrients like fat and carbohydrates [158]. These values are
published by the FDA as the amount of each nutrient required for normal functioning in
“average adults.” These percentage values are designed to help consumers plan their daily food
intake. Serving sizes referenced on the labels are derived from a comprehensive list of foods
referred to as the FDA’s “Reference Amounts Customarily Consumed” (RACC) [159]. Table 1
presents a list of NFP optional and required nutrients [158].

32

Figure 3 - A typical Nutrition Facts Panel (NFP). Image retreived from
http://www.fda.gov/ucm/groups/fdagov-public/documents/image/ucm153391.png

33

Table 1 - Optional and Required nutrients on Nutrient Fact Panels (NFP)

Macronutrients
(Required)

Micronutrients
(Required)

Other
(Optional)
Calories from saturated fat
Stearic acid (meat &
poultry only)

Total Fat (Trans fat,
Saturated fat)
Total Carbohydrate
(Sugars, Dietary
Fiber)
Protein

Sodium

Polyunsaturated fat

Vitamin A

Monounsaturated fat

Vitamin C

Potassium

Calcium

Soluble and Insoluble fiber

Iron

Sugar alcohol

Cholesterol

Percent of Vitamin A
present as beta-carotene
Other essential vitamins &
minerals

The Impact of the NLEA
In an early opinion article which pondered the potential impact of the NLEA,
Pappalardo [160] suggested that divergent views existed at the time. Different sections of the
society offered different views of the Act and its potential benefits, some agreeing that it would
result in more consumers opting for healthier options [161] and others declaring that, while the
Act had its good features, it would not spur healthy product innovation [162]. Other
publications at the time indicated that despite the industry incurring around $ 2 billion in
implementation costs, the projected benefits would be as much as $ 100 billion [163]. Others
estimated that the NLEA would result in the saving of almost 1.2 million life years within the
first 20 years of its passing [164]. Nutrition labeling was further updated in 2003, the FDA
mandated that all manufacturers would be required to display trans-fat per serving in the NFP
34

for all products. The FDA estimated at that time, that within the first 3 years of its
implementation, the trans-fat rule would prevent at least 240-480 Coronary Heart Disease
deaths per year and resulted in savings of up to $8.3 billion per year [165]. Confirmation or
rejection of these estimates were not found in the reviewed literature.
Provided below is a summary of studies that have assessed the NLEA’s impact from
various aspects.
Product Sales and Formulation:
Peer reviewed studies analyzing trends in specific product categories with respect to the
passage of the NLEA are rare. The published body of literature regarding the effect of nutrition
information (in the form of the NFP) on purchase decisions is largely inconclusive. Of the
limited studies available, results and conclusions about the usefulness of the presence of the
NFP vary significantly.
One study by Mojduska in 2001, which assessed sales in the frozen meal category, found
no apparent changes (α = 0.05) in buying behaviors before and after the passage of the NLEA,
and concluded that the demand for frozen prepared meals was affected more by taste and
price preferences of consumers than the new mandatory labeling policy [166].
By contrast, Mathios analyzed salad dressing sales (pre and post NLEA implementation),
and concluded that high-fat dressings experienced a decline in their sales after the NLEA
labeling was mandated (a decline of about 4 to 5% in sales for high-fat dressings as compared
to less than 3% for low-fat dressings). It was also reported that prior to the passage of the NLEA
when nutrition labeling was voluntary, a majority of low-fat dressings had nutrition labels while
high-fat dressings did not [167].
35

A popular trade publication reported that after the passage of the NLEA, sales of
unhealthy bologna went down and low-fat sliced turkey went up [168]. An article in USA Today
claimed that the sales of premium varieties of ice cream had gone down [169] and another
newspaper reported a drop in the sales of peanut butter [170]. A survey of supermarket deli
executives revealed that the sales of high-fat, non-nutritious items stayed the same despite the
fact that several new items had been introduced to promote sales of healthier products [171].
One indirect goal of the NLEA was to encourage manufacturers to reformulate products
in an effort to be competitive and promote public health in the process [12]. In a publication
dated 2012, Moorman et al. mention that in response to the NLEA, firms were more likely to
reduce negative nutrients in new brands and increase positive nutrients in existing brands, i.e.,
existing brands would be fortified with vitamins and fiber to increase perception of
healthfulness but their fat and sugar content would be untouched. New products on the other
hand would be formulated with reduced sugar and fat [172]. Readers are directed to Moorman
et al’s 2012 paper for a detailed analysis of the NLEA’s effect on nutrition content over a vast
array of product categories. It comprehensively addresses findings and addresses reformulation
from standpoint of firm size and overall changes in nutrition of the nation’s food supply [173].
The results of one study released by the US Economic Research Service (ERS, a division
of the US Department of Agriculture) concluded that there had been no significant change (α =
0.05) in the nutritional value of products (as measured by a Nutrition Quality Index) offered by
manufacturers in five common food categories (entrées, soups, salted snacks, cookies and
bacon) after the passage of the NLEA. This study was conducted by calculating the Nutrition

36

Quality Index (NQI) for all product offerings in the five categories in one New England
supermarket between 1992 and 1997 [174].
Another ERS study presented statistics on the number of new reformulated offerings by
manufacturers in the salty snack category in the 1994 to 1999 time period. In 1994 there were
1,914 and in 1995, 2,076 new, healthier products introduced in the salty snack category, but by
1999 this number had reduced to 481. The authors attribute this to the fact that the
manufacturer response to labeling legislation dwindled based on consumer response [175].
Information Search and Label Use:
Another important case for the NLEA was that it would increase the efficiency with
which consumers processed nutrition information. Balasubramaniam and Cole reported that
according to focus group studies they conducted, consumers began to pay more attention to
negative food attributes, like fat and cholesterol content, as a result of the Act [176].
The USDA’s ERS reports that, in general, the use of nutrition labels declined between
1995 and 2005. This study was conducted by performing probit regression analyses on the data
obtained from the US Government’s Diet and Health Knowledge Survey or DHKS (n=1045) and
data obtained from the National Health and Nutrition Examination Survey (NHANES). During
that 10 year period, the use of the NFP reportedly declined by 3% and the use of the ingredient
list by approximately 11%. The decrease was greatest for individuals between the ages of 18
and 29 and individuals whose primary language was Spanish. The use of information pertaining
to sodium, fat and sugar content decreased whereas the utilization of nutrition information
pertaining to fiber content increased dramatically. The authors attribute this to an increased
awareness of the importance of the high-fiber diet [177].
37

Keller et al. performed a mail-in survey (n=468) to study the interaction between
consumer’s motivation to process nutrition information and positive attitude towards healthy
products. Researchers reported higher purchase likelihoods, better overall nutrition and
product attitudes for healthy products when the motivation to process nutrition information
was high (F=3.4, p<0.02) [178].
Moorman performed a survey of supermarket shoppers in pre (n=554) and post NLEA
(n=558) periods, specifically, October 1993 and October 1994. The survey specifically
addressed 20 product categories and revealed that consumers acquired more nutrition
information after the NLEA was implemented (p<0.01). Discussing the impact of consumer
motivation on time spent processing nutrition information, Moorman reports a significant
positive relationship (p<0.05). This was stronger in the post NLEA subjects (β=2.050) than the
pre-NLEA subjects (β=0.751) [179].
Szykman et al. analyzed data from the FDA’s Food Label use and education survey
(FLUNES) and then performed a telephone survey (n=1945). Both analyses revealed a positive
relationship between nutrition knowledge (as measured by consumer ability to recall dietdisease relationships for three diseases – blood pressure, diabetes and cancer) and the use of
NLEA nutrition labels (p<0.01) [180].
Several studies cited and described above suggest that the NFP has been beneficial to
consumers who were either already very motivated towards healthy diet choices [178, 179] or
knowledgeable about nutrition labeling [180-182] . One probable reason for this is that when
new label designs are introduced, their immediate effects are felt only on those regularly

38

reading labels already. Drichoutis et al. suggest that making inferences on the transition to a
mandatory labeling system is extremely hard [183].
A study by Byrd-Bredbenner et al. published in 2000 examined the differences between
NLEA style NFPs and food labels created in accordance with voluntary European nutrition label
standards. A sample of women (n=50) aged 25 to 45 years of age was recruited from among
people visiting prenatal and dental clinics in the UK. Surveys revealed no statistically significant
difference between the ability of subjects to locate and manipulate nutrition information when
comparing the forms of food package nutrition labeling (p>0.05) [184].
Dietary Choices and Behavior:
The ultimate intention of the NLEA was to cause a change in the attitudes of the general
public, prompting them to choose healthier alternatives. There have been several studies that
have analyzed the impact of the NLEA on dietary choices.
In a recent quasi-experimental study that evaluated the NLEA's impact on dietary
changes, the author made use of the ‘natural experiment’ created by NLEA’s exemption of food
at restaurants. At the time when the study was performed, consumers were generally only
exposed to nutrition information when purchasing foods from a grocery store. By comparing
food choices at home between label users and non-users and creating an empirical model of
label use, Variyam concluded that the calories in the diets of label users were significantly
lesser than those of non-users for all NFP nutrients except sodium (p<0.01). Even in the food
away from home segment, where there are generally no nutrition labels present, label-users
were known to consume significantly less amounts of all nutrients except protein and sodium
(p<0.01). This led to a conclusion that label-users consume a better quality diet than label non39

users irrespective of whether the NFP was available or not. The study utilized data from the
CSFII (Continuing Survey of Food Intake of Individuals, n=5765) and the DHKS (Dietary Health
Knowledge Survey, n=1045) [147] to reach this conclusion.
The data from these surveys (the CSFII and the DHKS) have been used to perform a
variety of analyses. Kim et al. analyzed this data (n=5343) and performed a probit analysis
where they looked at the different variables that were associated with use of the NFP. Use of
the NFP was found to be positively associated with an increased HEI (Healthy Eating Index, a
measure of diet quality developed by the USDA) in consumers who had high incomes, were
younger, were female, lived in non-metro locations and were generally aware of the link
between diet and health (α=0.01) [185].
Another study that analyzed the results of the CSFII (1994-1996)/ DHKS focused on the
sugar intake of US consumers. Researchers concluded that around 32% of all US consumers
always used the NFP to acquire information about a product’s sugar content. The model
suggested estimates of their consumption of added sugars to be 1.1% less than other
individuals (α=0.05). Other dependent variables associated with reduced sugar consumption
were high levels of education and special diets [186].
Lin et al. used CSFII results to generate and study predictors of fat-intake among men
and women. It was revealed that the search for nutrition information, among other variables,
was a strong predictor of reduced fat intake (p<0.01) [187].
Neuhouser et al. conducted a survey (n=1450) in the state of Washington; concluding
that NFP use was higher in women, individuals younger than 35 years and those with higher

40

than a high school education (α=0.01). The authors also reported that NFP use was not
associated with an increased intake of fruits and vegetables [188].
Fitzgerald interviewed a sample consisting of diabetic (n =100) and non-diabetic (n=101)
Hispanic females and performed several statistical analyses relating the use of NFP labels and
nutrition knowledge. It was reported that the use of nutrition labels was positively associated
with increased nutrition knowledge, leading to better overall dietary choices (p<0.05). There
was, however, no effect of a positive or negative diabetes diagnosis on the use of nutrition
labels and healthy dietary habits (p>0.05) [189] .
A study performed in a laboratory setting by Kral et al. served meals to normal weight
women (n=40) with and without energy per serving information (provided to them in the form
of an NFP). Each meal had a choice of three entrees which had different calorie values (1.25
cal/g, 1.5 cal/g, 1.75 cal/g) but rated the same in terms of palatability. It was revealed during
statistical analyses that women who were provided with NFP’s had an energy intake that was
22% less than that of the others (p<0.0001) [190].
A study by Kristal et al. that spanned two years analyzed the various predictors of good
dietary practices in relation to cancer risk. A convenience sample (n=336 men, 502 women) of
subjects recruited by random digit dialing was interviewed on their level of nutrition label use
and diet-cancer risk awareness. The same subjects were interviewed two years later (19951996 initially and then 1997-1998 as a follow up) and changes in their diets, if any, were
recorded. Several predictors were revealed but the strongest predictor for a decreased intake
of fat in the diet was the usual use of nutrition labels (NFPs) (p=0.001, β1=-0.44) [191].

41

Variyam published an ERS study that empirically analyzed the results of the National
Health Information Survey (NHIS, n=90,000 to 125,000 each year). The NHIS results of 1991 and
1993 were considered to be the pre-NLEA case and the results of 1995 and 1998 to be the postNLEA case. Several results emerged from this analysis. Among them was the conclusion that
overall label use has risen only 3% after the passage of the NLEA (67% to 70% respectively). The
percentage of the population who were obese fell after the passage of the NLEA (p<0.01) both
in the overall sense and in several demographic groups (non-Hispanic white men, black men,
white women). However, there was no evidence of a statistically significant effect in the
percentages of black women who were obese before and after the passage of the NLEA [11].

Consumer Characteristics and Nutrition Label Use
A review of research regarding consumer attention to, and comprehension of nutrition
information is germane to this study, which analyzes consumers attention to and
comprehension of varied designs of FOP nutrition labeling. There are various consumer
characteristics that may impact the use of on-pack nutrition information and over the years
several of these have been studied closely – age, education, gender, income, location, work
status and other ancillary factors such as having a certain medical condition or having prior
knowledge of nutrition labels. The following subheadings review some of the research on
consumer nutrition label use. Readers are directed to Drichoutis et al. for extremely thorough
reviews on the subject [183].
Age:
Age related effects on the perception of label information have been documented since
the early days of labeling. However, like many of the other consumer characteristics analyzed
42

below, the effect of age on nutrition information use is not yet been fully understood [183].
Some researchers have reported that nutrition label use increases with age [181, 192, 193]
while others have reported the opposite [182, 185, 194] (see Table 2 for descriptions and
results of studies cited here)

43

Table 2 - Research investigating the effect of Age on Nutrition Label use

Method

Result

Positive Relationship
Dietary and Nutrition habit Survey of British
Age is positively correlated with use of
psychology students aged 17 to 52 (mean age = nutrition labels (r=0.21, p<0.05)
22.2±7.32 years, n=165, Females = 88% of n)
[192]
Empirical model fitted to data gathered in a
survey (n=330) administered in fifteen
supermarkets in Athens, Greece [181]

Nutrition label use highest for consumers
above the age of 55 (β=0.236, p=0.05)

Empirical model fitted to data gathered from
surveys administered in five grocery outlets in
New Jersey (n= 291) [193]

Individuals aged between 51 and 65 had
the highest likelihood of using nutrition
labels (estimated 19% higher than anyone
under the age of 36, p<0.01).

A cross sectional survey of nutrition label use
among African American consumers (n= 458,
mean age = 43.9±11.6 years) [195].

Positive association between age and
nutrition label use with older consumers
more likely to use nutrition labels (p<0.05)

Negative Relationship
Interview and observation of consumers (n=79)
in chain stores in a mid-western US university
town. Subjects randomly assigned to either
directed or non-directed product search tasks,
breakfast cereal being product of interest. Chain
store study was followed by laboratory
experiment to determine effect of age on
working-memory capacity using an established
working memory measurement task. [194]

Significant effect of age on search
intensity (measured by counting the
number of cereal boxes picked and panels
examined) with younger (20-59 years)
consumers inspecting more number of
boxes and panels than older consumers
(F=4.42, p<0.01). Younger adults also
inspected more number of boxes than
older adults in directed search tasks
(p<0.05). Age revealed to negatively
correlate with working memory capacity
(r=-0.42, p<0.007). Age however not
found to affect with motivation to process
nutrition information

Empirical regression analysis of data obtained
from 1994-1996 US Government Continuing
Survey of Food Intakes of Individuals (CSFII)
and 1995 US Government Diet and Health
Knowledge Survey (DHKS, n=5343) [185]

Use of nutrition fact panel negatively
associated with Age (β1=-0.0061, p<0.01)

44

From an information processing standpoint it is generally acknowledged that with age
comes a reduced capability for information acquisition and processing [196]. In an interview
based study by Cole and Gaeth, published in 1990 (n=48), consumers over the age of 60 were
reported to be less accurate than consumers aged lower than 60 when processing nutrition
information (F=37.64, p<0.01, consumers specifically were asked to distinguish between two
cereal products and pick out the healthier one), although age was not found to correlate
significantly with existing nutrition knowledge [197].
Using a mail survey (n=191), Burton et al. assessed label use and understandability.
Older consumers (mean age of 70 years) found the information on the NLEA NFP label harder to
understand (F=6.4, p<0.02) than younger consumers (mean age of 48 years). Another
significant effect was that the information in the NFP was more easily processed by consumers
when dealing with unhealthy foods than healthy foods (p<0.01) [198].
Education:
It is generally reported in the literature that more educated consumers use nutrition
labels more and indulge in more complex information search behaviors, drawing information
from both ingredient lists and nutrition fact panels (see Table 3 below for detailed descriptions
of studies).

45

Table 3 - Research investigating the effect of Education on Nutrition Label use

Method

Results

Empirical model fitted to data gathered in a survey Nutrition label use highest for
(n=330) administered in fifteen supermarkets in
consumers with university education or
Athens, Greece [181]
higher (β1=0.243, p=0.033).
Empirical regression analysis of data obtained from
1994-1996 US Government Continuing Survey of
Food Intakes of Individuals (CSFII) and 1995
DHKS (n=5343) [185]

Higher probability of individuals to use
information from Nutrition Fact Panels
if they possessed college education of
some form (β1=0.1259, p<0.01)

Empirical analysis of data from 1989 CSFII
(n=2214) [199]

Nutrition label use highest among
consumers with at least some college
degree (β1=0.2212, p<0.01)

Probit analysis of data obtained from telephonic
Label use found to increase with
interviews with adult US women between the ages education in number of years
of 20 and 59 (n=1265) [200]
(β1=0.084, p<0.01)
Telephone survey of grocery shoppers in Southern Label use greater among respondents
United States (n=1421) [201]
with college education or higher
(p=0.0003)
Survey administered in four supermarkets in four
different socio-economic localities in Middlesex
county, New Jersey (n=200) [202]

Consumers with college education
have a higher likelihood of using
nutrition information on packages at
home (p<0.05). Education not found to
be associated with label use in the
grocery store (α=0.05).

Consumer utilization of food labeling as a source
of nutrition information studied by analyzing
results of the US Government’s Nationwide Food
Consumption Survey administered between 19871988 (n=4250) [203]

Level of education of household head
positively associated with nutrition
label use (p<0.01). 56% of households
with heads who had completed four or
more years of college reported to use
nutrition information for purchases as
opposed to only 30% of households
with heads who had less than a college
education.

Gender:
Research suggests that female subjects have a higher probability of reading nutritional
labels and applying nutrition information to dietary choices; it has been suggested that this is
46

likely related to the fact that they tend to have primary responsibility for meal planning and
shopping. Table 4 provides descriptions of studies that have examined the link between gender
and nutrition information use.
Table 4 - Research investigating the effect of Gender on Nutrition Label use

Method

Results

Empirical model fitted to data gathered from
Males less likely to make frequent use of
survey administered five grocery retail outlets in nutritional labeling (β1=-0.5174,p<0.10)
New Jersey (n= 291) [193]
Empirical regression analysis of data obtained
from 1994-1996 US Government Continuing
Survey of Food Intakes of Individuals (CSFII)
and 1995 DHKS (n=5343) [185]

Males less likely to make use of all forms
of nutrition information on packages –
ingredient lists, health claims, NFP,
serving size information (p<0.01).

Empirical Regression Analysis of data obtained
from 1995 US Government Diet and Health
Knowledge Survey (DHKS, n=5343) [182]

Level of label use for men lower than that
for women (β1=-0.1449, p<0.01)

Empirical analysis of data from 1989 CSFII
(n=2214) [199]

Self reported label use greater among
women than men (β1=-0.2515, p<0.01)

Telephone survey of grocery shoppers in the
Southern United States (n=1421) [201]

A higher percentage of women use
nutrition labels than men; 82% as opposed
to 77%. This difference is statistically
significant at p=0.0393

In a study by Nayga that analyzed the data from the DHKS, it was reported that the
likelihood of a subject reporting that they found nutrition label information not useful was
higher if they were male (β1=-0.284, p<0.05) [204]. A subsequent study by the same authors
reported that female subjects were more likely than men to attend to and use all forms of
nutrition information on packages (p<0.05)[205].

47

In a similar study that fitted empirical models to data gathered in a survey (n=330)
performed in fifteen supermarkets in Athens, Greece, it was revealed that females were more
likely to gather information from calorie disclosures (p<0.01), sugar disclosures and
vitamin/mineral disclosures (p<0.05), but were less likely to gather information from and use
ingredient lists (p<0.05) [181].
Time Pressure:
While the general consumer behavior literature frequently cites time pressure and time
availability as one of the major factors affecting shopping behaviors [200, 206-208] no studies
were located in a synthesis of the peer reviewed studies, specifically analyzing the relationship
between time pressure and nutrition label use. With an increasing number of media outlets and
consumer organizations now citing low time availability as a reason for declining nutrition label
use, it is envisioned that future editions of the DHKS and CSFII will contain questions pertaining
to nutrition label use in time sensitive situations.
Income:
Like the effect of age on label usage, peer reviewed publications investigating the effect
of income have yielded mixed results, with some studies indicating a positive relationship and
others a negative one. See Table 5 for detailed descriptions of studies and results.

48

Table 5 - Research investigating the effect of Subject Income on Nutrition Label use

Method

Results

Positive Relationship
Telephone survey of grocery shoppers in
Southern United States (n=1421) [201]

Higher likelihood of individuals
belonging to households with an annual
income over $35,000 using nutrition
labels (p=0.005) as compared to those
belonging to households with annual
incomes below that amount.

Consumer utilization of food labeling as a source Annual household income positively
of nutrition information studied by analyzing
associated with probability of nutrition
results of the Nationwide Food Consumption
label use (β=0.0632,p<0.05)
Survey administered between 1987-1988
(n=4250) [203]
Empirical regression analysis of data obtained
from 1994-1996 US Government Continuing
Survey of Food Intakes of Individuals (CSFII)
and 1995 DHKS (n=5343) [185]

The probability of the use of the
information from the nutrition facts
panel higher if household annual
income over $10,000
(β=0.0360,p<0.01)

Mail survey administered to randomly selected
households in Louisiana (n=730) to gauge use of
nutrition labels while purchasing meat products
[209]

Participants from households with
annual incomes of over $15000 more
likely to report using nutrition labels to
check for nutrients when comparing
between processed meat products like
packaged salami when compared with
households with incomes below that
amount (p<0.05)

Negative relationship
Empirical model fitted to data gathered in a
survey (n=330) administered in fifteen
supermarkets in Athens, Greece [181]

Probability of label use negatively
associated with household income.
Label use statistically lowest when
annual household income is over 20,000
Euros (β=-0.0315, p<0.001).

Mail survey administered to randomly selected
households in Louisiana (n=617) to gauge use of
nutrition labels while purchasing meat products
[210]

Households with family incomes over
$60,000 were less likely to read
nutrition labels while purchasing
processed meat products like salami
(p<0.05)

49

High household incomes are also known to affect other aspects of nutrition label usage.
In the study by Nayga detailed in a previous section, it was reported that consumers from high
income groups, along with perceiving nutrition labels to be more helpful, also perceived
themselves to be able to make better choices based on their knowledge of nutrition
information (p<0.05) [204]. Another study by the same author reported that income also
affects the utilization of specific information in the NFP, with high income consumers using
information about calories, sodium, fiber and fat significantly more than their low income
counterparts [205].
Working Status:
Working status as a possible effect on nutrition label usage has also been studied. See
Table 6 below for detailed descriptions and results of studied that have investigated this link.
Table 6 - Research investigating the effect of Work Status on Nutrition Label use

Method

Results

Survey administered in four supermarkets in
four different socio-economic localities in
Middlesex county, New Jersey (n=200) [202]

According to the statistical model
unemployed consumers report using
nutrition labels more, both while shopping
and at home to choose and compare between
products (β=0.64 to 0.95, p<0.05)

Sample of consumers surveyed in four
supermarkets in New Jersey (n=200) [211]

Nutrition label use measured in two cases –
when consumer has nutrition knowledge and
when he/she doesn’t. Unemployed
consumers are more likely to use nutrition
labels in either case (β=0.726 & 1.167,
p<0.05)

Empirical model fitted to data gathered in a
survey (n=330) administered in fifteen
supermarkets in Athens, Greece [181]

The probability of label use significantly
higher in employed consumers (p<0.05)

50

Household Label Usage:
The relationship between household type and nutritional label usage has also been
studied. Some studies have shown the size of the household to be negatively related to
nutrition label use, i.e. the smaller the household, the more nutrition label use is likely but
other studies have suggested the opposite relationship. Results are presented below (see Table
7)
Table 7 - Research investigating the effect of Household type on Nutrition Label use

Method

Results

Positive Relationship
Empirical model fitted to data gathered in a
survey (n=330) administered in fifteen
supermarkets in Athens, Greece [181]

Consumers who are primary shoppers
within the household report using labels
more than the other household members
(p<0.05)

Empirical model fitted to data gathered from
survey administered at five grocery retail
outlets in New Jersey (n= 291) [193]

Household size of consumer negatively
associated with nutrition label use with
consumers in smaller households reporting
a higher level of nutrition label use
(p<0.01)

Negative Relationship
Consumer utilization of food labeling as a
source of nutrition information studied by
analyzing results of the Nationwide Food
Consumption Survey administered between
1987-1988 (n=4250) [203]

Larger households are more likely to use
nutrition labels on a daily basis (β=-0.3300,
p<0.01)

Location:
Several studies detailed previously in this report have suggested that consumers
residing in rural areas are known to use nutrition information more than consumers living in

51

urban areas, although related causalities, such as the time pressures associated with living in
big cities, have yet to be explored. Table 8 contains detailed descriptions and results

Table 8 - Research investigating the effect of Geographical location on Nutrition Label use

Method

Results

Empirical model fitted to data gathered from
survey administered at five grocery retail
outlets in New Jersey (n= 291) [193]

Nutrition label use while shopping largest
among consumers from suburban localities
(p<0.01)

Consumer utilization of food labeling as a
source of nutrition information studied by
analyzing results of the Nationwide Food
Consumption Survey administered between
1987-1988 (n=4250) [203]

Higher likelihood of consumers using
nutrition labels on packages if residing in
non-metropolitan areas (p<0.01)

Mail survey administered to randomly selected
households in Louisiana (n=730) to gauge use
of nutrition labels while purchasing meat
products [209]

Rural residents used nutrition labels on
meat products to check for undesirable
nutrients and perform comparison between
products more than urban residents
(α=0.05)

Special diets & Medical Conditions:
It is fairly well documented that consumers following special diet regimens pay
significantly more attention to, and use, nutrition information than those without special
dietary needs. Table 9 contains descriptions.

52

Table 9 - Research investigating the effect of Special Diets and Medical Conditions on Nutrition
Label use

Method

Results

Empirical regression analysis of data obtained
from 1994-1996 US Government Continuing
Survey of Food Intakes of Individuals (CSFII)
and 1995 DHKS (n=5343) [185]

Following a special diet is positively
associated with the use of all types of food
product information found on a package
including information from the nutrition
fact panel (p<0.01)

Survey administered at retail outlets in the Cork 9% of all subjects surveyed reported using
region of Ireland (n=200) [212]
nutrition labels to enable them to follow a
special diet

Prior Knowledge:
Prior knowledge of nutrition information has been known to increase the perceived
efficiency of label usage among consumers. Consumers who are generally aware of the
importance of healthy diets have been indicated to use nutritional labels and nutrition
information more [180, 182, 199]. Limited studies have not indicated this to not be the case
[211]. See Table 10 below for descriptions.

53

Table 10 - Research investigating the effect of Prior Knowledge on Nutrition Label use

Method

Results

Empirical model fitted to data gathered in a
survey (n=330) administered in fifteen
supermarkets in Athens, Greece [181]

Prior nutrition knowledge is positively
associated with nutrition label use
(β=0.384, p<0.01)

Empirical Regression Analysis of data obtained Consumers with high levels of health
from 1995 US Government Diet and Health
knowledge are more likely to use nutrition
Knowledge Survey (DHKS, n=5343) [182]
labels (p<0.01)
Regression analysis of data from the FDA’s
Food Label use and education survey
(FLUNES) and an additional performed a
telephone survey (n=1945) [180]

Health knowledge measured as awareness
of diet-disease link healthy eating and three
diseases (blood pressure, heart disease). All
measures except knowledge of blood
pressure positively associated with
nutrition label use (p<0.01).

Sample of consumers surveyed in four
supermarkets in New Jersey (n=200) [211]

No statistically significant effect of
nutrition knowledge on overall nutrition
label use (α=0.05)

Note that the above review is restricted to studies and findings that assess consumer
use of ‘on-pack’ nutrition information. As mentioned in the previous chapter, consumers also
receive nutrition information from a variety of other sources, especially in the FAFH (Food away
from Home) sector. Authors have also published studies that assess consumer use and effects
of nutrition information found in dining hall and cafeteria signage [213], menus at restaurants/
fast food outlets [214-219], the internet [220] and from shelf-labels in stores [221-223].

Discussion
Overall, the foregone review suggests a decline in the use of the NFP in the decade
following its introduction. Further, results regarding the effectiveness of on-pack nutrition
label are quite mixed [177]. In a 2009 speech, Commissioner Hamburg of the FDA mentioned
that obesity costs “had doubled since 1998” and that food labeling was “a concern that has not
54

been substantially addressed since the FDA implemented the [NLEA] more than 16 years
ago[224]” A report from the CSPI released soon after analyzes the current format of the NFP
(Figure 3) and suggests several changes, mentioning that the “current nutrition facts label was
not specifically designed to prevent obesity and needs to be revised to reverse [the increase in
obesity rates] [225].”
It is obvious from the foregone review and research investigating the relationship
between NFP usage and diet finds that the labels have a positive benefit when used [185].
While people with health conditions such as high blood pressure or high cholesterol and those
on special diets (Table 9) and those with prior knowledge (Table 10) are significantly more likely
to use nutrition labels [188, 226-228], the overall percentage of consumers who access the
nutritional information in the NFP when making food purchasing decisions is low [229-231].
Thus, it seems that the labels are capable of achieving the desired effect, but fail to do so
because too few consumers access and use the information. This pattern led the FDA’s August
2003 obesity working group to conclude that “...it is clear that consumers would benefit if they
were to pay more attention to and make better use of information…on food labels. Providing
encouragement and making it as easy as possible for consumers to do so are worthy public
health objectives.” [18].

55

Chapter 4
Front-of-Pack Nutrition Labeling

Introduction & Classification
As described before, FOP labels represent the practice of moving key nutrient
information to the fronts of packages in an effort to capture consumer attention and simplify
product comparisons (See Figure 1). There is also some belief that displaying simplified
information by making use of color-coding and symbols may help less-educated consumers
utilize nutrition information and improve their diet [225] [29, 30] [232] [233].
Due to their position on the front of the package, FOP labels are thought to be more
noticeable than the NFPs that are present on the side or the back of the packages, serving as
‘health beacons’ to consumers as they peruse shopping aisles [234]. FOP labels also attempt to
address issues of time pressure and complexity in cross-product nutritional comparison. The
assumption is that a uniform FOP system will enable consumers to make nutritional
comparisons between products at a glance, thereby relieving them of time pressures and
improving their ability to plan a healthy diet [91]. It has also been suggested that the symbolic
and suggestive nature of FOPs would enable consumers with limited literacy skills to use
nutrition information [24, 25] [31]. Thus, FOP’s are said to be an important step forward in using
product labeling to solve the problem of escalating obesity [235-237].
Over the years FOP’s have taken on a variety of forms, and range from single icons to
large colorful panels containing both absolute and percentage nutrient values. Two major
modes of classification have emerged.
56

Based on Information Contained:
The US Institute of Medicine (IOM), in its in-depth, Phase 1 report on FOP labels,
classifies FOP labels into three types based on information contained in them [30]. Table 11
provides examples of each of these types.
(i)

‘Nutrient Specific Systems’ that simply provide information about various
nutrients (Eg: a Traffic Light Label)

(ii)

‘Summary Indicator systems’ in which the product is certified to carry a certain
FOP label claiming it as a healthy choice. (Eg: A Heart Check Icon)

(iii)

‘Food Group Information’ systems in which the presence of a certain food group
is claimed, almost like a graphical version of a health claim (Eg: 100% Whole
Grain stamp)

57

Table 11 – Classification of FOP label schemes by information contained. Classification scheme
sourced from Wartella et al.[30].
Type

Example

Nutrient Specific Systems
e.g.: A traffic Light label that
simply presents consumers with a
few major nutrients and their %
Daily Value - usually Fat,
Saturated Fat, Sugar, Sodium and
in some cases. Dietary fiber and
protein
Summary Indicator Systems
e.g.: A Canadian Heart Check Icon
indicating that the product meets
guidelines set by the Canadian
Heart & Stroke Foundation[238]

Food Group Information
Systems
e.g.: A ‘Whole Grains’ symbol that
the product uses to claim that it
has more than 16g of Whole grain
per serving [239].

