University of Southern California Dissertations and Theses

Food choice dimensions and the relationship with BMI

(USC Thesis Other)

Food choice dimensions and the relationship with BMI

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: FOOD CHOICE DIMENSIONS AND BMI 1

Food choice dimensions and the relationship with BMI
Elyse Plotin
University of Southern California

FOOD CHOICE DIMENSIONS AND BMI 2

Table of Contents
Abstract……………………………………………………………………………………………3
Food choice dimensions and the relationship with BMI………………………………………….4
Method…………………………………………………………………………………………...11
Participants…………………………………………………………………………….....11
Materials and Procedure…………………………………………………………………13
Food Choice Dimensions………………………………………………………………...14
BMI………………………………………………………………………………………16
Results……………………………………………………………………………………………16
Discussion………………………………………………………………………………………..22
References………………………………………………………………………………………..29
Tables…………………………………………………………………………………………….31
Appendix A………………………………………………………………………………………38
Appendix B……………………………………………………………………………………....49

FOOD CHOICE DIMENSIONS AND BMI 3

Abstract
This study set out to determine whether people of different BMIs focus on different Food Choice
Dimensions when selecting which foods to eat, and additionally, if these differences could
predict BMI. The study consisted of 581 participants ranging in age from 18-75, and ranging in
BMI from 18-44. Seven Food Choice Dimensions – caloric content, convenience, familiarity,
nutritional value, price, social demands, taste – were examined with three different measures all
asking participants to, in some way, introspectively identify which dimensions, or factors, are
important in selecting food. The three measures consisted of: a Food Choice Questionnaire, a
rank ordering task, and a relative weighting task to disperse 100 points across the seven
dimensions. Between-subjects effects verified our hypotheses that the main differences between
BMI groups would lie in weight management factors and an external factor such as price.
Nutritional Value and Price proved to be the main dimensions of important differences among
healthy weight individuals and obese individuals, whereby healthy weight individuals placed
more importance on Nutritional Value and obese individuals placed more importance on Price.
Consistent with the ANOVA results, stepwise multiple regression analyses showed Nutritional
Value and Price as the two main contributing factors for predicting individual differences in BMI
when FCQ scores and Relative Weights were used as the predictor variables.

FOOD CHOICE DIMENSIONS AND BMI 4

Food choice dimensions and the relationship with BMI
Why do people eat the foods they do? What factors influence people’s decisions when
selecting food? The present study was designed to analyze whether people who are of a healthy
weight base their food selection choices on different factors compared to those who are
overweight or obese. By examining the influence of “food choice dimensions,” the present study
also aimed to determine whether the factors people utilize in their food selections can predict
their body mass index (BMI), as categorized by the National Institutes of Health, as either within
a normal range (18.5-24.9), overweight (25.0-29.9), or obese (>30). It is hypothesized that
people at a normal, healthy weight will utilize food choice dimensions – factors describing the
prominent features of the food itself or the external attributes of food – primarily concerning
weight management compared to those who are overweight or obese, who might be more
focused on the aesthetic features or on the external attributes such as price. Specifically, it is
predicted that those at a healthy weight will be concerned with the factors of caloric content and
nutritional value, thereby reflecting a negative relationship with BMI. In contrast, it is
hypothesized that obese individuals will be less focused on the factors relating specifically to
weight management and that this lack of importance will be reflected in a stronger negative
relationship predicting high BMI. It is also hypothesized that the obese individuals will focus
more on an external factor such as price, thereby demonstrating a positive relationship with BMI.
Overweight and obesity in America has become a gargantuan problem for individuals and
society alike. With the prevalence of overweight and obesity having risen substantially over the
past three decades, 69% of Americans are now overweight (BMI > 25), and 36% of Americans
are medically classified as obese (BMI >30) (Flegal, Carroll, Kit, & Ogden, 2012). Not only does
being overweight place an individual at higher risks for a range of diseases including heart
FOOD CHOICE DIMENSIONS AND BMI 5

disease, cardiovascular disease, diabetes, hypertension, hypercholesterolemia, stroke, and even
certain kinds of cancers (Flegal, Carroll, Ogden, & Curtin, 2010; World Health Organization
[WHO], 2000), but it can also be a strain on the economy and society. The most recent estimates
indicate that obesity-related medical illness is responsible for almost 10% of total U.S. medical
expenditures, or, in fiscal terms, about $147 billion (Finkelstein, Trogdon, Cohen, & Dietz,
2009). Furthermore, the external costs to society are enhanced by the fact that government health
insurance programs often fund at least 20% of these costs (Finkelstein et al., 2009). This growing
percentage of the adult overweight and obese population is not only straining country
expenditures, but it is also setting a negative precedent for the younger population, who, if
following the course of U.S. adults aged 20 years or older, will likely see themselves in the
categories of overweight and obese as well. Even worse is the number of overweight and obese
school children – current estimates suggest 40% of US children are overweight, with 20%
classified as obese (Organisation for Economic Co-operation and Development [OECD], 2012) –
who not only face health risks at early ages, but also display decreased verbal skills, social skills,
and poor academic performance (Cawley, 2011).
What is leading to these astronomical numbers of growing obesity and medical
expenditures in the United States? From spending habits and food prices, to food consumption
behaviors, the rise in obesity in the United States is a multifaceted problem. In the simplest of
terms, weight gain occurs due to a mismanagement of calorie intake versus calorie expenditure
(Cutler, Glaeser, & Shapiro, 2004; Finkelstein & Zuckerman, 2008; WHO, 2000). Excess
calories without increased energy expenditure inevitably add up to weight gain – 3500 excess
calories translates into one pound of weight gained (Cutler et al., 2004; Finkelstein &
Zuckerman, 2008). The excess calories in Americans’ diets come from the combinations of
FOOD CHOICE DIMENSIONS AND BMI 6

eating more, eating the wrong foods, and moving less. In the past few decades, the observed
increase in median American weights required a net caloric imbalance of 100-150 calories per
day (Cutler et al., 2004). Whether this imbalance is from the can of regular soda consumed or the
absence of the mile and a half walk, the changed American lifestyle leaves little mystery as to
why the country has become overweight.
The food industry has laid a lot of the foundation for the changes in American diets.
From lower food prices of energy-dense, low-nutrient foods to increased portion sizes, there is a
vast availability and variety of unhealthy food options. As Finkelstein and Zuckerman (2008)
highlight, the prices of healthy foods such as fish, fruits, and vegetables have increased
dramatically since 1983 compared to energy-dense foods that contain a lot of fat, oil, and sugar.
In fact, compared to the prices of all foods, the unhealthy alternatives have become cheaper.
These cheaper, unhealthier options are also more widely available at fast-food and restaurant
locations and more widely consumed than ever before – almost 50% of food expenditures comes
from foods eaten away from the home (Bowman & Vinyard, 2004; Cutler et al., 2004). Yet
placing the blame on the food industry suggests people have no control in what they themselves
are consuming. The food industry is not forcing the average American to eat at the fast food
restaurant across the street from work or use the convenient vending machine for a high-calorie
midday snack. People make many more food decisions a day than they even realize (Wansink &
Sobal, 2007) and the growing trends of overweight and obesity suggest they are simply making
the wrong food decisions as far as weight management is concerned. Nevertheless, there has
been limited research directed at these specific decisions people make concerning food choices.
While prices, mass production of easily available prepackaged foods, and the widespread
fast food options increase the temptation and ease of poor food choices, Downs and
FOOD CHOICE DIMENSIONS AND BMI 7