58

Based on Format:
Based on format, FOPs can be classified into two basic types.
(i)

‘Iconic’ FOPs are single icons or symbols that certify the product as being
endorsed by an organization or meeting certain nutritional criteria (The Heart
check icon shown in Table 11 would be an example of this type).

(ii)

‘Tabular’ FOPs are more detailed and provide information on various nutrients.
They frequently feature color-coding and interpretive symbols (The Traffic Light
label shown in Table 11 or the FOP in Figure 1 would be an example of this type).

Herein, we focus our review and efforts on Tabular, Nutrient Specific FOPs. Specifically,
to investigate the effect of color and icon use, two of the elements that have been utilized and
contested in the systems employed throughout the world.
FOPs are currently being used on food products in a large number of countries including
Australia, parts of Central and South East Asia, Africa and Oceania. For the purpose of this
report, this regulatory history section restricts itself to Europe, UK and USA. Readers are
directed to Hersey et al’s comprehensive report on FOP policy prepared for the US Health and
Human Services Department for detailed reviews on FOP policy around the world [240].

Usage and Regulatory History
United Kingdom:
In March 2006, the now common Multiple Traffic Light (MTL) tabular FOP form (shown in
Figure 4) was introduced in the United Kingdom (UK) for the first time. Quoting from a press
59

release issued by the UK Food Standards Agency (FSA, their equivalent of the US FDA) it was
suggested that these MTL FOPs would,
1. “Make it easier for consumers to eat more healthily
2. Encourage consumers to look for and demand healthier foods
3. Provide businesses with an incentive to produce foods that are lower in fat, salt or
sugars [241]”
As shown in Figure 4, these Multiple Traffic Light (MTL) FOPs feature color-coded
representations of four nutrients (Fat, Saturated Fat, Salt or Sodium and Sugar) modeled after
traffic lights; a green panel conveying a low nutrient level, red conveying a high nutrient level
and amber falling in between. Four specific nutrients: fat, saturates, sugar and salt, were
chosen as being the ones most commonly implicated in causing diet-related diseases among
consumers.

Figure 4 - Examples of Multiple Traffic Light (MTL) FOP labels in different configurations.
Retrieved from http://www.eatwell.gov.uk/foodlabels/trafficlights/
This FOP system was completely voluntary and initial adopters included several large
retailers like Sainsbury’s, Waitrose, Marks & Spencer’s and McCain who began using these MTL
60

on their private-label products [242]. Influential consumer groups and independent
organizations like Netmum’s and the Royal College of Physicians (RCP) expressed support for
this form of labeling. The introduction of these MTLs onto packages was accompanied by a
multi-million pound ad campaign that aired on primetime television across the UK [243].
Since the inclusion and design of FOP nutrition label was entirely voluntary at this time,
food and beverage manufacturers headed by the massive Food and Drink Federation (FDF)
created their own format devoid of colored traffic lights. This format, commonly known as the
Percent Guideline Daily Amount Label (%GDA label), contained both absolute and percentage
amounts of nutrients but no colors (see Figure 5). The percentage values are based on the
Guideline Daily Amount for an “average adult” in good health. Note that this form of GDA
representation was already being used in many food packages prior to this (one source puts
their origin date at around 1998), but only on the back of the package [244].

Figure 5 – A typical %GDA label with absolute and percentage amounts of nutrients but no
traffic light colors.

61

In an effort to draw from the benefits of both formats - MTL and %GDA, combination
FOP schemes have also emerged and been used. These hybrids use both traffic light colors and
absolute/percentage values of guideline daily amounts (GDAs)

Food manufacturers have typically been against colored “traffic light” schemes of any
sort, believing that a row of red patches on the front of a package would demonize their
products, not providing consumers with an accurate representation of the product’s nutrients in
relation to the rest of their diet [245]. Quoting from the Food and Drink Federation (FDF)’s
policy statement on traffic light colors on FOPs,
“FDF is committed to working constructively with [the] FSA on more informative
nutrition labeling. FDF believes that labeling should be objective, to allow
consumers to interpret it in relation to their own requirements, and consistent
with the concept of encouraging consumers to achieve a balanced diet within the
context of their personal lifestyle.
FDF objects to the use of red, amber and green color coding to categorize
products as ‘high’ ‘medium’ or ‘low’ in fat, sugar and salt. This could mislead
consumers as ‘traffic lights’ fail to take account of portion sizes and the
consumption of a particular food in the context of the whole daily diet…
FDF supports the use of GDAs as a more objective way of providing nutritional
information which helps consumers construct a healthy, balanced diet and welcomes its
member’s ongoing move to put GDA information on the front of pack. [246]”

Since their introduction in 2006, both groups, the FDF and FSA, have promoted their
own FOP formats fiercely; funding and performing their own consumer research studies. By the
year 2010, varied FOP label formats were being used, with manufacturers often tailoring their
design and size to suit their own packages. This was counter to the primary purpose driving the
FOP initiative to begin with; that of providing uniform, at-a-glance summary information about
key nutrients within a given product [247, 248]. Inside the UK, this dispute, sometimes referred
62

to as the ‘label war [247]’ by the media, culminated in a vote performed in the UK FSA in March
2010 . The following recommendations were put forth in this session [249].
An ideal FOP should contain,
•

Textual cues – presence of words High, Medium, Low (AND)

•

% GDA Values (AND/OR)

•

Traffic Light Coloring

When the recommendations mentioned above were published, several manufacturers
came out saying that they would wait for the EU wide recommendations before changing their
FOP formats to conform to the above three guidelines [249]. An EU wide recommendation, the
EU 1169/2011 document was tabled in the EU parliament in late 2011; In it FOP labeling was
still voluntary [250].
A few months later in May 2012, the UK government launched a joint consultation (a
large scale public comments gathering phase) with the public and supermarkets [251]. The aim
of the consultation was to put forward a recommendation for a uniform FOP label that would
be adopted by all retailers and alleviate consumer confusion. After three months of consultation
in which over 190 responses from individuals and organizations were received, the FSA zeroed
in on a ‘Hybrid System’ similar to that depicted in Figure 6, containing all three features - colorcoding, %GDA values and textual cues (high, medium, low) [252]. The UK Public Health Minister
Anna Soubry announced in December 2012 after the recommendation was put forward that,
“The UK already has the largest number of products with front-of-pack labels in Europe,
but research has shown that consumers get confused by the wide variety of labels used. By
having a consistent system we will all be able to see, at a glance, what is in our food. This will
help us all choose healthier options and control our calorie intake [253, 254].”

63

It is worth noting that despite this consultation and recommendation by the FSA, FOP
labels are still voluntary in the UK; mandates would require legislation at the EU level, but as of
the time of writing, the FSA remains optimistic that the top ten supermarkets/retailers in the UK
would begin uniformly using the Hybrid label by Summer 2013 [254].
These recommendations have not been met with unanimous acceptance. The National
Farmers union in the UK protested immediately that color coding of FOPs would demonize
cheese, beef and ham as they contain enough fat to label them as being ‘red’ foods while
containing iron, calcium and other vital nutrients [255, 256].

Figure 6 - An artist's rendering of the Hybrid FOP system being proposed by the Food Standards
Agency in the UK.
The European Union:
When the FOP labeling debate reached a fever pitch in the UK around 2008, the EU
published a food labeling proposal, parts of which suggested that the standardization of
nutritional labels, in general, and that of FOP labels was being considered for the whole of the

64

EU (Note that both the UK and the EU do not currently have binding laws regarding nutrition
labeling similar to the NLEA in the US) [257].
The aforementioned food labeling proposal’s many postulates were put to vote in June
2010. While several postulates in the proposal were approved (e.g. those pertaining to health
claims, imitation foods etc), traffic light color coding of FOP labels was rejected outright, with
the committee voting in favor of non-color coded %GDA labels. A press release issued after this
vote revealed that several factors contributed to this rejection of color coded traffic light labels.
“[traffic light color coding] was rejected by two thirds of the [European
Parliament] for several reasons. First, this color coding system was created for
ready-made meals in England and that means that it does not function for basic
food. For example, because Coke Light is made with sweeteners instead of sugar,
it would get a green light for sugar because it contains none, while natural fruit
juice with no added sugar, would get a red light because of its natural sugar
content. Also, the traffic light cannot distinguish between butter and half-fat
margarine, everything is red because it contains a lot of fat. It would give a red
light to healthy products like nuts that are full of minerals and vitamins.
These examples show that the system doesn't work for basic products. Our task is
to find a labeling system to suit to each and every food and no- alcoholic
beverage.
Another reason is that we have the UK experience where producers tend to
reformulate their products to have better traffic lights [sic]. For example they
substitute sugar with starch, so the product seems to have fewer calories. Or they
substitute sugar with sweeteners, which means artificial ingredients, or salt with
sodium glutamines, which is not really healthy. It is cheating the consumer.
[258]”

This rejection was met with dismay from both the FSA and consumer groups across the
EU, with several trade press articles attributing the intense multibillion euro lobbying by food
and beverage companies as the main reason for the downfall of MTL labels [259, 260].
A further document, The EU Draft legislation 1169/201, the first major overhaul of food
labeling in the EU, was published in November 2011. The document declared nutrition labeling
65

mandatory in the EU for the first time since the formation of the European Union. By December
2014, all products that already bear nutrition information would have to alter their nutrition
information to display six nutrients - fats, saturates, carbohydrate, sugar, protein and salt in this
exact order expressed per 100g/100ml of the product. Products with no nutrition information
displayed on them as of November 2011 will be given until December 2016 to meet this
requirement. Other major postulates included labeling of allergens and country of origin but
Front-of-Pack labeling is still voluntary [250, 261].
United States:
Following their introduction in the EU and UK, several US manufacturers began using
FOPs in the early 2000s. In early 2004, PepsiCo published a set of nutritional thresholds for
healthy foods loosely based on existing FDA Recommended Daily Intakes (RDIs) and began using
the SmartSpot®, a type of iconic summary indicator, to draw consumer attention to products
and product lines that met those guidelines. Reports published by PepsiCo in the same year
mention that the sales of the products that carried the SmartSpot(R) icon grew by 13 % and
contributed to almost 40% of the company's total sales [262].

66

Likely as a result of the success of the “Smart Spot,” in 2007, Kellogg began using an FOP
labeling system similar to the EU’s %GDA system (see Figure 5 for the original %GDA system)
with the added tag line, ‘Nutrition at a Glance,’ on its leading cereals. [263]. Snack food giant,
MARS also began including %GDA labels on the front of its candy bar packs in the US in early
2009 [264].

In 2008 CocaCola, Con Agra, General Mills, PepsiCo, Kellogg, Post and several other large
food manufacturers came together with public health experts, research institutions and
nutritionists to create the Smart Choices® FOP, another summary indicator icon. Companies that
became a part of the consortium by paying a fee, could then use the “Smart Choices” Icon on
products that met certain nutritional criteria (apart from the membership fee mentioned above,
companies had to pay the consortium to use the icon on individual products). The Smart
Choices program enjoyed a high profile start, appearing on several high-volume products but
soon succumbed to heavy criticism.
It was in the midst of this rapid proliferation of FOP symbols and systems, that the FDA
began showing significant interest in FOP nutrition labeling. During a media briefing in October
2009, two months after the suspension of the ‘Smart Choices’ program, the FDA announced
that they would be working on a Draft Guidance for FOP label design and use that would be
released in the near future. The FDA also mentioned that it was working with the IOM in
conducting an 18 month long research study to look into the same [265].

67

Another noteworthy development was a press release by the FDA and an open letter to
the industry released in March 2010 [266]. The FDA expressed its support for the practice of
FOP labeling, in general. As part of the release, Commissioner, Dr. Margaret Hamburg, indicates,
“The use of front-of-pack nutrition symbols and other claims has grown
tremendously in recent years, and it is clear to me as a working mother that such
information can be helpful to busy shoppers who are often pressed for time in
making their food selections.
I believe we now have a wonderful opportunity to make a significant
advancement in public health if we can devise a front-of-pack labeling system
that consumers can understand and use [267].”

The FDA also warned manufacturers to stay within the rules of existing acts such as the
Fair Packaging and Labeling Act and the Nutrition Labeling and Education Act while devising
their own front of pack labeling schemes and not be misleading in any way [267].
In early October 2010, the IOM published its first report on the FOP label study, a joint
venture of FDA and IOM. While backing FOP labeling strongly, the report also analyzes the
current state of FOP labeling around the world and performs a comprehensive review of existing
FOP literature [30]. The report classified FOPs into the three types referred to in Table 11 and
mentioned that a ‘summary indicator system’ that was based on the amounts of nutrients
rather than a complex algorithm was the most promising. The report concluded that along with
a Summary indicator Icon, the most critical components to include on an optimum FOP label
would be calories, saturated fat, trans fat and sodium and that there was not enough data to
determine whether consumers would benefit from having sugars displayed on the FOP [30].
In its Phase 2 report released in mid 2012, the IOM made two major points – firstly a set
of several characteristics that an ‘ideal’ FOP would have and secondly that the “implementation
68

of any new FOP system should include a multi-stakeholder, multi-faceted awareness and
promotion campaign that includes ongoing monitoring, research and evaluation [29]”
The Phase 2 - IOM report also makes a strong case for an FOP symbol system that would
“move beyond simply informing the consumers about nutrition facts [29]” and not just present
consumers with nutrition information at the point of purchase, something that the NFP had
already been doing for the past 20 years. After a review of existing FOP systems, the IOM
concluded that a simple ‘rating’ system that would award ‘points’ to products based on
nutrition content would have the maximum impact. The IOM cited the Department of Energy’s
Energy Star® program as a similar approach that has enjoyed widespread success. The IOM also
mentioned that any widespread change to nutrition labeling would have to include a multistakeholder advertising and awareness campaign. See Figure 7 for a representation of the IOM’s
example 3-point FOP format [29].

69

Figure 7 – Proposed Model IOM 3-point FOP label format. The product on the far-left does not
qualify for points while the product on the far right qualified for points and meets pre-determined
criteria for all three nutrients - saturated fat+trans fat, sodium and added sugar. Reproduced from
the IOM Phase 2 report on Front of Pack Nutrition Rating Systems and Symbols [29].
As shown in Figure 7, under the proposed system all products would be required to
display calorie and serving size information. Products that contain less than certain threshold
amounts of all three nutrients – Saturated Fat + Trans Fat, Sodium and Added Sugar would
qualify for ‘points’. Based on pre-determined criteria, these products would be awarded 1-3
points which would be highlighted on the FOP label. In order to draw attention to the NFP and
to present these nutrients in the context of the overall product formulation, these points would
70

be highlighted within the NFP as well. Readers are directed to the IOM reports for detailed
information on proposed eligibility and points criteria [29, 30].

In a move that many believe attempts to preempt the FDA’s FOP initiative, the Grocery
Manufacturers Association (GMA) released a statement two weeks after the publication of the
first IOM report [30], stating that the food industry was moving forward with its plans to include
FOP labeling on “many of the country’s most popular food and beverage products [268]”. The
press release indicated that the new ‘Facts-up-Front (FUF)’ FOP labels would go on packages as
early as 2011 [268] [269]. Since then, several products have begun using the FUF FOP label. As
shown in Figure 8 , the FUF label is very similar to another industry created label, the %GDA
label in the EU and the UK. Subsequent press releases from the GMA also indicate plans to
spend at least $50 million in a consumer education campaign to inform consumers as to how to
use these new labels [268] [270].

Figure 8 - The GMA's Facts-Up-Front (FUF) FOP label scheme [269].

71

Consumer groups like the Center for Science in the Public Interest (CPSI) have reacted
cautiously to the GMA announcement, likening it to the ‘Smart Choices’ industry initiative that
failed [271]. Conversely, the CSPI’s reaction the IOM’s proposed 3-point label has been positive,
calling the 3-point system simpler and easier to use when compared to the GMA’s FUF label.
However, the CSPI mentioned that one of the drawbacks of the IOM label was that it did not
take into consideration any positive nutrients, like vitamins and minerals, when assigning scores
to products [272]. The GMA reacted negatively to the IOM’s proposed format saying that the
FUF system provides real value and that
“Consumers have said repeatedly that they want to make their own judgments, rather
than have government tell them what they should and should not eat. That is the
guiding principle of Facts Up Front, and why we have concerns about the untested,
interpretive approach suggested by the IOM committee. [273]”
Other reactions to the GMA’s FUF initiative have been very acerbic. Marion Nestle, an
academic/author of popular public-health books indicated,
“There can only be one reason GMA and [the Food Marketing Institute] are doing
this: to head off FDA action on front-of-package labels. The Institute of Medicine
has just filed a report recommending that FOP labels list calories, sodium, transfat and saturated fat. These are all negatives…
Packaged food companies want to list positives like vitamins and fiber, so they
can advertise the products as healthy [regardless of salt, sugar and fat content].
This action is a flagrant attempt to undermine the FDA's regulatory authority
over FOP symbols, and it is telling that the announcement comes so soon after
the IOM report and gives no details on what, exactly, they plan to do…
This self-serving proposal is all the evidence anyone needs to demonstrate the
need for mandatory FOP standards, not voluntary…The FDA should not take this
lightly and should start working on mandatory standards. [271]"
Brownell and Koplan in a ‘perspective’ piece published in the New England Journal of
Medicine, called the timing of the GMA initiative ‘suspicious’ and theorized various flaws in the
GMA’s Facts-up-Front (FUF) label including but not limited to its lack of scientific basis and
72

information overload. They mention that a color-coded multiple traffic light system may help
consumers’ healthier choices more easily [26].
Additionally FOPs in general have not been met with universal acclaim in the American
academic community. Bellati and Simon argue that FOPs do not introduce any new information
and are simply repeating information that has already been on packages for a long time. They
mention that there is neither any evidence to suggest that consumers in the US are in need of
more information on packages, nor is there any evidence to suggest that labeling systems have
any role to play in obesity prevention. They argue that with FOPs, food companies may try to
promote the consumption of highly processed foods even more when the real focus should be
on educating the public about the benefits of fresh fruits, vegetables and legumes.
Another argument that is put forward in their article is that the impetus for
reformulation of products that FOPs are said to bring with them is actually dangerous and that,
in the past, food companies that launched ‘reduced fat’ offerings have actually replaced fats
with more harmful components like Omega-6 Fatty acids. The authors also point to the food
industry’s history of misleading health claims and deceptive self-regulation practices as an
argument against FOP labeling of any kind [274] .
With the GMA positive that their FUF labels would be on 70% of on-the-shelf products
by the end of 2013 and with the FDA set to receive recommendations from the IOM on their
new format soon, the stage seems set for an industry-government debate similar to the
aforementioned ‘label war’ in the EU and UK.

Peer Reviewed Literature

73

The practice of including nutrition information on the FOP is relatively new. As such, the
majority of the research into FOP labels was performed and published only after the year 2000.
Because of the intense debate surrounding FOP label formats and their use worldwide, there
has been a great demand for scientific research into the same, and a wealth of information has
been published in the last decade (2002 to 2012). Varied aspects of these labels have been
researched, from design and colors to consumer acceptance and optimal size. Provided below is
a summary of FOP label related research from various countries.
The United Kingdom (UK):
As described before, there was rapid propagation of FOP label use in the UK between
2004 and 2008. This catalyzed the FSA to commission a series of research studies, of which the
final study was published in mid 2009, a time when the debate about FOP label formats had
begun to gain international prominence.
It is generally believed that the upsurge in public health consciousness in the UK, and
research into food labeling, began with the publication of the “Choosing Health: Making
Healthy Choices easier” white paper by the UK Department of Health in early 2004. The report
dealt with several pressing public health issues (viz. smoking, sexually transmitted diseases,
drug abuse and obesity) and proposed new approaches to promote healthier lifestyles among
the populace. In this report, Traffic light labeling (sometimes referred to in the report as
Nutrient Signposting) was mentioned as an important tool to empower people into making
healthy choices. The report also called for greater collaboration between the industry and the
government in developing standards for FOP signposting and nutrition profiling of products
[275].
74

Concept testing for these new ‘signposting’ formats began in early 2005. As part of
these efforts, the FSA commissioned a total of 20 focus group sessions featuring 7- 8 consumers
in each. Consumers were spread across different life stages - Single or Married with Cohabiting
Children (18-34 years), Older Families with Cohabiting Children (22 to 44 years), Empty Nesters
(45 to 64) and Retired (65+). Along with these focus groups, four ‘mini-friendship groups’
consisting of four consumers in the age group of 16 to 18 were recruited from schools and
colleges. There was almost unanimous support for FOP labeling in general, with consumers
expressing their support for a standard system. Out of the five FOP label formats tested (see
Figure 9), none received unanimous approval or disapproval, but the FSA notes that the ‘Simple
Traffic light (see Figure 9c)’ and the ‘Key Nutrients Format (See Figure 9b)’, an early version of
the Multiple Traffic Light (MTL) label “had significantly more promise than the others [276]”.
Consumers agreed that these two formats were simple and straightforward and at the same
time capable of grabbing attention when placed on the front of a package.

75

Figure 9 - FOP label concepts tested in preliminary FSA focus groups circa 2004.
a. %GDA Label Format b. Key Nutrients Format (an early version of the Multiple Traffic Lights
Format) c. Simple Traffic Light d. Healthy Icon Format e. Extended Traffic Light Format.
Formats b and c were noted by the FSA to have significantly more promise than competing
designs
76

After this initial concept testing was concluded, four FOP label formats were designed
and carried forward (see Figure 10) for quantitative testing.

Figure 10 - Four FOP label designs quantitatively tested by the FSA following concept testing of
6 potential designs (See Figure 24) a) Multiple Traffic Light (MTL) b) Simple Traffic Lights c)
Color-Coded %GDA d) Monochrome %GDA

The results of this quantitative study were published in late 2005 [277]. The two formats
that performed well in this quantitative study (Multiple Traffic Lights and a hybrid Color Coded
77

%GDA labels, Figure 10a and Figure 10c) were retested in 16 focus groups across the UK [278].
The results of these two studies (the quantitative study and the subsequent focus group study)
are described in Table 12 below.

78

Table 12 - Food Standards Agency (FSA, UK) consumer research evaluating Front of Pack
Label Formats (see Figure 10 a,b,c,d)

FSA Quantitative Study
Method

Results

Subject interviews (n=2676, between 16 and
70 years of age) administered in England,
Scotland, North Ireland and Wales in June
2005. Five products, four bearing the FOP
label formats shown in Figure 10 and a fifth
bearing no FOP label, i.e. a control case
shown to subjects. Subject interviews
included three types of questions – product
comparisons, nutrient level estimations and
FOP label preference [277].

Performance: Averaged across age groups, in
product comparison tasks, the Multiple Traffic
Light (MTL) format received the maximum
percentage of correct answers (79%). This is
significantly higher than the other formats (α =
0.05). The simple traffic light and the Control
condition received the lowest percentage of
correct answers (37%, 38%). The Color-Coded
%GDA format produced the most correct
responses for individuals under the age of 55.
Between the ages of 55 and 70, the MTL format
had the maximum correct responses. Speed of
interpretation of label information was
significantly higher for the MTL and the ColorCoded %GDA formats (p<0.05).
Preference: 95% of consumers reported
preference for color-coded formats like the MTL
and Color-Coded %GDA formats
Additional Results: 96% of consumers reported
that they found front-of-pack labeling useful.
29% of consumers reported that would like frontof-pack labeling implemented on all food and
products. Results of consumers belonging to
lower socio-economic groups were consistent
with the average results of the total sample.

FSA Focus group study
The two FOP label formats that tested the
best both in performance and preference in
the quantitative studies (MTL and Color
Coded %GDA) taken forward and evaluated
in 16 focus group sessions with consumers
aged 18 years and over. Topics covered in
focus group sessions included the use of FOP
labeling to make purchases, positioning of
FOP labels, shortcomings in FOP design etc.

Maximum use of FOP labels was reported with
women and men who were primary food
shoppers for the house. Both formats tested as
being simple and straightforward with a high
level of support for color coding of nutrients. The
Multiple Traffic Lights format was sometimes
criticized for being too simplistic and not
carrying enough information. The Color Coded
GDA format was sometimes criticized for not
explaining serving sizes and GDAs completely.

79

After these studies were completed and published, multiple traffic light labels were
introduced into the UK market and the FSA began a large-scale media campaign to promote
them (approximately mid 2006) but as mentioned before, the debate over label formats soon
began, and the FSA commissioned another study to evaluate FOP label formats in 2009. The
report from this study titled, “Comprehension and Use of UK Nutrition signpost labeling
schemes” was published in May 2009 and opened with the lines, “This is the most
comprehensive and robust evaluation of FOP nutrition signpost labeling published to date [20].”
The tabulation below summarizes the research questions and methods used during this study
(see Table 13). On the whole, it was the hybrid label (Figure 6), i.e. an FOP label that contains all
three features – traffic light colors, %GDA values and text indicators (high, med, low) that was
recommended as the best [20].

80

Table 13 – Description of FSA's comprehensive FOP label evaluation performed in 2009
Question Do consumers use FOP labels while shopping?
Methods

Shopping bag audits (n=112) & observed shopping tasks (n=112) followed by
subject interviews

Results

FOP was not heavily used during shopping. The only times FOP labels were used
heavily was when the consumer had a medical condition that required a special
diet, or when shopping for children. Among FOP label users, label use poor when
purchasing indulgent foods (treats) and repeat purchase products. Reasons for
limited use: Low awareness of FOP labels, General mistrust of FOP labels,
difficulty in interpretation of FOP information etc.

Question Does self-reported use of FOP labels coincide with observed results?
Methods

Survey (n=2932)

Results

Self-reported use of FOP labels much higher than actual observed usage (58% of
shoppers reported using FOP labels while shopping. Consumers who were
observed during shopping tasks were not generally found to use FOP labels).
Among those that reported not using FOP labels were seniors and consumers from
minority populations.

Question Which FOP label format works best when consumers are asked to evaluate nutrient
content and overall healthiness of a single product?
Methods

Survey (n=2932), Depth Interviews (n=25, n=100). Three FOP label design
elements (Traffic Light colors, %GDA, Indicator Text) were toggled to create 9
designs

Results

When consumers were asked to evaluate the nutrient content of individual
products, the FOP label design with all three features - %GDA values, traffic light
colors and indicator text (high, med, low) had the highest percentage of correct
responses (72%). The presence of ‘Indicator text’ in an FOP combination always
led to a higher percentage of correct responses (65%+).

Question Which FOP label formats work the best when consumers are asked to make
comparisons between products?
Methods

Survey (n=2932), Interviews (n=25, n=100). Three aspects of FOP label (Traffic
Light colors, %GDA, Indicator Text) were toggled to create 9 designs

Results

Results were inconclusive and no single FOP label was revealed to be better than
the other in enabling product comparisons

Question What is the effect of multiple FOP formats in the marketplace on the consumer?
Methods

Survey (n=1602), Depth Interview (n=50)

Results

Consumers appealed for consistency in FOP label schemes. The authors mention
that the presence of monochrome %GDA labels and %GDA labels with traffic light
label overlays alongside each other in the market confuses consumers.
81

Along with the aforementioned study, the FSA also commissioned a series of Citizen’s
Forums that took place across the UK. The major results from this ‘Citizen’s Forum’ report are
listed below [279].
(i) Awareness of FOP labels was generally high, although awareness did not
translate into high levels of usage. People who used FOP labels were more
likely to be doing weekly or monthly grocery shopping rather than buying a
quick snack or treat [279].
(ii) FOP label users were also more likely to be those that ate a lot of processed
food, like ready to eat meals. Other high use groups included parents who
shopped for their children and people on special diets [279].
(iii) Consumers mentioned that each FOP label format had its own advantages
and disadvantages, mentioning that the traffic light format showed a useful
‘at-a-glance’ summary of the product [279].
(iv) FOP labels were seen “as a way to save time and [make] informed healthy
decisions [279]” and that “labels offered a convenient way to identify
healthier products [279]”.
(v) Consumers strongly encouraged the government and the industry to come
together to create a uniform FOP label system that is used on all packages.
This, they believed, would make them easier to notice, more transparent and
result in increased awareness [279].

82

The Food and Drink Federation, the main proponent of the %GDA label scheme (See
Figure 5 ) also commissions FOP label research on a regular basis. GDA labeling surveys
conducted in early 2008 revealed that out of 560 people surveyed, 90% reported using GDA
labels and 85% found them easy to understand [280]. Another research study (survey with
n=500) commissioned by the FDF reports the following about %GDAs and FOP labeling, in
general [281].
(i) Awareness of %GDA labels was generally high but lower among consumers
from low socio economic strata (highest awareness of approximately 80% in
females from high social classes but 68% among males from low socioeconomic classes). Collective usage rates were estimated to be 56% as of
2008 [281].
(ii) In 2008, 85% of consumers reported that %GDA FOP labels were easy to use.
Comprehension levels were lower among consumers from low socioeconomic classes [281].
(iii) Multiple Traffic Light (MTL) labels have awareness levels comparable to those
of the %GDA labels. Comprehension tests revealed %GDA labels to be
marginally better than MTLs [281].

In addition to the aforementioned researched by the FSA and the FDF, academic
institutions have also performed FOP research and published in peer-reviewed journals.
A UK study published in early 2007 investigated the effects of nutrition education on
primary school children aged between 5 and 7 years (n=69). Although FOP labels were not used,
83

the facilitators labeled foods as red foods, amber foods and green foods to be eaten less, in
moderation and more, respectively. The authors reported that asking behaviors for ‘red’ foods
dropped three weeks after nutrition education, but this result was not statistically significant
when compared to the asking behaviors for green foods at the 95% significance level [282].
Murphy et al. (2008) administered surveys involving parents from two schools in
different socio-economic strata (n=53 each). Following this, two groups of four to six people
were brought together in focus group discussions. The Multiple Traffic Light (MTL) label system
(see Figure 4) was preferred by more subjects for its ‘at-a-glance’ features but performance
questions in the survey indicated the %GDA format (see Figure 5) to produce significantly more
correct answers during product comparisons (p<0.01). FOP label use was not significantly
associated with socio-economic status (α=0.05), although the author mentions that people in
poorer families found FOP labels harder to use (as revealed during focus group sessions). [283].
In more recent work, published by Sacks, Swinburn and Rayner (2009) investigators
studied the sales of sandwiches (n=12 brands) and salad dressing (n=6 brands) before and after
the introduction of traffic light labels in the UK (sales data was collected 4 weeks before and
after label introduction). The researchers were not able to conclude that the introduction of
FOP’s resulted in lowered sales of products in any way (p>0.1), and attributed the results to the
short-term nature of the study and the fact that they were studying only two product categories
[237].
A survey of FOP label designs conducted by the Food Safety Authority of Ireland (FSAI)
obtained responses from Irish residents in 2009 (n=1029). 53% of the people preferred the
%GDA labeling scheme and 38% preferred traffic light color coded FOPs. Unlike the results of
84

the studies by the FSA in the UK, this survey indicated that only a small proportion of the people
(only 8%) preferred hybrid FOPs that contained both traffic light colors and %GDA values (Figure
6). The FSAI mentions in its report that there has been a rise in the proportion of consumers
who consulted nutrition information (25%; up from only 8% in 2004) [284].
In a recent study by Herpen et al.(2012), student participants from the UK (n=186) and
The Netherlands (n=197) performed a virtual shopping task. Participants purchased a pizza
product that had one of four possible FOP label treatments - a no FOP control group, a Multiple
Traffic Light (MTL) label, a simple FOP icon and a %GDA label. The authors explain that the four
treatments were chosen in order to study the effect of familiarity on FOP label use. The MTL
label would be familiar to the UK subjects but not to the Dutch subjects; the reverse would be
true for the FOP icon which has widespread use in the Netherlands. Self reported familiarity
ratings for all FOP labels were consistent with these hypotheses (p<0.01). However, actual use
of the label in order to make a healthier choice was not affected by familiarity (p>0.01). The
%GDA labels were rated as being more difficult to use when compared with the MTL label and
the FOP Icon (p<0.01) [285].
The European Union (EU):

85

Following the EU government’s food labeling proposal in early 2006, a review of
nutrition information on EU food packages was performed by Grunert et al. and published in
2007. The authors discuss FOP labeling and review studies performed up to 2002. According to
the authors, FOP label studies typically fall into one of two categories [286].
(i) Research performed in academia (Universities, Professors, Graduate Students)
(ii) Research performed by other stakeholders like the government and food manufacturers
(Usually outsourced to market research firms)

The authors suggest that both types of studies offer valuable insights into FOP labels but
Type i studies typically have smaller sample sizes and feature advanced statistical methods to
explain results; the opposite being true with Type ii studies [286].
Following the publication of this review, Grunert et al. researched the awareness and use
of GDA labels by performing consumer interviews and in-store observations in six European
Countries (n=2019 in UK, n=1858 in Sweden, n=1804 in Hungary, n=2337 in France, n=1800 in
Poland and n=1963 in Germany). Eight percent of consumers cited nutrition value as a reason
for purchasing a particular product and 16.8% of consumers reported looking for nutrition
information while making a purchase. The three main sources of on-pack nutrition information
were the %GDA label, the back of pack nutrition panel and the ingredients list. A regression
analysis was performed on this data to reveal that label awareness was significantly affected by
the ‘country’ variable with awareness of FOP labels highest in Sweden and the UK (p<0.05).
Differences between countries were attributed to several factors, both demographic and
political with the “role of nutrition in public debate” playing an important role. Self-reported
86

understanding of the %GDA label was highest in the UK (61%) and lowest in France (26.8%)
[287].
Grunert et al.’s latest study was in collaboration with the EU’s Food information Council
(EUFIC), and was published in 2010 [39]. This large scale study first involved in store
observations and interviews with consumers (n=2019). The results of this study revealed that up
to 27% of consumers actively look for and use nutrition information while shopping and front of
pack labels were found to be one of the major sources of nutrition information. Consumers
considered ‘fat content’ to be the most important piece of nutrition information and the
probability of looking for nutrition information while purchasing foods that are generally
considered to be healthy (example: yoghurt) was highest (p<0.05). Owing to the multi-store
nature of the study it was not possible to interpret on the effectiveness of different FOP label
schemes (%GDA, Traffic Lights etc). Out of the original 2,019 consumers, 921 agreed to
participate in an in-home survey of nutrition information knowledge and understanding of FOP
labels. In the results of this study, Grunert et al. classify understanding of FOP labels into two
forms [39].
(i) Conceptual Understanding – This refers to the ability to understand the
nature of the format, i.e. what the GDA’s in the %GDA labels actually mean
and what the colors in the traffic light label stand for [39].
(ii) Substantive Understanding – This refers to the ability to use FOP labels in
comparing foods and making healthy diet choices [39].