Loewenstein (2011) suggest people are overweight and obese by choice, as “they have made a
deliberate decision favoring the pleasures of eating over the advantages of lower body weight”
(p. 139). Murphy (2006) adds to this “rational choice” notion of obesity by assuming that obese
people have made a deliberate decision that accounting for caloric intake with increased
expenditure does not justify the benefits of weight loss (as cited in Downs & Loewenstein, 2011,
p. 139). Murphy’s point of view is closer to the general economic notion that people allocate
their scarce resources of time and money to maximize their utility. In order to maximize utility,
“The Economics of Obesity” suggests people work to satisfy the “last dollar” rule, by which the
scarce resource of money is optimally allocated when utility cannot be increased from
rearranging monetary expenses (Cawley, 2011). In terms of food choices, the “last dollar” rule
helps explain the appeal of cheaper, more energy-dense foods by suggesting people obtain the
greatest utility from allocating their monetary resources to these kinds of foods, in contrast to the
more expensive but healthier options available.
Just as the “last dollar” rule argues for the increase in overweight and obesity by an
economic framework of allocating resources for unhealthy food options, the “last hour” rule
works to explain the increased weight trend in terms of time rather than money. Since excess
calories can come simply from the lack of exercise, the “last hour” rule suggests people
maximize their utility for time by increasing sedentary leisure compared to all other activities
(Cawley, 2011). This maximization of leisure time is reflected in recent estimates of Americans’
physical activity levels, as only 32% of adults get at least the minimum amount of exercise as
recommended by the United States Department of Health and Human Services (Carlson, Fulton,
Schoenborn, & Loustalot, 2010).
FOOD CHOICE DIMENSIONS AND BMI 8

The economic frameworks stemming from maximizing utility also seem to be driven
largely by people’s biases of future consequences, or, what is referred to in decision analysis
literature as present biased preferences, or delay discounting. By this notion, people are biased
towards the present, overweighing immediate costs and benefits compared to those even slightly
delayed in the future (Cawley, 2011; Rasmussen, Lawyer, & Reilly, 2010; Weller, Cook, Avsar,
& Cox, 2008). Weight management is a pertinent example of delay discounting as the reward of
eating is immediate, and the initial cost is minimal, but the potential negative consequences of
mismanagement are delayed at some point in the future. Additionally, the costs of continuous
discounting are often intangible from the present biased viewpoint, such that once the costs are
realized, the problem seems insurmountable. Studies have revealed that compared to healthy
weight controls, obese people show steeper discount rates, even when the rewards are
hypothetical money or food (Rasmussen et al., 2010; Weller et al., 2008). In terms of actual food
choices, an unhealthy, gluttonous meal will not have a great effect on weight immediately, but
continuous discounting with unhealthy options or continuous slips in a healthy diet are clear
determinants for weight gain. As Wing & Phelan (2005) disclose, inconsistent diet has proven to
be a strong determinant of weight regain in the National Weight Control Registry – a registry of
over 4000 self-selected adults who have lost at least 30 pounds and kept it off for at least one
year – while frequent monitoring of weight was a common characteristic of registry members.
Diet consistency and frequent monitoring of weight divert focus away from only the immediate
present by making the importance of future consequences salient. Noticing the slips in a healthy
diet and the resulting incremental weight gains makes these negative consequences more
tangible. In the realm of weight management, delay discounting is a very costly bias that not
FOOD CHOICE DIMENSIONS AND BMI 9

only emphasizes poor decision making, but also makes one more susceptible to the changed food
industry.
In an effort to capitalize on people’s irrational decision making in the context of weight
management, Downs and Loewenstein (2011) suggest that rather than exploiting people’s
decision making biases, the food industry should play to people’s propensity to opt for the
default option. Rather than allowing the prices of energy-dense foods and unhealthful
prepackaged snacks to decline, a helpful policy initiative would entail lowering the prices of the
healthy food alternatives thereby leaving less for the economic consumer to trade-off and less
decision making for them to do. By changing the default to lower priced, healthful options, there
is less room for flawed decision making to take effect, and a greater likelihood that people would
alter their eating habits to benefit weight management. This default bias, in accordance with
present biased preferences and the economic framework, highlights the importance of “food
choice dimensions” that underlie the decisions people make about food.
The food industry’s modern reliance on mass production and declining prices of energy-
dense food plays to people’s adherence to the economic framework built on maximizing utility
and their susceptibility to present biased preferences and the default bias. The mass production of
prepackaged food has largely driven the snack industry, an industry whose growth has closely
paralleled the rising pattern of obesity (Cutler et al., 2004), and highlights several prominent
factors that are apparent in food choices. Perhaps the most detrimental factors that the food
industry preys on are convenience and price. Both of these factors underlie the economic
framework and are undoubtedly salient factors in the modern world of working families and a
downtrodden economy. Due to the fact that the majority of foods available for convenience and
low cost are of the unhealthy variety, people are already at a disadvantage when they utilize
FOOD CHOICE DIMENSIONS AND BMI 10

these factors. The desire for immediate gratification and the demands of a working lifestyle often
overshadow the negative consequences of acquiescing to convenience and low cost. While much
research has been done relating weight management to decision making biases and an economic
framework, limited research has attempted to uncover the motivations behind people’s food
selection decisions.
Steptoe and Pollard (1995) sought to develop a measure of the fundamental motivational
factors of dietary choices that would be applicable to individuals of all weights. As many efforts
before their 1995 study were limited in scope, Steptoe and Pollard’s main goal was to establish a
measure that incorporated a variety of factors that would enable comparisons for levels of
importance. What became the Food Choice Questionnaire (FCQ) first started as a 68 item
questionnaire to assess a variety of focal points that are attended to when people select food; the
items were descriptions such as “Is high in protein,” “Is easy to prepare,” “Contains natural
ingredients.” Using the initial 68 items, participants rated the statement “It is important to me
that the food I eat on a typical day:” (e.g., “Is easy to prepare;” “Is not expensive;” “Is what I
usually eat,” etc.) for each item on a scale of importance from one to four – “not at all
important,” “a little important,” “moderately important,” and “very important.” Factor analysis
was conducted on the participants’ Importance Ratings to parse out a group of factors that
accounted for the most variance and to determine which of the 68 items had the highest loading
on the concluding nine factors. What emerged was a Food Choice Questionnaire of 36 items
categorized by their relationship to nine factors: health, mood, convenience, sensory appeal,
natural content, price, weight control, familiarity, and ethical concern (Steptoe & Pollard, 1995).
Steptoe and Pollard’s (1995) multidimensional model of food choice dimensions was the
first of its kind to successfully determine the most crucial elements people consider when
FOOD CHOICE DIMENSIONS AND BMI 11

selecting food. In creating a model that highlights the fundamental features that are important in
food decisions, they have widened the scope of circumstances for which the model can be
applied. While their analyses focused on group differences among age, gender, and income, it is
clear that the model applies to all individuals of all weights.
The motivation for the present study stemmed largely from the interest in decision
making as it relates to weight management, and the study was built off the groundwork laid by
the fields of economics and decision analysis, and the contributions of Steptoe and Pollard. We
expanded the intention of current research to investigate how food choice dimensions actually
relate to BMI. Our study aimed to develop the use of food choice dimensions to discern whether
healthy weight and overweight people make different decisions concerning food options, and
whether these differences can help predict BMI. It was hypothesized that healthy weight
individuals would utilize the factors concerning weight management – nutritional value and
caloric content – more so than those who are overweight, who, in contrast, would utilize factors
concerning external attributes, such as price.
Method
Participants
Two participant groups were recruited to take part in the survey and all participants were
18 years of age or older. One group consisted of students from the University of Southern
California’s (USC) Psychology Subject Pool and the second group was a more general
representation of US adults, consisting of Mechanical Turk users. USC participants were able to
sign up to participate in the online study from the Subject Pool website and received 0.5 extra
credit points for their psychology class as compensation for their participation. US Mechanical
Turk participants were able to select the study from the list of available Human Intelligence
FOOD CHOICE DIMENSIONS AND BMI 12