87

While using a 1-10 self-reported scale to measure conceptual understanding, the traffic
light labels and the %GDA labels rated almost the same (Averaging 6.9 and 7.0 respectively). The
authors mention that while 61% of consumers correctly interpreted GDAs as guideline daily
amount values, a majority of them overestimated the severity of the amber (57%) and red
colors (73%) in Traffic light labels. While measuring substantive understanding, a majority of
consumers (up to 92%) interpreted label information correctly and performed correct
comparisons, irrespective of which FOP label format they were using [39].
In a similar study, Feunekes et al. recruited men and women (n=1630, aged 18 to 55
years) from four European countries (UK, Germany, Italy, Netherlands) and performed a survey
about FOP nutrition labels and compared different formats based on the following descriptors –
credibility, likability, consumer friendliness, comprehension, ease of use in decision making.
Almost all labels tested (Multiple traffic light, %GDA etc; See Figures 15 to 17) had high levels of
consumer friendliness (ranging from 3.4 to 4.0 on a scale of 1 to 5). According to the results of
the survey, the MTL format was more credible and significantly more easy to comprehend than
the other formats (p<0.01). A main effect of country was also reported on FOP label
comprehension and credibility (p<0.01) [235].

A study commissioned by the EU Food Information Council (EUFIC) looked at simple
front of pack icons that contained just calorie information and eschewed all other nutrients like
fats, sugars and salts. The authors, Van-kleef et al., studied eight different formats of ‘front of
pack calorie flags’ that varied in complexity from a simple circle containing the number of
calories per 100 g of product to a more complex table containing calorie information along with
88

the percentage of daily caloric needs. They were evaluated in a total of 12 focus groups, 3 each
in UK, France, Germany and Netherlands. It was reported that the simplest format was received
better than the more complex formats (F=9.01, p<0.001) [36].
Borgmeier et al. examined consumer understanding of FOPs by conducting a pair-wise
comparison experiment with 420 subjects. Consumers were shown 28 pairs of food products
with one of five possible FOP label conditions - (i) an MTL, (ii) a %GDA label, (iii) a Choices icon,
(iv) a Hybrid label i.e. an MTL that also displays %GDA information and, (v) a no FOP control. In
each pair-wise comparison they were asked to pick out the healthier product. The average
number of correct Reponses was highest for the Multiple Traffic Light label (24.8, p<0.01). There
was no evidence of statistically significant differences in task accuracy between the %GDA and
the Hybrid label (p>0.01). The authors mention that despite these differences being statistically
significant, the actual difference was only moderate with the average number of correct pair
wise comparisons was 20.2 for the no label case, 22.8 for the %GDA label, 23.1 for the Hybrid
label and 21.5 for the choices icon [288].
Studying FOP labeling in a larger scale, Vyth et al. performed a search of all FOP related
research papers (n=31) published between 1990 and 2011 and assigned ‘quality percentage
scores’ to each of the 31 studies obtained during the search. The authors mention that these
quality scores were assigned based on established criteria by Sirriyeh et al. They discuss that the
lowest quality scores were assigned to studies that relied on simple self-reports to assess FOP
effectiveness and the highest quality scores were assigned to studies that made use of
epidemiological modeling and health outcomes in areas where FOP labels were being used. The
highest score (88%) was assigned to a study that made use of bio-markers (sodium content in
89

urine of subjects) to assess actual sodium intake as affected by FOP labels. Studies that made
use of observational methodologies (eye tracking, shopper observation) and sales data analyses
(reformulation studies, in-store promotion studies) fell in the middle ground. Almost all the 31
studies that were reviewed by Vyth et al. are described in various parts of this chapter [21].

The FLABEL (Food Labeling to Advance Better Education for Life) foundation was set up
in EU following the upsurge in interest on nutrition information. The foundation aims to conduct
scientific research into nutrition labeling in order to inform policy. Their first study was
published in 2010. It reports that out of 37,000 products audited across the 27 EU states and
Turkey, 48% of them had some form of FOP labeling on them. The %GDA format was the most
common, appearing on 25% of all products studied. Among individual products, breakfast
cereals had the highest FOP label penetration of around 70% [289].

Citing an overabundance of research that was attempting to compare different FOP
labels and establish ‘winners’, Hodgkins et al. performed a small exploratory study in the EU that
assessed the underlying consumer understanding of FOP formats. Sixty subjects from four
countries (15 each from Turkey, UK, Poland and France) viewed a set of 22 cards, each bearing
an FOP label. Subjects were asked to perform a ‘free sort’ with these cards, i.e. they were asked
to create groups of cards “so that all the cards in one group were similar to each other in some
important way and different from other groups”. Participants were asked to perform up to three
such sorts. The FOPs present on the cards were diverse, representing the sum total of FOP
formats found across the EU and the US; among them were %GDA labels, Multiple Traffic Light
90

(MTL) labels, FOP icons in a variety of formats, the Guiding Stars FOP system etc. The authors
described and performed a Multiple Scalogram Analysis, commonly used to determine which
‘constructs’ are used by consumers during free sort tasks. In this study, the construct used by
the majority of subjects for their first sort (explained as having the highest salience) was
‘information content’, i.e. Iconic systems having the least amount of information in one group,
Nutrient Claim (e.g. high in fiber) type FOP icons in a second group and a large third group
containing all the %GDA and MTL label types. The authors explain how this ‘information
content’ construct can also be described in terms of its ‘directiveness’, i.e. how iconic systems
already make a decision about healthfulness for the consumers and are ‘direct’ as opposed to
the high-information systems like the MTL and %GDA FOPs that require the consumer to make a
decision for themselves. This kind of free sort task provides some insight into consumer
evaluation and processing of FOP labels. The authors describe the need for a pan-European FOP
label that contains both high and low directiveness elements [290].
Australia:
In Australia and New Zealand, where the government’s estimate that 1 in 5 adults are
obese, reviews and reports have shown that the existing back-of-pack Nutrition Information
Panel (The NIP, their version of the USA’s NFP) is not being used to its potential and that
consumers are confused by it [291]. Similar to the UK and the EU, The MTL (Figure 4) and %GDA
(Figure 5) formats have been the two major competitors in the Australian retail market as well.
In 2008, Louie et al., from the University of Wollongong, New South Wales, reviewed
these two competitors – The Multiple Traffic light Label (MTL) and the %GDA label. While
commenting on the advantages and disadvantages of both formats, they mention that the
91

traffic light label format would be most suited for inclusion on packages in Australia. The
following tabulation (See Table 14), adapted from their report, examines the advantages and
disadvantages of both formats [233].
Table 14 - Advantages and Disadvantages of two major FOP label formats.
Reproduced from Louie et al.[233]
“Requires no calculation by the consumer to interpret
information, therefore more equitable
Easy to understand
Indication of nutrient profile at a glance
Eye-catching and immediately noticeable
Quick to interpret”
“Some potential to confuse: e.g. if two green and two
red lights appear on the same product
Does not take into account other positive nutrients
(e.g. fiber, protein)”

Advantages

“Provides more detailed, ‘factual’ information
Widely supported by the food industry”

Disadvantages

%GDA label

Advantages

Disadvantages

Multiple
Traffic Light
Label

“No guidance on relative amounts (i.e. what is ‘a
lot’/‘a little’)
Not relevant for children and adolescents
Requires consumer education to be useful
Very difficult to interpret by less educated consumers”

Also in late 2008, Maubach and Hoek presented a highly-cited study on FOP labels from
the same University, Wollongong. In this study, 294 consumers participated in a survey which
examined their evaluations of a brand of breakfast cereal that was presented to them in one of
two possible health levels bearing one of three possible label treatments - (i) A Multiple Traffic
Light Label, (ii) a Percent Daily Intake (PDI) label which is very similar to a %GDA label (iii) a no
FOP control. Participants were asked a series of questions to determine their perceived attitude
towards the cereal product. Mean ‘attitude scores’ by nutrition format and health level were
calculated and there was a significant difference between the mean attitude scores between
92

health levels only for the Multiple Traffic Light FOP condition (p<0.01) indicating a higher
influence on consumer perception of products than both the control and PDI cases [292].
The heart foundation of Australia published reports in early and late 2008 stating its
policy stance on FOP labeling. The authors mentioned that, based on their consumer survey
(n=600), no clear preference was revealed for either the traffic light label or the %GDA type
label (p>0.05). Both labels seemed to work the same for all consumers irrespective of socioeconomic status (p>0.05) [293].
A study conducted in 2009 by Kelly et al. examined the preference of Australian grocery
shoppers (n=790) for four different FOP label systems – A multiple traffic light system, a multiple
traffic light system along with an overall rating color, a %GDA scheme and a %GDA scheme with
color coding. The authors reported that participants were more likely to choose healthy foods
while using the two Traffic Light label formats than while using %GDA labels (F=9.20, p<0.001).
Perceived ease of use of FOP labels was also higher for the Traffic Light label formats (p<0.01)
Participants generally expressed support for FOP labeling and welcomed the use of a uniform
standardized label (90% positive response to question in survey) [236].
Amidst discussions in early 2013 between Australian food manufacturers and the
government regarding self-regulated vs. mandatory FOP labeling, Carter et al. performed an
audit of 365 Energy Dense Nutrient Poor (EDNP, containing more than 6g of saturated fat and/or
over 15g of sugar per serving) snacks by 48 manufacturers in the top three supermarket chains
in Australia. They specifically looked for the presence of the industry’s voluntary Daily Intake
Guide (DIG) FOP labels. These DIGs are very similar to the EU FDF’s %GDA labels and the US
GMA’s Facts-up-Front labels. Authors report that 66% of the products surveyed contained the
93

DIG label. 74% of these products contained the most minimal form of the DIG, displaying only
calories per serving information and not information on any of the other nutrients. Another
interesting finding reported was that store or generic brands were more likely to display the fullversion of the DIG than mainstream brands (p<0.001). Authors comment that while the
widespread use of the DIG label seems to add credence to the argument that the industry is
capable of self regulation, while the reluctance of a majority of manufacturers to display sugar
and saturated fat information on EDNPs “casts serious doubt over the industry’s claims of selfregulation and, if anything, points to the need for more government regulation, not less. [294]”
United States:
As suggested elsewhere in this document, the recent political climate in the USA and the
EU has resulted in a lot of interest in FOP related research.
It was mentioned in a previous section of this chapter that after the voluntary, industry
driven ‘Smart choices’ nutrient profiling scheme received its cease and desist notice, the FDA
began considering a uniform FOP label/icon. Following this, the initial ‘PHASE 1’ IOM-FDA
report on FOP labeling schemes was published in November 2010 [30].
The report examined 20 different FOP icons and schemes, several of which are
referenced in this document, to examine the merits and demerits of each. The authors
classified FOP labels into three types as shown in Table 11 [30].
Along with expressing general support for FOP labeling, the report contains several
conclusions and recommendations. These are presented below.
The authors mention that “the most useful primary purpose of front-of-package rating
systems and symbols would be to help consumers identify and select foods based on the
94

nutrients most strongly linked to public health concerns for Americans [30].” Regardless of FOP
label design, the authors conclude that calorie information and serving size are the most
important pieces of information on FOP label systems. In terms of nutrients in FOP systems –
saturated fat, trans-fat and sodium are the three most important. On the subject of sugars and
fibers as nutrients in FOP labels, the authors mention that there is currently no universally
agreed upon stance on the role of added/natural sugars in the American diet. The authors plan
that Phase 2 of the IOM research project would involve consumer testing to evaluate the best
FOP label formats for introduction onto American packages [30].
With the publication of the IOM’s report and the need for FOP related research being
made public, there were a large number of FOP related studies conducted in the USA after mid
2010.

In a comprehensive review of FOP label research, Hawley et al.cite FDA surveys that
report that 67% of consumers use FOP labels and for this reason, mention how critical it is that
a uniform and useful FOP system be implemented as soon as possible. The authors performed a
search and review of research papers published prior to February 2011 that have studied FOP
labeling (n=28), nutritional labeling and shelf and restaurant labeling. In concluding their review,
the authors conclude that the Multiple Traffic Light Label has “the highest empirical support”
and that labeling systems that feature ‘on-the-shelf’ icons also “hold promise”. Regarding the
information that should be included on FOPs, the authors recommend that FOP labels should
bear both calorie and nutrient information along with high/med/low text. Almost all the studies

95

that were reviewed by Hawley et al.on FOP labeling have been described in the various parts of
this chapter [23].

Yale researchers Roberto et al. conducted several studies and published several papers in
2012, most of them dealing with FOP label comparisons. Their first study assessed the
controversial ‘Smart choices (SC)’ labeling icon discussed earlier in this report. Participants
(n=216) took part in a breakfast cereal consumption task where they were presented with a
package of store brand cereal that either bore an SC icon or did not. The authors reported that
the presence of the SC icon (containing calories and serving size information) significantly
affected the ability of consumers to recall calorie information, with participants who saw the SC
icons on packages twice as likely to recall calorie information correctly as compared to those
who saw packages from the control group (p=0.001). However, the SC icon did not affect
perceptions of healthfulness (p=0.2), amount of cereal consumed (p=0.46) or purchase
intentions (self p=0.22, for children p=0.69) [31].
In a follow up study to the one referenced above, Roberto et al.chose 100 products that
bore the smart choices icon and applied their nutrition information to the Nutrient Profile
Model (NPM), an evaluation method used in the UK to determine which products cannot be
advertised to children on television. The authors report that 60 of the 100 products sampled
tested as being 'not healthy’. Among the product categories sampled, only 6/19 (31%) of snacks
and 5/32 (15%) of cereals were healthy. The authors, similar to their previous publication
summarized above, reacted skeptically to industry self-regulation [24].

96

Roberto et al. next performed an online study (n=480) to evaluate consumer
understanding of FOP labels. Participants were first shown a PSA about label consciousness and
then shown an instructional video on how to use one of five possible FOP label treatments - (i) a
no FOP control, (ii) a Multiple Traffic Light label, (iii) A Choices icon, (iv) A multiple traffic light
label with calories per day information and (v) a special Traffic light label that displayed only
nutrients that received the ‘red’ color coding, i.e. high level. After this instructional phase,
participants were shown 15 pairs of products each containing the FOP label treatment assigned
to them and were asked to pick out the healthier option from among the two. Participants were
more likely (p<0.001) to give correct answers in this task if their assigned FOP treatment was
the Multiple Traffic Light label with calories per day information or the Choices icon, i.e. options
(iv) and (iii) when compared to when they were in the control group (no FOPs). Providing the
most information, the three types of Traffic Light labels were also more likely (p<0.001) to result
in more accurate nutrients and calories per serving estimates. But similar to other studies,
traffic light labels resulted in reduction in perceived healthfulness when compared to the
choices icon (p<0.001) [25].
Roberto and her colleagues continued their work investigating the effectiveness of
various FOP labeling systems, in another online instrument that required a task; participants
(n=703) first viewed a PSA about label consciousness and then answered a series of questions
based on an FOP treatment they were assigned. This study focused on consumer understanding
of Traffic Light labels in comparison with the GMA’s Facts up Front (FUF) label - the five possible
FOP treatments being (i) No FOP (ii) A multiple Traffic light label (iii) an extended multiple traffic
light label that also displayed the amount of Protein and Fiber (iv) A standard Facts-up-Front
97

label and (v) An extended Facts-up-Front label that also displayed protein and fiber. After
viewing the aforementioned PSA, participants took part in a seventeen-item forced choice task,
in which they were asked to compare nutrient levels between pairs of products presented. Both
products had the same FOP Type. The non-control group participants (i.e. those whose
packages had FOPs significantly outperformed the control group participants in terms of
percent correct answers (p<0.05). The authors report that participants assigned to them the
Extended Traffic Lights treatment performed the best (>80% correct) [22]

In a study that specifically studied FOP use and perceptions in low socio-economic areas,
an in-store survey task (n=110) was administered by Marco in one of the lowest socio-economic
areas of Washington D.C. Participants were shown a box of Honey-nut Cheerios that bore a
%GDA FOP label similar to the one shown in Figure 8. Overall, 30% of consumers reported that
their purchase would be influenced by the FOP label ‘very much’ and 14% answered ‘not at all’.
There was a significant effect of age reported on this measure. 40% of consumers in the older
age group (46+ years) reported being influenced ‘very much’ by the FOP and only 18% of
younger consumers (18-45 years) being influenced on the same level. Overall, 75% of the
participants recalled seeing the FOP label before [32].

In a study focusing on beverages alone, Kim et al. conducted an online survey (n=629) in
which all participants were first shown labels of 13 different beverage products and asked to
rate their healthiness using Likert scales (0 being unhealthiest and 9 being healthiest).
Consumers rated water, 100% Fruit Juices, 100% Fruit/Vegetable blends and milks as the
98

healthiest beverages (average baseline rating= 6.7 to 8.5) and a regular soft drink (average
baseline rating: 2.3) as the least healthy beverage. Fruit cocktails and other fruit drinks received
intermediate ratings. After this ‘baseline’ rating was established, participants were shown the
same 13 different product labels (differing brands) with either a single calories per bottle FOP
icon or a detailed GMA “Facts-up-Front” FOP label and asked to rate the healthiness again.
When presented with this additional information, authors report that for those participants
who saw the calories per bottle FOP icon, the new ratings did not differ from the baseline
significantly “indicating that they did not derive any new information from the [calories per
bottle icon]”. However, healthiness ratings decreased significantly for milk and 100% juice and
increased for diet sodas for participants who saw the more detailed GMA’s “Facts-up-Front”
label (p<0.05) [33].

99

Chapter 5
Background

The previous chapter introduced FOP labels and reviewed research conducted using
varied FOP labels from the US and around the world. While it is evident from the review that
there have been a large number of FOP-related studies conducted and published over the last
two decades, a majority of them restrict themselves to concepts like understanding,
comprehension and perceived helpfulness. In doing so, they presuppose an important stage in
the flow of information, assuming that the labels will be noticed. Furthermore, the majority of
the FOP studies published thus far in the US and around the world restrict themselves to
qualitative research methods: focus groups (EU - [35] [36] , US - [37]), assessment of in store
behaviors (UK- [38] [39]), surveys (US - [32] [33] ) , depth interviews (UK - [38]). Only a very
limited portion of studies conducted used US consumers and actual packages. The vast majority
were conducted using photographs [33] or images of labels on computer screens [22, 25].
In addition to the above need to study FOP labels from an ‘attention capture’
perspective, there is also the need to study the FOP nutrition label in tandem with the other
nutrition label found on US packages, the Nutrition Facts Panel (NFP). As mentioned before,
these standard NFPs have been mandatory on almost all US food packages since the passing of
the Nutrition Labeling education Act in 1990 [156].
In essence, FOP labels simply repeat or offer summaries of the information already
present in the NFP on the front of the package (the Principal Display Panel or PDP as defined by
the Fair Packaging and Labeling Act [295]). One possibility is that during a consumer’s search for
100

nutrition information on a package, the FOP label may serve as a short-cut or ‘substitute’,
obviating the need for examination of the more complete information found in the form of the
existing NFP. Conversely, the FOP label may catalyze consumer interest in nutrition information,
and its presence may increase the amount of time spent analyzing the nutrition information.
As mentioned before, the Government has also recently become more explicit in its
recommendations for labeling. In 2010, The Whitehouse’s Task Force on Childhood Obesity
recommended the development and implementation of a “standard system of nutrition
labeling for the front of packages” based on scientific research [19]. FDA also recently
identified “the exploration of FOP nutrition labeling opportunities” as a key initiative in its
2012-2016 Strategic Plan [27] [28] and reported that FDA surveys estimate that almost 67% of
consumers use FOP labels [23]. In 2010, The Institute of Medicine was commissioned by the
Center for Disease control and prevention working under direction from the Congress to
“examine and provide recommendations regarding FOP nutrition systems and symbols [29]”
Many of these official announcements have also referenced the importance of having any
standardized system rely on rigorous, scientific work [27, 29, 30, 267].

101

Understanding the effect of FOP label design on the early stages of information
processing (i.e. attention and encodation) as well as measuring the effect that the addition of
the FOP label has on attention to and use of the already present nutrition information on the
packages (NFP) are important, and missing, pieces of information. To this end, four experiments
were conducted in an effort to objectively evaluate various Front-of-pack nutrition label designs
[40].

Background
Information Processing Models:
Before describing the four experiments and their methods in detail, introducing some
concepts and terms may be helpful. Although the ultimate goal is to demonstrate that effective
labels impact people’s nutritional choices and reduce obesity, the proximal goal is to develop a
labeling strategy that is likely to convey nutrition information to the greatest number of people.
Studies evaluating labels have frequently made use of models that organize information
transfer into several stages [296-298]. Any information that is contained on a package (e.g. a
warning statement or a health claim) must pass through these stages to be successful. A
common information processing model [299] explains that for any package information to be
effective and alter consumer behavior, it must be,
(i)

Noticed/attended by consumers

(ii)

Encoded into their memory

(iii)

Comprehended

(iv)

Acted upon (choice) [299]

102

For instance, for a health claim on a package to be effective, it must capture the
attention of a consumer as he/she walks down a supermarket aisle. Encoding of the
information in the health claim can take place only if the consumer has adequate cognitive
resources available to devote to this task. Once this encoding process has taken place, the
information must be comprehended. Only if these three stages are successfully completed, the
information in the health claim can be acted upon by the consumer.
The application of this type of model to the design of labels represents a translational
bridge between basic research on visual cognition and the applied problem of designing
effective labels.
Identifying optimal FOP design strategies that facilitate early stages in the processing
model (attention and encodation) is a necessary prerequisite to evaluating the extent to which
nutritional labeling can impact later stages (dietary choice). That is, if people fail to attend to
the FOP on a package, it would not be encoded and the nutrition information cannot have the
desired impact on people’s behavior.
Change and Change Blindness:
Experiment 1 in this study, described in detail in the next chapter makes use of a ‘flicker
task paradigm’ to evaluate the visual salience of various FOP label designs.
Even though an in-depth discussion of perception research is beyond the scope of this
study, it is important to understand that perception is a complex process. It is not simply the
transference of visual stimuli to the brain for processing, but an intricate series of steps that
frequently results in key features in our field of view not being processed (for example: failing
to recognize and react to a road sign even though it is within the field of vision) [297, 300, 301].
103

It seems easy enough to suppose that a change in our field of view is easily detectable
when conspicuous. However, the striking ability of individuals to miss large sweeping changes,
‘change blindness’ is a recognized phenomenon [302]. Even though change blindness has been
observed since the beginning of visual attention research, the modern wave of change
detection research began in the early 1990’s where change blindness was first studied in realworld scenes and objects, raising interesting questions into our understanding of how we
perceive the world around us [303-305].
Focused attention is needed to see change. Rensink explains that any change in our
visual field takes place over a background of general noise (other objects in the scene,
distractions etc). The ability of an observer to overcome change blindness and spot the change
depends on the strength of the change ‘signal’. When the strength is high, the change will stand
out amidst the noise and be spotted. Of course, there is also the corollary, that when the
background noise is increased to extreme levels and changes are introduced into the system,
only changes to those objects in the scene that demand the most attention will be detected
first [302, 304, 306, 307].
This forms the basis of what is known as a ‘flicker task’. Developed by Resink et al., a
typical flicker task involves showing subjects a loop of four images (see Figure 11).
(i)

An original (appearing for 240 milliseconds)

(ii)

An interleaving blank grey screen (appearing for 80 milliseconds)

(iii)

The original with a change (appearing for 240 milliseconds)

(iv)

An interleaving blank grey screen (appearing for 80 milliseconds)

104

When these four images are shown in a continuous loop, the change ‘signal’ is masked
by the noise generated by the interleaving grey screen. As a result, only changes to regions and
objects in the scene that demand the most focused attention will be noticed [302, 304, 306,
307]. This form of flicker task experiment has now found application in the field of packaging to
study the salience of packaging features, like labels and warnings [300, 308].
In change detection studies, a stimulus loop (flicker task) is shown on a computer screen
until the subject responds (usually by pressing a key) that they have spotted a change. Subjects
are then asked to identify the location of the change. This ensures that researchers obtain
information as to whether subjects actually detected the change or were indicating they did
merely due to visual confusion or boredom during the testing sequence. This type of
‘localization’ testing is typical and most commercially-available flicker task software allows
researchers to input change coordinates along with stimuli images [302].

105

Figure 11 - The Flicker task paradigm, a loop of 4 images - an original and a changed image with
an interleaved grey screen. On looping, the change itself will appear to flicker, giving this
paradigm its name. Text within image is not meant to be readable and is for visual reference only
Eye Tracking:
Experiment 2, also described in detail in the next chapter, uses eye tracking to perform
an objective evaluation of varied FOP nutrition label designs.
Eye tracking has long been used by those in the fields of the Psychology and
Neuroscience as a means to understand how the human visual process and the central nervous
systems work. With eye trackers reducing in cost and complexity, they are being applied in a
variety of fields, including packaging, to study attentive behaviors [309].
In a typical setup for an eye tracking experiment, two video cameras, one capturing the
field of view (often termed a scene camera) and the other capturing the eye itself (often

106

termed the eye camera) record simultaneous video feeds. Following a standard calibration
procedure in which a subject is asked to look at different points in his/her field of view subjects
are handed packages to examine. The eye tracking software uses advanced image processing
techniques to superimpose a set of cross hairs over the scene-video feed, providing up to the
moment information about the subject’s eye movements.
Eye tracking analysis involves going through these videos and defining regions of
interest known as ‘Lookzones’. The analysis software is then capable of computing many
metrics, e.g. time spent by a subject on a zone, number of visual fixations on a zone, the order
in which the zones were looked at, the time taken until a particular zone was seen etc. Due to
its ability to directly measure behavior, eye tracking has found applications in Design research
(websites, books, magazines), Neuropsychology (visual attention, perception), Marketing
(advertising, persuasive messages, supermarket layouts) and more recently, Packaging
(usability, label design, warning effectiveness). Readers are directed to reviews by Duchowski
for detailed descriptions of studies [309]. Readers are also directed to Graham et al. and a 2009
report published by the UK Food Standards Agency for detailed reviews of eye tracking in
nutrition labeling research [310] [311].
Design Features in FOP labels:
Reviewed literature has suggested that no single scheme has clearly emerged as
superior to the plethora of existing systems that resonates with all consumers in all processing
stages. Our synthesis of the literature suggested the effect of color to be a central issue of
debate.

107

Further, there was no indication in the literature that the question of whether the FOP is
a “short cut” (removing the need for the NFP), or “catalyst” (leading to more detailed
examination of the NFP) for nutrition information; a question of undoubted importance in
terms of policy decisions.
Within the field of basic visual science, it has been suggested that facial icons are
immune to the phenomenon of change blindness previously discussed. There is abundant data
suggesting that face stimuli are given extremely high attentional priority and that the
processing of facial expressions of emotion requires very few cognitive resources. Research
indicates that facial stimuli capture attention even when people are engaged in another task
[312] i.e. they are relatively immune to the phenomena of inattentional blindness [313]. Facial
expressions of emotion are readily evaluated in the near absence of attention [314-316].
Finally, the inclusion of pictorial icons in an FOP label may increase the ease with which the
information is encoded and remembered [317]. People extract the basic meaning from pictures
extremely rapidly [318] and form relatively long lasting memory of them [319], even when
people are not actively attempting to form memories of the pictures [320]. As such, we were
interested in studying the effect of facial icons on the information processing of FOPs. The final
factor of interest incorporated into the study design was the effect the presence of facial icons
had on the stages of information processing.

108

Chapter 6
Experiments
Materials
Front of Pack labels:
Conducted studies examined the effect of color (2 levels) crossed with icon (2 levels)
(see Figure 12). As such, four FOPs were developed for inclusion in the experiments:
•

Color + No Facial Icon

•

No Color + No Facial Icon

•

Color + Facial Icon

•

No Color + Facial Icon

109

Figure 12 - Traffic Light colors & facial icons were manipulated to create 4 FOP label designs
110

As seen in Figure 12 , the nutrients chosen for inclusion in the FOPs used in this study
were representative of those most commonly found in FOP labels around the world – fat,
saturated fat, salt/sodium and sugars; those that are most commonly implicated in diet-related
illness. Note that the IOM’s 3-point FOP system (in Figure 7) also features the same nutrients
i.e. Fat, salt/sodium and sugar (‘Fat’ in the 3-point system is expressed as a combination of
saturated and trans-fats). Nutrients were categorized into red, amber and green based on the
Traffic Light Label guidelines released by the Food Standards Agency [321]. Note that these
original guidelines were based 100 grams of a food but a ‘per serving’ value was calculated for
this study (see Table 15 and Figure 13). As typical around the world, calories were not colorcoded for any of the tested designs.

Table 15 - Categories for nutrients per 100 grams of product [321]. See Figure 13 for graphical
representation.
Nutrient

Amount of nutrient in category (in grams per 100 gram amount)

Fat
Saturated
Fat
Sugars

Green
Less than 3

Amber
3 to 20

Red
More than 20

Less than 1.5

1.5 to 5

More than 5

Less than 5

5 to 12.5

More than 12.5

Sodium

Less than 0.3

0.3 to 0.9

More than 0.9

111

Sodium
Sugars
Saturated Fat
Fats
0

5
10
15
20
25
Grams of nutrient per 100 grams of product

30

Figure 13 - Graphical representation of data in Table 15; nutrient categories per 100 grams of
product.

Product Types and Health levels:
Product Types:
Three product types: breakfast cereal, crackers and prepared meals were chosen for
several reasons. Breakfast cereal was chosen because of the great amount of research that has
been devoted to understanding the impact of its on-pack graphics on choice and commercial
attention [31, 292, 322]. Further, commercial cereals have a wide range of nutrient densities
ranging from the indulgent to the nutritious; this was also the case for frozen prepared meals.
Both cereals and crackers have simple, consistent and structurally similar packaging (crackers).
This reduces potential study confounds like surface area and shape. Frozen, prepared meals
were also chosen for similar reasons.
Care was taken to design packages for each product type that generally conformed to
brands available in the market while not being too suggestive of nutrient value. The brands
designed for this study were created to be free of attention-grabbing elements (like spokescharacters and overly complicated artwork). Since the analysis of health claims is beyond the
112

scope of this study, the packages designed for use in the study did not include them. Four
brands of cereal, four brands of crackers and two brands of prepared meals were created for
this study and used in each experiment.
Health Levels:
For Experiments 1, 2, and 3 each brand was created with nutritional composition at two
‘health levels’ – healthy and unhealthy. The ‘healthy’ case was defined as having three ‘green’
components in the FOP and the unhealthy case having three ‘red’ components (see Figure 14).
The fourth component within the FOP would either be green or yellow. All nutrients that were
not present in the FOP were kept the identical across brands within a product type, i.e. all
brands of breakfast cereal had the same value for Vitamin A, Polyunsaturated fat, Potassium
etc. These values were derived by averaging the nutrient values from at least ten randomly
chosen commercially-available brands of the product type.

113

Figure 14 - BranBlast, a brand of breakfast cereal created for this study represented with a Color
+ Facial Icon FOP in both health levels - healthy and unhealthy. Text within package is not
meant to be readable and is for visual reference only
The ingredients list was kept constant within a product type, i.e. all brands of breakfast
cereal had the same ingredients list and so did all brands of crackers. The ingredients for the
two prepared meals brands were not kept the same due to the vast differences in their
composition (Spaghetti and Meatballs vs. Teriyaki chicken). Calories were kept constant within
a health level for each product.