Tasks (HITs) from the home page once the user was logged in and qualified participants received
$1.00 to their accounts as compensation for their participation. HITs on Mechanical Turk were
stratified into four age blocks and grouped by gender within each stratum in an effort to obtain
equal numbers of men and women spanning many years of age. The age strata consisted of 18-30
year olds, 31-40 year olds, 41-50 year olds, and 51 years and older.
Overall, 581 participants were retained (N=360 females), ranging in age from 18 years
old to 75 years old, with a mean age of 32.22 years (SD=13.99) and BMIs ranging from 18-44,
with mean BMI of 24.82 (SD=5.21). Participants were excluded for missing two out of four
quality control questions within the survey that were meant to ensure they were reading carefully
and following all instructions, and participants were also excluded for either being classified as
underweight (BMI < 18) or for having an excessively high BMI (cutoff BMI > 45). These cut
off points for BMI were used to generate a more representative sample of the general population
consisting of people who are at a healthy weight, overweight, obese, even morbidly obese.
Twenty-two participants were excluded for unrepresentative BMI. Using standard BMI
categorizations for healthy weight, overweight, and obese led to group sizes of 335, 158, and 88,
respectively.
Our sample is 58% healthy individuals and only 42% overweight individuals, contrasting
population statistics largely due to the high number of young college students in the study
(N=240) – mean age of 20.5 years (SD=2.04) and mean BMI of 22.38 (SD=3.06). This younger
population is not only healthier than the general population – only 22% are overweight – but they
also have less variation in BMI and only span a range of BMI from 18-37. On the other hand, the
second group surveyed as a more general representation of US adults consisted of 341
individuals with a mean age of 40.5 years (SD=12.86), a mean BMI of 26.54 (SD=5.70), and
FOOD CHOICE DIMENSIONS AND BMI 13

spanning a range of BMI from 18-44. In contrast to the younger participants, this representative
adult sample consisted of 57% overweight individuals, with 42% being obese.
Materials and Procedure
The study was conducted online via Qualtrics.com
®
. Participants all received the same
Qualtrics instrument, which was expected to take between 15 and 20 minutes to complete. No
personal identifiers were collected. All participants were provided with an information sheet at
the beginning of the questionnaire to state the purpose of the study, the extent of their
involvement, their rights to withdraw from the study without penalty, the confidentiality of the
data collected, the respective compensation, and the principal investigator and IRB contact
information. The study was reviewed and approved by the USC IRB.
The survey (see Appendix A) was designed for participants to introspectively examine
their eating behaviors, exercise behaviors, beliefs related to weight management, and to identify
the importance of certain “Food Choice Dimensions” as they relate to selecting food options.
The questionnaire collected data on demographics (age, gender, education, race, SES, and family
size); BMI information (weight and height) and beliefs about medical classifications for weight;
beliefs about weight management and exercise; frequency of exercise; and food consumption
patterns (typical breakfasts, lunches, and dinners; consumption of various food groups such as
lean meat and fish, fruits and vegetables, desserts, fast food meals, junk food, and regular soft
drinks).
Three sections within the survey were designed to explicitly measure the participants’ use
of food choice dimensions and the relative levels of importance for each dimension. Food choice
dimension data was captured through a questionnaire (see Appendix B), a task of rank ordering
the dimensions on level of importance in food selection, and a point weighting task where each
FOOD CHOICE DIMENSIONS AND BMI 14

dimension was assigned a relative number of points based again on the dimension’s importance
in food selection.
Food Choice Dimensions. Three tasks were used to capture the element of “food choice
dimensions” and their relative importance in the participants’ food selection processes. The food
choice dimensions are factors describing the prominent features of the food itself, such as caloric
content or taste, or the external attributes of food, such as price or convenience. These factors are
thought to influence decisions about selecting which foods to eat. The first task was a
questionnaire adapted from Steptoe and Pollard’s (1995) “Food Choice Questionnaire (FCQ).”
The current questionnaire asked the subjects to rate the degree to which certain facets of food
characteristics were important or unimportant, using the statement “It is important to me that the
food I eat on a typical day:” on a 1 (not at all important) to 7 (very important) scale –
intermediate values used were: 2 (unimportant), 3 (somewhat unimportant), 4 (neither important
nor unimportant), 5 (somewhat important), and 6 (important). This questionnaire was stratified
by each dimension in which three to five items were used to assess each food choice dimension:
caloric content, convenience, familiarity, nutritional value, price, social demands, and taste. The
statements for “nutritional value,” “convenience,” “price,” “caloric content,” “taste,” and
“familiarity” were taken from Steptoe and Pollard (1995), although we used several different
labels for the dimensions themselves. Where we used the labels “nutritional value,” “caloric
content,” and “taste,” Steptoe and Pollard respectively used “health,” “weight control,” and
“sensory appeal.” The substituted label names were thought to be conceptually simpler for
people to understand, while also making the attribute of interest salient (e.g., “taste” compared to
overall aesthetics in “sensory appeal”). Additionally, for two dimension sets – nutritional value
and taste – one statement from each group was omitted from the Steptoe and Pollard instrument
FOOD CHOICE DIMENSIONS AND BMI 15

as the statements did not explicitly apply to the new dimension labels. Furthermore, the current
study added the dimension of “social demands” and the corresponding statements for the
questionnaire. Three factors (ethical concern, mood, and natural content) from Steptoe and
Pollard’s questionnaire were omitted because they were not thought to be explicitly related to
people’s food selections for purposes of BMI differences. An exploratory factor analysis was run
on the data collected in this study to verify the newly formed questionnaire structure (Table 1).
The second task required participants to rank order the seven dimensions to reflect the
relative level of importance in the food selection process. This task was designed to assess a rank
order of these integral factors in food decisions. While the FCQ measured the levels of
endorsements of several statements for each factor in terms of importance, the rank ordering
forced participants to make tradeoffs of importance among the different dimensions. The rank
ordering task was the first time in the study the participants were made aware of the explicit
dimension labels: Caloric Content, Convenience, Familiarity, Nutritional Value, Price, Social
Demands, and Taste. The dimensions were randomized within the Qualtrics instrument so the
participants did not arrive at this task with the dimensions in any consistent, predetermined order
that might have affected their rank ordering. The participants were given instructions clarifying
that the order could be rearranged as many times as needed to appropriately reflect the level of
importance for all seven factors.
The third task required participants to assign relative weights to each of the seven
dimensions so the weights added to 100. Like the rank ordering task, assigning points to each
dimension required the participant to make tradeoffs in terms of relative importance, but unlike
rank ordering, this task created a meaningful interval scale of measurement. Again, the
dimensions were presented in a randomized order and participants were instructed to rearrange
FOOD CHOICE DIMENSIONS AND BMI 16

the values as many times as was necessary to appropriately reflect the weights of all seven
dimensions in their food selection processes. While they were instructed the relative weights, or
points, must sum to 100, it was not required that each dimension was allotted some of the points.
Participants were simply instructed to distribute the 100 points among the seven dimensions
based on their relative importance in their food choices.
All three tasks were expected to be highly correlated with one another, as they were all
measuring the relative importance of the seven factors in decisions for food choices. Adapting
the Steptoe and Pollard questionnaire seemed to be a logical progression from the previous
research to confirm the efficacy of such a measure while also providing detail about the various
dimensions by including several descriptive statements. The rank ordering and point assignment
tasks were both used to have participants introspectively make tradeoffs for the importance of
each of the seven factors, and the point assignment task added the element of creating an interval
scale of measurement to assess quantitative differences between the factors.
BMI. BMI was calculated by taking the self-reported weight in pounds, dividing by the
self-reported height in inches squared, and multiplying by the conversion factor of 703. This
conversion factor is used to transform the measurements for weight and height from kilograms
and meters to pounds and inches.
Results
The exploratory factor analysis in Table 1 was run to verify the newly formed structure of
the Food Choice Questionnaire. All of the items loaded highly onto the respective dimensions we
hypothesized for the instrument, with the exception of one item. One item dropped out of
Familiarity and loaded onto Taste. The item in question was, “Is what I like to eat” which we
originally judged to be an item of Familiarity, but clearly the participants saw it as a statement of
FOOD CHOICE DIMENSIONS AND BMI 17

Taste, based on the factor analysis results. Accordingly, subsequent analyses proceeded to group
this Familiarity item with Taste as suggested by the factor analysis and our judgment of the
subjective content of the item.
Descriptive statistics for BMI group responses on each measure are presented in Table 2.
Table 2 reveals mean values on each dimension to show the varying degrees of importance of the
food choice dimensions across BMI groups for each measure. For all three measures, Table 2
reveals the greatest differences in mean importance ratings across the BMI groups on the
responses for Nutritional Value and Price; healthy weight individuals placed more importance on
Nutritional Value compared to overweight and obese individuals, and obese individuals placed
more importance on Price compared to overweight and healthy weight individuals. It is
important to note that by examining the means of the rank ordering task, we have, in effect,
treated this measure as an interval scale rather than an ordinal scale of measurement. Table 3
presents the correlations of the three measures of food choice dimensions with one another
(FCQ, Ranks, Relative Weights). As can be seen in Table 3, all of the measures were statistically
correlated with one another at the .001 level across all seven factors. The measures of Ranks and
Relative Weights were the most highly correlated.
A 3X7 mixed model ANOVA was conducted to determine whether the seven factors,
within each of the three measures, were significantly different from one another. This 3X7 mixed
model ANOVA was conducted on Ranks and Relative Weights, although these measures were
constrained by an allotment of rank order positions (7) and points (100), respectively. The
ANOVA results confirmed that there are statistically significant differences across all seven food
choice dimensions. Table 4 reveals the statistically significant effects that are apparent in each of
the three ANOVA measures: FCQ, F(6, 570) = 201.99, p < .001, η
2
= .680; Ranks, F(6, 570) =
FOOD CHOICE DIMENSIONS AND BMI 18

981.35, p < .001, η
2
= .912; Relative Weights, F(6, 567) = 629.89, p < .001, η
2
= .870. One-way
ANOVAs further revealed statistically significant between-subjects effects for the various food
choice dimensions across the BMI groups (Table 5). Additionally, the ANOVA results presented
statistically significant interaction effects for the seven food choice dimensions with BMI
groups; the significant pairwise comparisons for BMI groups are clearly shown in Table 6. The
interactions were statistically significant in each of the three measures: FCQ, F(12, 1142) = 2.72,
p < .001, η
2
= .028; Ranks, F(12, 1142) = 3.89, p < .001, η
2
= .039; Relative Weights, Roy’s
Largest Root = .030, F (6, 568) = 2.82, p < .01, η
2
= .029. Finally, the ANOVA contrast results
revealed that the seven food choice dimensions were significantly different from one another (p
< .001) when comparing the means of each dimension importance ratings within each of the
three measures, although there were a few exceptions. Each measure had at least one pair of
dimension contrasts that were not significantly different – FCQ: Convenience and Price; Ranks:
Convenience, Nutritional Value, and Price; Relative Weights: Convenience and Caloric Content,
and Nutritional Value and Price (see Table 2 for details of means).
Table 5 shows the statistically significant between-subjects effects for BMI groups for the
food choice dimensions that emerged in one-way ANOVAs. These results support our
hypotheses by revealing that the most consistent significant differences were found on the
dimensions of Nutritional Value and Price. Nutritional Value was a statistically significant DV in
each of the three measures, while Price was statistically significant for Ranks and Relative
Weights (and approached significance for FCQ, p = .073). The Familiarity food choice
dimension was significant only within the Ranks measure. All between-subjects effects were
statistically significant at the .05 level, as shown in Table 5, with many being significant at far
lower levels (.01 and .001).
FOOD CHOICE DIMENSIONS AND BMI 19

Table 6 expands on Table 5 by showing statistically significant pairwise comparisons of
mean differences for BMI groups by food choice dimensions. Table 6 clearly supports our
hypotheses that people at different ends of the BMI spectrum utilize different food choice
dimensions – healthy individuals focus on weight management (Nutritional Value dimension),
and obese individuals pay less attention to weight management but more attention to external
factors, such as Price. Healthy weight participants rated Nutritional Value very highly, assigning
a mean relative weight of 20.2 points to this dimension, compared to obese individuals, who
assigned only 15.15 points. On the other hand, the obese individuals placed a lot of importance
on the Price dimension, assigning a mean relative weight of 21.35 points, while the healthy
weight individuals assigned only 15.83 points (see Table 2 for details). These differences
between healthy and obese individuals on the dimensions of Nutritional Value and Price are
present in each of the three measures. Although not displayed in Table 6, a comparison between
the healthy and obese groups on the Familiarity dimension within the Ranks ANOVA
approached significance (p = .058) – mean rank of 4.91 for healthy individuals versus 4.58 for
obese individuals.
Table 6 additionally reveals more intermediate differences between healthy and
overweight individuals, and overweight and obese individuals. The differences between healthy
and overweight individuals are only present in ANOVAs on Ranks, where the differences
appeared within the Familiarity and Nutritional Value dimensions. The mean ranks assigned to
the Familiarity dimension were 4.91 for the healthy weight individuals compared to 4.57 for the
overweight individuals (Table 2). The Nutritional Value dimension shows a mean rank of 2.97
for the healthy weight group compared to 3.35 for the overweight group (Table 2). Statistically
significant differences between overweight and obese individuals on the Nutritional Value
FOOD CHOICE DIMENSIONS AND BMI 20