114

Note that the above descriptions of nutrients, calories and health levels only apply to
Experiments 1 through 3. A detailed labeling table was created for the packages and FOPs
generated in Experiment 4 which will be described later.
For all experiments, package blanks were generated using ESKO ArtiosCAD® and
graphics were designed using Adobe Photoshop CS3®. Package graphics were printed on a HP
Designjet 4520ps large format plotter on Economy Photo Satin paper supplied by Graphix
Universal. These printed graphics were mounted on 16 pt coated Solid Bleached Sulfate (SBS)
board donated for this study by MeadWestVaco® using a Xyron 4400 Laminator. These package
blanks were then cut using a Flatbed Kongsberg 1930 sample cutting table and pasted using
Grainger® low melt thermal glue.

Experiments
In order to test the four FOP label designs described (Figure 12) previously for attentioncapture, encoding and comprehension, four experimental studies were conceived and
conducted (Table 16). All methods were approved under IRB #10-459 by the Institutional
Review Board of Michigan State University.

115

Table 16 - Summary of experimental procedures
Information
processing step

Experiment

Method

Attention Capture

1

Change Detection Study to evaluate which FOP
label designs captures the most attention and have
most visual salience

2

Eye Tracking study to determine if presence of
FOP label leads people to more detailed nutrition
information on NFP and evaluate noticeability of
an FOP label

Encoding

3

Recall study to determine which FOP label design
has the highest ability to cause consumers to
encode and retain information about a product’s
nutrition, even when people are not attempting to
access and remember nutritional information, i.e.
evaluate the relative amount of incidental
encodation that different FOP label designs are
capable of

Comprehension

4

Sort task study to determine whether FOP labels
are noticed and used when consumers are
specifically performing nutrition comparisons
between products and if noticed, which FOP
labels enables easiest/fastest comparisons

Experiment 1 was conducted in late 2011. A new set of subjects were recruited for
Experiment 2 which was conducted in early and mid 2012. Experiments 3 and 4 were conducted
within a single testing session in late 2012 - early 2013. Note that all four experiments in this
study were limited disclosure studies, i.e. participants were not told about the true nature of
the experiment (i.e. an investigation of FOP labels until after their tasks had been completed).
Detailed breakdowns of numbers of subjects recruited and demographics will be presented at
the beginning of the Results chapter (Chapter 7).

116

Experiment 1 – Using Change detection to evaluate Attentional Priority:
Overview:
Experiment 1 utilized the change detection flicker-task method described previously to
evaluate the effect of FOP design on allocation of attention. This experiment also provided
some insight into the amount of attention directed to the NFP relative to the FOPs. There were
two working hypotheses while this experiment was being planned.
(i)

The presence of color and facial icons on the FOP would result in a statistically
higher rate of successful detections than those without and, when successfully
detected would result in faster detection times.
Changes occurring within FOP locations would result in a greater rate of successful

(ii)

detections (finding the change prior to timing out) and when detected successfully,
would be found more quickly than those occurring within the NFP.
Stimuli:
As the experiment was being designed, the IOM published its Phase 2 report on FOP
labels [29]and described the 3-point design as an example FOP system (Figure 7). As a result,
the 3-point system was incorporated into the four FOP designs already prepared for the study.
Hence, two additional FOP labels were designed for this experiment
•

Color + Checkmark

•

No Color + Checkmark
See Figure 15 for the Color + Checkmark and No Color + Checkmark labels in both

healthy and unhealthy forms.

117

Figure 15 - The IOM's 3-point labels added to Experiment 1
As described before, each trial in the flicker task consists of a loop of images – an
original, a changed image and an interleaving gray screen (Figure 11). In Experiment 1, the
images in the flicker task trials were of the flattened front and right side panels of each package
(i.e. front panel including the traditional NFP; see Figure 16a for example images and the three
brands used for this study). Each subject took part in a change detection flicker task that
consisted of 216 such trials. Within each of the tested brands, there were 12 critical “FOP trials”
(12 => 6 types of FOPs x 2 Health levels, unhealthy and healthy) in which the entire FOP label
disappeared, and 12 critical “NFP trials” in which the change occurred on the traditional NFP.
The surface area of the FOP and the changes occurring in the NFP were the same area in order
to minimize any effect of size. Within each brand, 48 filler trials were also conducted in which
the changes occurred to non-nutritional aspects of the packaging (e.g. brand name, graphics).

118

The filler trials served to keep participants from preferentially attending to nutritional
information. The distribution of trials on a flattened image is shown in Figure 16b below.
As such, each subject participated in 216 trials => 3 brands x (12 critical FOP trials + 12
critical NFP trials + 48 filler trials). See Figure 16a for the three brands used in this experiment
and Figure 16b for the distribution of trials on a flattened stimulus image.

119

Figure 16 – a. The Three Brands used in this Change Detection Experiment b. Example Change
detection image divided into 10 sectors. Per brand, The FOP and NFP sectors featured 12 critical
changes each. The filler sectors featured 12 changes each to non-critical aspects of packaging
(brand names, graphics etc). Spread over the three brands this resulted in a total of 72 * 3 = 216
trials per subject. Text within package images is not meant to be readable and is for visual
reference only

The presentation of stimuli for this change detection study used a custom module
written using a commercially available software package, EPrime®[323]. Researchers prepared
120

image sets (stimuli for encodation tasks, original and changed images for the flicker tasks) using
Adobe Photoshop CS3® and input said images and change coordinates to EPrime®.
Recruiting:
Power and sample size calculations for Experiment 1 used the concept of effect size as
proposed by Cohen J [324], and further developed by Stroup [325] and Tempelman [326].
These made use of previous data [308], which indicated reasonable differences between
extreme treatments of 2-4 seconds. A sample size of ~50 subjects was determined to grant 80%
power to detect a true effect size of ∆ = 1 .4 at a Type I error rate of 5% [327].
A total of 63 participants were recruited from the East and South Lansing communities.
Recruitment was performed through email announcements and printed flyers. Testing for this
experiment was carried out in three locations – the Packaging building (Michigan State
University), the Department of Psychology (Michigan State University) and the Southside
Community Center in South Lansing, Michigan.
The flyers and email announcements mentioned three screening criteria. Participants
had to be above 18 years old, not be legally blind and not have any history of seizure. Subjects
with any history of seizure were screened.
Procedure:
When participants arrived at the testing center, informed consent was first obtained
during which the screening criteria (participant should above 18 years old, not have history of
seizure and not be legally blind) for this experiment was reiterated. Participants were seated at
a table with a laptop that was loaded with the change detection module and images. The

121

laptops were set to a screen resolution of 1024x768. The following instructions were first given
to the subjects.
“You will see two images separated by a brief blank. The images are identical
except for one change. Your task is to detect the change. As soon as you see the
change, press the space bar. Then the cursor will appear. Use the mouse to click
on the location where the change occurred. The task is timed until you hit the
space bar. Using the mouse to indicate the change location is not timed. If you
fail to find the change within 18 seconds the trial will time out.”

After a set of six practice trials featuring packages that were not used in the main
experiment and not containing FOPs, participants were asked if they were comfortable with
performing the point and click operation with the mouse on their own. If they indicated that
they were not familiar with the mouse, they were asked to simply point by hand and a
researcher was assigned to help them with the point and click operation.
The main experiment consisting of 216 change detection trials was then administered.
After completing the task, all participants were requested to complete a brief Demographic
survey and a Brief Block Food Frequency Questionnaire (BFFQ), a standardized tool providing
information about their diet [328]. Subject height and weight were recorded using a digital
scale and standard stadio-meter.
Experiment 2 – Using eye tracking to evaluate attentional priority and label use:
Overview:
In Experiment 2, eye gaze position, a converging measure of attention, was monitored
while participants interacted with novel, product packages manufactured for use in this
research. The color/facial icon FOP label was the focus of this experiment and the only FOP
design tested. Two hypotheses drove the experimental design.
122

(i)

When present, the FOP label will be noticed

(ii)

The inclusion of an FOP label increases the likelihood of people attending to
nutritional information.

(iii)

The presence of an FOP label will encourage people to access the more detailed
nutritional information presented in the traditional NFP [329] [330].

From an eye tracking perspective the above two hypotheses would mean that the NFP
label would receive a higher number of hits and have a higher probability of being noticed
when the FOP label is present on the package. Additionally, the time to first fixation on
nutrition information will be lower when the FOP label is present on the package due to its high
noticeability. The Applied Science Laboratories (ASL) 501® Head Mounted Optics eye tracking
system used in this study employs head-mounted optics. As described before, the eye camera
and scene camera capture simultaneous video feeds and use advanced image processing to
determine fixations and saccades [331]. Readers are directed to the author’s masters’ thesis
and several publications from the Packaging HUB research group for a detailed description of
this system’s construction, calibration set up and system employed to obtain accurate tracking
while interacting with three dimensional objects (e.g. packaging) [332] [333].
The GazeTracker® analysis software allows eye tracking video recordings to be broken
into regions of interest known as ‘look zones’. Various look zone metrics can be obtained
subsequent to date collection (total time spent on a zone, number of visual hits for a zone,
order that the zones appear in the gaze trail, time taken to first fixation etc.).
Stimuli:
123

Each participant examined 8 packages corresponding to a 3-way factorial experimental
design i.e. 8 packages => 2 FOP conditions (Color + Facial Icon vs. Control, NFP only) x 2 health
levels (healthy vs. unhealthy) x 2 products (cereal vs. crackers).
Four novel brands of crackers and four novel brands of cereal were created for this
study. Treatments were assigned to brands randomly and counterbalanced between subjects
See Figure 17 for an example 8 package set featuring the 8 brands that were designed for this
experiment.

124

Figure 17 – An example 8 package set that a participant viewed in Experiment 2 (Eye Tracking)
8 packages => 2 Product Types (cereal vs. crackers) x 2 FOP Types (Color + Facial icon vs. NFP
only) x 2 Health Levels (healthy vs. unhealthy). The assignment of brands to FOP type and
Health level were randomized and counterbalanced across subjects. Note that participants
examined actual packages. Text within package images is not meant to be readable and is for
visual reference only

125

Recruiting:
Power calculations performed during the planning stages indicated that a sample size of
55 subjects would be adequate to detect a conservative effect size of ∆ = 1.2 with 80% power
at a 5% Type I error rate. This effect size was based on anticipated treatment differences
(between 3 and 5 seconds) and variance components estimated from previous work by the
Packaging HUB Research Group here at the Michigan State University School of Packaging.
[333].
74 subjects, those who had not participated in Experiment 1, were recruited for
Experiment 2. Recruitment was again performed using IRB approved email announcements and
printed flyers (see Appendix 2). 75% of participants in this experiment (55 out of 74) were
recruited through the local Supplemental Nutrition Assistance Program (SNAP), Expanded Food
and Nutrition Education Program (EFNEP) and Women Infants and Children (WiC) programs.
25% of Participants (19 out of the total 74) were recruited using listservs and word of mouth
advertising at the Michigan State University. Testing was carried out at two locations – the
packaging building (Michigan State University) and the MSU Extension Offices at the Ingham
County Health Department in Lansing, MI.
Screening Criteria for this experiment included that participants had to be above 18
years old and not legally blind. Participants were also screened out if they wore hard contact
lenses because of the inability of the ASL 501 eye tracker to track wearers of hard contact
lenses with optimum accuracy.
Procedure:

126

After obtaining informed consent, participants were seated at a table fitted with a chin
rest and instrumented with ASL 501 optics. After adjusting chair and chin rest height, a custom
made plywood and acrylic pane was placed in front of them on the table and participants were
asked to place their chin on the rest and move as little as possible (See Figure 18).

Figure 18 – The custom built plywood and acrylic pane used for running Experiment 2

After a standard calibration procedure, participants were given the following instructions
“We are interested in how people perceive box labels. We’re going to show you
the packages for some new crackers and cereals. Please look at these packages
as if you were thinking of buying them. You will see 8 packages. Please continue
to look at each until I tell you to stop. You may look at any side of the package,
but please press the package against the plastic pane while viewing so we can
accurately track your eyes.”

Following this, packages were presented to subjects one at a time in random order and
participants viewed them for 20 seconds each (as timed by a stop watch). During the viewing
127

period, participants were free to manipulate the package in any way they wished. Similar to
Experiment 1, after completing the task, all participants were requested to complete a brief
demographic survey, the Brief Block Food Frequency Questionnaire (BFFQ) and their heights
and weights were recorded by nutrition students using an electronic scale and stadiometer.
Participants were then debriefed and incentives distributed.
Experiment 3 – Using recall and recognition tasks to evaluate incidental encoding:
Overview:
Ideally, a label that is designed to increase attention to nutritional information will also
increase the amount of information that people encode and retain about a product’s nutrition,
even when people are not attempting to access and remember nutritional information. This is
particularly important given evidence that most consumers do not explicitly seek nutritional
information when making their purchasing decisions [286, 334]. Thus, the ability of FOP designs
to inspire encodation of nutritional information into memory when the participant’s goal is not
explicitly related to nutritional information (i.e. better incidental memory) is worth
investigating.
During Experiment 3 participants interacted with packages under the guise of rating
their aesthetics. A post interaction questionnaire that consisted of binary forced-choice tasks
(using only the product name) was used to determine if any incidental encoding of nutrition
information had occurred during the interaction. The two working hypotheses for this
experiment were:
(i)

There will be more incidental encodation of nutrition information for packages
that contain FOPs relative to those packages that contain only traditional NFPs
128

(ii)

Color coding and presence of facial icons in the FOPs will increase the amount of
incidental encoding of nutrition information.

Stimuli:
Each subject interacted with 6 packages designed specifically for this experiment. The
six packages comprised three product types (cereal, crackers and prepared meals) across two
health levels. Across subjects, these 3 food types were combined with 3 FOP label formats
(Color + Facial Icon, No Color + No Facial Icon, and the NFP only control). Each subject was
exposed to all 3 food types and all 3 FOP types, but only 3 of the 9 possible food-type-by-labelformat combinations. For a given food type (e.g. crackers), the two packages presented to
subjects were composed of two different brands and differed by health level (one healthy, one
unhealthy). Thus, 6 packages were viewed by each participant, in three different pairs and each
pair was a given product type (cereal, prepared meals and crackers) with each member of the
pair at a different health level (one healthy, one unhealthy). Each of the three product types
were presented in one of three FOP label types (Color + Facial icon, No Color + No Facial icon
and NFP only Control), with both products within the group having the same label (at differing
levels of health).
Note that only two types out of the four total FOPs: Color + Facial icon and the No Color
+ No Facial icon were studied during this experiment. This ‘Replicated Latin Square design’ was
well-suited for this research question as it allowed for solid inference on effect of nutritional
label formats across a variety of food types while limiting the number of packages that a
participant had to remember. If all four FOP types were tested in one experiment, it would
result in participants interacting with 10 packages instead of 6. This approach would have likely
129

resulted in memory overloading and increased interference effects. It was proposed that if
there was a strong evidence of enhanced incidental encoding with the presence of facial icons
and color-coding i.e. with the Color + Facial Icon label, this experiment would be replicated with
a new set of treatments, namely Color + No Facial icon, Color + Facial Icon and No Color + Facial
Icon. Thus, this proposed two-stage approach kept memory capacity demands and interference
effects low while allowing an in-depth investigation of additive and/or interaction effects of
facial icons and color coding.
Recruiting:
Power calculations performed before the experiment indicated that a total of
approximately 80 subjects would provide 80% power to detect a true effect of nutrition label
format on the probability of a correct choice in the forced-choice-task at a Type I error rate of
5%.
As mentioned before, a total of 100 participants were recruited for Experiments 3 & 4
together. Recruitment was performed using IRB approved flyers and email announcements.
Participants in this experiment were recruited from two sources – (i) the local Supplemental
Nutrition Assistance Program-Education Program (SNAP-ED), Expanded Food and Nutrition
Education Program (EFNEP) and Women Infants and Children (WiC) programs, (ii) The Family
Resources Center mailing list at Michigan State University.
Testing was performed at two locations – the Packaging building (Michigan State
University) and the MSU Extension Offices at the Ingham County Health Department at Lansing,
MI.

130

Screening criteria for this experiment included that participants had to be above 18
years old and not legally blind. Participants were also verbally asked to bring any prescription
eye wear that they generally used to read information on packages.
Procedure:
After providing informed consent, participants were seated at a table and given the
following instructions
“We are interested in how people perceive box labels. We’re going to show you
the packages for some new crackers, cereals and prepared meals. Please look at
these packages as if you were thinking of buying them. You will see 6 packages.
Please continue to look at each until I tell you to stop. You may look at any side
of the packages we show you.”

Participants were handed the three pairs (i.e. 2 cereals, 2 crackers, 2 prepared meals) of
packages in random order and were given 20 seconds to examine each package. After this task
was completed, a laptop computer was placed in front of them. The two brand names in the
breakfast cereal product type were displayed on the screen and participants were told to
identify the healthier among the two brands based on the packages they had just examined, i.e.
a binary forced-choice task. As explained at the beginning of this chapter, the design of
nutritional labels for this study specified precise "health levels" for each product to allow for
such comparisons. The responses to these forced-choice tasks were recorded. After completing
the above incidental encoding task (Experiment 3), participants performed the packaging sort
task (Experiment 4).
Experiment 4 – Using sort tasks to evaluate FOP ease of use:
Overview:
131

This experiment investigated whether participants noticed and used FOP labels when
their explicit goal was to make nutrition comparisons between packages. Participants were
given a series of speeded, nutrient-based sort tasks where they were asked to sort 4 packages
from lowest to highest based on specific nutrients.
Two working hypotheses for this experiment were
(i)

FOPs will be noticed and used for nutritional comparisons

(ii)

For those participants who notice and use the FOPs for sort tasks, sorts will be
fastest for tasks that involve packages bearing FOPs that have color-coding and
facial icons) as opposed to those that do not

Stimuli:
In this experiment each participant performed a total of 25 trials in a factorial
experimental design=> 5 FOP formats (Color + Facial icon vs. Color + No Facial icon vs. No Color
+ Facial icon vs. No Color + No Facial icon vs. an NFP only control) x 5 nutrient dimensions (Fat
vs. Saturated Fat vs. Salt/Sodium vs. Sugars vs. Protein). The five nutrient dimensions represent
the four nutrients found in each FOP label, plus Protein, a nutrient found only on the NFP. Each
sort task consisted of one FOP label format/nutrient dimension combination.
A nutrition composition scheme was specifically generated for this experiment. As
shown in Table 17, for the four FOP nutrients, each brand’s FOP label had a ‘high’ amount of
one nutrient, ‘low’ amount of one nutrient and ‘medium’ amounts of two nutrients, i.e. one
“red” component, one “green” component and two “yellow” components in the color coded
FOP label types. This was also true for each nutrient across the four brands as shown.

132

Table 17 – Labeling scheme generated for Experiment 4. For the four FOP nutrients, each
brand’s FOP label has a ‘high’ amount of one nutrient, ‘low’ amount of one nutrient and
‘medium’ amounts of two nutrients, i.e. resulting in one red panel, one green panel and two
yellow panels in the color coded FOP label types. This was also true for each nutrient across the
four brands

Fat

Sat Fat

Sugar

Salt

Protein

Brand A

2

1.8

16

0.4

8

Brand B

7

6

10.5

0.1

4

Brand C

13

1.5

3

1.2

12

Brand D

22

0

6

0.8

2.5

There are currently no established high/med/low criteria for protein and amounts for
protein for each brand were chosen arbitrarily. The calorie values and values of all other
nutrients on the NFP were kept the same for all brands, as were the ingredients lists. The use of
this labeling scheme allowed each sort task irrespective of FOP-nutrient dimension combination
to have a precise and unique response.
Procedure:
After subjects provided informed consent and completed the incidental encoding task
and questionnaire (Experiment 3), participants were taken to a table with the sorting station.
To administer the sort tasks, a tray-and-slot apparatus that monitored package
locations, sort times and accuracy of responses via Radio Frequency Identification (RfID) was
designed. The setup was constructed from high-strength double-wall corrugated board and
consisted of a platform and four slots labeled ‘lowest’, ‘second lowest, ‘third lowest’ and
‘highest’ as shown in the photograph (see Figure 19).

133

Figure 19 - RfID sorting setup designed for Experiment 4. Text within image is not mean to be
readable and is for visual reference only

Each slot contained an RF antenna (contained within white folding cartons as shown in
Figure 19) that was connected to an Impinj Speedway Revolution RF Reader. The Reader was in
turn connected to a laptop computer running the Impinj® Multireader RFID software. All
packages used in the experiment were given a unique RF Tag. The RF Tags used in this
experiment were manufactured by LOWRY and resembled small mailing labels. These tags were
invisible to the participants and were affixed on the inside of the packages. The trays used to
present package sets to participants for sort tasks were also equipped with these RF tags; this
enabled precise start times to be recorded for each participant.
Participants were given the following instructions before the experiment began.

134

“We are going to ask you to sort packages from lowest to highest based on a
specific nutrient. For example, if I asked you to sort based on fat, you would put
the one with the lowest fat here, 2nd lowest here, third lowest here, and the
highest fat here. Each of these packages has nutritional information that will tell
you how much of each nutrient a product has. Use the packages’ information to
do the sort. Please do these sorts as quickly as possible. When you’re happy
with your sort please say “DONE”.
On other trials I might ask you to sort on a different nutrient, like saturate fat,
sugar, salt/sodium, or protein. So if I asked you to sort on protein, you would put
the product with the lowest protein here, 2nd lowest here, 3rd lowest here, and
highest protein here. Again please do these sorts as quickly as possible while
trying to sort them in the correct order, and say “DONE” when finished sorting.
Let’s do a practice trial to see if you have any questions?”

A practice trial was first administered and any questions that participants had were
answered. After the practice trial the main 25-trial experiment was begun. Prior to each sort
trial a tray was loaded with the packages to be sorted. The order of trials was randomized using
EPrime® as was the order in which trays were loaded with the four packages needed for each
trial. The participant was told which nutrient dimension to sort the four packages by (eg:
“please sort these four packages from lowest to highest by fat”) and the tray containing the
packages was dropped onto the platform (see Figure 19). The tray’s RFID tag was recognized by
one of the RF antennae and a timestamp was recorded (this was the start time). The participant
then examined each package as he or she pleased and performed the sort, dropping the
packages into the appropriate slots. The RF antenna in each slot was used to record final
package positions and package-drop time stamps for each trial.
A windows application written in Python 3.1 was used to tabulate responses for each
trial and calculate time taken for each trial from the timestamps obtained.

135

After finishing the 25 sort tasks (administered in random order), all participants were
requested to complete a brief demographic survey, and the Brief Block Food Frequency
Questionnaire (BFFQ) that obtained information about their diet and with their consent, and
had their height and weight recorded using a digital scale and standard stadio-meter.
Subject Characterization & Demographics:
As mentioned before, all participants were asked to complete the Block Food
Frequency Questionnaire (BFFQ) [335] to assess usual food intake. It is envisioned that this
detailed information will be used for future analyses.
The weight and height of each participant was measured discreetly by a trained
researcher in a private/screened area following standard procedures using a digital scale and
standard stadiometer [336]. Using these measures, the Body mass index (BMI) for each subject
was calculated to determine weight status based on the CDC classification of underweight
≤18.5, normal = 18.5-24.9, overweight 25-29.9 and obese >30 [8] . Other data collected
included physician diagnosed diet-related diseases of self and family, as well as general
demographics (age, gender, income, education level, size of household, number of children).
Participant education was categorized into two levels ‘high school or less’ and ‘more
than high school’ for ease of statistical inference. Participant annual income was also
categorized into two levels – ‘Less than 20,000$’ and ’20,000$ or more’.
In addition to the responding to the above surveys, for Experiments 2 and 3/4, each
subject was also tested for visual acuity using a Near-point visual acuity card (Dow Corning
Opthalmics®) and for risk of red-green color blindness using a set of pseudo-isochromatic color

136

plates (Richmond Products®). All demographics and subject characterization data are presented
and described in detail in the next chapter.

137

Chapter 7
Results
Experiment 1
Subjects:
As mentioned before, a total of 63 participants were recruited from the East and South
Lansing communities for Experiment 1. Five participants chose to discontinue participation
before they had completed the experiment and data from two participants was rendered
unusable because of a computer error. One participant did not take part in the task because
he/she had a history of seizure. Thus, usable data was obtained from 55 participants.
Participants whose data were usable were between 18 and 74 years old (M=31.8±16.25
years). Three out of the 55 usable participants chose not to disclose their ages and are not
included in the above average. Among the usable participants, 38 were female and 17 were
male. The demographic information obtained from subjects was thoroughly characterized (see
Figure 20).

138

Exp 1 Subjects - Age Distribution
30
25
20
15
10
5
0
18 to 25
years

26 to 35
years

36 to 45
years

46 to 55
years

56 to 65 65 years Did not
years
and over Disclose

Exp 1 Subjects - Gender

Male
Female

Exp 1 Subjects - Education Level

High School or less
More than High School
Exp 1 Subjects - Income Level

<20,000$
>20,000$
Did Not Disclose

Figure 20 - Demographic Information for 55 usable participants of Experiment 1

139

Data Processing:
Two response variables were obtained for this experiment from the EPrime® software
suite for each change detection trial.
(i)

Binary Variable: Successful detection of change (Yes/No) prior to timing out at 18
seconds

(ii)

Continuous Variable: Time to successful change detection (milliseconds) prior to
timing out at 18 seconds.

Results:
Binary Variable – Change Detected (Yes/No):
A generalized linear mixed model was fitted to this binary variable - change detected
(yes/no or timeout at 18 seconds) using a Bernoulli distribution and a logit-link function to
model the probability of change detection (in %). Only critical trials were analyzed i.e. changes
to the FOP and NFP sectors (See Figure 16b for distribution critical and distracter trials). Linear
predictors in this model were FOP Type (Color + Facial Icon vs. No Color + Facial Icon vs. Color +
No Facial Icon vs. No Color + No Facial Icon vs. Color + Checkmark vs. No Color + Checkmark, see
Figure 10 and Figure 15), Health Level (healthy vs. unhealthy, see Figure 14), Location of change
(FOP vs. NFP) and all possible 2-way and 3-way interactions.
From the demographic information collected, only age and education level (high school
or less vs. more than high school) was retained in the final model based on their Type III pvalues (α=0.05). Additional random blocking effects of Brand Name (three unique brands were
designed specifically for this experiment see Figure 16a) and computer (experiment was run six

140

different computer workstations) used were also removed from the final model because they
provided no evidence of significance.
The model was fitted using the GLIMMIX procedure of SAS (Version 9.2, SAS Institute,
Cary, NC). Relevant pair-wise comparisons were conducted using either Tukey-Kramer or
Bonferroni adjustments to avoid inflation of Type I error rate due to multiple comparisons.
Estimated least square means and 95% confidence intervals are reported below.
There was no evidence for a main effect of FOP type (p=0.46) or health level (p=0.52) on
the probability of change detection, suggesting that the tested designs did not impact the
likelihood of whether a subject would detect a change prior to timing out.
However, there was a strong main effect of the location of the change on the probability
of change detection (p<0.0001), with as shown in Figure 21, participants more likely to detect a
change to the FOP label (LSM=99.16%, SEM=0.29%) than those in the NFP label (LSM=85.45%,
SEM=4.02%).

141

Probability of Change Detection (%)

100
95
90
85
80
75
70
FOP

NFP
Location of Change

Figure 21 - Effect of Location of Change on Probability of Change detection. Participants were
more likely to detect changes to the FOP than changes to the NFP (p<0.0001)

There was also evidence for an effect of participant ‘education level’ on the probability
of change detection (p=0.0099). Regardless of FOP Type, Health level and Location of change,
participants who fell in the ‘more than high school’ category were more likely to detect a
change (LSM=96.82%, SEM=1.48%) than participants who fell in the ‘high school or less’
category (LSM=86.06%, SEM=4.16%).
Age was also indicated to significantly impact whether the probability of change
detection (p<0.0001) with older subjects on average, less likely to detect changes, regardless of
FOP type and Location of change.
Continuous Variable – Time to detect change (milliseconds):
For critical changes that were successfully detected prior to timing out at 18 seconds, a
second variable, "time to detect change," was recorded in milliseconds. In order to meet
142

necessary model assumptions, it was expressed in the log scale. Similar to the analysis for the
previous variable, linear predictors in this model were FOP Type (Color + Facial Icon vs. No Color
+ Facial Icon vs. Color + No Facial Icon vs. No Color + No Facial Icon vs. Color + Checkmark vs. No
Color + Checkmark), health level (healthy vs. unhealthy), location of change (FOP vs. NFP) and
all possible 2-way and 3-way interactions.
Again, only age and education level (high school or less vs. more than high school) were
retained out of all the demographic factors and possible covariates that were collected during
the experiment based on Type III p-values (α=0.05).
The model was fitted using the GLIMMIX procedure of SAS (Version 9.2, SAS Institute,
Cary, NC). Estimated least square means (M) and corresponding 95% confidence intervals (LCL
and UCL) are reported in the original millisecond scale. Relevant pair wise comparisons were
conducted using either Tukey-Kramer or Bonferroni adjustments to avoid inflation of Type I
error rate due to multiple comparisons.
It was estimated that changes that occurred to the FOP location were detected
consistently faster (M=2865.70 ms, LCL=2556.13, UCL=3212.76) than those that occurred in the
NFP location (M=4614.30, LCL=4111.73, UCL=5178.30) (p<0.0001). It was estimated that
participants took 61% longer to detect changes to the NFP location than the FOP location. This
is shown in Figure 22 below.

143

Estimated Time to detect change
(ms)

6000
5000
4000
3000
2000
1000
0
FOP

NFP
Location of Change

Figure 22 - The effect of Location of Change (p<0.0001) on Time to detect change. It is
estimated that participants took 61% longer time to detect changes to the NFP Location when
compared to the FOP Location
There was also evidence of a 2-way interaction between FOP type and location of
change on the ‘time to detect change’ response (p<0.0001). This was to be expected, as the
comparisons were not parallel. Changes that occurred at the FOP location were the actual
disappearance of that design of label, while those that occurred at the NFP represented the
disappearance of an area of that label that was equal in size to the FOP. We would not expect
the design in the FOP location to appreciable influence the viewer’s ability to detect the
disappearance of information occurring in the NFP. By contrast, we do expect the design of the
FOP to influence time to detect changes that happened there, and it did.
When only changes to the FOP location were considered, there was evidence for a
strong effect of color on time to change detection (p<0.0001). Participants took significantly
144

less time to detect changes to the color-coded FOPs i.e. Color + No Facial Icon, Color + Facial
Icon and Color + Checkmark when compared to their no-color counterparts i.e. No Color + No
Facial Icon, No Color + Facial Icon and No Color + Checkmark (α=0.05). Corresponding estimated
least square means and pair-wise comparison p-values are shown in Figure 23 below.