dimension are present in all three measures, but the difference between these groups on the Price
dimension are only present when using Ranks and Relative Weights. Table 2 reveals the means
of both groups on the Nutritional Value dimension across all three measures – FCQ: 5.16
(overweight) versus 4.61 (obese); Ranks: 3.35 (overweight) versus 4.11 (obese); Relative
Weights: 19.17 (overweight) versus 15.15 (obese). The descriptive statistics of the overweight
and obese groups on the Price dimension show mean ranks of 3.82 and 3.18, respectively, and
mean relative weights of 15.56 and 21.35, respectively (Table 2 for details). Although not shown
in Table 6, two comparisons between overweight and obese individuals approached significance
at p = .057 in ANOVAS on FCQ scores on the dimensions of Caloric Content and Price. Table 2
reveals these mean FCQ scores – 4.80 for the overweight group compared to 4.42 for the obese
group on the Caloric Content dimension, and 5.17 for the overweight group compared to 5.49 for
the obese group on the Price dimension.
Stepwise multiple regression analyses (Table 7) were consistent with ANOVA findings
of Nutritional Value and Price being the main dimensions of importance in predicting individual
differences in BMI. Regression analyses were run on each of the three measures, even though
one measure was based on rank ordering.
With FCQ scores on the seven food choice dimensions predicting BMI, Nutritional Value
entered first (β = -.176, R
2
= .031, p < .001) and Price entered second (β = .143, R
2
= .051, p <
.001). The negative beta for Nutritional Value indicates the higher the score on this factor in the
questionnaire, the lower the participant’s BMI. On the other hand, the beta for Price indicates a
direct relationship with BMI, such that the higher the score on this dimension, the higher the
individual’s BMI. With only these two variables entering the regression model, 5% of the
variance in BMI is explained (see Table 7).
FOOD CHOICE DIMENSIONS AND BMI 21

With Relative Weights of the seven food choice dimensions predicting BMI, the order of
entry into the regression model was reversed for the two dimensions compared to the first model
– Price entered first (β = .155, R
2
= .024, p < .001) and Nutritional Value entered second (β = -
.104, R
2
= .033, p < .02). Relative Weights as the predictor variables accounted for only 3.3% of
the total variance in BMI. Whereas the second variable in the FCQ model accounted for an
additional 2% of the variance, in the model of Relative Weights, the second variable accounted
for only an additional 0.9% of new variance. Although this model does not explain a lot of the
variance in BMI, the pattern of relationships between the variables and BMI are similar to those
found in the first regression model, such that higher scores on Nutritional Value display a
negative relationship with BMI, whereas higher scores on Price display a positive relationship
with BMI. Additionally, both models are consistent with the various ANOVA results, which
reliably showed statistically significant effects for BMI groups on these two dimensions (Tables
5 & 7).
Although regression analyses presuppose an interval scale of measurement, we broke
convention and ran a stepwise multiple regression using Ranks as the independent variables to
predict BMI. As the other two measures had already revealed statistically significant results, we
were curious what a model using Ranks as predictor variables would show. Accordingly,
Nutritional Value entered into the equation. Interestingly, Convenience entered into the model
when it had not previously shown to be statistically significant. Nutritional Value entered first (β
= .254, R
2
= .065, p < .001) and Convenience entered second (β = .132, R
2
= .079, p < .01).
Using Ranks as the predictor variable accounted for 7.9% of the variance in BMI – a greater
amount than either of the two previous regression models. In terms of absolute value of the beta
weights for Nutritional Value across all three regression models, this model using Ranks
FOOD CHOICE DIMENSIONS AND BMI 22

revealed the greatest weight for Nutritional Value (.254), perhaps a result of the ordinal scale of
measurement. Since the ranks ranged from one to seven, one was considered to be the most
important factor. Consequently, Nutritional Value has a positive beta weight in this model to
reflect the higher rank order, compared to the FCQ and Relative Weights models where a higher
value for the factor was considered more important (See Table 7 for details).
Discussion
This study was designed to examine the ways in which people select which foods to eat
on the basis of several food choice dimensions. It was hypothesized that people within a normal
BMI range would utilize different factors than people who are overweight or obese. Specifically,
it was hypothesized that healthy weight individuals would utilize factors concerning weight
management, whereas overweight or obese individuals would pay less attention to weight
management dimensions and pay more attention to external factors such as price. Furthermore, it
was expected that these differences would help predict BMI, with a utilization of weight
management factors relating negatively to BMI and a utilization of external factors relating
positively to BMI. The results supported the hypotheses by revealing BMI group differences
across the dimensions of Nutritional Value and Price. Additionally, stepwise multiple regression
analyses further supported these findings as both Nutritional Value and Price consistently entered
the model as the only two variables to predict BMI.
Three measures in the study were used to identify individuals’ perceived utilization of
food choice dimensions, and all three measures were significantly correlated with one another (p
< .001) (Table 3). Accordingly, 3X7 ANOVAs for each measure revealed statistically significant
differences for the seven food choice dimensions across BMI groups. The between-subjects
effects and pairwise comparisons presented the crucial information for revealing the specific
FOOD CHOICE DIMENSIONS AND BMI 23

differences among the dimensions and across the three BMI groups. As hypothesized, Nutritional
Value proved to be a consistent contributing factor to the differences among the groups,
revealing significant differences between the groups in each of the three measures. In all cases,
these differences showed that normal weight participants placed more importance on the
Nutritional Value of the foods they chose to eat as compared to the obese participants.
Additionally, the specific external factor hypothesized, Price, was significant in a Ranks
ANOVA and Relative Weights ANOVA, and accompanied by another external factor,
Familiarity, for Ranks. In all cases where significant differences were found between BMI
groups on Price and Familiarity, the obese participants placed more weight on these factors in
choosing what to eat, as compared to the normal weight participants. These patterns of mean
differences can be clearly seen in Table 2.
Pairwise comparisons revealed the detailed differences among the three BMI groups
across each of the seven dimensions for each of the three measures (Table 6 presents the
statistically significant differences). These results supported the main hypothesis that there would
be differences in food choice dimension utilization between healthy and overweight or obese
individuals, and these differences consistently showed in the Nutritional Value and Price factors
for each of the three measures. Additionally, the pairwise comparisons revealed statistically
significant intermediate differences between healthy and overweight individuals and between
overweight and obese individuals. These intermediate differences attest to the discreteness of all
three BMI groups and add more validity to this study of differences among BMIs as significant
differences were found for every combination of group comparisons – healthy versus
overweight, overweight versus obese, and healthy versus obese. Had the intermediate group,
overweight individuals, been more similar to one of the extreme groups, the only significant
FOOD CHOICE DIMENSIONS AND BMI 24

differences would have been between the extremes of healthy and obese, but this was not the
case. While the comparisons of the healthy and overweight groups did not consistently reveal
significant differences, the Ranks measure did reveal statistically significant differences on the
dimensions of Familiarity and Nutritional Value for these groups (see Table 6 for details). The
use of the Familiarity factor revealed that overweight individuals were more inclined to rank
Familiarity higher in their food choices than were healthy individuals (p = .016) Additionally,
statistically significant differences between overweight and obese individuals were found on the
Nutritional Value dimension within each of the three measures, as well as on the Price dimension
within the Ranks and Relative Weights measures. These differences between the overweight and
obese groups highlight the distinctiveness of these two unhealthy weight groups, reflecting the
fact that they are indeed different from one another. Furthermore, these differences were evident
in the crucial dimensions of Nutritional Value and Price, reflecting the varying levels of
importance across the BMI spectrum.
The multiple regression analyses confirmed the various ANOVA results by finding that
the importance of Nutritional Value and Price in people’s reported food choices were the two
dimensions that consistently predicted individual differences in BMI. Nutritional Value and Price
were the sole entries into the regression models based on FCQ scores and Relative Weights,
respectively. However, the regression models were only modest predictors of BMI from the food
choice dimension variables as only 5.1% of the variance in BMI was explained by the best
model. The regression model using Ranks as predictor variables accounted for 7.9% of the
variance in BMI. The Ranks regression model also found that the Convenience dimension
entered as a significant predictor of BMI, explaining 1.4% of new variance when entered after
Nutritional Value. Overall, the regression results are in line with the hypothesis that healthy
FOOD CHOICE DIMENSIONS AND BMI 25

weight individuals will focus on nutritional value, thereby revealing a negative relationship with
BMI. In other words, the more a person focused on Nutritional Value, the lower their BMI, while
the less they focused on Nutritional Value, the higher their BMI.
While the regression analyses were statistically significant, it was expected that more of
the variance in BMI would be accounted for by the food choice dimensions in the models, but
only relatively weak effects were found. This could partly be due to the fact that unlike the
national population, our sample was very healthy – 58% within a normal BMI range and only
15% obese. The high percentage of healthy weight individuals in the overall sample is explained
by the high percentage of healthy, young college students surveyed through USC – 78% of the
USC group was within a healthy BMI range. Consequently, these individuals accounted for 56%
of the healthy BMI group overall. Thus, it could be that the preponderance of young age college
students was a confounding variable that limited the amount of the variance we found for BMI in
this study. Due to the fact that our sample had a high composition of young, healthy weight
adults with low variability in BMI, the sample as a whole was not fully representative of the
general population. Unfortunately, the large proportion of young college aged adults did not have
as much variability in BMI as was expected. In the future, to better predict BMI, it is necessary
to obtain a sample that is more representative of the general population in terms of variability in
BMI.
Additionally, the amount of variance in BMI unexplained by the various regression
models suggests there are more facets to a person’s behavior that influence BMI aside from the
decisions in selecting food. For example, as questioned in the Qualtrics® survey, other factors
that influence BMI could include amount of exercise and portion sizes of food. Improvements on
this research should examine the relationships of food choice dimensions, exercise, and portion
FOOD CHOICE DIMENSIONS AND BMI 26

sizes to help get a clearer picture of all the factors that influence BMI as a whole, and the relative
sizes of their contributions to individual differences in BMI.
In an effort to gain a better understanding of the influences of BMI for participants in this
study, an improvement on this investigation could examine the correlations between participant
BMI and their reported common meals and food group consumption. Since the statistically
significant differences were limited to primarily two food choice dimensions – Nutritional Value
and Price – analyzing the reports of common meals eaten and typical eating patterns and
behaviors would provide more insight into the dimensions the participants might be using. For
example, people might not have reported Convenience as an important factor in their decisions,
but if they report often eating leftovers, this might suggest otherwise. This leads to two
limitations of this study: data is based on self-report measures for a sensitive subject and
participants were, in effect, primed with the seven food choice dimensions when asked to
introspectively identify the important factors in their food choices.
The problem of priming effects with the food choice dimensions is that it gave the
participants a basis for which to think about their eating behaviors and food selections in
retrospect. Consequently, it was easier for people to, in hindsight, attempt to identify the
important factors they use in their food choice selections, although they might have misidentified
factors that do actually come into play. Reflecting on food choices with these particular
dimensions already in mind could possibly lead participants to believe that certain factors were
important in their decision making when this might not actually be the case. Consequently, to
deal with these possibly extraneous influences, improvements on this research should investigate
people’s reported beliefs about weight management to determine if their beliefs and behaviors –
food consumption – are in fact correlated. If beliefs and behaviors are in line with one another,
FOOD CHOICE DIMENSIONS AND BMI 27

then this could reflect the fact that people do have insight into the judgment policies they are
using for weight management, and this could downplay effects of priming or social desirability.
If beliefs and behaviors are not correlated, then it would be interesting to see which BMI
group(s) displayed this pattern, and this could reflect the fact that the dimensions are likely not
salient in the food selection process for these individuals.
In all, this study added valuable new information to the field of research. This study
improved on the questionnaire originally developed by Steptoe and Pollard (1995) by not only
modifying their original questionnaire to produce a clean factor structure with only seven
dimensions, but additionally relating these seven dimensions directly to BMI. Furthermore, in
addition to the modified FCQ, we utilized simplified approaches to investigate food choice
dimensions that required participants to make tradeoffs of importance values in their choices. All
three measures assessing food choice dimension importance were highly correlated with one
another and all three measures yielded highly similar results in identifying statistically significant
differences in the seven dimensions and the BMI groups. Accordingly, it is suggested that further
research expand the use of the simplified approaches to investigate food choice dimensions
without requiring the administration of the current study’s 26-item questionnaire. The use of the
Relative Weights measure was perhaps the most significant contribution for assessing food
choice dimensions as it was not only simpler than the FCQ, but it created an interval scale of
measurement to evaluate quantitative differences between the dimensions and the assigned levels
of importance. Further research could attempt to capitalize on the efficacy and simplicity of this
Relative Weights measure in examining food choice dimensions and BMI.
Overall, this present study made strides in examining the underlying factors that
influence people’s food choices. This study began with the questions, “Why do people eat the
FOOD CHOICE DIMENSIONS AND BMI 28

foods they do?” and “What factors influence people’s decisions when selecting food?” and in the
end determined that Taste, Nutritional Value, Price, and Convenience were of the greatest
importance to people’s decisions (see Table 2 for details), with Nutritional Value and Price
revealing significant differences across the three BMI groups (see Table 6 for details).
Additionally, this study added to the scarce literature of food choice dimensions and expanded
on this notion by directly relating these factors to BMI. We sought to determine how these
factors are actually influencing people’s weights and found, in accordance with our hypotheses,
the people who pay attention to the weight management factor of Nutritional Value are more
likely to be at a healthy weight whereas people who pay attention to external factors like Price
are more likely to be obese. Our findings suggest that people may lack the motivation or the
knowledge to pay attention to factors such as nutritional value, as this factor was not highly
regarded as important in people’s decisions within the overweight and obese groups. Since this
study did not manipulate any variables to make any causal inferences about the direction of these
relationships between food choice dimensions and BMI, a future direction of this field of
research could attempt an interventional design by assigning people to groups to receive
information about the various food choice dimensions and their respective positive or negative
effects on BMI. Overall, this study improved on the basis of literature surrounding food choice
dimensions and consequently, set a foundation for future research to investigate the relationships
between these dimensions and BMI.