145

Figure 23 - The effect of Color (p<0.0001) on the estimated time to detect change. There was
evidence at the 5% statistical significance level to conclude that the non color coded FOPs took
non-color
longer to detect than their color coded counterparts. P
P-values for pair-wise comparisons between
wise
color coded and non-color coded FOP Type pairs are shown above each pair of bars. Alphabets
color
bars
indicate statistically significant pair wise comaparisons between all bars at the 5% significance
pair-wise
level
Regardless of FOP type or location of change, health level had an effect on the time to
detect change (p=0.0176). Unhealthy labels were detected faster ( =3549.68 ms, LCL=3164.9
(M=3549.68
ms, UCL=3981.23 ms) than healthy labels (M=3725.18 ms, LCL=3321.46 ms, UCL=4177.97 ms).
.23
(M=3725.18
Note that despite there being a statistically significant difference here, the actual difference is
te
less than 200 milliseconds, of little practical significance.
146

After accounting for the effects of FOP type, location of change and health level, an
effect of education level on time to successful detection was evident (p=0.0034). Participants
who fell into the ‘high school or less’ category took significantly longer to detect changes
(M=4130.17 ms, LCL=3634.21 ms, UCL=4693.8 ms) when compared to participants who had
gone beyond high school (M=3201.61 ms, LCL=2770.41 ms, UCL=3699 ms).
A significant negative association was revealed between age and the time to change
detection (p<0.0001); with the time to detect change significantly increasing with an increase in
age.
Discussion:
Although it has been presumed that FOP labels are superior to a more traditional
approach to nutritional labeling, as seen from the review of the literature, a large portion of
what is known is derived from methods that have the potential to influence findings. A
majority of studies that investigate FOP label efficacy rely on self-reports[337], surveys [32] [33]
and focus groups[37].
Findings from this study have the potential to impact policy debates that have occurred
throughout the world[247, 261]. Both of the dependent variables of interest, probability of
successful detection and the time to successful detection were significantly impacted by the
location of change i.e. FOP vs. NFP (α=0.05). The use of change detection requires focal
attention, filling an important gap in our understanding regarding the FOP’s saliency in the
attentional scan paths of viewers. It has long been presumed that nutrition information placed
in FOP locations were superior to those placed in more traditional location with regard to

147

garnering attention, but we are the first to objectively evaluate this to the best of the author’s
knowledge.
Participants were not only more likely to detect a change to the FOP than a change to
the NFP, (M=99.16%, SEM=0.29% and M=85.45%, SEM=4.02% respectively, See Figure 21), but
there was also evidence to show that the estimated time to detect a change in the NFP was
significantly (p<0.0001) higher than the estimated time to detect changes that occurred in the
FOP (M=4614.30ms, LCL=4111.73ms, UCL=5178.30ms and M=2865ms, LCL=2556.13ms,
UCL=3212.76ms respectively, See Figure 22) This suggests that people may be preferentially
attending information presented on the PDP as compared to the NFP location. Further, it
should be noted that this is a conservative estimate as we have presupposed exposure to the
NFP information by “flattening” the graphics for this experiment (see Figure 16).
Another issue of debate that has occurred throughout the world centers on issues of
optimal FOP design [246, 247, 251, 268]. Although much of this debate is beyond the scope of
our study (e.g. what “cutoffs” are most appropriate for color coding, or that nutrient empty
foods can be “green-lighted” at the same time more nutrient rich foods receive reds [256,
337]), data presented here suggests color to be a significant factor in garnering consumer
attention. Changes to colored FOPs were detected significantly faster than their non-color
coded counterparts (α=0.05) (see Figure 23 for pairwise comparisons).
To further inform the debate regarding optimal FOP design, we were also interested in
the effect the presence (or absence) of an icon will have on attention. Specifically, we
hypothesized that facial icons would enhance attention. As mentioned previously, published
data from the field of perceptual psychology suggests that faces are given high attentional
148

priority, require few cognitive resources and are immune to inattentional blindness and
attentional blink [313, 338, 339]. However, no evidence for an effect of facial icon was
indicated when time to detect change were analyzed. That said, it is important to realize that
this experiment only evaluated early stages of the information processing model [340] and it is
quite possible that incorporation of icons would assist consumers in later stages of information
processing (encodation and comprehension).
The effect of demographic information was also incorporated into the analysis.
Participants who indicated that they had received education beyond high school had a higher
probability (p=0.0099) of detecting changes when compared to their counterparts who had an
education level of high school or less (M=96.82%, SEM=1.48% and M=86.06%, SEM=4.16%
respectively). They also successfully detected changes significantly faster (p=0.0034) than their
less educated counterparts (M=3201.61 ms, LCL=2770.41 ms, UCL=3699 m and M=4130.17 ms,
LCL=3634.21 ms, UCL=4693.8 ms respectively). The finding is consistent with the CDC’s 2012
National Health reports that suggest that in the US, adults with a higher education are less likely
to be obese and higher education is associated with a lower incidence of chronic diseases [341].
The CDC also reports that college educated women are less likely to be obese than their low
educated counterparts [342].

Experiment 2
Subjects:
In the subsequent experiment, 74 subjects who had not participated in the previous
experiment were recruited. Of these, 12 subjects were eliminated because of calibration
difficulties. A further seven subjects were not included in the analysis because of difficulties
149

with the eye tracking software. Thus, usable eye tracking data was obtained from 55
participants. Out of these 55 usable participants – 52 viewed all eight packages in the
experiment. Two participants complained of discomfort and quit after viewing seven out of the
eight packages. The eye tracking software allowed analysis of only the first three packages for
one subject. All 55 usable subjects including these aforementioned ‘incomplete’ ones were
included in every statistical analysis unless specifically noted.
The participants included in the analysis were, on average 36.7 years of age (±14.33),
and 24 were male and 31 were female. Figure 24 further characterizes the demographics of test
participants.

150

Exp 2 Subjects - Age Distribution
20
15
10
5
0
18 to 25
years

26 to 35
years

36 to 45
years

46 to 55
years

56 to 65 65 years
years and over

Exp 2 Subjects - Gender
Male
Female

Exp 2 Subjects - Education Level
Less than High School
High School or More

Exp 2 Subjects - Income

<20,000$
>20,000$
Did Not Disclose

Figure 24 - Experiment 2 Subject Demographics for 55 usable subjects. Note that the Education
demographic is interpreted differently for this experiment.

151

Data Processing:
Two lookzones were defined for each package (see Figure 25).
(i)

An FOP label zone (if present for the package being analyzed)

(ii)

An NFP zone

Figure 25 - Schematic representation of the two eye tracking Lookzones on a flattened package.
Note that the red borders around the zones are for visual reference and were not actually present
on the packages that were used in the Experiment. Text within the package is not meant to be
readable and is for visual reference only

152

Participants were handed eight novel packages in sequence (see Figure 17 in Chapter 6),
and allowed to examine each for a period of 20 seconds. For all packages, nutrition information
was present in both the FOP and NFP look zones. Total time spent on nutrition information was
any of the presented nutritional information (the sum of the time spent on the FOP and NFP
look zones).
For each look zone, various metrics were recorded
(i)

Binary Variable: Zone Noticed (Yes/No)

(ii)

Continuous Variables:
a. Number of visual hits on zone (hits),
b. Total time spent on a zone (seconds), and
c. Time to first visual hit on a zone (seconds)
Responses were also considered in aggregate (for instance, total time spent on ANY

nutritional information) as well as in isolation (e.g. time spent on an NFP).
Results:
From the eye tracking videos it was possible to determine that out of 52 participants
who viewed all 8 packages, 38 fixated upon some form of nutrition information (either the FOP
or NFP or both) on all 8 packages. As many as 7 participants did not view the NFP on all 8
packages, but there was not even one participant who went without fixating upon all 4 FOPs. A
detailed breakdown is provided in Table 18 below. Note than only the Color + Facial Icon FOP
was studied in this experiment.

153

Table 18 - Breakdown of participants who fixated upon ANY nutritional information, NFPs and
FOPs on Packages

Number of Participants

Fixated upon ANY Nutritional Information

38

on All 8 Packages

4

on 7 Packages

3

on 6 Packages

6

on 4 Packages

3

on 3 Packages

Number of Participants

Fixated upon the NFP

30

On All 8 Packages

11

On 7 Packages

2

On 6 Packages

1

On 5 Packages

1

On 4 Packages

7

On None of the Packages

Number of Participants

Fixated upon FOP Label

41

On All 4 Packages

9

On 3 Packages

2

On 2 Packages

The one participant who examined only 3 packages and the two participants who
examined 7 packages before discontinuing the experiment fixated upon all NFPs on their
packages.
The responses involving ANY nutrition information on the package are first examined
below. This is followed by a detailed examination of the NFP zone. As is common with eye
tracking analyses, the probability of fixation is first analyzed, followed by the total time spent
and the time to first fixation.

154

Response to ANY nutritional information:
Binary variable – noticed ANY nutrition information on the package (yes/no):
A generalized linear mixed model was fitted to the binary variable ‘ANY (FOP or NFP)
nutrition information noticed (yes/no)’ using a Bernoulli distribution and a logit-link function to
model the probability of detection. The linear predictors considered in this model were the
fixed effects of Label Type (FOP + NFP vs. NFP only), Product Type (crackers vs. breakfast cereal)
and Health Level (healthy vs. unhealthy) and all 2- and 3-way interactions. Note that only the
Color + Facial Icon FOP label was studied in this experiment.
Other covariates and demographics – BMI category, age, gender, income level,
education level, color blindness, visual acuity, primary shopper status (yes/no) and household
size were not included in the final model because of high Type III p-values (α=0.05). Estimated
Least square means and standard errors are presented below.
There was no evidence for a main effect of product type (p=0.76) or health level
(p=0.93) on the probability of noticing any nutrition information on a package and no evidence
of significant 2 or 3 way interactions were revealed in the analysis (p>0.05). However, a main
effect of label type (p=0.0013) was identified, with a higher probability of fixation on nutrition
information on packages that had the FOP (the color + facial icon was the only FOP design
tested in this experiment) label (M=99.6%, SEM=0.36%) compared with those that only had an
NFP (M=90.9%, SEM=3.3%), See Figure 26 below.

155

Estimated Probability of Fixation on
Nutrition Information (%)

100
95
90
85
80
75
70
FOP + NFP

NFP Only
Label Type

Figure 26 - Effect of Label type, FOP + NFP vs. NFP only and Probability of fixation on
nutrition information (p<0.0001)

Continuous Variable - Total time spent on ANY nutrition information (seconds):
A general linear mixed model was fitted to this continuous variable total time spent on
any nutrition information (in seconds) in the square root scale. As before, the linear predictors
included the fixed effects of label type (FOP + NFP vs. NFP only), product type (crackers vs.
breakfast cereal) and health level (healthy vs. unhealthy) and all 2- and 3-way interactions. Only
non-zero data points were considered for this analysis, i.e. packages in which subjects had not
fixated upon nutrition information at all were excluded. Note that the only FOP studied in this
experiment was the color + facial icon type.

156

Of all the demographic covariates, age (in years), BMI category, Children status (yes/no),
visual acuity and income category (>20,000$ per year vs. <20,000$ per year) improved model fit
(α=0.05) and, thus, were included in the final model. Estimated least square means and 95%
confidence intervals are provided below. Pair-wise comparisons were performed using the
Bonferroni adjustment to avoid inflation of the Type I error rate.
There was no evidence for any 2 or 3 way interactions between the main factors.
Furthering the evidence of the FOP’s ability to garner attention, a main effect of its presence
was identified on total time spent on nutritional information (p=0.0032) (see Figure 27a). Time
spent was greater for packages that had the Color + Facial Icon FOP label (M=6.11s, LCL=5.18s,
UCL=7.13s) than those that only contained the NFP (M=5.23s, LCL=4.36s, UCL=6.19s). A main
effect of product type was also identified (p=0.0070), with breakfast cereals (M=6.10s,
LCL=5.15s, UCL=7.13s) garnering more time on nutritional information than crackers (M=5.25s,
LCL=4.37s, UCL=6.20s) (see Figure 27b). Health level was also indicated as a significant effect
(p=0.0096); time spent on the nutrition information was greater for the unhealthy packages
(M=6.00s, LCL=5.07s, UCL=7.00s) when compared to the healthy packages (M=5.34s, LCL=4.47s,
UCL=6.29) (see Figure 27c).

157

Estimated Total Time
spent on Nutrition
Information (seconds)

8

p = 0.0032

6
4
2
0
FOP + NFP

NFP Only

Estimated Total Time
spent on Nutrition
Information (seconds)

Label Type
8

p = 0.0070

6
4
2
0
Cereal

Crackers

Estimated Total Time
spent on nutrition
information (seconds)

Product Type
8

p = 0.0096

6
4
2
0
Healthy

Unhealthy
Health Level

Figure 27 - The effect of Label type (p=0.0032), Product Type (p=0.0070) and Health Level
(p=0.0096) on the total time spent on nutrition information
158

After adjusting for the other demographic covariates that were chosen for inclusion in
this model (BMI category, Visual acuity, Income range and children - yes/no), there was a
significant negative association between age and the total time spent on nutrition information
(p=0.0015). As age increased, the time spent on nutrition information was found to decrease.
A significant effect of BMI category on the time spent on nutrition information (p=0.03) was
also identified by the analysis. It was estimated that after adjusting for the effects of the other
demographic covariates, obese subjects (BMI of 30+) spent 56% more time on nutrition
information when compared to the overweight subjects (BMI of 25 to 30). Figure 28 below
displays least square means and 95% confidence intervals.

159

Figure 28 - The effect of Subject BMI category on the time spent on nutrition information
(p=0.03) Obese subjects (with a BMI of 30+) spent an estimated 56% more time on nutrition
information than Overweight subjects (with BMI of 25 to 29.9). Alphabets represent statistically
significant pair-wise comparisons at α=0.05
In addition to the children-yes/no variable, the total time spent on nutrition information
was also affected by the self-reported income category (p=0.0042). After adjusting for the other
demographic covariates, subjects who had an income of less than $20,000 per year spent
significantly less time on the nutrition information (M=4.54s, LCL=3.69s, UCL=5.48s) than those
subjects who had an income of more than $20,000 (M=6.92s, LCL=5.48s, UCL=8.52s). See Figure
29b
Another interesting effect was revealed while examining the effect of each demographic
covariate. Participants who answered positively to the ‘Do you have Children?’ question in the
demographic survey, i.e. those whose child status was ‘Yes’ spent an estimated 80% more time
160

on the nutrition information than participants who’s child status was ‘No’ (p=0.0008; M=7.40s,
LCL=5.71s, UCL=9.31s vs. M = 4.16s, LCL=3.43s, UCL=4.96s). See Figure 29a

161

Figure 29 - The Effect of Children Status (p=0.0008) and Income Range (p=0.0042) on
estimated total time spent on nutrition information
162

Additionally, a significant effect of visual acuity on the total time spent on nutrition
information was revealed (p=0.0030). As shown in Figure 30 below, there is evidence to
conclude that participants who had visual acuities of 20/50 spent almost 120% more time on

Estimated Time spent on
Nutrition Information (seconds)

nutrition information than participants with visual acuity of 20/30.

14
b

12
10

ab

8
a

a

6
4
2
0
20/20

20/30
20/40
Subject Visual Acuity

20/50

Figure 30 - The effect of subject visual acuity on time spent on nutrition information (p=0.003).
Bars and alphabet are indicative of statistically significant differences at α =0.05

Continuous Variable – Time to first fixation of ANY nutrition information (seconds):
This continuous variable, time to first fixation on nutrition information was modeled as a
general linear model and was expressed in a square root scale. As before, the linear predictors

163

included the fixed effects of Label Type (FOP + NFP vs. NFP only), product type (crackers vs.
cereal) and health level (healthy vs. unhealthy) and all 2- and 3-way interactions.
None of the demographic covariates and factors were found to improve model fit based
on their Type III p-values (α=0.05) and were excluded from the final model. Estimated least
square means of main effects along with 95% confidence intervals are provided below in the
original scale.
As with the binary variable and the total time variable, a significant main effect of FOP
presence i.e. Label Type (p<0.0001) was indicated on the time to the first fixation of any
nutrition information; packages that had an FOP incorporated had faster times to fixation of
any nutritional information (M=2.39s, LCL=1.85s, UCL=2.97s) than those that only had an NFP
(M=7.88s, LCL=7.03s, UCL=8.79s) (see Figure 31). There was no evidence for an effect of health
level (p=0.42) or product type (p=0.62) on the time to first fixation of nutrition information, nor
evidence for any 2 or 3 way interactions among the main effects (p>0.05).

164

Estimated Time to first fixation of
nutrition information (seconds)

10
9
8
7
6
5
4
3
2
1
0
FOP + NFP

NFP Only
Label type

Figure 31 - Effect of Label type, FOP + NFP vs. NFP only on Time to first fixation of nutrition
information (p=0.0013).

To this point, analyzed metrics have examined varied dependent variables relative to
ANY nutritional information. Analyses specifically to only the NFP follow, as before with the
probability of noticing the zone being examined first, followed by the total time spent on the
zone and the time to first fixation.
Response to NFP information:
Binary variable – Noticed the NFP (yes/no):
The binary response, Noticed the NFP (Yes/No) was fitted as a generalized linear mixed
model using a Bernoulli distribution and logit-link function to model the probability of noticing

165

the NFP. The linear predictors were the fixed effects of Label Type (FOP + NFP vs. NFP only),
Product Type (crackers vs. breakfast cereal) and health level (healthy vs. unhealthy) and all 2
and 3-way interactions. None of the demographic factors were found to improve model fit
based on their type III p-values.
There was no evidence for any effect of Label Type (p=0.59), Product Type (p=0.60) or
Health Level (p=0.58) or any 2 or 3-way interactions.
Continuous Variable – Total Time on the NFP (seconds):
This response was also modeled as a general linear mixed model in the square root
scale. Linear predictors were again, the fixed effects of Label Type (FOP + NFP vs. NFP only),
product type (crackers vs. breakfast cereal) and health Level (healthy vs. unhealthy) and all 2
and 3-way interactions.
Similar to the results of the total time spent in ANY nutrition information (described
previously), four demographic covariates were found to improve model fit: age, children
(yes/no), visual acuity and income. These were included in the final model and their effects are
also discussed below.
The model was fitted using the GLIMMIX procedure of SAS (Version 9.2, SAS Institute,
Cary, NC). Estimated least square means and corresponding 95% confidence intervals are
reported in the original scale. Relevant pairwise comparisons were conducted using either
Tukey-Kramer or Bonferroni adjustments to avoid inflation of Type I error rate due to multiple
comparisons.
Statistical analysis suggested evidence for a two way interaction between product type
and label type (p=0.0486) on the total time spent on the NFP. This suggested that the time

166

spent on the nutritional information spent on the NFP depended on both the design of the FOP
present on the package’s front and the type of product that was being examined (cereal or
crackers). Specifically, for cereal products, the time spent on the NFP was greater when the
package had no FOP label (M=6.18s, LCL=5.15s, UCL=7.29s) when compared to cereal boxes
that had an FOP (M=5.16s, LCL=4.24s, UCL=6.17s). This statistically significant effect of FOP
Type existed within the Cereal product (p=0.0085) but not in the Crackers Product Type

Estimated time spent on the
NFP Zone (seconds)

(p=0.95); hence the two way interaction. See Figure 32 below.

8
7
6
5
4
3
2
1
0

Cereal
Crackers

FOP + NFP
NFP only
Label Type
Figure 32 - Estimated time spent in the NFP by product type by label type. There is a statistically
significant effect of Label Type for the Cereal Product Type (p=0.0085) but not for the Crackers
Product Type (p=0.95).

167

There was also an effect of health level on the total time spent on the NFP zone
(p=0.0053), with participants spending more time on the NFP in unhealthy packages (M=5.68s,
LCL=4.79s, UCL=6.64s) than healthy packages (M=4.96s, LCL=4.13s, UCL=5.87s).
With respect to the demographic effects, after adjusting for the children (yes/no)
variable, and income range, there was a significant negative association of age with time spent
on the NFP zone (p=0.0015). As age increased, the time spent on the NFP label was estimated
to decrease.
There was also evidence to show that subjects who answered ‘Yes’ to the Children
Yes/No question in the demographic survey), after adjusting for the other demographic factors,
spent an estimated 83% more time on the NFP label zone than subjects who did not have
children at home (p=0.0005, M=3.84s, LCL=3.13s, UCL=4.62 vs. M=7.03s, LCL=5.45s, UCL=8.81s).
A similar statistically significant effect was revealed when the impact of income on the time
spent on the NFP zone was examined (p=0.0094). After adjusting for the effects of the other
demographic factors, subjects who earned more than 20,000$ a year spent more time on the
NFP (M=6.42s, LCL=5.06s, UCL=7.94s) when compared to subjects who earned less than
20,000$ a year (M=4.32s, LCL= 3.47s, UCL=5.26s). These two effects are shown in Figure 33 a
and b below.

168

Figure 33 - The effect of children status (p=0.0005) and income per year (p=0.0094) on the
estimated time spent on the NFP zone (seconds)

169

Similar to the effect observed before on the Total time in ANY nutrition information
variable, there was also an effect of visual acuity on the time spent on the NFP zone (p=0.0027).
The time spent on the NFP increased with diminishing visual acuity. Pairwise comparisons are
shown in Figure 34 below and were somewhat similar to the effect of visual acuity on Total

Estimated Time spent on the NFP Zone
(seconds)

time spent on Nutrition information (as shown in Figure 30).

14
b

12
10
ab
8
6

a

a

4
2
0
20/20

20/30
20/40
Subject Visual Acuity

20/50

Figure 34 - The effect of Visual Acuity on the time spent on the NFP zone (p=0.0027).
Alphabets indicate statistically significant differences at α=0.05

170

Continuous Variable – Number of visual hits in the NFP Zone:
The number of visual hits on a zone typically increases with an increase in time spent on
a zone and analyzing this response is a good way to corroborate a total time spent analysis for
that zone. For this reason, the number of visual hits to the NFP was modeled using a
generalized linear mixed model using a negative binomial distribution and a log link function. As
before, the linear predictors included the fixed effects of Label type (FOP + NFP vs. NFP only),
product type (crackers vs. breakfast cereal) and health level (healthy vs. unhealthy) and all 2and 3-way interactions.
All the demographic covariates were considered for inclusion in this model but only
education level (less than high school vs. high school and above) was included in the final model
based on its contribution to data fit (α=0.05).
The model was fitted using the GLIMMIX procedure of SAS (Version 9.2, SAS Institute,
Cary, NC). Estimated least square means and 95% confident intervals or standard errors are
provided below. Pair wise comparisons were performed using the Bonferroni adjustment.
As with the response variable time spent on the NFP, there was evidence for a
significant 2-way interaction between product type and label type (p=0.038). Thus, the impact
of the presence or absence of the FOP (FOP + NFP vs. NFP only) on the number of visual hits to
the NFP depended on whether the participants were viewing cereal or crackers. For cereal
packages, participants had estimated greater number of visual hits to the NFP in packages that
did not have an FOP (M=3.16, LCL=1.86, UCL=5.38) compared to packages that did (M=2.243,
LCL=1.30, UCL=3.78; p=0.0002). However, there was no evidence of significant differences in
the number of hits to the NFP for crackers as a function of label type (p=0.4) (see Figure 35).

171

Estimated number of visual
hits on the NFP Zone

6
5
4
3

Cereal

2

Crackers

1
0
FOP + NFP
NFP only
Label Type

Figure 35 - The effect of label type on number of visual hits to the NFP by product type. As
shown, there is significantly (p=0.0002) higher number of the NFP zone visual hits for cereal
packages that have no FOP when compared to cereal packages that have the Color + Facial Icon
FOP label. There is no such statistically significant difference for crackers (p=0.4)

There was also a marginal effect of health level on the number of visual hits on the NFP
zone (p=0.057). Unhealthy packages were estimated to have more visual hits on the NFP zone
(M=2.60, SEM=0.577) when compared to healthy packages (M=2.23, SEM=0.67). Despite the
marginal significance, Note that the magnitude of the difference is very low.
Analysis suggested that the education level of participants significantly impacted the
number of visual hits to the NFP zone (p=0.03). Participants who had completed high school or

172

more registered more visual hits to the NFP zone (M=4.277, SEM=0.73) when compared to
those participants in the ‘less than high school’ category (M=1.36, SEM=0.657).

Continuous Variable – Time to first fixation on the NFP zone (seconds):
The response variable, time to first fixation on the NFP, was modeled using a general
linear mixed model only for those times when the NFP was fixated (i.e. non-infinity values). This
variable was expressed in log scale during modeling. The linear predictors included the fixed
effects of label type (Color + Facial Icon vs. NFP only), product type (crackers vs. cereal) and
health level (healthy vs. unhealthy) and all 2 and 3-way interactions. None of the demographic
covariates improved model fit and, therefore, were not included in the final model.
A main effect of label type was identified by this analysis (p<0.0001); the time to first
fixation on the NFP was lower in packages that did not have the FOP (M=7.48s, LCL=6.80s,
UCL=8.23s) when compared to those that had the Color + Facial Icon FOP (M=9.27s, LCL=8.42s,
UCL=10.2s).
Discussion:
On its surface, it seems intuitive that an FOP label, given its prominent positioning on
the package’s PDP, would be an effective strategy for delivering information. That said, studies
that directly measure behaviors related to the processing of front-of-pack nutritional
information, particularly those that focus on US audiences, are very limited [25] [23]. Further, it
is possible that FOPs may serve to either “catalyze (lead to more detailed examination of)” or
“short cut (reduce examination of)” access to the more complete nutritional information found
in the form of the NFP. Recognizing these issues and others, both federal officials and

173

researchers have indicated that any FOP system that is implemented should be informed by
well-designed, ecologically valid research studies that directly measure behavior [23] [29] [224,
267].
Experiment 2’s results afford insights into how FOP labels impact information processing
in individuals from varied backgrounds. Results from this experiment also lend credence to the
argument that the FOP label enhances the likelihood that nutritional information will be
attended. A significant effect of label type (NFP only or FOP + NFP) was noted for all responses:
probability of noticing nutrition information on the package (p<0.0013, Figure 26), time to first
fixation of nutrition information (p<0.0001, Figure 31) and total time spent on nutrition
information (p=0.0032, Figure 27) i.e. when the FOP was present, participants were more likely
to fixate the nutritional information, hit nutritional information faster, and spend longer time
viewing the information. Strikingly, it was estimated that nutrition information was noticed
almost 230% faster on packages that had FOPs present than those that only had NFPs (See
Figure 31).
Another factor that affected the time spent on both nutrition information as a whole
and the NFP was the health level of the package. There was evidence to conclude that
participants spent an estimated 12% more total time on the nutrition information on unhealthy
packages (p=0.0096). This effect of health level was also indicated on the number of visual hits
and total time on the NFP; irrespective of the product or label type, participants spent more
time on the NFP of unhealthy packages (See Figure 27). It was initially speculated that this
effect of health level may have occurred differently for those packages with FOPs (i.e.
participants examined nutritional information for the unhealthy packages that contained the
174

multiple red circles and frowning facial icons) more seriously. However, the analysis suggested
that the effect of health level on the time spent on the NFP (p=0.0053) as well as the total time
spent on nutrition information of any kind (p=0.0096) occurred irrespective of label type (FOP +
NFP vs. NFP only) or product type (cereal vs. crackers). In other words, participants exhibited a
tendency to examine nutrition information more elaborately on packages with an unhealthy
nutritional composition regardless of the package or labeling scheme that they were presented
with.
It has been proposed that the presence of an FOP on food packages may result in a
‘substitution’ effect, whereby the FOP label is used as a source of nutrition information instead
of the NFP when it is present. Analysis that utilized the eye movement behaviors related to only
the NFP, suggests that this may be dependent on the product that is being examined.
A statistically significant 2-way interaction between product type (crackers vs. cereal)
and label type (NFP + FOP vs. NFP only) on both the time spent on the NFP (p=0.0486) and the
number of visual hits to the NFP (p=0.0384) was revealed. Evidence for both response variables
supported the idea that the FOP label served as a short cut for the more complete NFP
information in the case of the cereal only. There was no evidence of this happening for the
crackers (see Figure 32 and Figure 35). However, it is important to note that based on the
analysis of the total time spent on nutrition information of any kind, irrespective of label type
and health level, participants spent more time (about 16% more) on the nutrition information
on the cereal packages as opposed to the cracker packages (p=0.0070, Figure 27b). It is
possible that this is due to the fact that they view cereal as a meal and crackers as a snack;
allocating more attention accordingly. A larger study into the effects of product type on
175

information processing and label utilization is recommended with a wider range of products
than comprised in this study.
Evidence also suggested that specific demographic attributes of the participants
impacted the way that they interacted with the nutritional labeling information. As participants
became older, they spent more total time on nutritional information (p=0.0015) and more time
within the NFP itself (p=0.02).
Perhaps of greater interest from a policy perspective was the finding that participants
with lower incomes (<$20,000 dollars annually) spent significantly less total time on nutritional
information (p=0.0042, Figure 29) and on the NFP zone (p=0.0094, Figure 33) than participants
who reported that they earned more than 20,000 dollars a year, with participants in the higher
income bracket spending almost 49% more time on the NFP zone. These findings are in line
with CDC reports which mention that children from low socio-economic strata are more likely
to be obese [343]. For adult women, the CDC reports that “obesity prevalence increases as
income decreases [342]” but for adult men obesity prevalence is “generally similar at all income
levels [342].”
Additionally, participants who reported that they had children, spent significantly more
time on nutrition information on the whole (p=0.0008, Figure 29) and more time on the NFP
zone (p=0.0005, see Figure 33) than those that did not have children.
BMI status also significantly affected the total time that participants spent on nutrition
information (p=0.03). As shown in Figure 28, obese subjects (BMI 30+) spent an estimated 56%
more time on nutrition information when compared to overweight subjects (BMI 25 to 30). By
contrast, overweight subjects spent significantly less time on the nutrition information than
176

either those characterized as obese or normal. It is possible that the obese participants are
engaging in interventional strategies discussed herein, specifically, changes in diet, as a result of
underlying health conditions and, as such, are more engaged with nutritional information than
those classified as overweight or normal (See Table 9 for a review of related research).

Experiments 3 & 4
Subjects – Experiments 3:
One hundred participants were recruited to take part in Experiments 3 and 4 together.
For Experiment 3, one subject’s data was rendered unusable because of a computer error. As
such, analysis included responses collected from 99 for Experiment 3, the encodation task. Of
these, 46 subjects were male and 53 were female with an average age of 45.09±12.63 years.
See Figure 36 for further characterization of the study population.

177

Exp 3 Subjects - Age Distribution
40
30
20
10
0
18 to 25 26 to 35 36 to 45 46 to 55 56 to 65 65 years
years
years
years
years
years and over
Exp 3 Subjects - Gender

Male
Female

Exp 3 Subjects - Education
High School or
Less
More than High
School
Did not Disclose
Exp 3 Subjects - Income
<20,000$
>20,000$
Did Not Disclose

Figure 36 - Demographic information for Exp 3 Participants
178

Data Processing – Experiment 3:
After viewing six packages, i.e. three pairs of products (in each pair one at the unhealthy
level; the other at the healthy level) with each pair representing a different product type (cereal
vs. crackers vs. prepared meals) and having a different FOP scheme (color+facial icon vs. no
color+no facial icon vs. NFP only), subjects participated in a forced choice task. During the task
only the brand names of the pair were displayed and subjects were asked to choose the
healthier of the two. This task was administered using EPrime® and the responses of each
participant were tabulated. There was one response of interest for this experiment.
(i)

Binary Variable – Correct/Healthier Package chosen (Yes/No)

Results – Experiment 3:
Binary Variable – Correct Package Chosen (Yes/No):
This binary response, correct package chosen (yes/no) was modeled as a Bernoulli
distribution and a logit-link function with in order to estimate the probability of choosing the
correct package. The linear factors included the fixed effects of FOP type (color + facial icon vs.
no color + no facial icon vs. NFP only), product type (cereal vs. crackers vs. prepared meals) and
their two way interactions.
Covariates and demographic factors considered for inclusion in this model included BMI
category, age, gender, income range, ethnicity, education range, color blindness, primary
shopper status (Yes/No), children status (yes/no), visual acuity, position of correct answer
package on screen (Left/Right) and trial run order (1st, 2nd, 3rd). Only age and run order were
included in the final model as both indicated evidence of improved model fit (α=0.05).

179

The model was fitted using the GLIMMIX procedure of SAS (Version 9.2, SAS Institute,
Cary, NC). Pairwise comparisons were performed with the Tukey-Kramer adjustments to avoid
inflation of Type I error rate.
There was no evidence for an effect of the two main factors product type (p=0.75) and
FOP type (p=0.30) or an interaction between them (p=0.8656).
Regardless of run order, product type or FOP type, there was a marginal negative
association of age on the probability of choosing the right package (p=0.0588). As participants’
age increased the probability of choosing the right package decreased.
A significant effect of run-order on the correct response was evident (p=0.0166). after
adjusting for age effects, subjects had a higher probability of choosing the right package on the
third trial. Pairwise comparisons were conducted using the Tukey Kramer adjustment and are
shown in Figure 37 below. There was a statistically significant difference in the estimated
probability of selecting the correct package between the first and third trial (p for this pair-wise
comparison = 0.017). In other words, no matter what packages they saw and what pairs were
presented in each trial, participants had a higher probability of getting the third trial correct
every time (p=0.016). Estimated least square means and 95% confidence intervals are displayed
in Figure 37 below.

180

Probability of Choosing the Correct
Package (%)

100

b

90

ab

a

80
70
60
50
40
30
20
10
0

Trial 1

Trial 2
Trial Number

Trial 3

Figure 37 - The effect of run order on the probability of selecting the correct/healthier package in
the incidental encoding task (p=0.016). Lines and alphabets of indicative of statistically
significant differences at α=0.05

Discussion – Experiment 3:
None of the evidence presented suggested that FOP design had a significant effect on a
participant’s ability to encode nutritional information. Although future research is needed, this
result would suggest that although evidence provided via the previous experiments suggests
that FOPs enhance the likelihood that nutritional information will be noticed, the time it takes
to notice it and the amount of time that is spent on it, the presence of these FOP designs do not
enhance the ability of a consumer to encode the information into memory.
Subjects - Experiment 4:

181

Upon completion of experiment 3, participants were asked to participate in experiment
4, a sorting task. Of the 100 participants recruited for experiment 3, six chose not to take part in
the sorting task, and another subject’s data was not included in the analysis because he/she
chose to discontinue after the first three sorting trials. This resulted in 93 usable subjects for
this experiment. Of these, 42 were male and 51 were female with an average age of
44.53±12.23 years. Figure 38 further characterizes subjects demographically.

182

Exp 4 Subjects - Age Distribution
40
30
20
10
0
18 to 25 26 to 35 36 to 45 46 to 55 56 to 65 65 years
years
years
years
years
years and over
Exp 4 Subjects - Gender

Male
Female

Exp 4 Subjects - Education
High School or Less
More Than High
School
Did not Disclose

Exp 4 Subjects - Income
<20,000$
>20,000$
Did Not Disclose

Figure 38 - Demographic information for Experiment 4 Subjects
183

Data Processing – Experiment 4:
The RFiD system used to administer the sorting tasks in Experiment 4 has been
described in detail in the Methods section (see Figure 19). A parsing script written in Python 3.1
was used to interface with the RFiD software and calculate two measures for each trial:
(i)

The binary variable , sorted correctly (Yes/No)

(ii)

The continuous variable , time taken to sort (seconds)
Analyses presented here include data related to the FOP nutrients only (sugar, fat,

saturated fat and sodium); trials related to protein, which was only present on the NFP, were
not included.
Results – Experiment 4
Binary Variable - Sorted Correctly (Yes/No):
Out of the 93 subjects tested, 26 subjects correctly sorted 100% of their trials and 76
subjects had error rates of 25% or lower.
A generalized linear mixed model was fitted to the binary response, “sorted correctly”
(yes/no), using a Bernoulli distribution and logit-link function to model the probability of correct
sorting. The linear predictors in this model were the fixed effects of FOP type at five levels:
(NFP only vs. color + no facial Icon vs. color + facial icon vs. no color + facial icon vs. no color +
no facial icon) (See Figure 12 for all FOP types), nutrient at four levels: (fat vs. salt/sodium vs.
saturated fat vs. sugar) and the 2-way interactions. As mentioned before, upon completion of
all twenty five trials in the experiment (5 nutrients by 5 FOP types; data from protein, which
was only on the NFP, was not included in the analysis), participants were asked whether they

184

used the FOPs or not and their response was recorded. This self-reported FOP use (yes/no)
variable was also considered in the analysis as an explanatory factor.
Demographic factors were confounded with this explanatory factor of self-reported FOP
use and were, thus, not considered in the final model. This explanatory factor is analyzed in on
its own later. The model was fitted using the GLIMMIX procedure of SAS 9.2 (SAS Institute, Cary
NC). Pair-wise comparisons were performed using the Tukey-Kramer adjustment to avoid
inflation of the Type-1 Error rate.
There was evidence that subjects that self-reported using the FOP for sorting were more
likely to correctly sort packages than those who did not (P=0.02), with subjects that reported
using the FOP system more likely to correctly sort the set (M=95.3%, LCL=89.7%, UCL=97.9%)
than those that did not report using it. as opposed to those that did not report using the FOP
(M=87.2%, LCL=80.8%, UCL=91.6%), (See Figure 39).

185

Estimated Probability of Correct Sort
(%)

100
95
90
85
80
75
70
65
60
55
50
Used FOP (NO)
Used FOP (Yes)
Self Reported FOP Use

Figure 39 - Estimated Probability of correctly sorting as determined by self reported use of the
FOP (p=0.02)

That said, there was no evidence for an effect of FOP type (p=0.51), or presence of an
FOP (p=0.83), on the probability of correctly sorting. A significant effect of nutrient on the
probability of correctly sorting was suggested by the data (p<0.0001). Participants had a
significantly lower likelihood of correctly sorting for saturated fat (M=85.8%, LCL=77.8%,
UCL=91.3%) as compared to the other FOP nutrients: fat (M=92.5%, LCL=87.4%, UCL=95.7%),
sugar (M=93.88%, LCL=98.4%, UCL=96.5%), salt/sodium (M=94.5%, LCL=89.4%, UCL=96.5%).
See Figure 40.

186

b

b

b

a

Figure 40 - The effect of nutrient on the probability of correctly sorting (p<0.0001). There was
an estimated lower likelihood of sorting packages by sat-fat when compared to the other
nutrients. Alphabets are indicative of statistically significant differences at α=0.05

Continuous Variable - Time to sort (seconds):
The continuous variable, time to correctly sort by nutrient, was analyzed using a
generalized linear model. This response was expressed in the log-scale for analysis. The linear
predictors in this model were the fixed effects of FOP type at five levels (NFP only vs. color + no
facial Icon vs. color + facial icon vs. no color + facial icon vs. no color + no facial icon) and
nutrient at four levels: (fat vs. salt/sodium vs. saturated fat vs. sugar) as well as all 2 and 3-way
interactions. As before, also considered was the additional explanatory factor, self-reported
FOP use (Yes/No). As with the binary variable (correctly sorted), only non-protein trials were
considered in these analyses.
187

Of all the demographic factors considered, age, education range (high school or less vs.
more than high school) and visual acuity were found to improve model fit and were included in
the final model. The model was fitted using the GLIMMIX procedure of SAS 9.2 (SAS institute,
Cary, NC). Estimated least square means and 95% confidence intervals are reported below in
the original scale i.e. after backtransformation. Tukey-Kramer or Bonferroni adjustments were
used in pair-wise comparisons to avoid inflation of Type 1 error-rate.
Regardless of label type, nutrient or status reported for FOP use (yes, no), there was a
positive relationship between age and time to sort (p=0.0055). As participant age increased, the
time to correctly sort was also found to increase. Subject visual acuity was also found to have a
significant impact on time to correct sort (p=0.0144). As expected, participants who had a nearpoint visual acuity of 20/50 took the most estimated time to sort (M=22.1909s, LCL=19.2296s,
UCL=25.6082s), significantly higher (p=0.0080 for this particular pair-wise comparison) than
subjects with a near point visual acuity of 20/40 (M=16.8439s, LCL=14.9971s, UCL=18.9181s).
The effect of subject visual acuity on time to sort is shown in Figure 41)

188

Figure 41 - The effect of subject visual acuity on time to correct sort (p=0.0055)
Education status also provided evidence of a significant effect (p=0.0091); participants
(p=0.0091)
who had only completed high school or less (M=21.12s, LCL=19.05s, UCL=23.42s) took
d
significantly longer time to correctly sort packages than participants who had more than high
school education (M=17.49s, LCL=15.84s, UCL=19.31s) (see Figure 42).

189

Estimated time to correct sort
(seconds)

25
20
15
10
5
0
High School or Less
More than High School
Education Range
Figure 42- The effect of the Education Range demographic factor on time to correct sort
(p=0.0091)
Besides the effects of these demographic factors, A significant three-way interaction
(p=0.0329) was revealed between FOP type, nutrient and whether or not the participant
reported using the FOP (Yes/No). Due to this 3-way interaction, the effects of each predictor
are interpreted below within levels of the other two predictors.
Figure 43 examines the effect of FOP type on time to sort (sec) by self-reported use
status (yes/no) and nutrient. Irrespective of nutrient, FOP type suggest statistically significant
differences (α=0.05) in time to sort ONLY when participants reported that they used the FOPs
i.e. answered ‘Yes’ to the ‘Did you use the FOPs?’ question. Further, when participants reported

190

using the FOPs, the package sets that contained only the NFP always took the longest time to
sort.
Among those self-reporting use of the FOP, facial icons significantly lowered sort times
in trials involving fat and saturated fat, with the color + facial icon, taking the least amount of
time for fat (M=13.23 s, LCL==11.13s, UCL=15.73s) and saturated fat (M=13.87s, LCL=11.63s,
UCL=16.54s) (See Figure 43 a and b). However, for salt/sodium, color was the design feature
influencing sorting time, leading to faster sorts, with the two color-coded FOPs, color+facial
icon (M=13.87s, LCL=11.63s, UCL=16.54s) and color + no facial icon (M=15.077s, LCL=12.70s,
UCL=17.89s) leading to the fastest sorts (See Figure 44a). When sorting for sugar there was no
evidence for an effect of either facial icons or color coding on time to sort among those that
reported using the FOPs (see Figure 44b). In Figure 43 and Figure 44 the code used on the Xaxes for FOP type is interpreted as follows – (C+NF = Color + No Face, C+F = Color + Face, NC+F
= No Color + Face, NC + NF = No Color + No Face)

191

Fat
25

a

a

a

d

a

a

bcd

20

b

cd

bc

15
10
5
0
C + NF

C+F

NC + F NC + NF

NFP

C+F

NC + F NC + NF C + NF

Used FOP (No)

NFP

Used FOP (Yes)

Saturated Fat
35
30

e

e

e

e

e

g

25

g

g

g

20

f

15
10
5
0
C + NF

C+F

NC + F NC + NF

NFP

Used FOP (No)

C+F

NC + F NC + NF C + NF

NFP

Used FOP (Yes)

Figure 43 – The effect of FOP Type on the estimated time to sort (seconds). Effects are
interpreted for each level of the Used FOP (Yes/No) factor for the a) Fat and b) Saturated Fat
nutrients because of the significant 3-way interaction between Nutrient, Self Reported FOP Use
and FOP Type (p=0.0329). Alphabets are indicative of statistically significant pair-wise
comparisons within each chart at α=0.05. Y-axes represent ‘Estimated time to sort (seconds)’

192

Salt/Sodium
30

h

h

h

25

h

h

j

20

i

i

i

ij

15
10
5
0
C + NF

C+F

NC + F NC + NF

NFP

C+F

Used FOP (No)

C + NF NC + F NC + NF

NFP

Used FOP (Yes)

Sugars
30

k

k

k
k

25

n
k

mn
l

20

lm

lm

15
10
5
0
C + NF

C+F

NC + F NC + NF

NFP

Used FOP (No)

C + F NC + NF NC + F C + NF

NFP

Used FOP (Yes)

Figure 44 – The effect of FOP Type on the estimated time to sort (seconds). These effects are
interpreted for each level of the Used FOP (Yes/No) factor for a) Salt/Sodium and b) Sugars
nutrients. because of the significant 3-way interaction between Nutrient, Self Reported FOP Use
and FOP Type (p=0.0329). Alphabets are indicative of statistically significant pair-wise
comparisons within each chart at α=0.05. Y-axes represent ‘Estimated time to sort (seconds)’

193

Figure 45 and Figure 46 examine this complex three-way interaction between nutrient,
FOP type and Self reported FOP use from a different perspective, specifically looking at
significant effects of nutrient on time to sort (seconds) for each of the four FOP Types.
For the color + facial icon FOP type, there were significantly faster sorts for all four
nutrients (p<0.05 for all four pair-wise comparisons) when FOP use was reported as opposed to
when it was not (see Figure 45a). For the no color + facial icon FOP type there were significantly
faster sorts for all nutrients (p<0.05 for all pairwise compairons), with the exception of the sorts
related to sugar when FOP use was reported (see Figure 45b). Thus, for both FOP types that had
facial icons, there were faster sorts for at least three out of the four nutrient types when FOP
use was reported, as opposed to when FOP use was not reported (See Figure 45)
Evident from Figure 46 is that the two FOP types that did not have facial icons were not
as effective in speeding up sorts as much. The color+ no facial icon FOP type led to significantly
faster sorts only in the salt/sodium trials and the no color + no facial icon FOP type led to
significantly faster sorts for salt/sodium and sugar only.

194

Figure 45 - The effect of self reported FOP use on time to sort (seconds) for each nutrient type
for the two FOP Types that had facial icons. Significant differences in time to sort at α=0.05 for
each nutrient within each FOP Type between FOP Used (Yes) and FOP Used (No) cases are
shown in bold.

195

Figure 46 - The effect of self reported FOP use on time to sort (seconds) for each nutrient type
for the two FOP types that did not have facial icons. Significant differences in time to sort at
α=0.05 for each nutrient within each FOP Type between FOP Used (Yes) and FOP Used (No)
cases are shown in bold.

196

Binary Variable – Self Report of FOP label use (Yes/No)
Out of the 93 subjects that were used in the analysis for this experiment, only 24
reported that they used the FOPs while sorting (slightly over 25%). For the purposes of policy
creation with the purpose of a healthier society, it is important to understand the variables that
affect FOP use. The self-reported binary response, FOP label used (yes/no) was fitted with a
generalized linear mixed model with a Bernoulli distribution and a logit-link function to model
the probability of FOP label use.
The predictors in this model were the demographic factors: BMI category, education
range, ethnicity, income range, household size, children status (yes/no) and age (categorized
into four levels for ease of interpretation – Less than 35 years, 35 to 45 years, 45 to 55 years
and 55 years and over). Out of these demographic factors, only age category and education
range were retained in the final model based on improvement of model fit. Pair-wise
comparisons are performed using the Tukey-Kramer adjustment to avoid inflation of the type-1
error rate.
Both education range (p=0.0009) and age group (p=0.0437) were found to have a
significant impact on FOP use. Participants who had more than a high school education were
much more likely to report using the FOP(M=46.2%, LCL=29.07%, UCL=64.41%) than
participants who had only completed high school or less (M=10.1%, LCL=4.18%, UCL=22.45%).
(see Figure 47).

197

Estimated Probability of FOP label
use (%)

70
60
50
40
30
20
10
0
High School or Less
More than High School
Subject Education Range

Figure 47 - Impact of education range on FOP label use (p=0.0009). Participants who had more
than a high school education were more likely to self-report use of the FOP then participants who
had less than a high school education.

Figure 48 examines the estimated probabilities of FOP label use for individual agegroups. The estimated probability of FOP use was greatest for individuals in the 35 to 45 age
group (M=48.52%, LCL=21.6%, UCL=22.16) and lowest for those in the 45 to 55 age group
(M=6.5%, LCL=5.02%, UCL=17.68%); of all possible comparisons, the comparison of these two
groups was the only one suggesting significance (p=0.02).

198

a
ab
ab

b

Figure 48 - Effect of Age group on the probability of FOP label use. Alphabets are indicative of
statistically significant pair-wise comparisons at α=0.05
Discussion – Experiment 4:
The purpose of this sort task experiment was two-fold. Firstly, to study whether
participants, when explicitly told to make nutrient comparisons between products would use
the FOPs to do so, and secondly, if used, to examine the effect that varied FOP designs had on
cross-product comparisons, as measured by the ability to correctly sort, as well as the time it
takes to correctly sort.
Just over 25% of participants reported using the FOP labels as they sorted products;
education and age group influenced the probability of FOP use, both of which have been
reported as factors that have an impact on obesity prevalence [65, 342]. In the study reported
199

herein, participants with more than a high school education were significantly more likely to
use the FOP labels (M=46.2%, LCL=29.07%, UCL=64.41%) for their sorts compared with those
that had education levels of only high school or lower (M=10.1%, LCL=4.18%, UCL=22.45%,
Figure 47). The more educated participants also performed their sorts significantly faster than
those that reported high school or less as their educational status (See Figure 42). This is an
important finding, CDC reports that as education increases, the incidence of obesity and other
chronic diseases is reduced [341, 342]; further, these are the very consumers that are most
likely to benefit from the simplified system.
Along with having a positive relationship with the time to sort (p=0.0055, as age
increased time to sort increases), age also influenced the probability of FOP label use. The
probability of FOP label use was estimated to be highest for the 35 to 45 years age group
(M=48.52%, LCL=21.6%, UCL=22.16, See Figure 48). The results of the most recent National
Health and Nutrition Education Survey (NHANES) show that obesity prevalence is the lowest
(27.5%) for men in the 20 to 39 age group, an age range that overlaps the age group estimated
to have the higher probability of FOP use [344]. This lends further credence to the finding that
nutrition labels have a positive benefit when used [185].
While the design of the FOPs (FOP Type) was not found to influence probability of
sorting packages correctly (p=0.51), a complex 3-way interaction between FOP Type, nutrient
and self-reported FOP use that affected time to sort was revealed (p=0.032, Figure 43, Figure
44, Figure 45 and Figure 46) making the nature of FOP design effects on time to sorting hard to
interpret in a general manner. However, FOP-use was found to have a positive impact on the
probability of correctly sorting (p=0.02), with participants that used the FOP more likely to sort
200

correctly (M=95.3%, LCL=89.7%, UCL=97.9%) as opposed to those that did not use the FOP
(M=87.2%, LCL=80.8%, UCL=91.6%, Figure 39).
The results of this experiment begin to create a picture that FOP-use levels may be low
with only 25% of consumers using FOPs, but those that do use them are able to make faster
and quicker nutrition comparisons between products. There are several plausible explanations
as to why usage of the FOP was low. Many of the participants who reported that they did not
use FOPs during the sort task, mentioned that they used the NFPs simply because they were
‘used to seeing them on packages’. It is possible that when FOPs are eventually mandated,
there will be an accompanying media blitz both by the government and by major food
companies and the resulting rise in consumer awareness may lead to increased use levels. It is
also speculated that the low use of FOPs may have been an artifact of our sample. As
mentioned before, the majority of participants for this study were recruited through the local
Supplemental Nutrition Assistance Program (SNAP), Extended Food and Nutrition Education
Program (EFNEP) and Women, Infants and Children (WiC) initiatives. It is possible that these
subjects were a part of detailed nutrition education seminars that concentrated on how to use
the NFP, leading them to use the NFPs for their sorts.

201

Chapter 8
Conclusions, Limitations & Future work
Conclusions
Six FOP labels (Shown in Figure 12 and Figure 15) were designed based on three major
design features – color, facial icons and checkmarks. Four experiments were conducted to
assess the efficacy of these six FOP label designs. The experiments were modeled based on a
commonly used information processing model [340] and utilized four different methodologies.
(i)

Experiment 1 - A Change detection experiment to evaluate which FOP label designs
capture the most attention and have most visual salience

(ii)

Experiment 2 - Eye Tracking study to determine if presence of FOP label leads people
to more detailed nutrition information on NFP and evaluate noticeability of an FOP
label

(iii)

Experiment 3 - Recall study to determine which FOP label design has the highest
ability to cause consumers to encode and retain information about a product’s
nutrition, even when people are not attempting to access and remember nutritional
information, i.e. evaluate the relative amount of incidental encodation that different
FOP label designs are capable of

(iv)

Experiment 4 - Sort task study to determine whether FOP labels are noticed and
used when consumers are specifically performing nutrition comparisons between
products and if noticed, which FOP labels enables easiest/fastest comparisons.

202

Overall, this study adds to the existing body of knowledge about FOP labels, making a
strong case for the noticeability of the FOP, especially color-coded FOPs in comparison to the
NFP. Despite the clear finding of the enhanced noticeability of FOPs as compared with NFPs,
there was no indication that they were more easily encoded than NFP information when
participant’s tasks were not explicitly related to nutrition information. The presence of FOP
labels was beneficial to sort tasks, when they were used, but a large number of participants did
not use them (~75%). Perhaps more concerning was the fact that people that were most likely
to benefit from their presence, those with little education, were less likely to use them during
the sort trials. As such, if the Federal government does choose to move forward with a
standardized system for FOP labeling, an important factor will be awareness/education
campaigns.

Limitations
Despite representing an important addition to the existing body of knowledge on FOP
labels and the first step in understanding FOPs from an information processing model using US
consumers, limitations are present. These are listed below
(i)

In Experiment 2, despite being told that they were to examine packages as if
they were considering purchasing them, participants were sitting in a chair with
a chin-rest, wearing an eye tracking device (See Figure 18). This is typical of highaccuracy eye tracking studies. In Experiment 4, whether the participant used the
FOPs for his/her sorts or not was determined by a self-report. Even though this
was easily verifiable based on the observations of the two researchers

203

conducting the experiment, there is always a chance that those who reported
not using the FOPs used them on a few trials. Furthermore, this Used FOPs –
Yes/No self report was obtained after the entire experiment. Even participants
who reported that they had used the FOPs may have only noticed them and
begun using them after the first 10 or 15 trials. A future study that keeps track of
this using eye tracking or some similar method would be needed to overcome
this limitation.
(ii)

The subject pool for this study, while diverse, consisting of people from various
ethnicities and socio-economic groups and ages, was entirely sourced from in
and around the Greater Lansing area. Before federal legislation regarding FOPs is
attempted a similar study with a larger sample conducted across state
boundaries would be needed.

Future Work
Exploring Color-Coding further
As mentioned before, this study adds to the body of evidence that favors Multiple
Traffic Light Labels (color coding). Despite this, industry groups, both in the USA [269] and in
the EU [245] have chosen to endorse mono-chrome FOP labels (see Figure 8). Some
commentators in the UK have also mentioned that widespread adoption of these labels will
alienate typically nutrient-dense foods like meats and cheeses [256], harming these industries.
While differences in perceived healthfulness of products that are induced by FOP
presence have been analyzed before [33], this alleged ability of Multiple Traffic Light color

204

coded FOPs to alienate products has not been studied to the best of the author’s knowledge. A
study that would shed some light on this would represent an important step forward and serve
to remove some of this uncertainty about color coding.
Exploring Product Type Effects
Including this one, FOP label studies conducted in the US and abroad seem to focus
entirely on the FOPs themselves. Product types are chosen mostly as placeholders and there is
never any explicit attempt made to study the interaction of FOP and Product Type. The Phase 2
IOM report on FOPs described previously talks about how different criteria may need to be
defined for each Product type[29]. In this study, the Eye Tracking Experiment revealed a strong
Product Type by Label Type interaction (p<0.05, Figure 32, Figure 35) on the time spent on the
NFP and an effect of Product Type on the total time spent on nutrition information (p<0.05,
Figure 27) which suggests that FOPs may be used differently on different products. Considering
the large variety of product types that a mandatory FOP system will eventually find its way on,
studies that focus specifically on FOP use on different products are needed.
Location, FOP Size, Package Form and other design effects
In commercially available packages, FOPs have been used in various locations - top right
of PDP, bottom right of PDP etc. In this study, in order to achieve ease of inference and
uniformity, all FOP labels were of the same size and were present on the bottom right of the
PDP. Additionally, all previously published studies reviewed in this document kept FOP size and
location constant across their stimuli. Before mandating FOPs on commercial packages, it is
essential to consider the effect that FOP size and location has on salience and usefulness.

205

A wide variety of product types means a wide variety of package sizes and package
forms. For instance, In the US, FOPs are already present on cans of soda, folding cartons filled
with cookies, pouches of potato chips etc. A study that investigates both product and package
type would be most helpful in studying the potential benefits of mandating FOPs nationwide.
The brands that were designed for this study were stripped clean of all the other
information that is generally found on package fronts – health claims, special offers,
promotions, tie-ins with sporting events, cross-branding with movie releases etc. In the real
world, the FOP will have to compete with all these distracters. One current study here at the
Michigan State University repeats Experiment 1 (Change Detection), examining the effects of
FOPs on commercially available packages, the results of which, it is anticipated, will be
published in mid 2013.
Working with Children and speakers of other languages
At the time of this writing, Childhood obesity continues to be one of the most significant
public health problems in the USA [6]. One possible benefit of using color-coding and symbols
on FOP labels is that they would eventually become useful to children and those with a limited
command of the English language (or any primary language used on the package for that
matter) to make responsible dietary choices. It is envisioned that future studies with FOP labels
will be performed using samples of children in various age groups (3-5 years, 5-7 years and 8 to
12 years) and non-native speakers of English to see if this is practical.

206

APPENDICES

207

Appendix 1 - Consent Forms:

Michigan State University
School of Packaging
INSTRUCTIONS AND RESEARCH CONSENT FORM – PACKAGE DESIGN EXPERIMENT
You are being asked to participate in a research study. The entire study should last no more than
2 hours. In exchange for your participation, you will receive 1 HPR credit for each half hour (or
any part of a half hour) of participation. You may discontinue participation at any time and still
receive extra credit.
If you have any questions at any time please ask.
Change Detection Experiment
This experiment makes use of a Change Detection software program. You will be seated in front
of a computer system and take part in what is known as a ‘Flicker Task’. In this, an original and
an altered image alternate on screen separated by a blank grey frame. Your objective is to spot
the change in the altered image and press a key as instructed. The flicker tasks in this study shall
feature images of packages of novel brands of food products.
You will be performing a total of 9 sets of thirty two trials each.
After the trials, researchers will collect some information about you. You will be led to a
screened area where your weight and height will be measured discreetly using standard
procedures. A color blindness test will be administered by asking you to view a series of cards
and asking you to decipher images to the best of your ability. A short interview will be used to
gather information about your normal dietary habits, history of diet-related illness and
demographics (age, level of education, gender).
Risks and Benefits
There is little or no risk associated with this research. It is possible that you will experience some
discomfort while you are performing the change detection tasks. Please let us know if you do
experience any discomfort at any time. You may discontinue at any time and still receive extra
credit. There is no direct benefit to you for participating in this study. However, we are hopeful

208

that we can learn what constitutes well designed packages so that this information is used to
create future package designs that are more effective.
Confidentiality

Study results will be treated in strict confidence. Your privacy will be protected to the maximum
extent allowable by law. Within these restrictions, results of this study will be made available to
you at your request. Data may be stored in the lab on secure systems (server and local
computers in a closed network), on tape back-up, and on the experimenter's passwordprotected computers. Data may be stored for the amount of time required by the American
Psychological Association (typically 5 years). Collected data will only be stored by subject
number and cannot be tied to your identity.
If you have any concerns or questions about this research study, such as scientific issues, or if
you believe you have been harmed because of the research, please contact Professor, Mark
Becker, 517-432-3367; 285B, Psychology Building, East
Lansing MI 48824 or email becker54@msu.edu
If you have questions or concerns about your role and rights as a research participant, would
like to obtain information or offer input, or would like to register a complaint about this study,
you may contact, anonymously if you wish, the Michigan State University's Human Research
Protection Program at 517-355-2180 or e-mail irb@msu.edu or regular mail at 207 Olds Hall,
MSU, East Lansing, MI 48824.
I voluntarily agree to participate in this Packaging Design Study.
Sign:
Date:
You will be provided with a copy of your signed consent form if you ask for it.

209

Michigan State University
School of Packaging
INSTRUCTIONS AND RESEARCH CONSENT FORM
EVALUATING FRONT OF PACKAGE LABELING- EYETRACKING
You are being asked to participate in a research study. Participation in this study is voluntary.
The entire study should last no more than 2 hours. In exchange for your participation, you will
receive $20 OR, if you are in psychology courses, you can opt to receive 1 HPR credit for each
half hour (or any part of a half hour) of participation. These credits can be used for a
psychology course that either requires research participation or offers extra credit for
participation. You may discontinue participation at any time and still keep the $20 OR HPR
credit.
To participate in this study you must:
•
•
•
•

Be at least 18 years of age
Not be legally blind
Not wear hard contact lenses
Not have participated in other aspects of this experiment, which involve food labeling

If you have any questions at any time please ask.
Eye Tracking Experiment
In this experiment you will be inspecting and rating package aesthetics of novel brands of food
products. This experiment makes use of an ASL MobileEye® Eye tracking device. This device sits
on your nose like a pair of eye glasses and is connected to a video recorder and records your
eye movements as you inspect the packages given to you.
You will be seated at a table and made to wear the eye tracker. Once the device is adjusted and
all images brought into focus, a short calibration procedure will be administered where you will
be asked to look at different points in the room while keeping your head still. Once a
satisfactory calibration is established you will be handed twelve packages, one after the other.
You are free to manipulate the packages and examine them in any way you please.
After this procedure is complete, you will be asked to rate the aesthetics of each package.
Once this eye tracking task is complete, researchers shall collect some information about you.
You shall be led to a screened area where your weight and height shall be measured discreetly
210

using standard procedures. A color blindness test shall be conducted by asking you to view a
series of cards and asking you to decipher images to the best of your ability. A short interview
shall gather information about your normal dietary habits, history of diet- related illness and
demographics (age, level of education, gender).
Risks and Benefits
There is little or no risk associated with this research. There is no direct benefit to you for
participating in this study. However, we are hopeful that we can learn what constitutes well
designed packages so that this information is used to create future package designs that are
more effective.
If you are injured as a result of your participation in this research project, researchers from
Michigan State University will assist you in obtaining emergency care, if necessary, for your
research related injuries. If you have insurance for medical care, your insurance carrier will be
billed in the ordinary manner. As with any medical insurance, any costs that are not covered or
in excess of what are paid by your insurance, including deductibles, will be your responsibility.
The University’s policy is not to provide financial compensation for lost wages, disability, pain or
discomfort unless required by law to do so. This does not mean that you are giving up any legal
rights you may have. If you believe you have been harmed as a result of this study, please
contact please contact Laura Bix, 517-355-4556; 153 Packaging Building, East Lansing MI 48824
or email bixlaura@msu.edu
Confidentiality
Study results will be treated in strict confidence. Your confidentiality will be protected to the
maximum extent allowable by law. Within these restrictions, results of this study will be made
available to you at your request. Data may be stored in the lab on secure systems (server and
local computers in a closed network), on tape back- up, and on the experimenter's
password-protected computers. Data may be stored for the amount of time required by the
American Psychological Association (typically 5 years). Collected data will only be stored by
subject number and cannot be tied to your identity.
If you have any concerns or questions about this research study, such as scientific issues, or if
you believe you have been harmed because of the research, please contact Professor Laura Bix,
Ph: 517-355-4556; 448 Wilson Road #153 Packaging Building East Lansing MI 48824 or email
bixlaura@msu.edu.
If you have questions or concerns about your role and rights as a research participant, would
like to obtain information or offer input, or would like to register a complaint about this study,
you may contact, anonymously if you wish, the Michigan State University's Human Research
Protection Program at 517-355-2180 or e-mail irb@msu.edu or regular mail at 408 W. Circle,
Room 207 Olds Hall, MSU, East Lansing, MI 48824.

211

I voluntarily agree to participate in this Packaging Design Study.
Sign:
Date:
You will be provided with a copy of your signed consent form.

212

Michigan State University
School of Packaging
INSTRUCTIONS AND RESEARCH CONSENT FORM
PACKAGE DESIGN EXPERIMENT- INCIDENTAL ENCODING AND SORTING TASK
You are being asked to participate in a research study. Participation in this study is voluntary.
The entire study should last no more than 2 hours. In exchange for your participation, you will
receive $20. You may discontinue participation at any time and still keep the $20.
To participate in this study, you must:
•
•
•

Be at least 18 years of age
Not be legally blind
Not have participated in other aspects of the experiment, which involve food labeling

If you have any questions at any time please ask.
Packaging Study
You will have several tasks to complete in this experiment.
One of them will be setting sets of packages in order based on nutritional content.
You will be seated at a table and a tray of packages for novel brands of food products shall be
handed to you. You shall be timed as you sort them based on the instructions given to you by
the researcher (for example: Order these products highest to lowest based on Vitamin content).
You will perform a series of these sort tasks based on different nutritional dimensions.
During another task you will be inspecting and rating package aesthetics of novel brands of food
products. You will be viewing images of food packages of novel brands on a computer screen
and rating them for aesthetics and design.
You will be seated in front of a computer screen and asked to view nine packages of novel food
products one after another. After examining all packages you will be asked to answer a few
questions based on the packages you just interacted with.
Once these tasks are complete, researchers will collect some information about you. You will be
led to a screened area where your weight and height shall be measured discreetly using
213

standard procedures. A color blindness test shall be conducted by asking you to view a series of
cards and asking you to decipher images to the best of your ability. A short interview will be
used to gather information about your normal dietary habits, history of diet- related illness and
demographics (age, level of education, gender).
Risks and Benefits
There is little or no risk associated with this research. There is no direct benefit to you for
participating in this study. However, we are hopeful that we can learn what constitutes well
designed packages so that this information is used to create future package designs that are
more effective.
If you are injured as a result of your participation in this research project, researchers from
Michigan State University will assist you in obtaining emergency care, if necessary, for your
research related injuries. If you have insurance for medical care, your insurance carrier will be
billed in the ordinary manner. As with any medical insurance, any costs that are not covered or
in excess of what are paid by your insurance, including deductibles, will be your responsibility.
The University’s policy is not to provide financial compensation for lost wages, disability, pain or
discomfort unless required by law to do so. This does not mean that you are giving up any legal
rights you may have. If you believe you have been harmed as a result of this study, please
contact please contact Laura Bix, 517-355-4556153 Packaging Building, East Lansing MI 48824
or email bixlaura@msu.edu.
Confidentiality
Study results will be treated in strict confidence. Your confidentiality will be protected to the
maximum extent allowable by law. Within these restrictions, results of this study will be made
available to you at your request. Data may be stored in the lab on secure systems (server and
local computers in a closed network), on tape back- up, and on the experimenter's
password-protected computers. Data may be stored for the amount of time required by the
American Psychological Association (typically 5 years). Collected data will only be stored by
subject number and cannot be tied to your identity.
If you have any concerns or questions about this research study, such as scientific issues, how to
do any part of it, or if you believe you have been harmed because of the research, please
contact the researcher Laura Bix 517-3554556; 448 Wilson Road #153 Packaging Building East Lansing MI 48824 bixlaura@msu.edu.
If you have questions or concerns about your role and rights as a research participant, would
like to obtain information or offer input, or would like to register a complaint about this study,
you may contact, anonymously if you wish, the Michigan State University's Human Research
Protection Program at 517-355-2180 or e-mail irb@msu.edu or regular mail at 408 W. Circle
Drive, 207 Olds Hall, MSU, East Lansing, MI 48824.

214

I voluntarily agree to participate in this Packaging Study.
Sign:
Date:
You will be provided with a copy of your signed consent form.

215

Appendix 2 - Recruitment advertisements:

Participants Wanted for
Packaging Design Study
Receive Extra Credit in exchange for your participation
Procedure: This experiment makes use of a Change Detection software program. You will be
seated in front of a computer system and take part in what is known as a ‘Flicker Task’. In this,
an original and an altered image alternate on screen separated by a blank grey frame. Your
objective is to spot the change in the altered image and press a key as instructed. The flicker
tasks in this study shall feature images of packages of novel brands of food products. You will
be performing nine sets of thirty two trials each.
After the trials, researchers will collect some information about you. You will be led to a
screened area where your weight and height shall be measured discreetly using standard
procedures. A color blindness test will be administered by asking you to view a series of cards
and asking you to decipher images to the best of your ability. A short interview after that will be
used to gather information about your normal dietary habits, history of diet-related illness and
demographics (age, level of education, gender).
Risks and Benefits: There are no known risks associated with this study. There is no direct
benefit to you. If at any time you are uncomfortable with the testing or wish to discontinue the
data collection process. You will receive 1 HPR credit for each half hour (or any part of a half
hour) of participation you may discontinue participation without penalty and still receive extra
credit. The entire testing procedure will not take more than 2 hours.
If you are ABOVE 18 YEARS OF AGE, have NO HISTORY OF SEIZURE and are interested in
pursuing this opportunity, please contact Raghav Prashant Sundar at sundarra@msu.edu or
cell phone 517-898-9029 to make an appointment. If you have questions or comments

216

regarding this study, please contact Dr. Mark Becker, Professor of Psychology at Michigan
State University at 517-432-3367 or becker54@msu.edu

217

The School of Packaging and the Department of Psychology at Michigan State University
invite you to participate in a study of label design
To participate in this study you must: Be at least 18 years of age, not legally blind, not wear
hard contact lenses, not have participated in other portions of the study
You will be asked to: View package designs while we track your eye movements using an eye
tracker. After viewing the packages you will be asked to rate their aesthetics. At the end we will
also collect some demographic information about normal dietary habits, history of diet-related
illness and demographics (age, level of education, gender). You will be paid $20 for your input
and time.
Duration: Not more than 2 hours (~1.5 hours)
For more information or to schedule a convenient time: Contact Chad Peltier by emailing
peltie11@msu.edu or calling (517) 353-6622

218

The school of packaging and the department of psychology at the Michigan state university
invite you to participate in a study of label design
To participate in this study you must: Be at least 18 years of age, not legally blind, not have
participated in any other aspects of this study
You will be asked to: View package designs, rate their aesthetics and perform simple sorting
tasks. At the end we will also collect some information about your normal dietary habits,
history of diet-related illness and demographics (age, level of education, gender). You will be
paid $20 for your input and time.
Duration: Not more than 2 hours (~1.5 hours)
For more information or to schedule a convenient time: Contact Chad Peltier by emailing
peltie11@msu.edu or calling (517) 353-6622

219

Appendix 3-Products Designed for Study:

Figure 49 - Bran Blast (Breakfast Cereal). Text within image is not meant to be readable and is
for design reference only

220

Figure 50 - Golden Harvest (Breakfast Cereal). Text within image is not meant to be readable
and is for design reference only
221

Figure 51 - Spyros (Breakfast Cereal). Text within image is not meant to be readable and is for
design reference only
222

Figure 52 - Sunrise (Breakfast Cereal). Text within image is not meant to be readable and is for
design reference only
223

Figure 53 - Earls (Crackers) Text within image is not meant to be readable and is for design
reference only
224

Figure 54 - Fish Eye (Crackers) Text within image is not meant to be readable and is for design
reference only
225

Figure 55 - Marvel (Crackers). Text within image is not meant to be readable and is for design
reference only

226

Figure 56 - Snappers (Crackers). Text within image is not meant to be readable and is for design
reference only

227

Figure 57 - Best Cuisine (Prepared Meals) Text within image is not meant to be readable and is
for design reference only

228

Figure 58 - Read Meals (Prepared Meals) Text within image is not meant to be readable and is
for design reference only

229

REFERENCES

230

REFERENCES

1.

Maguire, T. and D.W. Haslam, Obesity epidemic and its management. 2010, London:
Pharmaceutical Press. xi, [4], 264 p.

2.

Gard, M. and J. Wright, The obesity epidemic : science, morality, and ideology. 2005, London ;
New York: Routledge. 218 p.

3.

Gill, T., et al., Obesity in the Pacific: Too big to ignore. 2002, Secretariat of the Pacific
community.

4.

Carson, C. Public health care enemy number 1: Childhood Obesity. 2009 [cited 2009 November
22]; Available from: http://www.huffingtonpost.com/carole-carson/public-health-careenemy_b_361583.html.

5.

Koplan, J.P. and Institute of Medicine (U.S.). Committee on Progress in Preventing Childhood
Obesity., Preventing childhood obesity: Health in the Balance. 2005, National Academies Press:
Washington, D.C. p. 1 electronic text.

6.

Smith, J.C., Understanding childhood obesity. Understanding health and sickness series. 1999,
Jackson: University Press of Mississippi. xii, 160 p.

7.

Wang, Y. and M.A. Beydoun, The obesity epidemic in the United States - Gender, age,
socioeconomic, Racial/Ethnic, and geographic characteristics: A systematic review and metaregression analysis. Epidemiologic Reviews, 2007. 29: p. 6-28.

8.

Center for Disease Control and Prevention . About BMI for Adults. 2013 [cited 2013 Feb 21st];
Available from: http://www.cdc.gov/healthyweight/assessing/bmi/adult_bmi/index.html.

9.

Eknoyan, G., Adolphe Quetelet (1796-1874) - the average man and indices of obesity.
Nephrology Dialysis Transplantation, 2008. 23(1): p. 47-51.

10.

CDC. About BMI for Children and Teens. 2009 [cited 2009 November 22]; Available from:
http://www.cdc.gov/healthyweight/assessing/bmi/childrens_bmi/about_childrens_bmi.html.

11.

Variyam, J.N. and J. Cawley, Nutrition labels and Obesity. 2008, National Bureau of Economic
Research.

12.

Shapiro, R., Nutrition labeling handbook. 1995, New York: Marcel Dekker. xv, 704 p.

13.

Koplan, J.P., et al., Progress in preventing childhood obesity how do we measure up? 2006,
National Academies Press: Washington, D.C. p. 1 electronic text.

14.

Institute of Medicine (U.S.). Committee on Accelerating Progress in Obesity Prevention. and
Institute of Medicine (U.S.). Food and Nutrition Board., Measuring progress in obesity prevention
: workshop report. 2012, Washington, D.C.: National Academies Press. xiii, 144 p.

231

15.

Institute of Medicine (U.S.). Committee on Accelerating Progress in Obesity Prevention. and
Institute of Medicine (U.S.). Food and Nutrition Board., Assessing impact of marketing and
industry, in Measuring progress in obesity prevention : workshop report, Institute of Medicine
(U.S.). Committee on Accelerating Progress in Obesity Prevention. and Institute of Medicine
(U.S.). Food and Nutrition Board., Editors. 2012, National Academies Press: Washington, D.C. p.
xiii, 144 p.

16.

Ogden, C.L., et al., Prevalence of Obesity and Trends in Body Mass Index Among US Children and
Adolescents, 1999-2010. Jama-Journal of the American Medical Association, 2012. 307(5): p.
483-490.

17.

Ogden, C.L., et al., Prevalence and trends in overweight among US children and adolescents,
1999-2000. Jama-Journal of the American Medical Association, 2002. 288(14): p. 1728-1732.

18.

Obesity Working Group (2004) Calories count: Report of the working group on obesity. Volume,

19.

US White House Task Force on Childhood obesity, Solving the problem of Childhood obesity
within a generation. 2010.

20.

Malam, S., et al., Comprehension and use of UK nutrition signpost labelling schemes. 2009,
Project Management Panel, Food Standards Agency.

21.

Vyth, E.L., et al., Methodological quality of front-of-pack labeling studies: a review plus
identification of research challenges. Nutrition Reviews, 2012. 70(12): p. 709-720.

22.

Roberto, C.A., et al., Facts Up Front Versus Traffic Light Food Labels A Randomized Controlled
Trial. American Journal of Preventive Medicine, 2012. 43(2): p. 134-141.

23.

Hawley, K.L., The science on front-of-package food labels. Public Health Nutr, 2012.

24.

Roberto, C.A., et al., Choosing front-of-package food labelling nutritional criteria: how smart
were 'Smart Choices'? Public Health Nutrition, 2012. 15(2): p. 262-267.

25.

Roberto, C.A., et al., Evaluation of consumer understanding of different front-of-package
nutrition labels, 2010-2011. Prev Chronic Dis, 2012. 9(120015).

26.

Brownell, K.D. and J.P. Koplan, Front-of-Package Nutrition Labeling - An Abuse of Trust by the
Food Industry? New England Journal of Medicine, 2011. 364(25): p. 2373-2375.

27.

US Food and Drug Administration, FDA Foods and Veterinary Medicine Program Strategic Plan
(2012 – 2016). 2012, US FDA.

28.

Nestle, M. (2012) FDA releases strategic plan for 2012-2016. Food Politics Volume,

29.

Institute of Medicine (U.S.). Committee on Examination of Front-of-Package Nutrition Rating
Systems and Symbols., E. Wartella, and Institute of Medicine (U.S.). Food and Nutrition Board.,
Front-of-package nutrition rating systems and symbols : promoting healthier choices. 2012,
Washington, D.C.: National Academies Press. xvi, 163 p.

232

30.

Wartella, E., A. Lichtenstien, and C. Boon, Examination of Front-of-Package Nutrition Rating
Systems and Symbols, Phase 1 Report. 2010, Institute of Medicine.

31.

Roberto, C.A., et al., The Smart Choices front-of-package nutrition label. Influence on perceptions
and intake of cereal. Appetite, 2012. 58(2): p. 651-657.

32.

Marco, A., The impact of Front-of-Package labeling in low socioeconomic areas, in Health
Promotion Management. 2011, American University: Washington D.C.

33.

Kim, H., et al., Front-of-Package nutritional labels and consumer beverage perceptions. Applied
Economic Perspectives and Policy, 2012. 34(4): p. 599-614.

34.

Zhu, C., The effects of Front-of-Package Nutritional labeling on consumers and producers:
Evidence from two natural experiments. 2012, University of Connecticut: Storrs-Mansfield.

35.

Vyth, E.L., et al., A front-of-pack nutrition logo: a quantitative and qualitative process evaluation
in the Netherlands. J Health Commun, 2009. 14(7): p. 631-45.

36.

Van Kleef, E., et al., Consumer preferences for front-of-pack calories labelling. Public Health
Nutrition, 2008. 11(2): p. 203-213.

37.

Lando, A. and L.-W. J, Helping consumers make more healthful food choices: consumer views on
modifying food labels and providing point-of-purchase nutrition information at quick-service
restaurants. J Nutr Educ Behav, 2007. 39((3)): p. 157-63.

38.

Malam, S., et al., Comprehension and use of UK nutrition signpost labelling schemes, F.S. Agency,
Editor. 2009: London, UK.

39.

Grunert, K.G., J.M. Wills, and L. Fernandez-Celemin, Nutrition knowledge, and use and
understanding of nutrition information on food labels among consumers in the UK. Appetite,
2010. 55(2): p. 177-189.

40.

Laughery, K. and A. Hammond, Overview, in Warnings and Risk Communication, M. Wogalter, D.
DeJoy, and K. Laughery, Editors. 1999, Taylor and Francis:: London/Philadelphia. p. 3-12.

41.

D'Alessio, D.A., et al., Thermic effect of food in lean and obese men. J Clin Invest, 1988. 81(6): p.
1781-9.

42.

Zahorska-Markiewicz, B., Thermic effect of food and exercise in obesity. Eur J Appl Physiol Occup
Physiol, 1980. 44(3): p. 231-5.

43.

Center for Disease Control and Prevention . Overweight and Obesity. 2010 [cited 2010 August
11th]; Available from: http://www.cdc.gov/obesity/causes/index.html.

44.

Oliveria, S.A., et al., Parent-Child Relationships in Nutrient Intake - the Framingham Childrens
Study. American Journal of Clinical Nutrition, 1992. 56(3): p. 593-598.

45.

Ard, J.D., Unique perspectives on the obesogenic environment. Journal of General Internal
Medicine, 2007. 22(7): p. 1058-1060.

233

46.

Moore, L.L., et al., Influence of Parents Physical-Activity Levels on Activity Levels of YoungChildren. Journal of Pediatrics, 1991. 118(2): p. 215-219.

47.

Vandewater, E.A., M.S. Shim, and A.G. Caplovitz, Linking obesity and activity level with children's
television and video game use. Journal of Adolescence, 2004. 27(1): p. 71-85.

48.

Dietz, W.H. and S.L. Gortmaker, Do We Fatten Our Children at the Television Set - Obesity and
Television Viewing in Children and Adolescents. Pediatrics, 1985. 75(5): p. 807-812.

49.

Robinson, T.N., Does television cause childhood obesity? Jama-Journal of the American Medical
Association, 1998. 279(12): p. 959-960.

50.

Robinson, T.N., Television viewing and childhood obesity. Pediatric Clinics of North America,
2001. 48(4): p. 1017-+.

51.

Borjeson, M., Etiology of Obesity in Children - Study of 101 Twin Pairs. Acta Paediatrica
Scandinavica, 1976. 65(3): p. 279-287.

52.

Maffeis, C., Aetiology of overweight and obesity in children and adolescents. European Journal of
Pediatrics, 2000. 159: p. S35-S44.

53.

Stunkard, A.J., T.T. Foch, and Z. Hrubec, A Twin Study of Human Obesity. Jama-Journal of the
American Medical Association, 1986. 256(1): p. 51-54.

54.

Stunkard, A.J., T. Sorensen, and F. Schulsinger, Use of the Danish Adoption Register for the Study
of Obesity and Thinness. Research Publications-Association for Research in Nervous and Mental
Disease, 1982. 60: p. 115-120.

55.

Stunkard, A.J., et al., Adoption Study of Obesity. New England Journal of Medicine, 1986. 315(2):
p. 130-130.

56.

Stunkard, A.J., et al., An Adoption Study of Human Obesity. New England Journal of Medicine,
1986. 314(4): p. 193-198.

57.

Roberts, S.B., et al., Energy-Expenditure and Intake in Infants Born to Lean and Overweight
Mothers. New England Journal of Medicine, 1988. 318(8): p. 461-466.

58.

Weinsier, R.L., et al., Metabolic Predictors of Obesity - Contribution of Resting EnergyExpenditure, Thermal Effect of Food, and Fuel Utilization to 4-Year Weight-Gain of Post-Obese
and Never-Obese Women. Journal of Clinical Investigation, 1995. 95(3): p. 980-985.

59.

Bouchard, C., et al., Genetic Effect in Resting and Exercise Metabolic Rates. Metabolism-Clinical
and Experimental, 1989. 38(4): p. 364-370.

60.

Kaplan, A.S., B.S. Zemel, and V.A. Stallings, Differences in resting energy expenditure in
prepubertal black children and white children. Journal of Pediatrics, 1996. 129(5): p. 643-647.

61.

Morrison, J.A., et al., Determinants of resting energy expenditure in young black girls and young
white girls. Journal of Pediatrics, 1996. 129(5): p. 637-642.

234

62.

Caballero, B., The global epidemic of obesity: An overview. Epidemiologic Reviews, 2007. 29: p.
1-5.

63.

World Health Organization. Obesity & Overweight - Fact Sheet. 2012 [cited 2013 January 2nd].

64.

WHO. Obesity and Overweight. 2009 [cited 2009 November 22]; Available from:
http://www.who.int/dietphysicalactivity/publications/facts/obesity/en/.

65.

Flegal, K.M., et al., Prevalence and trends in obesity among US adults, 1999-2008. JAMA, 2010.
303(3): p. 235-41.

66.

Ogden, C., et al., Prevalence of Obesity in the United States, 2009-2010. 2012, National Center
for Health Statistics (NCHS).

67.

Center for Disease Control and Prevention . New Obesity Data Shows Blacks Have the Highest
Rates of Obesity. 2009, Center for Disease Control and Prevention

68.

Center for Disease Control and Prevention . U.S. Obesity Trends by State 1985–2009. 2009
[cited 2010 August 9th]; Available from: http://www.cdc.gov/obesity/data/trends.html.

69.

Pan, L., et al., Trends in the prevalence of extreme obesity among US preschool-aged children
living in low-income families, 1998-2010. 2012.

70.

Obama, B., Presidential Proclamation -- National Childhood Obesity Awareness Month, 2012,
W.H.O.o.t.P. Secretary, Editor. 2012: Washington DC.

71.

Woolf, M., Obesity is as serious a problem as climate change, says minister, in The Independnt.
2007: London.

72.

Eckel, R.H., et al., Preventing cardiovascular disease and diabetes - A call to action from the
American Diabetes Association and the American Heart Association. Circulation, 2006. 113(25):
p. 2943-2946.

73.

Schmieder, R.E. and F.H. Messerli, Does Obesity Influence Early Target Organ Damage in
Hypertensive Patients. Circulation, 1993. 87(5): p. 1482-1488.

74.

Colditz, G.A., et al., Weight-Gain as a Risk Factor for Clinical Diabetes-Mellitus in Women. Annals
of Internal Medicine, 1995. 122(7): p. 481-486.

75.

PinhasHamiel, O., et al., Increased incidence of non-insulin-dependent diabetes mellitus among
adolescents. Journal of Pediatrics, 1996. 128(5): p. 608-615.

76.

Salazar-Martinez, E., et al., Case-control study of diabetes, obesity, physical activity and risk of
endometrial cancer among Mexican women. Cancer Causes & Control, 2000. 11(8): p. 707-711.

77.

Shoff, S.M. and P.A. Newcomb, Diabetes, body size, and risk of endometrial cancer. American
Journal of Epidemiology, 1998. 148(3): p. 234-240.

235

78.

Ford, E.S., Body mass index and colon cancer in a national sample of adult US men and women.
American Journal of Epidemiology, 1999. 150(4): p. 390-398.

79.

Kono, S., et al., Obesity, weight gain and risk of colon adenomas in Japanese men. Japanese
Journal of Cancer Research, 1999. 90(8): p. 805-811.

80.

Shike, M., Body weight and colon cancer. American Journal of Clinical Nutrition, 1996. 63(3): p.
S442-S444.

81.

Chow, W.H., et al., Obesity and risk of renal cell cancer. Cancer Epidemiology Biomarkers &
Prevention, 1996. 5(1): p. 17-21.

82.

Goodman, M.T., et al., Diet, body size, physical activity, and the risk of endometrial cancer.
Cancer Research, 1997. 57(22): p. 5077-5085.

83.

Chow, W.H., et al., Body mass index and risk of adenocarcinomas of the esophagus and gastric
cardia. Journal of the National Cancer Institute, 1998. 90(2): p. 150-155.

84.

Li, S.D. and S. Mobarhan, Association between body mass index and adenocarcinoma of the
esophagus and gastric cardia. Nutrition Reviews, 2000. 58(2): p. 54-56.

85.

Bianchini, F., R. Kaaks, and H. Vainio, Overweight, obesity, and cancer risk. Lancet Oncology,
2002. 3(9): p. 565-574.

86.

Nomura, A.M., Body size and prostate cancer. Epidemiologic Reviews, 2001. 23(1): p. 126-131.

87.

American Institute for Cancer Research., Food, Nutrition, Physical Activity and the Prevention of
Cancer: a Global Perspective. 2007, American Institute for Cancer Research.

88.

Haslam, D.W., Obesity: Your Questions answered. 2006, Oxford: Radcliffe Medical Press.

89.

US Department of Health and Human Services (US DHHS), The Surgeon General's call to action to
prevent and decrease overweight and obesity:, P.H. Service, Editor. 2001: Rockville, MD.

90.

Finkelstein, E., et al., Annual medical spending attribute to obesity: Payer and Service-Specific
estimates. Health Affairs, 2009. 28(5): p. W822-W831.

91.

Koplan, J.P. and K.D. Brownell, Response of the Food and Beverage Industry to the Obesity
Threat. Jama-Journal of the American Medical Association, 2010. 304(13): p. 1487-1488.

92.

Prather, R.C., Improving the Long-Term Management of Obesity - Theory, Research, and Clinical
Guidelines - Perri,Mg, Nezu,Am, Viegener,Bj. Contemporary Psychology, 1993. 38(10): p. 10701071.

93.

Sandberg, D.E., Improving the Long-Term Management of Obesity - Theory, Research, and
Clinical Guidelines - Perri,Mg, Nezu,Am, Viegener,Bj. Journal of Developmental and Behavioral
Pediatrics, 1994. 15(2): p. 140-141.

236

94.

Schmidt, U., Improving the Long-Term Management of Obesity - Theory, Research and Clinical
Guidelines - Peri,Mg, Nezu,Am, Viegener,Bj. Behaviour Research and Therapy, 1995. 33(2): p.
236-236.

95.

Abenhaim, L., et al., Appetite-suppressant drugs and the risk of primary pulmonary hypertension.
New England Journal of Medicine, 1996. 335(9): p. 609-616.

96.

Connolly, H.M., et al., Valvular heart disease associated with fenfluramine-phentermine. New
England Journal of Medicine, 1997. 337(9): p. 581-588.

97.

United States Department of Agriculture, Dietary Guidelines for Americans. 2005, USDA.

98.

Parker, L., et al., Alliances for obesity prevention : finding common ground, workshop summary.
2011, Washington, D.C.: National Academies Press. x, 73 p.

99.

Economic Research Service, Food Expenditures. 2012, United States Department of Agriculture.

100.

McNeal, J., Tapping the three kids' markets. American Demographics, 1998. 20: p. 37-41.

101.

Strasburger, V.C., Children and TV advertising: nowhere to run, nowhere to hide. J Dev Behav
Pediatr, 2001. 22(3): p. 185-7.

102.

Kovacic, W., et al., Marketing Food to children and adolescents - A review of industry
expenditures, activiities and self-regulation. 2008, Federal Trade Commission.

103.

Holt, D., et al., Children's Exposure to TV advertising in 1977 and 2004: Information for the
obesity debate, Bureau of Economics Staff Report, Editor. 2007, Federal Trade Commision:
Washington, DC.

104.

Story, M. and S. French, Food Advertising and Marketing Directed at Children and Adolescents in
the US. International Journal of Behavioral Nutrition and Physical Activity, 2004. 1(3).

105.

Fischer, P.M., et al., Brand logo recognition by children aged 3 to 6 years. Mickey Mouse and Old
Joe the Camel. JAMA, 1991. 266(22): p. 3145-8.

106.

Center for Science in the Public Interest Pestering Parents: How Food Companies Market Obesity
to Children, M. Wootan, Editor. 2003, CSPI.

107.

Mangleberg, T., CHILDREN'S INFLUENCE IN PURCHASE DECISIONS: A REVIEW AND CRITIQUE.
Advances in Consumer Research, 1990. 17: p. 813-825.

108.

Jenkins, R., THE INFLUENCE OF CHILDREN IN FAMILY DECISION-MAKING: PARENTS'
PERCEPTIONS. Advances in Consumer Research, 1979. Volume 6: p. 413-418.

109.

Mello, M., Federal Trade Commission Regulation of Food Advertising to Children: Possibilities for
a Reinvigorated Role. Journal of Health Politics and Law, 2010. 35(2): p. 227-278.

110.

Children's Advertising Review Unit Home Page. 2010 [cited 2010 January 20th].

237

111.

Better Business Bureau. Children's Food and Beverage Advertising Initiative,. 2011 [cited 2011
February 27th]; Available from: http://www.bbb.org/us/children-food-beverage-advertisinginitiative/.

112.

Samuels, S., et al., Media, Marketing and Advertising and Obesity, in Handbook of obesity
prevention a resource for health professionals, S.K. Kumanyika and R.C. Brownson, Editors. 2007,
Springer: New York. p. xxvi, 537 p.

113.

Fried, E., Assessing Effectiveness of Self-Regulation: A Case Study of the Children’s Advertising
Review Unit. Loyola of Los Angeles Law Review, 2006. 39: p. 93-137.

114.

Federal Trade Commission Perspectives on Marketing, Self-Regulation, and Childhood Obesity.
2006, Federal Trade Commission and Department of Health and Human Services: Washington,
DC.

115.

Rotfeld, H.J., Power and Limitations of Media Clearance Practices and Advertising SelfRegulation. Journal of Public Policy & Marketing, 1992. 11(1): p. 87-95.

116.

Campaign for a Commercial Free Childhood. Home Page. 2013 [cited 2013 Jan 7th]; Available
from: http://commercialfreechildhood.org/.

117.

Ad Council . Coalition for Healthy Children. 2006 [cited 2009 December 6th]; Available from:
http://www.adcouncil.org/healthychildren/index.html.

118.

Levere, J.L., An Anti-obesity campaign takes to the airwaves, in New York Times. 2012: New
York.

119.

Gantz, W., et al., Food for thought: Television food advertising to children in the United States.
2007, Kaiser Family Foundation: Menlo Park, CA.

120.

Dhar, T. and K. Baylis, Fast-Food Consumption and the Ban on Advertising Targeting Children:
The Quebec Experience. Journal of Marketing Research, 2011. 48(5): p. 799-813.

121.

Urban, A. Advertising bans work: Quebec has lowest childhood obesity rate. 2012 [cited 2013
Jan 7th]; Available from: http://www.care2.com/causes/advertising-bans-work-quebec-haslowest-childhood-obesity-rate.html.

122.

Musemeche, C. Ban on advertising to children linked to lower obesity rates. 2012 [cited 2013
Jan 7th].

123.

Kretser, A.J. The Packaged food industry. 2006 [cited 2009 December 6th]; Available from:
http://www.ift.org/divisions/food_law/Slides_200601/KRETSER.pdf.

124.

Walmart Inc. Making healthier food a reality for all. Our Comittments 2011 [cited 2013 Jan
7th]; Available from: http://corporate.walmart.com/global-responsibility/hunger-nutrition/ourcommitments.

125.

Center for Disease Control and Prevention Do increased portion sizes affect what we eat? 2006,
National Center for Chronic Disease Prevention and Health Promotion.

238

126.

Rolls, B.J., et al., Increasing the portion size of a packaged snack increases energy intake in men
and women. Appetite, 2004. 42(1): p. 63-69.

127.

Wansink, B. and S.B. Park, At the movies: how external cues and perceived taste impact
consumption volume. Food Quality and Preference, 2001. 12(1): p. 69-74.

128.

Fisher, J.O., B.J. Rolls, and L.L. Birch, Children's bite size and intake of an entree are greater with
large portions than with age-appropriate or self-selected portions. American Journal of Clinical
Nutrition, 2003. 77(5): p. 1164-1170.

129.

McConahy, K.L., et al., Food portions are positively related to energy intake and body weight in
early childhood. Journal of Pediatrics, 2002. 140(3): p. 340-347.

130.

Nabisco. Nabisco 100 Calorie Packs. 2010 [cited.

131.

Wansink, B. and J. Peters, The food industry role in obesity prevention, in Handbook of obesity
prevention a resource for health professionals, S.K. Kumanyika and R.C. Brownson, Editors. 2007,
Springer: New York. p. xxvi, 537 p.

132.

Packaging Digest Could carrots replace junk food if branded well? New packaging aims to find
out, in Packaging Digest. 2010.

133.

Cohn, S.R. Away from home dining. 2006 [cited 2009 December 6th]; Available from:
http://www.ift.org/divisions/food_law/Slides_200601/COHN.pdf.

134.

Childs, N.M. Obesity - What are we doing about it? 2006 [cited 2009 December 6th]; Available
from: http://www.ift.org/divisions/food_law/Slides_200601/CHILDS.pdf.

135.

Baranowski, T., et al., Squire's quest! Dietary outcome evaluation of a multimedia game.
American Journal of Preventive Medicine, 2003. 24(1): p. 52-61.

136.

Cullen, K.W., et al., Squire's quest: intervention changes occurred at lunch and snack meals.
Appetite, 2005. 45(2): p. 148-151.

137.

Gorecki, S. Food Industry Uses Internet to Reach Kids. 2006 [cited 2009 December 6th];
Available from:
http://www.redorbit.com/news/technology/581113/food_industry_uses_internet_to_reach_ki
ds/index.html.

138.

Public Interest Media Group. Eat Smart, Grow Strong. 2009 [cited 2009 December 6th];
Available from: http://www.eatsmartgrowstrong.com.

139.

United States Congress, et al., Pure Food and Drug Act (1906) United States, United States
Statutes at Large (59th Cong., Sess. I, Chp. 3915, p. 768-772; cited as 34 U.S. Stats. 768). 2004,
NCBI: Bethesda, MD.

140.

United States. Congress. House. Committee of Conference., Food, Drug, and Cosmetic Act. 1938,
[s.n.]: Washington. p. 25 p.

239

141.

United States. Food and Drug Administration., Fair packaging & labeling act : FDA requirements
and interpretations manual. 1978, [Washington]: Dept. of Health, Education, and Welfare, Public
Health Service, Food and Drug Administration : for sale by the Supt. of Docs., U.S. Govt. Print.
Off. 320 p. in various pagings.

142.

United States. Congress. Senate. Committee on Labor and Human Resources., Nutrition Labeling
and Education Act of 1989 : hearing before the Committee on Labor and Human Resources,
United States Senate, One Hundred First Congress, first session on S. 1425 ... November 13, 1989.
1990, Washington: U.S. G.P.O. : For sale by the Supt. of Docs., Congressional Sales Office, U.S.
G.P.O. iii, 214 p.

143.

Center for Food Safety and Applied Nutrition (U.S.), Information for consumers Food Allergen
Labeling and Consumer Protection Act of 2004 : questions and answers. 2006, U.S. Food and
Drug Administration, Center for Food Safety and Applied Nutrition: [College Park, Md.]. p. 1
electronic text.

144.

Center for Food Safety and Applied Nutrition (U.S.), Food Allergen Labeling and Consumer
Protection Act of 2004, Public Law 108-282 report to the Committee on Health, Education, Labor,
and Pensions, United States Senate, and the Committee on Energy and Commerce, United States
House of Representatives. 2006, Dept. of Health and Human Services, Food and Drug
Administration, Center for Food Safety and Applied Nutrition: [College Park, Md.]. p. 33 p.

145.

Lancaster, K.J., NEW APPROACH TO CONSUMER THEORY. Journal of Political Economy, 1966.
74(2): p. 132-157.

146.

Stigler, G.J., The Economics of Information. Journal of Political Economy, 1961. 69(3): p. 213-225.

147.

Variyam, J.N., Do nutrition labels improve dietary outcomes? Health Economics, 2008. 17(6): p.
695-708.

148.

Heimbach, J.T. and R.C. Stokes, NUTRITION LABELING AND PUBLIC-HEALTH - SURVEY OF
AMERICAN-INSTITUTE-OF-NUTRITION MEMBERS, FOOD-INDUSTRY, AND CONSUMERS. American
Journal of Clinical Nutrition, 1982. 36(4): p. 700-708.

149.

Caswell, J.A., Current Information Levels on Food Labels. American Journal of Agricultural
Economics, 1992. 74(5): p. 1196-1201.

150.

Institute of Medicine (U.S.). Committee on the Nutrition Components of Food Labeling., D.V.
Porter, and R.O. Earl, Nutrition labeling : issues and directions for the 1990s. 1990, Washington,
D.C.: National Academy Press. xv, 355 p.

151.

Department of Health and Human Services Surgeon General's report on Nutrition and Health.
1988, DHHS.

152.

National Research Council (U.S.). Committee on Diet and Health., Diet and health : Implications
for reducing chronic disease risk. 1989, Washington, D.C.: National Academy Press. xiv, 749 p.

153.

Bendich, A. and R.J. Deckelbaum, Preventive nutrition : the comprehensive guide for health
professionals. 3rd ed. 2005, Totowa, N.J.: Humana Press. xxv, 958 p.

240

154.

Silverglade, B.A., Preemption - the Consumer Viewpoint. Food and Drug Law Journal, 1990. 45(2):
p. 143-149.

155.

Silverglade, B.A., A comment on public policy issies in health claims for foods. Journal of Public
Policy & Marketing, 1991. 54(55).

156.

United States. Congress. House. Committee on Energy and Commerce., Nutrition Labeling and
Education Act of 1990 report (to accompany H.R. 3562) (including cost estimate of the
Congressional Budget Office). 1990, U.S. G.P.O.,: [Washington, D.C.?

157.

United States. Congress. House. Committee on Energy and Commerce. Subcommittee on Health
and the Environment., Nutrition Labeling and Education Act hearing before the Subcommittee
on Health and the Environment of the Committee on Energy and Commerce, House of
Representatives, One Hundred First Congress, first session, on H.R. 3028, a bill to amend the
Federal Food, Drug, and Cosmetic Act to prescribe nutrition labeling for foods, August 3, 1989.
1989, U.S. G.P.O. : For sale by the Supt. of Docs., Congressional Sales Office, U.S. G.P.O.,:
Washington.

158.

Code of Federal Regulations Code of Federal Regulations. Title 21 Chapter 101.9. 2010 [cited
2010 August 17th]; Available from:
http://edocket.access.gpo.gov/cfr_2010/aprqtr/21cfr101.9.htm.

159.

Code of Federal Regulations Code of Federal Regulations. Title 21 Chapter 101.12. 2010 [cited
2010 August 17th]; Available from:
http://edocket.access.gpo.gov/cfr_2010/aprqtr/21cfr101.21.htm.

160.

Pappalardo, J.K., Evaluating the NLEA: Where's the beef? Journal of Public Policy & Marketing,
1996. 15(1): p. 153-156.

161.

Silverglade, B.A., The Nutrition Labeling and Education Act - Progress to date and challenges for
the future. Journal of Public Policy & Marketing, 1996. 15(1): p. 148-150.

162.

Petruccelli, P.J., Consumer and marketing implications of information provision: The case of the
Nutrition Labeling and Education Act of 1990. Journal of Public Policy & Marketing, 1996. 15(1):
p. 150-153.

163.

Zarkin, G.A., et al., Potential Health Benefits of Nutrition Label Changes. American Journal of
Public Health, 1993. 83(5): p. 717-724.

164.

Insel, P.M., R.E. Turner, and D. Ross, Nutrition. 2002 update. ed. 2002, Boston: Jones and
Bartlett Publishers. xxviii, 730, [200] p.

165.

US Food and Drug Administration, Trans fatty acids in nutrition labeling, Nutrient content claims
and health claims, in Code of Federal Regulations. 2003: US.

166.

Mojduszka, E.M., J.A. Caswell, and H. Michael, Consumer choice of food products and the
implications for price competition and government policy. Agribusiness, 2001. 17(1): p. 81-104.

241

167.

Mathios, A.D., The impact of mandatory disclosure laws on product choices: An analysis of the
salad dressing market. Journal of Law & Economics, 2000. 43(2): p. 651-677.

168.

Kramer, L., Ukrops is making nutrition labeling work, in Supermarket News. 1995.

169.

A Licking, in USA Today. 1994.

170.

Shoppers are reading and heeding labels, in Philadelpha Inquirer. 1995.

171.

Kramer, L., Nutrition labels help build deli, bakery sales (results of the Nutrition Labeling and
Education Act), in Supermarket News. 1995.

172.

Moorman, C., Market-level effects of information: Competitive responses and consumer
dynamics. Journal of Marketing Research, 1998. 35(1): p. 82-98.

173.

Moorman, C., R. Ferraro, and J. Huber, Unintended Nutrition Consequences: Firm Responses to
the Nutrition Labeling and Education Act. Marketing Science, 2012. 31(5): p. 717-737.

174.

Mojduszka, E.M., et al., Changes in the Nutritional Quality of Food Product Offerings and
Purchases: A Case Study in the Mid-1990's. 1999, Food and Rural Economics Division, Economic
Research Service, U.S. Department of Agriculture.

175.

Allshouse., J., B. Frazao., and J. Turpening., Are Americans Turning Away From Lower Fat Salty
Snacks?, in FoodReview. 2002.

176.

Balasubramanian, S.K. and C. Cole, Consumers' search and use of nutrition information: The
challenge and promise of the Nutrition Labeling and Education Act. Journal of Marketing, 2002.
66(3): p. 112-127.

177.

Todd, J. and J.N. Variyam, Decline in consumer use of nutrition labels, 1995-2006. 2008,
Economic Research Service, US Department of Agriculture.

178.

Keller, S.B., et al., The effects of nutrition package claims, nutrition facts panels, and motivation
to process nutrition information on consumer product evaluations. Journal of Public Policy &
Marketing, 1997. 16(2): p. 256-269.

179.

Moorman, C., A quasi-experiment to assess the consumer and informational determinants of
nutrition information processing activities: The case of the nutrition labeling and education act.
Journal of Public Policy & Marketing, 1996. 15(1): p. 28-44.

180.

Szykman, L.R., P.N. Bloom, and A.S. Levy, A proposed model of the use of package claims and
nutrition labels. Journal of Public Policy & Marketing, 1997. 16(2): p. 228-241.

181.

Drichoutis, A.C., P. Lazaridis, and R.M. Nayga, Nutrition knowledge and consumer use of
nutritional food labels. European Review of Agricultural Economics, 2005. 32(1): p. 93-118.

182.

Kim, S.-Y., R.M. Nayga, Jr., and O. Capps, Jr., Health Knowledge And Consumer Use Of Nutritional
Labels: The Issue Revisited. Agricultural and Resource Economics Review, 2001. 30(1): p. 10-19.

242

183.

Drichoutis, A.C., P. Lazaridis, and R.M. Nayga, Consumers' Use of Nutritional Labels: A Review of
research studies and issues. Academy of Marketing Science, 2006. 2006(9).

184.

Byrd-Bredbenner, C. and A. Wong, Consumer understanding of US and EU nutrition labels.
British Food Journal, 2000. 102: p. 615-629.

185.

Kim, S., R. Nayga, and O. Capps, Food label use, self-selectivity and diet quality. Journal of
Consumer Affairs, 2001. 35(346-363).

186.

Weaver, D. and A. Finke, The relationship between the use of sugar content information on
nutrition labels and the consumption of added sugars. Food Policy, 2003. 28(3): p. 213-219.

187.

Lin, C.T.J., J.Y. Lee, and S.T. Yen, Do dietary intakes affect search for nutrient information on food
labels? Social Science & Medicine, 2004. 59(9): p. 1955-1967.

188.

Neuhouser, M., A. Kristal, and R. Patterson, Use of food nutrtion labels is associated with lower
fat intake. Journal of the American Dietetic Association, 1999. 99: p. 45-53.

189.

Fitzgerald, N., et al., Nutrition knowledge, food label use, and food intake patterns among
Latinas with and without type 2 diabetes. Journal of the American Dietetic Association, 2008.
108(6): p. 960-967.

190.

Kral, T.V.E., L.S. Roe, and B.J. Rolls, Does nutrition information about the energy density of meals
affect food intake in normal-weight women? Appetite, 2002. 39(2): p. 137-145.

191.

Kristal, A.R., et al., Predictors of self-initiated, healthful dietary change. Journal of the American
Dietetic Association, 2001. 101(7): p. 762-766.

192.

Coulson, N.S., An application of the stages of change model to consumer use of food labels.
British Food Journal, 2000. 102(9): p. 661-668.

193.

Govindasamy, R. and J. Italia, The influence of consumer demographic characteristics on
nutritional label usage. Journal of Food Products marketing, 1999. 5(4): p. 55-68.

194.

Cole, C.A. and S.K. Balasubramanian, Age-Differences in Consumers Search for Information Public-Policy Implications. Journal of Consumer Research, 1993. 20(1): p. 157-169.

195.

Satia, J.A., J.A. Galanko, and M.L. Neuhouser, Food nutrition label use is associated with
demographic, behavioral, and psychosocial factors and dietary intake among African Americans
in north Carolina. Journal of the American Dietetic Association, 2005. 105(3): p. 392-402.

196.

John, D.R. and C.A. Cole, Age-Differences in Information-Processing - Understanding Deficits in
Young and Elderly Consumers. Journal of Consumer Research, 1986. 13(3): p. 297-315.

197.

Cole, C.A. and G.J. Gaeth, Cognitive and Age-Related Differences in the Ability to Use Nutritional
Information in a Complex Environment. Journal of Marketing Research, 1990. 27(2): p. 175-184.

198.

Burton, S. and J.C. Andrews, Age, product nutrition, and label format effects on consumer
perceptions and product evaluations. Journal of Consumer Affairs, 1996. 30(1): p. 68-89.

243

199.

Guthrie, J.F., et al., Who Uses Nutrition Labeling, and What Effects Does Label Use Have on Diet
Quality. Journal of Nutrition Education, 1995. 27(4): p. 163-172.

200.

Feick, L.F., R.O. Herrmann, and R.H. Warland, Search for Nutrition Information - a Probit Analysis
of the Use of Different Information-Sources. Journal of Consumer Affairs, 1986. 20(2): p. 173192.

201.

McLean-Meyinsse, P.E., An analysis of nutritional label use in Southern United States. Journal of
Food Distribution Researh, 2001. 32(1): p. 110-114.

202.

Nayga, R.M., D. Lipinski, and N. Savur, Consumers' use of nutritional labels while food shopping
and at home. Journal of Consumer Affairs, 1998. 32(1): p. 106-120.

203.

Wang, G.J., S.M. Fletcher, and D.H. Carley, Consumer Utilization of Food Labeling as a Source of
Nutrition Information. Journal of Consumer Affairs, 1995. 29(2): p. 368-380.

204.

Nayga, R.M., Jr., Toward an understanding of consumers' perceptions of food labels.
International Food and Agribusiness Management Review, 1999. 2(1): p. 29-45.

205.

Nayga, R.M., Jr., Determinants of consumers' use of nutritional information on food packages.
Journal of Agricultural and Applied Economics, 1996. 28(2): p. 303-312.

206.

Beatty, S.E. and S.M. Smith, External Search Effort - an Investigation across Several Product
Categories. Journal of Consumer Research, 1987. 14(1): p. 83-95.

207.

Katona, G.C. and E. Mueller, A study of purchase decisions, in Consumer Behavior: The dynamics
of consumer reactions, L.H. Clark, Editor. 1955, New York University Press: New York. p. 30-87.

208.

Park, C.W., E.S. Iyer, and D.C. Smith, The Effects of Situational Factors on in-Store Grocery
Shopping Behavior - the Role of Store Environment and Time Available for Shopping. Journal of
Consumer Research, 1989. 15(4): p. 422-433.

209.

Piedra, M.A., A.R. Schupp, and D.E. Montgomery, Consumer use of nutritional labels on
packaged meats. Journal of Food Distribution Researh, 1996. 27(2): p. 42-47.

210.

Schupp, A.R., J. Gillespie, and D. Reed, Consumer awareness and use of nutrition labels on
packaged fresh meats: A Pilot Study. Journal of Food Distribution Researh, 1998. 29(2): p. 24-30.

211.

Nayga, R., Nutrition knowledge, gender and food label use. Journal of Consumer Affairs, 2000.
341: p. 97-112.

212.

Shine, A., S. O'Rielly, and K. O'Sullivan, Consumer use of nutrition labelling. British Food Journal,
1997(99): p. 290-296.

213.

Driskell, J.A., M.C. Schake, and H.A. Detter, Using Nutrition Labeling as a Potential Tool for
Changing Eating Habits of University Dining Hall Patrons. Journal of the American Dietetic
Association, 2008. 108(12): p. 2071-2076.

244

214.

Roberto, C.A., et al., Evaluating the Impact of Menu Labeling on Food Choices and Intake.
American Journal of Public Health, 2010. 100(2): p. 312-318.

215.

Albright, C.L., J.A. Flora, and S.P. Fortmann, Restaurant Menu Labeling - Impact of Nutrition
Information on Entree Sales and Patron Attitudes. Health Education Quarterly, 1990. 17(2): p.
157-167.

216.

Burton, S., E. Howlett, and A.H. Tangari, Food for Thought: How Will the Nutrition Labeling of
Quick Service Restaurant Menu Items Influence Consumers' Product Evaluations, Purchase
Intentions, and Choices? Journal of Retailing, 2009. 85(3): p. 258-273.

217.

Roberto, C.A., et al., An observational study of consumer use of fast-food restaurant drivethrough lanes: implications for menu labelling policy. Public Health Nutrition, 2010. 13(11): p.
1826-1828.

218.

Swartz, J.J., D. Braxton, and A.J. Viera, Calorie menu labeling on quick-service restaurant menus:
an updated systematic review of the literature. International Journal of Behavioral Nutrition and
Physical Activity, 2011. 8.

219.

Tandon, P.S., et al., Nutrition Menu Labeling May Lead to Lower-Calorie Restaurant Meal
Choices for Children. Pediatrics, 2010. 125(2): p. 244-248.

220.

Brug, J., et al., The internet and nutrition education: challenges and opportunities. European
Journal of Clinical Nutrition, 2005. 59: p. S130-S137.

221.

Lang, J.E., et al., Use of a supermarket shelf-labeling program to educate a predominately
minority community about foods that promote heart health. Journal of the American Dietetic
Association, 2000. 100(7): p. 804-809.

222.

Jeffery, R.W., et al., Nutrition Education in Supermarkets - an Unsuccessful Attempt to Influence
Knowledge and Product Sales. Journal of Behavioral Medicine, 1982. 5(2): p. 189-200.

223.

Levy, A.S., et al., The Impact of a Nutrition Information Program on Food Purchases. Journal of
Public Policy & Marketing, 1985. 4: p. 1-13.

224.

Hamburg, M. Keynote Address: National Food Policy Conference. Speeches by FDA officials 2009
[cited 2013 January 21st]; Available from:
http://www.fda.gov/NewsEvents/Speeches/ucm182061.htm.

225.

Silverglade, B., Food Labeling Chaos: The case for reform. 2010, Center for Science in the Public
Interest: Washington D C.

226.

Kreuter, M., et al., Do nutrition label readers eat healthier diets? Behavioral correlates of adults'
use of food labels. American Journal of Preventive Medicine, 1997. 13: p. 277-283.

227.

Kristal, A., et al., Predictors of self-initiated healthful dietary changes. Journal of the American
Dietetic Association, 2001. 101: p. 762-766.

245

228.

Marietta, A., K. Welshimer, and S. Anderson, Knowledge, attitudes, behaviors of college students
regarding the 1990 nutrition labeling education act food labels. Journal of the American Dietetic
Association, 1999. 99: p. 445-449.

229.

Garretson, J. and S. Burton, Effects of nutrition facts panel values, nutrition claims, and health
claims on consumer attitudes, perceptions of disease-related risks, and trust. Journal of Public
Policy and Marketing, 2000. 19: p. 213-227.

230.

Levy, L., et al., How well do consumers understand percentage daily value on food labels?
American Journal of Health Promotion, 2000. 14: p. 157-160.

231.

Roe, B., A. Levy, and B. Derby, The impact of health claims on consumer search and product
evaluation outcomes: Results from FDA experimental data. Journal of Public Policy and
Marketing, 1999. 18: p. 89-105.

232.

Food Standards Agency Support grows for FSA’s recommended traffic light colour signpost
labelling, in FSA News. 2006, Food Standards Association: UK.

233.

Louie, J.C.-Y., et al., A comparison of two nutrition signposting systems for use in Australia, in
Faculty of Health & Behavioural Sciences - Papers. 2008, University of Wollongong: Sydney.

234.

Feunekes, G., et al., Front-of-pack nutrition labelling: Testing effectiveness of different labelling
formats front-of-pack in four European countries. Appetite, 2008. 50: p. 57-70.

235.

Feunekes, G.I., et al., Front-of-pack nutrition labelling: Testing effectiveness of different nutrition
labelling formats front-of-pack in four European countries. Appetite, 2008. 50(1): p. 57-70.

236.

Kelly, B., et al., Consumer testing of the acceptability and effectiveness of front of pack food
labelling systems for the Australian grocery market. Health Promotion International, 2009. 24(2):
p. 120-130.

237.

Sacks, G., M. Rayner, and B. Swinburn, Impact of front of pack traffic light nutrition labeling on
consumer food purchases in the UK. Health Promotion International, 2009. 24(4): p. 344-353.

238.

Heart and Stroke Foundation. What is Health-Check? 2010 [cited 2010 November 26th];
Available from: http://www.healthcheck.org/page/what-health-check.

239.

Whole Grains Council. Whole Grains Stamp. 2013 [cited 2013 February 5th]; Available from:
http://wholegrainscouncil.org/whole-grain-stamp.

240.

Hersey, J., et al., Policy Research for Front of Package Nutrition Labeling: Environmental Scan
and Literature Review. 2011, US Department of Health and Human Services.

241.

Food Standards Association Support grows for FSA’s recommended traffic light colour signpost
labelling, in FSA News. 2006, Food Standards Association: UK.

242.

Food Standards Association More big names adopt traffic light colour signpost labelling, in FSA
News. 2006, Food Standards Association: UK.

246

243.

Food Standards Agency Signing up to make healthier earing an easier option, in FSA News. 2007,
Food Standards Association: UK.

244.

Food and Drink Federation. GDA's Explained. 2010 [cited 2010 November 18th]; Available from:
http://www.gdalabel.org.uk/gda/explained.aspx.

245.

Food and Drink Federation. GDA Labels: Why not colour code? 2010 [cited 2010 November
18th]; Available from: http://www.fdf.org.uk/health_wellbeing/GDA_labels/colour.aspx.

246.

Food and Drink Federation. Nutrition Labelling and Signposting Schemes: Policy Position. 2010
[cited 2010 November 18th]; Available from: http://www.fdf.org.uk/keyissues.aspx?issue=628.

247.

Cellan-Jones, R., Supermarkets in label wars. 2006, BBC News: London.

248.

Cooper, B., Opinion still polarized on front of pack nutrition labeling, in Just-Food. 2010.

249.

Baynton, C. and C. Boville, Front of Pack (FOP) Nutrition Labeling: Annexes to Board Paper. 2010,
Food Standards Association: Cardiff.

250.

European Union, Food Labeling - EU 1169/2011. 2011.

251.

Food Standards Agency. Front of Pack Labeling - Consulation. 2012 [cited 2013 February 1st];
Available from: http://www.food.gov.uk/scotland/scotnut/signposting/#.UQrroichiuI.

252.

Food Standards Agency, Front of Pack nutrition Labelling - Joint response to consulation. 2013.

253.

Triggle, N. Food labelling: Consistent system 'to start next year'. 2012 [cited; Available from:
http://www.bbc.co.uk/news/health-20050420.

254.

Campbell, D. All supermarkets to adopt 'traffic-light labelling' for nutrition. 2012 [cited 2013
January 28th]; Available from: http://www.guardian.co.uk/business/2012/oct/24/supermarketstraffic-light-labelling-nutrition.

255.

Poulter, S. Farmers and dairy companies attack traffic light food label system that 'will demonise
meat and cheese' Mail Online 2012 [cited 2013 Jan 28th]; Available from:
http://www.dailymail.co.uk/news/article-2221622/Farmers-dairy-companies-attack-traffic-lightfood-label-demonise-meat-cheese.html.

256.

Blythman, J. Why traffic light labels on food will make us all fatter AND ruin our farmers. Mail
Online 2012 [cited 2013 Jan 28th]; Available from: http://www.dailymail.co.uk/news/article2221622/Farmers-dairy-companies-attack-traffic-light-food-label-demonise-meat-cheese.html.

257.

European Commission, Commission proposal to overhaul EU food labelling rules. 2008, European
Commission: Brussels.

258.

European Parliament. Food labels: rapporteur Renate Sommer speaks after vote. 2010 [cited
2010 November 18th].

247

259.

Hick, M., Victory for food lobby as MEPs reject new labelling rules, in The Independant. 2010:
London.

260.

Smithers, R., MEPs reject 'traffic light' food labelling system, in Guardian. 2010: London.

261.

EUFIC. Nutrition labelling becomes mandatory in Europe. Food Today 2012 [cited 2013 Jan
28th]; Available from: http://www.eufic.org/article/en/artid/Nutrition-labelling-becomesmandatory-in-Europe/.

262.

Starling, S., Pepsi's 'Smart Spot' enjoys strong growth, in Functional Ingredients. 2005, Penton
Media, Inc.

263.

Packaging Digest, Kellogg changes nutrition standards, adds front-of-pack labeling. 2007, UBM
Canon.

264.

Mars Snackfood Inc. Mars first confectionery company to launch front of pack nutrition labeling
globally. 2008 [cited 2010 November 17th]; Available from:
http://www.mars.com/global/news-and-media/press-releases/newsreleases.aspx?SiteId=94&Id=1473.

265.

Food and Drug Administration. Transcript for FDA's Media Briefing on Front-of-Pack Labeling.
Media Call Transcript FDA Position on Front-of-Package Labeling. 2009 [cited 2010 November
17th]; Available from:
http://www.fda.gov/downloads/NewsEvents/Newsroom/MediaTranscripts/UCM187809.pdf.

266.

Food and Drug Administration. FDA Calls on Food Companies to Correct Labeling Violations; FDA
Commissioner Issues an Open Letter to the Industry. 2010 [cited 2010 November 17th];
Available from:
http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm202814.htm.

267.

Hamburg, M. Open Letter to Industry from Dr. Hamburg. 2010 [cited 2010 November 17th];
Available from:
http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm202814.htm.

268.

Kennedy, B. Food and Beverage Industry Announces Front-of-Pack Nutrition Labeling Initiative to
Inform Consumers and Combat Obesity. 2010 [cited 2010 November 22nd]; Available from:
http://www.gmaonline.org/news-events/newsroom/food-and-beverage-industry-announcesfront-of-pack-nutrition-labeling-initi/.

269.

Grocery Manufacturer's Association. Facts-up-Front Overview. 2013 [cited 2013 February 3rd].

270.

Grocery Manufacturer's Association, Facts-up-Front - Fact Sheet. 2013, Grocery Manufacturer's
Association.

271.

Lukovitz, K., Industry To Implement Own Package Label System, in Marketing Daily. 2010.

272.

Watson, E. IOM calls for 'fundamental shift' in approach to front-of-pack nutrition labeling. 2011
[cited 2013 February 3rd]; Available from: http://www.foodnavigator-usa.com/Market/IOMcalls-for-fundamental-shift-in-approach-to-front-of-pack-food-labeling.

248

273.

Kennedy, B. Grocery Manufacturers Association Statement in Response to the Institute of
Medicine’s Committee on Examination of Front-of-Package Nutrition Rating Systems and
Symbols Report. 2011 [cited 2013 February 3rd]; Available from:
http://www.gmaonline.org/news-events/newsroom/gma-statement-in-response-to-the-iomfront-of-package-nutrition-rating-syst/.

274.

Bellatti, A. and M. Simon, Regulating Front of Package Labeling: An Exercise in Futility? Journal
of Hunger & Environmental Nutrition, 2012. 6(4): p. 513-523.

275.

UK Department of Health Choosing Health: Making healthy choices easier. 2004, National Health
Service, UK.

276.

Food Standards Agency, Concept Testing of Alternative Labelling of Healthy / Less Healthy Foods.
2004, Food Standards Agency,: Bucks, UK.

277.

Food Standards Agency, Quantitative evaluation of alternative food signposting schemes. 2005,
Food Standards Agency,: Bucks, UK.

278.

Food Standards Agency, Qualitative Signpost Labelling Refinement Research. 2005, Food
Standards Agency,: Bucks, UK.

279.

Food Standards Agency, Citizens' forums on food: Front of Pack Labeling (FOP) nutrition
labelling. 2009, Food Standards Agency,: Bucks, UK.

280.

Mirror Group GDA Mirror Group Survey. 2008, Mirror Group: London,UK.

281.

Malcom, K., et al., How has awareness, comprehension and usage of GDA labelling evolved?
2008, Millward Brown.

282.

Ellis, R.M. and R.C. Ellis, Impact of a traffic light nutrition tool in a primary school. J R Soc Promot
Health, 2007. 127(1): p. 13-21.

283.

Murphy, M., S. Fallows, and G. Bonwick, Parents use and understanding of front-of-pack food
labelling and the impact of socio-economic status. Proceedings of the Nutrition Society, 2008.
67(OCE8).

284.

Food Safety Authority of Ireland A research study into consumers' attitudes towards food
labelling. 2009, Food Safety Authority of Ireland.

285.

van Herpen, E., E. Seiss, and H.C.M. van Trijp, The role of familiarity in front-of-pack label
evaluation and use: A comparison between the United Kingdom and The Netherlands. Food
Quality and Preference, 2012. 26(1): p. 22-34.

286.

Grunert, K. and J. Wills, A review of European research on consumer response to nutrition
information on food labels. Journal of Public Health, 2007. 15(5): p. 385-399.

287.

Grunert, K.G., L. Fernandez-Celemin, and J.M. Wills, Use and understanding of nutrition
information on Food labels in six European countries. Journal of Public Health, 2009.

249

288.

Borgmeier, I. and J. Westenhoefer, Impact of different food label formats on healthiness
evaluation and food choice of consumers: a randomized-controlled study. Bmc Public Health,
2009. 9.

289.

Storcksdieck Genannt Bonsmann, S., et al., Penetration of nutrition information on food labels
across the EU-27 plus Turkey. Eur J Clin Nutr, 2010.

290.

Hodgkins, C., et al., Understanding how consumers categorise nutritional labels: A consumer
derived typology for front-of-pack nutrition labelling. Appetite, 2012. 59(3): p. 806-817.

291.

Mhurchu, C.N. and D. Gorton, Nutrition labels and claims in New Zealand and Australia: a review
of use and understanding. Australian and New Zealand Journal of Public Health, 2007. 31(2): p.
105-112.

292.

Maubach, N. and J. Hoek. The effect of alternative nutrition information formats on consumers'
evaluations of a children's breakfast cereal. in International Nonprofit and Social Marketing
Conference. 2008. University of Wollongong.

293.

Mackintosh, A.-M., Front of Pack Labeling, in The Tick Update. 2008, Heart Foundation of
Australia: Sydney.

294.

Carter, O., et al., An independent audit of the Australian food industry's voluntary front-of-pack
nutrition labelling scheme for energy-dense nutrition-poor foods. European Journal of Clinical
Nutrition, 2013. 67(1).

295.

Fair Packaging and Labeling Act in USC Title 15; Chapter 39. 1966: US.

296.

Bix, L., et al., Examining the conspicuousness and prominence of two required warnings on OTC
pain relievers. Proceedings of the National Academy of Sciences of the United States of America,
2009. 106(16): p. 6550-6555.

297.

Bix, L., et al., The use of change detection as a method of objectively evaluating labels. Packaging
Technology and Science, 2010. 23: p. 393-401.

298.

Sundar, R.P., The effect of color contrast on the noticeability and legibility of Prescription Drug
Warning Labels (PWL's), in Packaging. 2009, Michigan State University: East Lansing.

299.

Rogers, W.A. and G.K. Rousseau, Maximizing the effectiveness of the warning process:
Understanding the variables that interact with age, in Processing of medical information in aging
patients: Cognitive and human factors perspectives, R.W. Morrell, K. Shifren, and P. Denise,
Editors. 1999, Psychology Press.

300.

Dehenau, C., Applying Change Detection to Test the Noticeability of Components of Medical
Labels, in School of Packaging. 2010, Michigan State University: East Lansing.

301.

Gregory, R.L. and A.M. Colman, Sensation and perception. Longman essential psychology series.
1995, New York: Longman Group. xv, 135 p.

302.

Rensink, R.A., Change detection. Annu Rev Psychol, 2002. 53: p. 245-77.

250

303.

Henderson, J. and A. Hollingworth, The Role of Fixation Position in Detecting Scene Changes
Across Saccades. Psychological Science, 1999. 10(5): p. 438-443.

304.

Rensink, R.A., J.K. O'Reagan, and J.J. Clark, To See or Not to See: The Need for Attention to
Perceive Changes in Scenes. Psychological Science, 1997. 8: p. 368-373.

305.

Simons, D.J., In Sight, Out of Mind: When object representations fail. Psychological Science,
1996. 7: p. 301-305.

306.

Simons, D.J. and M.S. Ambinder, Change Blindness: Theory and Consequences. Current
Directions in Psychological Science, 2005. 14(1): p. 44-48.

307.

Simons, D.J. and R.A. Rensink, Change blindness: past, present, and future. Trends Cogn Sci,
2005. 9(1): p. 16-20.

308.

Bix, L., et al., The use of change detection as a method of objectively evaluating labels. Packaging
Technology and Science, 2010(23): p. 393-401.

309.

Duchowski, A.T., Eye tracking methodology theory and practice. 2007, Springer: London. p. xxiv,
328 p.

310.

Graham, D.J., J. Orquin, and V. Visschers, Eye tracking and nutrition label use: A review of the
literature and recommendations for label enhancement. Food Policy, 2012 37(4): p. 378–382.

311.

Rawson, D., I. Janes, and K. Jordan, Pilot Study to Investigate the potential of eye tracking as a
technique for FSA food labelling behaviour research. 2008, Food Standards Agency.

312.

Langton, S.R.H., et al., Attention capture by faces. Cognition, 2008. 107(1): p. 330-342.

313.

Mack, A., et al., What we see: Inattention and the capture of attention by meaning.
Consciousness and Cognition: An International Journal, 2002. 11(4): p. 488-506.

314.

Bishop, S.J., R. Jenkins, and A.D. Lawrence, Neural processing of fearful faces: Effects of anxiety
are gated by perceptual capacity limitations. Cerebral Cortex, 2007. 17(7): p. 1595-1603.

315.

de Gelder, B., et al., Non-conscious recognition of affect in the absence of striate cortex.
NeuroReport: For Rapid Communication of Neuroscience Research, 1999. 10(18): p. 3759-3763.

316.

Whalen, P.J., et al., Masked presentations of emotional facial expressions modulate amygdala
activity without explicit knowledge. The Journal of Neuroscience, 1998. 18(1): p. 411-418.

317.

Paivio, A., Mental comparisons involving abstract attributes. Memory & Cognition, 1978. 6(3): p.
199-208.

318.

Greene, M.R. and A. Oliva, The briefest of glances: The time course of natural scene
understanding. Psychological Science, 2009. 20(4): p. 464-472.

319.

Shepard, R.N., Recognition Memory for Words, Sentences, and Pictures. Journal of Verbal
Learning & Verbal Behavior, 1967. 6(1): p. 156-163.

251

320.

Becker, M.W. and I.P. Rasmussen, Guidance of attention to objects and locations by long-term
memory of natural scenes. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 2008. 34(6): p. 1325-1338.

321.

Food Standards Agency Front-of-Pack Traffic light signpost labelling technical guidance - Issue 2.
2007, Food Standards Agency.

322.

Ford, G.T., et al., Can consumers interpret nutrition information in the presence of a health
claim? A laboratory investigation. Journal of Public Policy & Marketing, 1996. 15(1): p. 16-27.

323.

Psychology Software Tools inc Eprime 2. 2011 [cited 2011 January 4th]; Available from:
http://www.pstnet.com/eprimepublications.cfm.

324.

Cohen, J., Statistical power analysis for the behavioral sciences. 2nd ed. 1988, Hillsdale, N.J.: L.
Erlbaum Associates. xxi, 567 p.

325.

Stroup, W.W., Power Analysis Based on Spatial Effects Mixed Models: A Tool for Comparing
Design and Analysis Strategies in the Presence of Spatial Variability. Journal of Agricultural,
Biological, and Environmental Statistics, 2002. 7(4): p. 491-511

326.

Tempelman, R.J., Assessing experimental designs for research conducted on commercial dairies.
Journal of Dairy Science, 2009. 92(1): p. 1-15.

327.

Bello, N., Power Calculations for School of Packaging, NIH r21 Grant Application. 2010: Lansing.

328.

Nutrition Quest. Block Food Frequency Questionnaire. 2013 [cited 2013 30th March]; Available
from: http://nutritionquest.com/assessment/.

329.

Wansink, B., How do front and back package labels influence beliefs about health claims? The
Journal of Consumer Affairs, 2003. 37(2): p. 305-31.

330.

Federal Register [Notices]. 229(74).

331.

Applied Science Laboratories Mobile Eye. 2011 [cited 2011 January 4th]; Available from:
http://www.asleyetracking.com/Site/Products/MobileEye/tabid/70/Default.aspx.

332.

Sundar, R., The effect of color contrast of text on the legibility and noticeability of prescription
drug labels, in Packaging. 2009, Michigan State University East Lansing.

333.

Bix, L., et al., Examining the conspicuousness and prominence of two required warnings on OTC
pain relievers. Proceedings of the National Academy of Sciences of the United States of America,
2009. 106(16): p. 6550-6555.

334.

EUFIC, EUFIC Press Release from 25/09/2008: One in four UK consumers look for nutrition
information on food labels. . 2008.

335.

Block, G., A. Hartman, and D. Naughton, A reduced dietary questionnaire: Development and
validation Epidemiology, 1990. 1(1): p. 58-64.

252

336.

Lohman, T., A. Roche, and R. Martorell, Anthropometric standardization reference manual, in
Human Kinetics Book. 1988: Champaign, IL.

337.

Schuldt, J.P., Nutrition Label Color Affects Perceptions of Healthfulness. Health Communications,
2013. 1(8).

338.

Becker, M.W., Panic search: Fear produces efficient visual search for nonthreatening objects.
Psychological Science, 2009. 20(4): p. 435-437.

339.

Becker, M.W. and B. Detweiler-Bedell, Early detection and avoidance of threatening faces during
passive viewing. The Quarterly Journal of Experimental Psychology, 2009. 62(7): p. 1257-1264.

340.

Rogers, W., G. Rousseau, and N. Lamson, Maximizing the Effectiveness of the Warning Process:
Understanding the Variables that Interact with Age, in Aging and Medication Information
Processing, R. Morrell and D. Park, Editors. 1998.

341.

Center for Disease Control and Prevention. Higher education and income levels keys to better
health, according to annual report on nation's health. 2012 [cited 2013 2nd April]; Available
from: http://www.cdc.gov/media/releases/2012/p0516_higher_education.html.

342.

Ogden, C., et al., Obesity and Socioeconomic Status in Adults: United States, 2005–2008, in NCHS
Data Brief. 2010, Center for Disease Control and Prevention.

343.

Ogden, C., et al., Obesity and Socioeconomic Status in Children and Adolescents: United States,
2005–2008, in NCHS Data Brief. 2010, Center for Disease Control and Prevention.

344.

Flegal, K., et al., Prevalence and trends in obesity among US adults, 1999-2008. Journal of the
American Medical Association, 2010. 303(3): p. 235-241.

253