FOOD CHOICE DIMENSIONS AND BMI 29

References
Bowman, S. A., & Vinyard, B. T. (2004). Fast food consumption of U.S. adults: Impact on
energy and nutrient intakes and overweight status. Journal of the American College of
Nutrition, 23(2), 163-168. Retrieved from http://www.jacn.org/content/23/2/163.long
Carlson, S. A., Fulton, J. E., Schoenborn, C. A., & Loustalot, F. (2010). Trend and prevalence
estimates based on the 2008 physical activity guidelines for Americans. American
Journal of Preventive Medicine, 39(4), 305-313. doi: 10.1016/j.amepre.2010.06.006
Cawley, J. (2011). The economics of obesity. In J. Cawley (Ed.), The oxford handbook of the
social science of obesity (pp.120-137). New York, NY: Oxford University Press, Inc.
Cutler, D., Glaeser, E., & Shapiro, J. (2004). Why have Americans become more obese. Journal
of Economic Perspectives, 17(3), 93-118. doi: 10.1257/089533003769204371
Downs, J. S., & Loewenstein, G. (2011). Behavioral economics and obesity. In J. Cawley (Ed.),
The oxford handbook of the social science of obesity (pp. 138-157). New York, NY:
Oxford University Press, Inc.
Finkelstein, E. A., Trogdon, J. G., Cohen, J. W., & Dietz, W. (2009). Annual medical spending
attributable to obesity: Payer-and service-specific estimates. Health Affairs, 28(5), 822-
831. doi: 10.1377/hlthaff.28.5.w822
Finkelstein, E. A., & Zuckerman, L. (2008). The fattening of America: How the economy makes
us fat, if it matters, and what to do about it. Hoboken, NJ: John Wiley & Sons, Inc.
Flegal, K. M., Carroll, M. D., Kit, B. K., & Ogden, C. L. (2012). Prevalence of obesity and
trends in the distribution of body mass index among US adults, 1999-2010. Journal of the
American Medical Association, 307(5), 491-497. doi:10.1001/jama.2012.39

FOOD CHOICE DIMENSIONS AND BMI 30

Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2010). Prevalence and trends in
obesity among US adults, 1999-2008. Journal of the American Medical Association,
303(3), 235-241. doi:10.1001/jama.2009.2014
The Organisation for Economic Co-operation and Development (OECD). (2012, Feb. 21).
Obesity and the economics of prevention: Fit not fat (Key facts – United States, update
2012). Retrieved from http://www.oecd.org/health/health-systems/49712881.pdf
Rasmussem, E. B., Lawyer, S. R., & Reilly, W. (2010). Percent body fat is related to delay and
probability discounting for food in humans. Behavioural Processes, 83, 23-30. doi:
10.1016/j.beproc.2009.09.001
Steptoe, A., & Pollard, T. M. (1995). Development of a measure of the motives underlying the
selection of food: The food choice questionnaire. Appetite, 25, 267-284. doi: 0195-
6663/95/060267+18
Wansink, B., & Sobal, J. (2007). Mindless eating: The 200 daily food decisions we overlook.
Environment and Behavior, 39(1), 106-123. doi: 10.1177/0013916506295573
Weller, R. E., Cook, E.W., Avsar, K. B., & Cox, J. E. (2008). Obese women show greater delay
discounting than healthy-weight women. Appetite 51, 563-569. doi:
10.1016/j.appet.2008.04.010
Wing, R. R., & Phelan, S. (2005). Long-term weight loss maintenance. The American Journal of
Clinical Nutrition, 82(1), 2225-2255. Retrieved from
http://ajcn.nutrition.org/content/82/1/222S.long
World Health Organization (WHO). (2000). Obesity: Preventing and managing the global
epidemic (WHO technical report series; 894). Retrieved from
http://libdoc.who.int/trs/WHO_TRS_894.pdf
FOOD CHOICE DIMENSIONS AND BMI 31

Table 1. Exploratory factor analysis of the seven dimensions within the Food Choice
Questionnaire.

Pattern Matrix
a

Component
1 2 3 4 5 6 7
nutr1 .882
nutr3 .837
nutr5 .830
nutr4 .808
nutr2 .762 .114 .193
conv2 .854 -.120
conv1 .853
conv3 .724 -.190 .101 .190 -.144
conv5 -.107 .695 .182 -.107 .200
conv4 .645 .166 -.115 .168
social2 .917
social3 .898
social1 .894
taste1 .771 -.116
taste3 .743 .119
taste2 -.167 .725 .121
taste4 .175 .652
price1 .938
price2 -.105 .888
price3 .164 .804 -.101
fam1 .849
fam2 .112 .832
fam3 .111 .734
cal1 .896
cal3 .850
cal2 .125 .829
Extraction Method: Principal Component Analysis.
Rotation Method: Oblimin with Kaiser Normalization.
a. Rotation converged in 7 iterations.

FOOD CHOICE DIMENSIONS AND BMI 32

Table 2. Descriptive statistics for food choice dimensions within all three measures.

Dimension

FCQ

Ranks

Healthy Overweight Obese Healthy Overweight Obese Healthy Overweight Obese
Taste 5.83
(0.8)
5.92
(0.8)
5.92
(0.9)
1.85
(1.2)
1.67
(1.1)
1.63
(1.0)
29.50
(16.6)
29.83
(15.2)
30.07
(17.0)
Nutritional Value 5.21
(1.2)
5.16
(1.1)
4.61
(1.4)
2.97
(1.7)
3.35
(1.6)
4.11
(1.8)
20.20
(14.6)
19.17
(13.9)
15.15
(14.7)
Price 5.15
(1.2)
5.17
(1.2)
5.49
(1.4)
3.91
(1.7)
3.82
(1.8)
3.18
(1.6)
15.83
(13.1)
15.56
(12.0)
21.35
(15.8)
Caloric Content 4.67
(1.5)
4.80
(1.4)
4.42
(1.7)
4.16
(1.7)
4.23
(1.6)
4.37
(1.7)
13.18
(12.04)
12.56
(9.8)
11.05
(10.4)
Convenience 5.29
(1.1)
5.23
(1.1)
5.34
(1.1)
3.61
(1.5)
3.75
(1.4)
3.53
(1.4)
13.04
(10.1)
13.39
(11.0)
13.17
(11.5)
Familiarity 4.16
(1.2)
4.06
(1.4)
4.20
(1.3)
4.91
(1.4)
4.57
(1.6)
4.58
(1.4)
6.34
(6.6)
7.31
(7.1)
6.80
(7.2)
Social Demands 3.22
(1.6)
3.38
(1.7)
3.48
(1.8)
6.58
(0.9)
6.61
(1.0)
6.58
(1.0)
1.91
(3.6)
2.17
(5.2)
2.42
(5.3)
Note. Mean response values on each dimension within each measure across BMI groups (SD).

Relative Weights
FOOD CHOICE DIMENSIONS AND BMI 33

Table 3. Correlations among the food choice dimensions across all three measures.

Correlated
Measures

Taste

Nutritional
Value

Price

Caloric
Content

Convenience

Familiarity

Social
Demands
FCQ &
Relative Wts.
.21* .51* .53* .55* .29* .32* .16*
FCQ & Ranks -.29* -.56* -.61* -.59* -.31* -.39* -.18*
Ranks &
Relative Wts.
-.53*

-.66*

-.71*

-.68*

-.51*

-.56*

-.49*

Note. *Correlation significant at the .001 level (2-tailed)
FOOD CHOICE DIMENSIONS AND BMI 34

Table 4. ANOVA multivariate statistics for each measure.

Measure df F Eta Squared
FCQ (6, 570) 201.99* .680
Ranks (6, 570) 981.35* .912
Relative Weights (6, 567) 629.89* .870
Note. *p < .001

FOOD CHOICE DIMENSIONS AND BMI 35

Table 5. Between-Subjects effects for the food choice dimensions that were statistically significant across the BMI groups; taken from
one-way ANOVAs.

Measure Dependent Variable df Mean Square F Eta Squared
FCQ Nutritional Value 2 12.54 9.01
◊
.03
Familiarity 2 7.84 3.80* .013
Ranks Nutritional Value 2 46.346 16.88
◊
.055
Price 2 18.56 6.32*** .022
Nutritional Value 2 886.04 4.25** .015
Price 2 1175.89 6.69
◊
.023
Note. Test was set at the .05 level, but many effects were far greater. Four significance levels are notated: *p < .05; ** p < .02; ***p
< .01;
◊
p < .001

Relative Weights
FOOD CHOICE DIMENSIONS AND BMI 36

Table 6. Pairwise comparisons for BMI groups by food choice dimensions.

95% Confidence Interval for
Difference

Measure Dependent Variable Mean Difference Lower Bound Upper Bound
(Healthy-Obese) = .591
◊
.313 .869
FCQ (Overweight-Obese) = .549
◊
.241 .858
Price (Healthy-Obese) = -.338* -.633 -.043
Familiarity (Healthy-Overweight) = .337** .063 .610
(Healthy-Overweight) = -.373* -.689 -.057
(Healthy-Obese) = -1.141
◊
-1.531 -.751
(Overweight-Obese) = -.767
◊
-1.201 -.334
(Healthy-Obese) = .725
◊
.322 1.129
(Overweight-Obese) = .639*** .190 1.087
(Healthy-Obese) = 5.049*** 1.649 8.449
(Overweight-Obese) = 4.025* .246 7.805
(Healthy-Obese) = -5.518
◊
-8.642 -2.395
(Overweight-Obese) = -5.795
◊
-9.266 -2.323
Note. Based on estimated marginal means. *p < .05; **p < .02, ***p < .01;
◊
p ≤ .001

Price
Nutritional Value
Price
Nutritional Value
Ranks
Relative Weights
Nutritional Value
FOOD CHOICE DIMENSIONS AND BMI 37

Table 7. Stepwise multiple regression statistics for predicting BMI from food choice dimensions.

Measure Predictor Variables Total R
2
Δ R
2
β ΔF
Nutritional Value
1
.031 .031 -.176*** 18.40
Price
2
.051 .020 .143*** 12.40
Nutritional Value
1
.065 .065 .254*** 39.89
Convenience
2
.079 .014 .132** 8.85
Price
1
.024 .024 .155*** 14.02
Nutritional Value
2
.033 .009 -.104* 5.50
Note. Superscript numbers on predictor variables indicate the order of entry into the regression model. *p < .02, **p <.01, ***p < .001

Relative Weights
Ranks
FCQ
FOOD CHOICE DIMENSIONS AND BMI 38 38

Appendix A
Qualtrics survey without Food Choice Questionnaire
Demographics:

FOOD CHOICE DIMENSIONS AND BMI 39

Height and Weight:

Beliefs concerning medical classifications of BMI:

Medical condition:

FOOD CHOICE DIMENSIONS AND BMI 40

Preliminary measures

Self-efficacy and weight satisfaction:

FOOD CHOICE DIMENSIONS AND BMI 41

Weight management beliefs:

Environmental support:

FOOD CHOICE DIMENSIONS AND BMI 42

Frequency of weighing and exercising:

FOOD CHOICE DIMENSIONS AND BMI 43

Food consumption behaviors

Frequency of meals; most common breakfasts:

FOOD CHOICE DIMENSIONS AND BMI 44

Most common lunches:

FOOD CHOICE DIMENSIONS AND BMI 45

Most common dinners:

FOOD CHOICE DIMENSIONS AND BMI 46

Portion sizes:

FOOD CHOICE DIMENSIONS AND BMI 47

Food groups:

FOOD CHOICE DIMENSIONS AND BMI 48

Food Choice Dimensions tasks

Rank Ordering Task:

Relative Weights Task:

FOOD CHOICE DIMENSIONS AND BMI 49

Appendix B
Food Choice Questionnaire

FOOD CHOICE DIMENSIONS AND BMI 50

Abstract (if available)

Abstract This study set out to determine whether people of different BMIs focus on different Food Choice Dimensions when selecting which foods to eat, and additionally, if these differences could predict BMI. The study consisted of 581 participants ranging in age from 18-75, and ranging in BMI from 18-44. Seven Food Choice Dimensions – caloric content, convenience, familiarity, nutritional value, price, social demands, taste – were examined with three different measures all asking participants to, in some way, introspectively identify which dimensions, or factors, are important in selecting food. The three measures consisted of: a Food Choice Questionnaire, a rank ordering task, and a relative weighting task to disperse 100 points across the seven dimensions. Between-subjects effects verified our hypotheses that the main differences between BMI groups would lie in weight management factors and an external factor such as price. Nutritional Value and Price proved to be the main dimensions of important differences among healthy weight individuals and obese individuals, whereby healthy weight individuals placed more importance on Nutritional Value and obese individuals placed more importance on Price. Consistent with the ANOVA results, stepwise multiple regression analyses showed Nutritional Value and Price as the two main contributing factors for predicting individual differences in BMI when FCQ scores and Relative Weights were used as the predictor variables.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Goals and weight: the interplay of goal perceptions in weight management

PDF

A measure of weight management strategies and evidence for its utility

PDF

The acute impact of glucose and sucralose on food decisions and brain responses to visual food cues

PDF

Assessing the psychological correlates of belief strength: contributing factors and role in behavior

PDF

Classically conditioned responses to food cues among obese and normal weight individuals: conditioning as an explanatory mechanism for excessive eating

PDF

Behabioral and neural evidence of state-like variance in intertemporal decisions

PDF

An exploratory study of physical activity across different adult groups

PDF

Obesity and healthy food accessibility: case study of Minnesota, USA

PDF

Psychological correlates of certainty strength: measuring the relative influence of evidence, opinion of personal contacts & importance to self-identity

PDF

Value-based decision-making in complex choice: brain regions involved and implications of age

PDF

Validation of a neuroimaging task to investigate decisions involving visceral immediate rewards

PDF

Probability assessment: Continuous quantities and probability decomposition

PDF

Homeostatic imbalance and monetary delay discounting: effects of feeding on RT, choice, and brain response

PDF

Choice biases in making decisions for oneself vs. others

PDF

Reacting to the negative consequences of your decisions: an ATSS comparison of cognitive and affective reactions in older versus younger adults

PDF

Sequential decisions on time preference: evidence for non-independence

PDF

Social exclusion decreases risk-taking

PDF

Do you see what I see? Personality and cognitive factors affecting theory of mind or perspective taking

PDF

Applying adaptive methods and classical scale reduction techniques to data from the big five inventory

PDF

Reframing parental psychological control: the role of insecure attachment

Asset Metadata

Creator Plotin, Elyse N. (author)

Core Title Food choice dimensions and the relationship with BMI

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychological Sciences

Publication Date 07/15/2013

Defense Date 06/27/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag BMI and food choices,food choices,food decisions,OAI-PMH Harvest,weight management

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Walsh, David A. (committee chair), John, Richard S. (committee member), Read, Stephen (committee member)

Creator Email elyseplotin@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-288797

Unique identifier UC11292602

Identifier etd-PlotinElys-1772.pdf (filename),usctheses-c3-288797 (legacy record id)

Legacy Identifier etd-PlotinElys-1772.pdf

Dmrecord 288797

Document Type Thesis

Format application/pdf (imt)

Rights Plotin, Elyse N.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA