You are on page 1of 15

Am J Community Psychol (2013) 51:299–313

DOI 10.1007/s10464-012-9535-5

ORIGINAL PAPER

Simulation Modeling of Policies Directed at Youth


Sugar-Sweetened Beverage Consumption
David T. Levy • Karen B. Friend

Published online: 19 July 2012


! Society for Community Research and Action 2012

Abstract Childhood obesity is a significant public health Keywords Simulation modeling ! Obesity ! Policies !
problem requiring innovative solutions. While recent School programs ! Youth ! Sugar-sweetened beverages
reviews indicate that some policies show promise, there is
a lack of information regarding which policies, and policy
combinations, work best. Low-nutrition, energy-dense Introduction
foods and beverages such as sugar-sweetened beverages
(SSBs) have been identified as a major contributor to the Obesity and its associated health consequences are widely
problem. The purpose of this paper is to use simulation recognized as significant public health issues facing
modeling to show how changes in three categories of SSB today’s youth. While recent reviews indicate that some
policies—school nutrition, school-based education, and policies directed at the problem show promise (Andreyeva
taxes—impact SSB and other food consumption. The et al. 2010; Brown and Summerbell 2009; Doak et al. 2006;
model shows that policies directed at SSBs, particularly tax Flodmark et al. 2006; Levy et al. 2011; Matson-Koffman
hikes, could lead to substantial reductions in the number of et al. 2005; Seymour et al. 2004), there is a lack of infor-
calories consumed by youth. The estimates, however, are mation regarding which policies and policy combinations
subject to a high degree of uncertainty. Estimates from work best to reduce this major health threat. The difficulty
school-based nutrition and school-based education policies, with policy selections lies in part with the fact that nutrition
while also helping to reduce caloric intake, generally show involves a complex interaction of individual and genetic
smaller effects than tax policies and considerable variation components, coupled with environmental issues related to
around parameter estimates for individual and combined food intake and energy output (Gortmaker et al. 2011).
policies. We conclude with a discussion of the limits of the Low-nutrition, energy-dense foods and beverages (LNED)
model, and suggest where additional information is needed. have been identified as a major contributor. They are broad
Limitations notwithstanding, simulation modeling is a in number, however, making it challenging to develop and
promising methodology that can help advance our under- implement interventions to target all such offenders. Fur-
standing of policy effects, thereby helping policymakers to ther, the role of specific LNED foods in promoting weight
better formulate effective policies to reduce obesity prev- gain is generally difficult to distinguish.
alence and the associated social harms. Despite this difficulty, one specific subset of LNED bev-
erages has been identified as a target for public health inter-
ventions_sugar-sweetened beverages (SSBs). SSBs lack
nutritional value, increase overall appetite, and are not only
D. T. Levy (&)
Cancer Control, Lombardi Comprehensive Cancer Center, associated with weight gain (Malik et al. 2006; Vartanian et al.
Georgetown University, Washington, DC, USA 2007), but have also been directly linked to diabetes (Malik
e-mail: dl777@georgetown.edu et al. 2010), heart disease (Brown et al. 2011) and dental
decay. These effects may be attributed in part to their dis-
K. B. Friend
PIRE, Pawtucket, RI, USA placing healthier options such as milk (Malik et al. 2006;
e-mail: kfriend@pire.org Vartanian et al. 2007). SSB consumption doubled in the US

123
300 Am J Community Psychol (2013) 51:299–313

between 1977 and 2002 (Duffey and Popkin 2007). Chil- (SMs) can be especially useful for considering the potential
dren and adolescents now derive 10–15 % of their total impact of an array of policies that will be required to tackle
kilocalories per day (kcal/day) from SSBs (Wang et al. 2008). the obesity problem. Simulation models (SMs) are now
Three types of policies have received considerable widely used to understand and predict behavior in complex
attention as ways to reduce youth SSB consumption. First are systems. The use of these models in public health is at a
policies that limit access. One popular option in this category nascent stage, but has tremendous potential to bridge the
is limiting the availability of SSBs in schools, where youth gap between research and practice (Homer and Hirsch
are a ‘‘captive audience.’’ Access in schools, however, only 2006; Levy et al. 2006; Sterman 2006). SMs combine
contributes to a portion of youth SSB consumption, and, information from different sources to provide a useful tool
unless schools can educate youth not to drink SSBs at home, for examining how the effects of public health policies
school policies will have limited effects. Second are school unfold over time in complex systems and impact popula-
education policies that aim to change consumption habits. tion health (Homer and Hirsch 2006; Levy et al. 2006).
Third are policies that affect all SSB prices inside and outside SMs can be particularly useful in understanding the
of school. Policies in this category can reduce SSB con- complex interaction of the relationships between public
sumption not only among youth, but also adults, which, in policy, diet and exercise, measures of overweight and
turn, can help promote better role models for youth. Adults obesity, and reduced health and other outcomes associated
also serve as gatekeepers for youth SSB access, since they with overweight and obesity. As shown in Fig. 1, obesity
purchase most of the SSBs consumed by youth, including depends on the energy balance equation, which in turns
beverages for lunches brought to school. A compelling body depends on caloric intake through food consumption and
of evidence shows that cigarette tax hikes are consistently calories burnt through physical activity, growth, and other
associated with reduced tobacco use (Chaloupka and Powell processes that influence metabolism. Separate public and
2009; Levy et al. 2000), especially among youth. private policies may affect nutrition and physical activity.
While the immediate aim of SSB policies is to reduce In this study, we only consider policies affecting nutrition,
SSB consumption, the ultimate goal from a public health focusing on SSB consumption and how SSB policies may
perspective is to lower the trajectory of the youth body affect the consumption of other foods.
mass index (BMI) and improve health. Some models and In this paper, we develop a simulation model of the
statistical analyses have examined the impact of policies on pathways of policy effects of school-based access to
BMI, but they examine a single policy or limited set of nutrition, school-based education and SSB tax policies on
policies. Many studies do not report the effect of policies the consumption of SSBs by youth, and the resulting
on BMI or are inconclusive. In addition, reduced SSB effects on caloric intake. The model will embody a sys-
consumption in schools may be compensated by increased tematic approach that considers the effect of various school
intake of other LNED or high calorie foods and beverages nutrition policies on the number of SSB calories consumed
inside or outside of schools. In general, past studies do not inside and outside of school, and how changes in SSB
yet provide sufficient information to isolate the linkages consumption affect the consumption of other foods (more
from policies to SSB consumption to BMI. This is espe- specifically, their associated caloric intake). We also con-
cially likely where policies affect only a portion of con- sider how a range of price changes through tax increases
sumption, such as in schools. The effects may also take influence SSB and non-SSB consumption, so that the dif-
time to develop, and studies often last short periods of time ferent policies can be compared. Because a range of effect
(Gortmaker et al. 2011; Hall et al. 2011; Levy et al. 2011). sizes are associated with each of the pathways to final
As emphasized in a recently released Institute of Med- caloric intake, sensitivity analysis is used to indicate the
icine report (Institute of Medicine 2010) simulation models uncertainty surrounding the relevant parameters and

Public policies Consumption Total caloric


of food intake

Energy BMI
balance distribution
equation

Public policies Physical Total caloric


activity expenditure

Fig. 1 Effects of policies on overweight and obesity

123
Am J Community Psychol (2013) 51:299–313 301

highlight priority areas for future research. Thereby, I. Limits on Access to SSBs
modeling provides an important tool for researchers, as a. Vending machines and snack bars
b. Limits on portion size or a la’ carte offerings
well as a guide to policymakers.
c. Cafeterias
i. All meals
ii. Lunch and breakfast programs
Methods d. Open vs. closed campus during lunch

II. Education/Information Policies


In a previous article (Levy et al. 2011), we reviewed the a. Education in health classes
literature on the direct effects of school nutrition policies b. Parent education programs
on SSB consumption, the effect of price on SSB con- c. Ingredient labelling
sumption (with a focus on youth), the effects of reduced
III. SSB Price Increase
consumption of SSBs in school on the consumption of a. Cafeteria
other foods in school (LNEDs or otherwise), the effects of b. Vending machines and snack bars
SSB consumption in school on the consumption of SSBs
and other foods outside of school, and the effects of SSB Fig. 2 School nutrition policies that can be directed at SSBs
consumption on caloric intake (both through direct reduc-
tions in the amount consumed and the effect on the con- consumption. Three studies (Cullen et al. 2008; 2006;
sumption of other foods). We apply those estimates where Cullen and Zakeri 2004) examined the effect of a variety of
feasible in our current modeling approach and incorporate different school nutrition policies in a Texas middle school
new studies conducted since that review. with high Hispanic and lower income populations. Cullen
In gauging the effect of caloric intake on BMI, evalua- et al. (2007) and Hartstein et al. (2008) conducted a pilot
tions sometimes apply the equation that a 3,500 yearly study in three states that examined serving size limits.
calorie reduction (roughly 10 kcal/day) reduces weight by 1 Johnson et al. (2009) conducted a study in Washington
pound (Hall et al. 2011). However, Butte and Ellis (2003) examined limits in a variety of venues (vending machines,
argued that the energy gap is higher for children, due to snack bars and lunch); and Woodward-Lopez et al. 2010)
greater energy expenditures. Studies indicate that the examined a variety of venues; Shi (2010) focused on
required calorie reductions to prevent weight gain vary from vending machines. A study of Boston schools (Cradock
as little as 100–140 kcal/day at younger ages (Plachta- et al. 2011) examined a general policy of reducing SSB
Danielzik et al. 2008; Wang et al. 2006) to 300–500 kcal/day availability. The most comprehensive study (Briefel et al.
at slightly higher ages (Butte et al. 2007; Butte and Ellis 2009) considered a range of access and education policies
2003; Swinburn et al. 2006) and to 600–1,100 kcal/day across a broad range of schools.
(Wang et al. 2006) for those who became overweight ado- Blum et al. (2008) found no difference in the effect in
lescents over 10 years. At present, in the absence of a well- the intervention and control groups from reduced á la carte
accepted model (Hall et al. 2011), considerable uncertainty and vending machines, and Forshee et al. (2004) and?
surrounds the changes in caloric intake needed to prevent Fletcher et al. (2010a, b) found little or no effect on weight
abnormal weight gain in youth. For that reason, we stop of reducing SSB vending machine availability. Because
short of estimating the effects of SSB policies on BMI and these studies did not obtain significant effects, they were
limit the analysis to the effect of SSB policies on overall excluded from this study. However, they suggest uncer-
caloric intake. tainty regarding the effect of policies, especially on BMI.

In-School Nutrition SSB Policy Effect Sizes School Education Policies

As shown in Fig. 2, in-school policies to reduce SSB may In 2006, 70 % of states required nutrition and dietary
involve (1) limits on access in school, (2) school education behavior to be taught at the elementary, middle, and high
policies, and (3) price policies. Studies that do not include school levels as part of the health education curriculum but
specific outcome measures useful for modeling, such as did not necessarily emphasize SSBs (Kann et al. 2007). For
the quantity of SSBs consumed or caloric intake, were example, Briefel et al. (2009) found nearly significant
excluded. effects of school education policies on SSB consumption in
9–12th grades. Because their study did not distinguish
School Access Policies programs targeting SSB consumption, it may be premature
to draw strict conclusions. Another study of education
Overall, published research has found that limiting access programs (Lo et al. 2008) failed to find significant effects,
to SSBs in school is associated with decreased SSB indicating uncertainty surrounding the effect of education

123
302 Am J Community Psychol (2013) 51:299–313

policies. In contrast, two randomized control trials of substitution of other SSBs, non-SSBs and other foods, as
school education programs targeted at SSB consumption described below. Price elasticities are assumed constant
showed encouraging results: one (James et al. 2004) in over the range of prices considered.
England with elementary school students and another None of the price studies explicitly examine youth, but
(Sichieri et al. 2009) in Brazil with students aged 9–12. SSBs consumed at home or in restaurants or brought from
home to school are often purchased by adults. Some studies
School Pricing Policies have found that youth and individuals with high BMIs are
particularly sensitive to food prices (Auld and Powell
No studies were found that directly examined in-school 2009; Epstein et al. 2006; Powell et al. 2007; Sturm and
SSB price policies. However, reducing the price of lower Datar 2005, 2008). To be conservative, we apply the esti-
fat snacks, fruits and vegetables in school has been found to mates developed above for adults to youth, since adults
increase sales (French 2003), suggesting that students may purchase much of the food for youth.
also be sensitive to SSB prices. We consider the effect of
price assuming that the same elasticities apply inside The Effect of SSB Taxes
school as outside of school, but confining the direct effects
to SSBs purchased in school. Currently, there are differences in the taxes that states
apply and differences in the beverages to which they are
SSB Tax Policy Effect Sizes applied, such as carbonated sodas versus sports drinks
(Chriqui et al. 2008; Powell et al. 2009). Possibly because
A tax on SSBs leads to higher prices, which in turn affects the form of a sales tax levied at the counter is small, past
consumer purchases and SSB consumption. Therefore, the studies that consider tax rates (Powell and Chaloupka
effects of tax policies occur through differences in the price 2009) have generally found small or weak effects. These
paid by the consumer, and statistical studies that consider studies are subject to limitations, including that soda taxes
prices are likely to provide more direct estimates of their tend to show limited variance and thus are unable to dis-
effect on consumers. tinguish effects; are often imposed on a restricted set of
SSBs (e.g., carbonated sodas only); and are taken at the
Demand Studies checkout counter rather than viewed as posted prices.
We assume a common per unit tax across the selected
Demand studies generally estimate the effects of price in SSB beverages sold at all locations. While we recognize
terms of price elasticity, defined as the percentage change in that prices and thus the percentage effect of a tax will differ
quantity consumed relative to a specific percentage change depending on location (e.g., SSB prices are generally
in price. A recent meta-analysis by Andreyeva et al. (2010) higher at restaurants than convenience stores), we do not
estimated a price elasticity for soft drinks of -0.79, with a distinguish the source of purchase due to the lack of suf-
95 % confidence interval of -0.33 to -1.24. However, the ficiently detailed data.
studies varied in their classifications of SSBs, with some The tax is assumed to directly affect the price. Besley
narrowly confining their analyses to sodas and others con- and Rosen (1999) found a 129 % pass through of SSB
sidering a broader spectrum of SSBs, such as fruit and taxes to price. We consider a range of tax shifting from 80
sports drinks. In studies conducted since Andreyeva’s to 130 %, with our best estimate as 100 % shifting. We
review, Finkelstein et al. (2010) and Duffey et al. (2010) note, however, that while a per unit tax is more likely to be
obtained elasticities in the range of -0.7 and -0.9. More passed along to consumers than an ad valorem (percentage)
extreme elasticities were obtained by Zheng and Kaiser tax (Brownell et al. 2009), a constant per unit tax per ounce
(2008) of -0.15 and for soft drinks of -1.9 (Dharmasena will be applied at different percentages for beverages at
and Capps 2011). Smith et al. (2010) obtained a price different prices (e.g., SSBs in restaurants are generally
elasticity of -1.3 for all SSBs. Using narrowly defined higher priced the same size SSBs bought at supermarkets).
categories, Zhen et al. (2011) obtained roughly comparable We consider a tax of $0.01/ounce of soda, which translates
price elasticities, with less price responsiveness among low- to a 20 % increase in price (Andreyeva et al. 2011) with
income than high-income consumers. 100 % shifting. We also consider $0.005/ounce and $0.02/
For our model, based on the cluster of recent analyses ounce tax increases.
between -0.3 and -1.4, we use a midpoint of -0.9, with
bounds of -0.4 to -1.4 to estimate the SSB price elas- Consumption Parameters
ticity. We use a slightly broader estimate of -0.3 to -1.5,
with a best estimate of -0.9, for the narrower category of In order to operationalize the above effect sizes, we apply
carbonated SSBs, i.e., sugar-sweetened sodas, but allow for the effect sizes to measures of average SSB consumption in

123
Am J Community Psychol (2013) 51:299–313 303

terms of kcals/day (kcal/day). The effect of decreased SSB demand studies, these effects are measured in terms of
consumption on BMI will depend further on how this substituting other products in reaction to price changes.
reduction is affected by the consumption of other foods and Many of the demand analyses are confined to soda, so that
beverages and their associated calories. We adjust the there may be substitution to other SSBs, such as sport
estimates of reduced consumption of SSBs to incorporate drinks and fruit drinks.
the consumption of other beverages and solid foods. Examining all SSBs, Smith et al. (2010) estimated that
a 20 % SSB price increase reduced SSB caloric intake by
Levels of US SSB Consumption 38.8 % for adults and 48.8 % for youth. They found off-
sets from juices and milk of 1.9 % for adults and 6.1 %
For the price analysis, the percentage change in price is for youth, thus suggesting little offset. Zhen et al. (2011)
multiplied by the average SSB consumption. Using also found limited substitution and that, as the price of
nationally representative data from the National Health and sugar-sweetened sodas increased, consumption of other
Nutrition Examination Survey (NHANES), Wang et al. beverages such as diet soda fell, especially among low-
(2008) found average daily consumption was 184 kcal/day income consumers. For low-income consumers, Yen et al.
(9 % of total caloric intake) for ages 6–11 years and (2004) found a small effect of soda prices on the con-
301 kcal/day (13 % of total intake) for ages 12–19 years. sumption of other beverages. Finkelstein et al. (2010)
For policies that target a subset of SSBs, such as carbon- obtained relatively large substitution effects of a tax on all
ated beverages, carbonated drinks contributed 55 % (53 % SSBs, but the overall reduction in calories was still only
for individuals ages 6–11 years and 66 % for those ages about 30 % lower when other non-SSBs were considered.
12–19 years) and fruit drinks 33 % of all SSB kcal/day Calories were, however, reduced another 40 % if the tax
consumed (Wang et al. 2008). was confined to carbonated sodas. Dharmasena and Capps
Policies that apply only to schools, such as school access (2009, 2011) found substantial substitution, but examined
policies, require data regarding consumption specifically a limited set of beverages (e.g., milk) Fletcher et al.
while at school. Using the 3rd School Nutrition Dietary (2010a, b) obtained a nearly 100 % offset of SSB con-
Assessment Study (SNDAS, 2004–2005), Briefel et al. sumption through increased milk consumption, but other
(2009) found that school children consumed an average of studies have found substantially less substitution toward
159 kcal/day on school days (with 209 kcal/day by sec- milk (Vartanian et al. 2007). Brownell et al. (2009) esti-
ondary school students), of which 93 were at home, 36 mated a 25 % offset in calories. Wang (2011) estimated a
were at school and 31 were at other locations. Another 39 % offset, assuming that, upon reducing SSB con-
study by Briefel et al. (2009) found that SSBs were con- sumption, a third each switched to water, diet soda and
sumed by 17 % of students in the elementary school, 32 % milk or juice.
in middle school and 36 % in high school, of which only a The studies above demonstrate that substitution effects
portion were actually purchased at schools (27, 67 and are highly dependent on the products considered and
74 %, respectively). Students purchased and consumed, on specification of the empirical model. Based on these
average, 3, 29 and 46 kcal/day in elementary, middle and results, our best estimate is a 20 % substitution for a tax on
high school, respectively. all SSBs, with estimates ranging from 0 to 60 %. substi-
School studies sometimes report the results in terms of tution of calories from other beverages. If the tax is con-
the absolute change in calories consumed in school and fined to carbonated SSBs, we estimate an additional 15 %
other times in terms of the percentage change in calories offset with a range of 5–40 %. The 40 % estimate reflects
consumed. We use an estimate of SSBs consumed and the potential that, in reaction to the tax, beverage manu-
purchased in school to convert the percentage change to facturers may develop new non-carbonated SSBs whose
an absolute change. We converted the results to average costs are favorable to consumers.
daily consumption by multiplying the average reduction As shown in Fig. 3, the effect of SSB school policies on
per school day by the percent of school days in the SSB caloric intake involves multiple pathways. When SSB
year (=180/365). The only exception is for Sichieri et al. access is limited in schools, youth may bring SSBs from
(2009), where results were reported in terms of all SSB home. For non-price policies, we use the estimated sub-
consumption. stitution effects from price studies, i.e., an estimated 20 %
reduction per 18 % increase in price, with a range of
Substitution to Other Beverages Inside and Outside 0–60 % around that estimate.
of School The effects of school nutrition access policies on overall
caloric intake also depend on how consumption in school
The most direct effect of a change in the consumption of affects consumption outside of school. Even if SSB pur-
SSBs is to replace other beverages. In the context of chases are reduced in school, youth may compensate with

123
304 Am J Community Psychol (2013) 51:299–313

Non-SSB consumption Results


in school

School SSB
SSB consumption in
The parameters used in our analysis are listed in Table 1.
policies
school Total caloric In conducting our analysis, we present best estimates, as
intake well as upper and lower bounds.
Non-school SSB consumption
SSB policies outside of school
School Access Policies
Non- SSB consumption
outside of school
Table 2 reports the results from studies of school access
studies, followed by the effects calculated in terms of
Fig. 3 Effects of youth-oriented SSB policies on caloric intake
caloric changes in SSBs. It then incorporates the effect
of reduced caloric intake due to consumption outside
of school, substitution to non-SSBs, and compensation/
increased consumption at home. However, studies that reduced consumption of non-liquids, and finally reports the
have considered overall consumption (Briefel et al. 2009; net effect with bounds after incorporating all substitution
Cullen et al. 2008; Fernandes 2008; Schwartz et al. 2009; and compensation.
Woodward-Lopez et al. 2010) have not found increased Cullen and Zakeri (2004) found that, when SSBs
SSB consumption at home in response to reduced SSB became more available to 6th graders, consumption
consumption in schools. Indeed, youth may further reduce increased 1.3 ozs/day (from 2.1 to 3.4 ozs), and fell only
consumption at home if drinking SSBs becomes less of a slightly in the next year. Upon converting to kcal/day, we
habit. We estimate no effect of reduced in-school SSB calculate a net increase in SSB consumption averaging
consumption on SSB consumption outside of school, but 7.1 kcal/day over the entire year, ultimately yielding a net
increase the variability of our estimates by 25 % to increase in caloric intake of 5.6 (with bounds of 3.2–8.8)
incorporate the increased uncertainty. kcal/day. Allowing for consumption outside of school,
substitution to non-SSBs reduced, and finally compensa-
SSB Consumption and Substitution to Solid Foods tion/reduced consumption of non-liquids, the simulation
model estimates a net increase in overall caloric intake of
Besides substituting for other beverages, reduced con- 5.6 (with bounds of 2.4–11.0) kcal/day due to the increased
sumption of SSBs may be replaced by the consumption of availability of SSBs.
solid foods. However, it has also been suggested that SSB Upon removing LNEDs from snack bars and in cafeteria
consumption may instead increase the appetite for other vending machines, Cullen et al. (2006) found that mean
foods (Bellisle and Rolland-Cachera 2001; Drewnowski SSB intake declined by 1.9 oz. in lunch and snack bars,
and Bellisle 2007; Hattersley et al. 2009; Mattes 1996; ultimately yielding a 4.9 (2.1–9.5) kcal/day reduction in
Yamada et al. 2008), especially high fat and high salt items overall caloric intake. Of three Texas middle school stud-
(e.g., French fries). Thus, reduced SSB consumption may ies, the largest effect was in schools limiting portion size
reduce or increase the intake of solid foods. and vending machine availability (Cullen et al. 2008),
Some studies have considered how SSB consumption whereby overall SSB fell 1.9 oz leading to 7.4 (3.1–14.5)
affects foods consumed and overall caloric intake. Lack of kcal/day reduced net caloric intake. Results from Hartstein
compensation is supported by a recent meta-analysis by et al. (2008) indicate that net caloric intake is reduced by
Vartanian et al. (2007), but an alternative view is presented 1.3 (0.6–2.5) kcal/day if all 6 schools in their sample are
in DiMeglio and Mattes (2000). In addition, Wang et al. included and by 2.6 (1.1–5.0) kcal/day if the 2 outlier
(2009) found that each additional 8 oz. serving of SSBs North Carolina schools are excluded.
was associated with a net increased intake of 106 kcal/day A large scale Washington study (Johnson et al. 2009)
on that day (p \ .001), similar to the 100 kcal/day from an yielded a net decline of 2.9 (1.2–5.6) kcal/day. Two studies
8-oz serving of sodas, indicating limited substitution to examined California policies, one (Shi 2010) of which
other foods and beverages. yielded a reduction of 4.9 (2.1–9.6) kcal/day and the other
In general, even while some research reports that SSB yielded a reduction of 1.4 (0.6–2.8) kcal/day. (Woodward-
consumption may affect the calories consumed from solid Lopez et al. 2010). Another large scale study for Boston
foods, but the effects appear limited and may be in either (Cradock et al. 2011) yielded a net reduction of 9.2
direction. Overall, we estimate no effect of changes in SSB (4.3–18) kcal/day.
consumption on caloric intake from solid foods, but Incorporating a vast array of policies, Briefel et al. (2009)
increase the uncertainty of the estimates to allow for an obtained a wide range of effect sizes for the different poli-
added 25 % variability in overall caloric intake. cies, many of which were not significant. Estimates for

123
Am J Community Psychol (2013) 51:299–313 305

Table 1 Estimates and parameters used to determine effects from studies


Parameter Population Best estimate Lower bounda Upper bounda

Consumption
SSBs in kcals Ages 2–19 years 224
Ages 2–5 years 24
Ages 6–11 years 184
Ages 12–19 years 301
Carbonated SSBs Ages 2–19 years 123.2
in kcals Ages 2–5 years 40
Ages 6–11 years 98
Ages 12–19 years 200
SSB consumption Elementary schools 19
in kcals Secondary schools 52
SSB consumption in kcals Elementary school 3
Purchase and consumed Middle school 29
In school High school 46
Tax policy
Size of tax/price increase $0.01/oz tax = 20 % $0.005 = 10 % price $0.02 = 40 % price
price increase increase increase
Shifting of the tax to price 100 % 80 % 130 %
Price elasticity All ages -0.9 -0.3 -1.5
Substitution to other beverages (in terms of the reduction in the effect on kcals)
Soda to other SSBs All ages 20 % 40 % 5%
SSBs to other beverages All ages 20 % 60 % 0%
Compensation Youth 0.0 % 25 % -25 %
(other foods)
In-school policies
Effect size (by individual study—see Table 1)
Substitute SSB consumption outside of school 0 25 % -25 %
Substitution (same as for tax policy)
Soda to other SSBs All ages 20 % 40 % 5%
SSBs to other beverages All ages 20 % 60 % 0%
Compensation Youth 0.0 % 25 % -25 %
(other foods)
a
95 % confidence interval

specific policies ranged from 6.3 kcal/day (lack of pouring School Education Policies
rights in middle school) to 16.2 kcal/day (not offering
French fries in high schools). Some of these reductions, Briefel et al. (2009) found that having school nutrition
however, were larger than the average SSB consumption of education in all grades had the potential to reduce SSB
students, which may reflect the large differences in con- consumption by 6.3 (2.7–12.3) kcal/day, although the
sumption between schools with and without restrictive effect did not quite reach statistical significance
practices and collinearity between the policy variables. Most (p = 0.51). In elementary schools in England, James and
relevant may be the results for their composite measures, Kerr (2004) reported that carbonated SSB consumption
where the model estimated that stricter availability policies over 3 days decreased by 0.3 servings (0.4 oz.), implying
(e.g., limits on vending machines and snack bars) reduces net the potential to reduce net caloric intake by 0.8 kcal/day
caloric intake by 18.9 (8.0–37.0) kcal/day in middle schools (0.3–1.5) kcal/day. For an education program aimed at
and by 22.1 (9.3–43.2) kcal/day in high schools. Stricter Brazilian students aged 9 to 12, results from Sichieri et al.
school lunch policies have the potential to reduce overall (2009) a reduced net caloric intake of 14.5 (6.3–28.4)
caloric intake by 30.4 (12.8–59.3) kcal/day. kcal/day.

123
306
Table 2 Studies of school nutrition interventions that may affect sugar-sweetened beverages
123
Study Setting year Major findings Reduced kcal With compen With With With With With With com- With com-
location grades consumption/ sation outside compensation compensation compensation compensation compensation pensation of pensation of
(year) day of school outside of of non-SSB of non-SSB of non-SSB of non-liquids non-liquids non-liquids
(lower bound) school (upper drinks (best drinks (lower drinks (upper (best estimate) (lower bound) (upper bound)
bound) estimate) bound) bound)

School nutrition policies


Cullen and TX SSBs increased 2.1–3.4 oz in first 7.1 5.3 8.8 5.6 3.2 8.8 5.6 2.4 11.0
Zakeri 1988–1989 year, decreased from 4.9 to 4.3 oz.
(2004) in second year
ES & early
MS
Cullen TX Mean SSB intake declined from 6.1 4.6 7.6 4.9 2.7 7.6 4.9 2.1 9.5
et al. 2001–2003 5.4 oz to 3.5 oz and sweetened soft
(2006) drink consumption declined from
MS 5.2 oz to 2.6. oz
Cullen TX Year 1–3 from 5.4 to 3.5 to 1.5 ozs 9.3 7.0 11.6 7.4 4.2 11.6 7.4 3.1 14.5
et al. 2001–2003 for all SSBs
(2008)
MS
Hartstein TX, CA, & SSBs reductions across the 6 schools 1.6, 3.2 1.2, 2.4 2, 4 1.28, 6 0.72, 1.4 2, 4 1.28, 2.6 0.54, 1.1 2.5, 5
et al. NC of -1 oz. If the 2 schools with (eliminating (eliminating (eliminating (eliminating (eliminating (eliminating (eliminating (eliminating (eliminating
(2008) 2003 increases are excluded = 0.5 oz. outliers) outliers) outliers) outliers) outliers) outliers) outliers) outliers) outliers)

MS
Johnson WA Net effect of 25 % reduction in SSB 3.6 2.7 4.5 2.9 1.6 4.5 2.9 1.2 5.6
et al. 2007–2008 consumption
(2009)
MS
Shi (2010) CA .181 fewer drinks in schools, and 0.16 6.2 4.6 7.7 4.9 2.8 7.7 4.9 2.1 9.6
2005 fewer drinks using a Kernel-based
propensity score
MS & HS
Woodwad- CA Soda consumption at school fell by 1.8 1.3 2.2 1.4 0.8 2.2 1.4 0.6 2.8
Lopez 2005–2008 7 % & sport drinks fell by about
et al. 5 % (NS) with non-significant
(2010) ES, MS & increases at home

Am J Community Psychol (2013) 51:299–313


HS
Craddock Boston, MA Soda consumption fell 0.16 and other 11.5 8.7 14.4 9.2 5.2 14.4 9.2 3.9 18.0
et al. 2004–2005 SSBs by -0.14 servings/day
(2011) Yielding -0.30 servings/day
HS
Briefel US
et al. 2004–2005
(2009)
MS Attending a school without snack 10.8 8.1 13.6 8.7 4.9 13.6 8.7 3.7 17.0
bars -22 kcal
MS Lack of pouring rights -16 kcal 7.9 5.9 9.9 6.3 3.6 9.9 6.3 2.7 12.3
MS No ala’ carte offerings -52 kcal 25.6 19.2 32.1 20.5 11.5 32.1 20.5 8.7 40.1
HS Attending a school without snack 13.8 10.4 17.3 11.0 6.2 17.3 11.0 4.7 21.6
bars -28 kcal
HS Stricter rules against VMs -40 kcals 19.7 14.8 24.7 15.8 8.9 24.7 15.8 6.7 30.8
HS Not offering french fries -41 kcal 20.2 15.2 25.3 16.2 9.1 25.3 16.2 6.8 31.6
Am J Community Psychol (2013) 51:299–313
Table 2 continued
Study Setting year Major findings Reduced kcal With compen With With With With With With com- With com-
location grades consumption/ sation outside compensation compensation compensation compensation compensation pensation of pensation of
(year) day of school outside of of non-SSB of non-SSB of non-SSB of non-liquids non-liquids non-liquids
(lower bound) school (upper drinks (best drinks (lower drinks (upper (best estimate) (lower bound) (upper bound)
bound) estimate) bound) bound)

MS, Stricter overall availability practices 23.7 17.8 29.6 18.9 10.7 29.6 18.9 8.0 37.0
composite 48 kcals
HS, Attending a school with stricter 27.6 20.7 34.5 22.1 12.4 34.5 22.1 9.3 43.2
composite overall availability practices -56
kcals
HS, Stricter school lunch policies - 38.0 28.5 47.5 30.4 17.1 47.5 30.4 12.8 59.3
composite 77 kcals
School education policies
Briefel HS School offering nutrition eduction in 7.9 5.9 9.9 6.3 3.6 9.9 6.3 2.7 12.3
et al. every grade-16 kcal, but p = 0.055
(2009)
James England Carbonated SSBs, change for 1.0 0.7 1.2 0.8 0.4 1.2 0.8 0.3 1.5
et al. 2001–2002 controls = 0.0 (95 % CI = -
(2004) 0.3–0.4) and for inter-vention
ES group = - 0.3 (95 % CI = -
0.6–0.1). The difference was 0.3
(95 % CI = -0.4–0.5).
Sichieri Brazil 2005 A decrease in the daily consumption 18.1 13.6 22.7 14.5 8.2 22.7 14.5 6.1 28.4
et al. Ages 9–12 of carbonated SSBs in the
(2009) intervention compared to control
group (mean difference = -56 ml;
95 % CI -119, -7 ml) a 20 %
reduction.
123

307
308 Am J Community Psychol (2013) 51:299–313

Effects of an SSB Tax studies consider substitution to non-liquid foods. The


estimates in Table 4 for a 0.01 tax/oz. on all SSBs range
Table 3 reports results for the effect of a tax increase using from 6.5 to 56 kcal/day. While we have assumed the same
best estimates of elasticities. We first consider the effects price elasticities for youth as adults in our analyses, the
of the same percentage tax applied to all SSBs. A tax results in Table 4 reflect similar ranges to our analysis. The
increase of $0.01 ($0.005, $0.02) per ounce is predicted to tax impacts, however, vary according to study methodol-
increase price by 20 % (10 %, 40 %) with 100 % tax ogies and subjects investigated (e.g., entire population vs.
shifting. This price increase in turn is predicted to reduce subgroups who may be more price sensitive, individuals
SSB consumption by 18 % (9 %, 36 %), with lower with lower vs. higher incomes, etc.).
bounds of 6 % (3 %, 12 %) and upper bounds of 30 %
(15 %, 60 %). In terms of SSB kcal/day, the $0.01 tax per
ounce is predicted to ultimately reduce SSB consumption Discussion
by 40.3 (13.4–67.2) kcal/day. The effects are greater for
those ages 12–19 years, a 54.2 kcal/day reduction, but less Our models predict that that policies directed at SSBs could
for those ages 6–11 years, a 33.1 kcal/day reduction. lead to substantial reductions in calories consumed. A tax
Under the assumption of constant elasticity, the effects of $0.01 per ounce on all SSBs is predicted to reduce
increase directly with the size of the tax; kcal/day reduc- consumption by 32 kcal/day. However, we estimate a
tions are half as much with a $0.005 tax/oz. and twice as variation of 4–84 kcal/day around that estimate, due to
much with a $0.02 tax/oz. compared to a $0.01 tax/oz. uncertainty about the effects of SSB prices on SSB con-
The effects also depend on tax shifting, where we sumption and their effect on the consumption of non-SSB
assumed 100 % shifting of the tax to price. The effects are beverages and foods. Estimates from school access and
reduced by 20 % (to 32.3 kcal/day) with 80 % tax shifting, education policies generally indicate smaller effects than
or increased by 30 % (to 52.4 kcal/day) with 130 % tax our estimates of tax policies, again with considerable var-
shifting. With substitution into other beverages and incor- iation around the effect for each policy and also for the
porating the consumption of non-liquid food, the kcal/day estimated effects of different policies.
reduction falls from 40 to 32.3, with the bounds from 4.0 to Our estimates are subject to important limitations. More
84.0 kcal/day. recent data is needed on SSB consumption. Parameters
If the tax were limited to carbonated soda rather than all estimates will also be improved as more evaluation studies
SSBs, the tax would directly affect only the consumption of are conducted. Detailed information is needed on diet at
carbonated SSBs. With the consumption of carbonated home as well as in school. For policies in school, it is
beverages at 123 kcal/day and the same elasticity, the important to understand how out-of-school consumption is
consumption of carbonated beverages is reduced to 22.2 related to in-school consumption and how policy-related
(7.4–37.0) kcal/day. Much larger effects are observed for changes in SSB consumption affect the consumption of
those ages 12–19 years, with initial consumption of non-SSBs, especially LNED foods. The mathematical
200 kcal/day, yielding a 36 kcal/day reduction. With a representation of the effects merits further consideration.
20 % reduction from substituting toward non-carbonated We have used absolute changes or percentage changes in
SSBs, substitution to other non-SSB drinks and incorpo- consumption based on the way that authors reported their
rating the consumption of non-liquid foods, the estimate results, but the effect on consumption may depend on the
declines to 14.2 kcal/day with the bounds expanded by initial level of consumption (i.e., dependent on relative
25 % to 4.0 kcal/day and 84 kcal/day. effect sizes). In examining the effect of tax increases on
We also calculated the effect of higher prices imposed in SSB consumption, the reaction of consumers purchasing at
all secondary school purchases. With average consumption different locations (e.g., restaurants versus school, where
of 37.5 kcal/day of SSBs bought and consumed in middle prices and elasticities may differ) merits further study.
and high schools, a 3.4 (1.1–5.6) kcal/day reduction in SSB The analysis above assumed the same elasticities for
consumption is estimated with a 10 % price increase, a 4.1 youth as for adults, because adults purchase much of the
(2.3–11.3) kcal/day reduction is estimated with a 20 % SSBs consumed by youth, However, based on the results
price increase, and a 8.3 (4.5–22.5) kcal/day reduction is for other types of food, especially for low income con-
estimated with a 40 % price increase. Allowing for sub- sumers (Auld and Powell 2009; Epstein et al. 2006; Powell
stitution to non-SSB beverages and solid foods, a 20 % et al. 2007; Sturm and Datar 2005, 2008), elasticities may
price increase yields a 3.3 (0.7–14.1) kcal/day reduction. be much higher for SSBs consumed at school and pur-
Table 4 summarizes other studies simulating the effect chased by teens outside of school. In that case, the effects
of an SSB tax. These studies, except Smith et al. (2010), do on SSB consumption will substantially increase, especially
not specifically consider youth. In addition, none of the for in-school consumption.

123
Am J Community Psychol (2013) 51:299–313 309

Table 3 Simulation of the effects of a $0.005, $0.01 and $0.02 per oz. tax increase on caloric intake
0.005 Tax/ounce 0.01 Tax/ounce 0.02 Tax/ounce

Effect of a tax on all SSBs on SSB consumptiona


Price increase assuming 100 % tax shifting 10 % 20 % 40 %
Percent reduction in SSB consumption (best estimate) 9.0 % (3.0 %, 18.0 % (6.0 %, 36.0 % (12.0 %,
15.0 %) 30.0 %) 60.0 %)
Change in SSB kcal/day for those ages 2–19 (best estimate) 20.1 (6.7, 33.6) 40.3 (13.4, 67.2) 80.6 (26.9, 134.4)
Change in SSB kcal/day for those ages 6–11 (best estimate) 16.6 (5.5, 27.6) 33.1 (11.0, 55.2) 66.2 (22.1, 110.4)
Change in SSB kcal/day for those ages 12–19 (best estimate) 27.1 (9.0, 45.15) 54.2 (18.1, 90.3) 108.4 (36.1, 180.6)
Tax shifting of 80 % and 130 %
Change in price with 80 % tax shifting 8% 16 % 32 %
Change in SSB kcals with 80 % tax shifting (best estimate) 16.1 (5.4, 26.9) 32.2 (10.8, 53.8) 64.5 (21.5, 107.5)
Change in price with 130 % tax shifting (best estimate) 13 % 26 % 52 %
Change in SSB kcals with 130 % tax shifting (best estimate) 26.2 (8.7, 43.9) 52.4 (17.5, 87.4) 104.8 (34.9, 174.7)
Effect on overall caloric intake of ages 2–19 with 100 % tax shifting
Best estimate with substitution to other non-SSB beverages, ages 2–19 16.1 (5.4, 67.2) 32.3 (10.8, 134.4) 64.5 (21.5, 107.5)
Best estimate with substitution to liquid and non-liquid consumption 16.1 (2.0, 42.0) 32.3 (4.0, 84.0) 64.5 (8.1, 168.0)
The effect of a tax on carbonated SSBs on carbonated SSB consumption w/100 % tax shifting
Price increase assuming 100 % tax shifting 10 % 20 % 40 %
Percent reduction in carbonated SSB consumption (best estimate) 9.0 % (3.0 %, 18.0 % (6.0 %, 36.0 % (12.0 %,
15.0 %) 30.0 %) 60.0 %)
Change in carbonated SSB kcal/day for those ages 2–19 (best estimate) 11.1 (3.7, 18.5) 22.2 (7.4, 37.0) 44.4. (14.8, 73.9)
Best estimate with substitution to other SSB beverages, ages 2–19 7.1 (1.3, 16.7) 14.2 (2.7, 33.4) 28.4 (5.3, 66.7)
Best estimate with substitution to other non-SSB foods, ages 2–19 7.1 (1.0, 20.8) 14.2 (2.0, 41.7) 28.4 (4.0, 83.4)
Effect on overall caloric intake of high school students of a price increase only in schools
Change in carbonated SSB kcal/day for those ages 12–19 (best 3.4 (1.1, 5.6) 4.1 (2.3,11.3) 8.3 (4.5, 22.5)
estimate)
Overall variation with substitution to home, liquid and non-liquid 2.7 (0.3, 7.0) 3.3 (0.7, 14.1) 6.6 (1.4, 28.1)
consumption
a
95 % confidence intervals shown in parentheses

Table 4 Summary of projected effects from other simulations of the effects of a tax on SSBs
Study Price increase Assumptions Population Kcal reduction

Andreyeva et al. (2010, 2011) 20 % Tax on all SSBs, w/out substitution All 50
Finkelstein et al. (2010) 20 % Tax on carbonated SSBs, w/substitution All 4.2
20 % Tax on all SSBs, w/substitution All 7
40 % Tax on carbonated SSBs, w/substitution All 7.8
40 % Tax on all SSBs, w/substitution All 12.4
Smith et al. 2010 20 % Tax on all SSBs, w/substitution Adults 34.2
20 % Tax on all SSBs, w/substitution Adults, low income 36.8
20 % Tax on all SSBs, w/substitution Adults, high income 33.3
20 % Tax on all SSBs, w/substitution Children 40
20 % Tax on all SSBs, w/substitution Children, low income 33.1
20 % Tax on all SSBs, w/substitution Children, high income 44.7
Sturm et al. (2008, 2010) 20 % tax Tax on all SSBs, w/out substitution 3rd to 5th graders 15.2
Duffey et al. 2010 20 % Tax on all SSBs, w/out substitution 56
Dharmasena and Capps (2009, 2011) 20 % Tax on all SSBs, w/out substitution 19.6, 31.2

123
310 Am J Community Psychol (2013) 51:299–313

In the above analyses, we used average consumption 2 years after the program ended, and Sichieri et al. (2009)
data, and conservatively assumed that the impacts were found that overweight girls showed a reduction in BMI.
distributed over the entire population. The effects, how- A widely-used metric is that each 3500 calorie reduction
ever, would be centered on those who already consume is associated with a one pound weight loss. This relation-
beverages. Of individuals ages 12–19 years, Wang et al. ship was formalized (Hall 2008) to predict an approximate
(2008) found that males consumed more than females 10 kcal/day reduction in food energy results in one pound
(357 kcal/day vs. 242 kcal/day), and found little difference of weight loss at steady state, but the reduction must be
in intake between those overweight or obese and those maintained. Larger reductions, as much as three times as
normal weight (e.g., for 12–19 year olds, 297 kcal/day for great, would be needed for older youth and as much as six
those less than the 85th percentile, 304 kcal/day by those times as great for youth at higher weight levels (Butte et al.
between the 85 and 95th percentile, 320 kcal/day by those 2007; Jordan and Hall 2008). As indicated above, there is
above the 95th percentile). Nevertheless, those overweight not yet a well-accepted model for youth to enable us to
may still be disproportionately affected through greater make those predictions. Initial weight and food consump-
reductions in consumption. In addition, those at low tion patterns as well changing patterns over time will need
income might be expected to be most affected by tax to be considered. Relatively modest behavioral modifica-
policies, since they are most likely to be sensitive to price. tions are required to prevent weight gain at younger ages,
For example, Sturm et al. (2010) found limited effects of but larger reductions may be required for older youth and
taxes on soda consumption by young school children, but for those at higher weights (Wang et al. 2006).
greater effects were observed in schools, and for those who Generally, simulation models that examine patterns in
are lower income, African American and higher BMI. consumption and BMI over time as individuals’ age will be
The effect of school policies at a national level will be needed to fully understand the effects of the different
less if other nutrition policies are already in place. Briefel obesity policies over time. Information will be needed on
et al. (2009) and the 2006 SHPPS data (O’Toole et al. current and past levels of BMI and about how policies
2007) found that SSBs were more accessible in higher might be anticipated to affect future BMI. Policy effects
grades. Similarly, the effect of a tax may depend on local may be strengthened over time as attitudes change, or
variations in price, and how the industry passes along unanticipated reactions by the individual or industry may
the tax changes in different locations. The effects of indi- counteract the effects of a policy. The effect of BMI on
vidual policies will also depend on the other nutrition disease outcomes is also complex. In addition to their
policies already in effect or those being simultaneously effect on BMI, SSB consumption may have a direct effect
implemented. on diabetes, heart disease and cancers (Jones et al. 2006;
We have not attempted to predict the effect of changes Lightwood et al. 2009; Milstein et al. 2007; Van Meijgaard
in total caloric intake on the distribution of BMI. Ludwig et al. 2009). Finally, in making policy choices, the costs
et al. (2001) found that, by replacing SSBs, BMI of those associated with each policy, as well as the feasibility,
ages 8–12 years can be reduced by between 0.1 and 0.24 sustainability and equity should be considered (Gortmaker
BMIs over 19 months, which translates to between 0.035 et al. 2011; Swinburn 2008).
and 0.08 kg/year, a substantial portion of the 0.25 kg The bottom line message of this paper is that the type of
reduction needed to maintain appropriate weight (Wang policy analysis needed to tackle the obesity problem
et al. 2006). Results from Ebbeling et al. (2006) imply a involves dealing with complexity, and admitting the
comparable reduction in BMI of 0.26 kg/m2 for every SSB uncertainty involved in any analysis. Does this mean that
serving (12 oz.) per day displaced by adolescents, and we should throw up our hands and await further study or just
Chen et al. (2009) found about twice that effect for adults. move on to other problems? In our opinion, the answer is
However, Fletcher et al. (2010a, b) found no effect of SSB no. The analysis above suggests that, while there is a wide
taxes on BMI. Powell et al. (2009) also found no effect of range of possible outcomes, there is considerable scope for
state-level soda taxes on adolescent BMI, but found a improving the current consumption habits. We expect that a
weak, though insignificant effect of vending machine soda multiplicity of policies will be needed, targeting different
tax rates on BMI of teens at risk for overweight. Assuming populations, such as in tobacco control policy (Levy et al.
a linear relationship, Sturm et al. (2010) found that an 18 2004). As we attempt various policies, outcomes must be
percent differential soda tax would correspond to 0.23 evaluated at the various levels in the chain from the initial
fewer BMI units, or a 1 % reduction in average BMI. At policy impact on behavior to population-level changes in
12 months, James et al. (2007) found that the percent of BMI. Simulation modeling provides a framework for sys-
overweight youth showed a net increase of 7.3 % relative tematically piecing together that information in order to
to the control group, but the difference disappeared within summarize the best information and provide direction on

123
Am J Community Psychol (2013) 51:299–313 311

the information still needed to better understand the prob- Chen, L., Appel, L. J., Loria, C., Lin, P. H., Champagne, C. M.,
lem and its solution. Elmer, P. J., et al. (2009). Reduction in consumption of sugar-
sweetened beverages is associated with weight loss: The
PREMIER trial. The American Journal of Clinical Nutrition,
Acknowledgments This study was funded by the Academy for 89(5), 1299–1306.
Educational Development, Project #4245-01 and the Robert Wood Chriqui, J. F., Eidson, S. S., Bates, H., Kowalczyk, S., & Chaloupka,
Johnson Foundation Healthy Eating Research Program, Contract F. J. (2008). State sales tax rates for soft drinks and snacks sold
#63048. through grocery stores and vending machines, 2007. Journal of
Public Health Policy, 29(2), 226–249.
Cradock, A. L., McHugh, A., Mont-Ferguson, H., Grant, L., Barrett, J.
L., Wang, Y. C., et al. (2011). Effect of school district policy
References change on consumption of sugar-sweetened beverages among
high school students, Boston, Massachusetts, 2004–2006. Pre-
Andreyeva, T., Chaloupka, F. J., & Brownell, K. D. (2011). venting Chronic Disease, 8(4), A74.
Estimating the potential of taxes on sugar-sweetened beverages Cullen, K. W., Hartstein, J., Reynolds, K. D., Vu, M., Resnicow, K.,
to reduce consumption and generate revenue. Preventive Med- Greene, N., et al. (2007). Improving the school food environ-
icine, 52(6), 413–416. ment: Results from a pilot study in middle schools. Journal of
Andreyeva, T., Long, M. W., & Brownell, K. D. (2010). The impact American Diet Association, 107(3), 484–489.
of food prices on consumption: A systematic review of research Cullen, K. W., Watson, K., & Zakeri, I. (2008). Improvements in
on the price elasticity of demand for food. American Journal of middle school student dietary intake after implementation of the
Public Health, 100(20), 216–222. Texas Public School Nutrition Policy. American Journal of
Auld, M. C., & Powell, L. M. (2009). Economics of food energy density Public Health, 98(1), 111–117.
and adolescent body weight. Economica, 76(304), 719–740. Cullen, K. W., Watson, K., Zakeri, I., & Ralston, K. (2006).
Bellisle, F., & Rolland-Cachera, M. F. (2001). How sugar-containing Exploring changes in middle-school student lunch consumption
drinks might increase adiposity in children. Lancet, 357(9255), after local school food service policy modifications. Public
490–491. Health Nutrition, 9(6), 814–820.
Besley, T. J., & Rosen, H. S. (1999). Sales taxes and prices: An Cullen, K. W., & Zakeri, I. (2004). Fruits, vegetables, milk, and
empirical analysis. National Tax Journal, 52(2), 157–178. doi: sweetened beverages consumption and access to a la carte/snack
10.1016/S0140-6736(00)04034-4. bar meals at school. American Journal of Public Health, 94(3),
Blum, J. E., Davee, A. M., Beaudoin, C. M., Jenkins, P. L., Kaley, L. 463–467.
A., & Wigand, D. A. (2008). Reduced availability of sugar- Dharmasena, S., & Capps, O. (2009). Demand interrelationships of
sweetened beverages and diet soda has a limited impact on at-home nonalcoholic beverage consumption in the United
beverage consumption patterns in Maine high school youth. States. Presented at the Agricultural & Applied Economics
Journal of Nutrition Education and Behavior, 40(6), 341–347. Association’s 2009 AAEA & ACCI Joint Annual Meeting,
Briefel, R. R., Crepinsek, M. K., Cabili, C., Wilson, A., & Gleason, P. Milwaukee, Wisconsin, July 26–29, 2009.
M. (2009a). School food environments and practices affect Dharmasena, S., & Capps, O., Jr. (2011). Intended and unintended
dietary behaviors of US public school children. Journal of consequences of a proposed national tax on sugar-sweetened
American Diet Association, 109(2 Suppl), S91–S107. beverages to combat the US obesity problem. Health Econom-
Briefel, R. R., Wilson, A., & Gleason, P. M. (2009b). Consumption of ics,. doi:10.1002/hec.1738.
low-nutrient, energy-dense foods and beverages at school, home, DiMeglio, D. P., & Mattes, R. D. (2000). Liquid versus solid
and other locations among school lunch participants and carbohydrate: Effects on food intake and body weight. Interna-
nonparticipants. Journal of American Diet Association, 109(2 tional Journal of Obesity and Related Metabolic Disorders,
Suppl), S79–S90. 24(6), 794–800.
Brown, I. J., Stamler, J., Van Horn, L., Robertson, C. E., Chan, Q., Doak, C. M., Visscher, T. L. S., Renders, C. M., & Seidell, J. C.
Dyer, A. R., et al. (2011). Sugar-sweetened beverage, sugar (2006). The prevention of overweight and obesity in children and
intake of individuals, and their blood pressure: International adolescents: A review of interventions and programmes.
study of macro/micronutrients and blood pressure. Hypertension, [Review]. Obesity Reviews, 7(1), 111–136.
57(4), 695–701. Drewnowski, A., & Bellisle, F. (2007). Liquid calories, sugar, and
Brown, T., & Summerbell, C. (2009). Systematic review of school- body weight. American Journal of Clinical Nutrition, 85(3),
based interventions that focus on changing dietary intake and 651–661.
physical activity levels to prevent childhood obesity: An update to Duffey, K. J., Gordon-Larsen, P., Shikany, J. M., Guilkey, D., Jacobs,
the obesity guidance produced by the National Institute for Health D. R., Jr, & Popkin, B. M. (2010). Food price and diet and health
and Clinical Excellence. Obesity Reviews, 10(1), 110–141. outcomes: 20 years of the CARDIA study. Archives of Internal
Brownell, K. D., Farley, T., Willett, W. C., Popkin, B. M., Chaloupka, Medicine, 170(5), 420–426.
F. J., Thompson, J. W., et al. (2009). The public health and Duffey, K. J., & Popkin, B. M. (2007). Shifts in patterns and
economic benefits of taxing sugar-sweetened beverages. The consumption of beverages between 1965 and 2002. Obesity
New England Journal of Medicine, 361(16), 1599–1605. (Silver Spring), 15(11), 2739–2747.
Butte, N. F., Christiansen, E., & Sorensen, T. I. (2007). Energy Ebbeling, C. B., Feldman, H. A., Osganian, S. K., Chomitz, V. R.,
imbalance underlying the development of childhood obesity. Ellenbogen, S. J., & Ludwig, D. S. (2006). Effects of decreasing
Obesity (Silver Spring), 15(12), 3056–3066. sugar-sweetened beverage consumption on body weight in
Butte, N. F., & Ellis, K. J. (2003). Comment on ‘‘obesity and the adolescents: A randomized, controlled pilot study. Pediatrics,
environment: Where do we go from here?’’. Science, 301(5633), 117(3), 673–680.
598. Author reply 598. Epstein, L. H., Dearing, K. K., Handley, E. A., Roemmich, J. N., &
Chaloupka, F. J., & Powell, L. M. (2009). Price, availability, and Paluch, R. A. (2006). Relationship of mother and child food
youth obesity: Evidence from bridging the gap. Preventing purchases as a function of price: A pilot study. [Article].
Chronic Disease, 6(3), A93. Appetite, 47(1), 115–118.

123
312 Am J Community Psychol (2013) 51:299–313

Fernandes, M. M. (2008). The effect of soft drink availability in Kann, L., Telljohann, S. K., & Wooley, S. F. (2007). Health
elementary schools on consumption. Journal of American Diet education: Results from the school health policies and programs
Association, 108(9), 1445–1452. study 2006. The Journal of School Health, 77(8), 408–434.
Finkelstein, E. A., Zhen, C., Nonnemaker, J., & Todd, J. E. (2010). Levy, D. T., Bauer, J. E., & Lee, H. R. (2006). Simulation modeling
Impact of targeted beverage taxes on higher- and lower-income and tobacco control: Creating more robust public health policies.
households. Archives of Internal Medicine, 170(22), 2028–2034. American Journal of Public Health, 96(3), 494–498.
Fletcher, J. M., Frisvold, D., & Tefft, N. (2010a). Taxing soft drinks Levy, D. T., Cummings, K. M., & Hyland, A. (2000). Increasing taxes
and restricting access to vending machines to curb child obesity. as a strategy to reduce cigarette use and deaths: Results of a
Health Affairs (Millwood), 29(5), 1059–1066. simulation model. Preventive Medicine, 31(3), 279–286.
Fletcher, J. M., Frisvold, D., & Tefft, N. (2010b). The effects of soft Levy, D. T., Friend, K. B., & Wang, Y. C. (2011a). A review of the
drink taxes on child and adolescent consumption and weight literature on policies directed at the youth consumption of sugar
outcomes. Journal of Public Economics, 94(2010), 967–974. sweetened beverages. Advances in Nutrition, 2, 182S–200S.
Flodmark, C. E., Marcus, C., & Britton, M. (2006). Interventions to Levy, D. T., Gitchell, J. G., & Chaloupka, F. (2004). The effects of
prevent obesity in children and adolescents: a systematic tobacco control policies on smoking rates: A tobacco control
literature review. [Review]. International Journal of Obesity, scorecard. Journal of Public Health Management & Practice,
30(4), 579–589. 10, 338–351.
Forshee, R. A., & Storey, M. L. (2004). Evaluation of the association Levy, D. T., Mabry, P. L., Wang, Y. C., Gortmaker, S., Huang, T. T.,
of demographics and beverage consumption with dental caries. Marsh, T., et al. (2011b). Simulation models of obesity: A
Food and Chemical Toxicology, 42(11), 1805–1816. review of the literature and implications for research and policy.
French, S. A. (2003). Pricing effects on food choices. Journal of Obesity Reviews, 12(5), 378–394. doi:10.1111/j.1467-789X.
Nutrition, 133(3), 841S–843S. 2010.00804.x.
Gortmaker, S. L., Swinburn, B. A., Levy, D., Carter, R., Mabry, P. L., Lightwood, J., Bibbins-Domingo, K., Coxson, P., Wang, Y. C.,
Finegood, D. T., et al. (2011). Changing the future of obesity: Williams, L., & Goldman, L. (2009). Forecasting the future
Science, policy, and action. Lancet, 378(9793), 838–847. doi: economic burden of current adolescent overweight: An estimate
10.1016/S0140-6736(11)60815-5. of the coronary heart disease policy model. American Journal of
Hall, K. D. (2008). What is the required energy deficit per unit weight Public Health, 99(12), 2230–2237.
loss? International Journal of Obesity (London), 32(3), 573–576. Lo, E., Coles, R., Humbert, M. L., Polowski, J., Henry, C. J., &
Hall, K. D., Sacks, G., Chandramohan, D., Chow, C. C., Wang, Y. C., Whiting, S. J. (2008). Beverage intake improvement by high
Gortmaker, S. L., et al. (2011). Quantification of the effect of school students in Saskatchewan, Canada. Nutrition Research,
energy imbalance on bodyweight. Lancet, 378(9793), 826–837. 28(3), 144–150.
Hartstein, J., Cullen, K. W., Reynolds, K. D., Harrell, J., Resnicow, Ludwig, D. S., Peterson, K. E., & Gortmaker, S. L. (2001). Relation
K., & Kennel, P. (2008). Impact of portion-size control for between consumption of sugar-sweetened drinks and childhood
school a la carte items: Changes in kilocalories and macronu- obesity: A prospective, observational analysis. Lancet, 357(9255),
trients purchased by middle school students. Journal of Amer- 505–508.
ican Diet Association, 108(1), 140–144. Malik, V. S., Popkin, B. M., Bray, G. A., Despres, J. P., & Hu, F. B.
Hattersley, L., Irwin, M., King, L., & Allman-Farinelli, M. (2009). (2010). Sugar-sweetened beverages, obesity, type 2 diabetes
Determinants and patterns of soft drink consumption in young mellitus, and cardiovascular disease risk. Circulation, 121(11),
adults: A qualitative analysis. Public Health Nutrition, 12(10), 1356–1364.
1816–1822. Malik, V. S., Schulze, M. B., & Hu, F. B. (2006). Intake of sugar-
Homer, J. B., & Hirsch, G. B. (2006). System dynamics modeling for sweetened beverages and weight gain: A systematic review.
public health: Background and opportunities. American Journal [Review]. American Journal of Clinical Nutrition, 84(2),
of Public Health, 96(3), 452–458. 274–288.
Institute of Medicine (2010). Bridging the evidence gap in obesity Matson-Koffman, D. M., Brownstein, J. N., Neiner, J. A., & Greaney,
prevention: A framework to inform decision making. M. L. (2005). A site-specific literature review of policy and
James, J., Thomas, P., Cavan, D., & Kerr, D. (2004). Preventing environmental interventions that promote physical activity and
childhood obesity by reducing consumption of carbonated nutrition for cardiovascular health: What works? [Review].
drinks: Cluster randomized controlled trial. BMJ, 328(7450), American Journal of Health Promotion, 19(3), 167–193.
1237–1239. Mattes, R. D. (1996). Dietary compensation by humans for supple-
James, J., Thomas, P., & Kerr, D. (2007). Preventing childhood mental energy provided as ethanol or carbohydrate in fluids.
obesity: Two year follow-up results from the Christchurch Physiological Behavior, 59(1), 179–187.
obesity prevention programme in schools (CHOPPS). BMJ, Milstein, B., Jones, A., Homer, J. B., Murphy, D., Essien, J., &
335(7623), 762–765. Seville, D. (2007). Charting plausible futures for diabetes
Johnson, D. B., Bruemmer, B., Lund, A. E., Evens, C. C., & Mar, C. prevalence in the United States: A role for system dynamics
M. (2009). Impact of school district sugar-sweetened beverage simulation modeling. Preventing Chronic Disease, 4(3), A52.
policies on student beverage exposure and consumption in O’Toole, T. P., Anderson, S., Miller, C., & Guthrie, J. (2007).
middle schools. The Journal of Adolescent Health, 45(3 Suppl), Nutrition services and foods and beverages available at school:
S30–S37. Results from the school health policies and programs study 2006.
Jones, A. P., Homer, J. B., Murphy, D. L., Essien, J. D., Milstein, B., The Journal of School Health, 77(8), 500–521.
& Seville, D. A. (2006). Understanding diabetes population Plachta-Danielzik, S., Landsberg, B., Bosy-Westphal, A., Johannsen,
dynamics through simulation modeling and experimentation. M., Lange, D., & Muller, M. J. (2008). Energy gain and energy
American Journal of Public Health, 96(3), 488–494. gap in normal-weight children: Longitudinal data of the KOPS.
Jordan, P. N., & Hall, K. D. (2008). Dynamic coordination of Obesity (Silver Spring), 16(4), 777–783.
macronutrient balance during infant growth: Insights from a Powell, L. M., Auld, M. C., Chaloupka, F. J., O’Malley, P. M., &
mathematical model. American Journal of Clinical Nutrition, Johnston, L. D. (2007). Access to fast food and food prices:
87(3), 692–703. Relationship with fruit and vegetable consumption and

123
Am J Community Psychol (2013) 51:299–313 313

overweight among adolescents. Advances in Health Economics in body weight in children. American Journal of Clinical
and Health Services Research, 17, 23–48. Nutrition, 83(4), 859–863.
Powell, L. M., & Chaloupka, F. J. (2009). Food prices and obesity: Van Meijgaard, J., Fielding, J. E., & Kominski, G. F. (2009).
Evidence and policy implications for taxes and subsidies. Assessing and forecasting population health: Integrating knowl-
Milbank Quarterly, 87(1), 229–257. edge and beliefs in a comprehensive framework. Public Health
Powell, L. M., Chriqui, J., & Chaloupka, F. J. (2009). Associations Report, 124(6), 778–789.
between state-level soda taxes and adolescent body mass index. Vartanian, L. R., Schwartz, M. B., & Brownell, K. D. (2007). Effects
The Journal of Adolescent Health, 45(3 Suppl), S57–S63. of soft drink consumption on nutrition and health: A systematic
Schwartz, M. B., Novak, S. A., & Fiore, S. S. (2009). The impact of review and meta-analysis. American Journal of Public Health,
removing snacks of low nutritional value from middle schools. 97(4), 667–675.
Health Education & Behavior, 36(6), 999–1011. Wang, Y. C. (2011). The potential impact of sugar sweetened
Seymour, J. D., Yaroch, A. L., Serdula, M., Blanck, H. M., & Khan, beverage Taxes in New York State Columbia (Ed.).
L. K. (2004). Impact of nutrition environmental interventions on Wang, Y. C., Bleich, S. N., & Gortmaker, S. L. (2008). Increasing
point-of-purchase behavior in adults: A review. [Article]. caloric contribution from sugarsweetened beverages and 100 %
Preventive Medicine, 39, S108–S136. fruit juices among US children and adolescents, 1988–2004.
Shi, L. (2010). The association between the availability of sugar- Pediatrics, 121(6), e1604–e1614.
sweetened beverage in school vending machines and its Wang, Y. C., Gortmaker, S. L., Sobol, A. M., & Kuntz, K. M. (2006).
consumption among adolescents in California: A propensity Estimating the energy gap among US children: A counterfactual
score matching approach. Journal Environmental Public Health, approach. Pediatrics, 118(6), e1721–e1733.
735613. Wang, Y. C., Ludwig, D. S., Sonneville, K., & Gortmaker, S. L.
Sichieri, R., Paula Trotte, A., de Souza, R. A., & Veiga, G. V. (2009). (2009). Impact of change in sweetened caloric beverage
School randomised trial on prevention of excessive weight gain consumption on energy intake among children and adolescents.
by discouraging students from drinking sodas. Public Health Archives of Pediatrics and Adolescent Medicine, 163(4),
Nutrition, 12(2), 197–202. 336–343.
Smith, T. A., Lin, B. H., & Lee, J.-Y. (2010). Taxing caloric Woodward-Lopez, G., Gosliner, W., Samuels, S. E., Craypo, L., Kao,
sweetened beverages: Potential effects on beverage consump- J., & Crawford, P. B. (2010). Lessons learned from evaluations
tion, caloric intake, and obesity. Washington, DC: US Depart- of California’s statewide school nutrition standards. American
ment of Agriculture, Economic Research Service. Journal of Public Health, 100(11), 2137–2145.
Sterman, J. D. (2006). Learning from evidence in a complex world. Yamada, M., Murakami, K., Sasaki, S., Takahashi, Y., & Okubo, H.
American Journal of Public Health, 96(3), 505–514. (2008). Soft drink intake is associated with diet quality even
Sturm, R., & Datar, A. (2005). Body mass index in elementary school among young Japanese women with low soft drink intake.
children, metropolitan area food prices and food outlet density. Journal of American Diet Association, 108(12), 1997–2004.
Public Health, 119(12), 1059–1068. Yen, S. T., Lin, B. H., & Smallwood, D. M. (2004). Demand for
Sturm, R., & Datar, A. (2008). Food prices and weight gain during nonalcoholic beverages: The case of low-income households.
elementary school: 5-year update. Public Health, 122(11), Agribusiness, 20(3), 309–321.
1140–1143. Zhen, C., Wohlgenant, M., Karns, S., & Kaufman, M. K. (2011).
Sturm, R., Powell, L. M., Chriqui, J. F., & Chaloupka, F. J. (2010). Habit formation and demand for sugar-sweetened beverages.
Soda taxes, soft drink consumption, and children’s body mass American Journal of Agricultural Economics, 93(1), 175–193.
index. Health Aff (Millwood), 29(5), 1052–1058. Zheng, Y., & Kaiser, H. M. (2008). Advertising and US nonalcoholic
Swinburn, B. A. (2008). Obesity prevention: the role of policies, laws beverage demand. Agricultural and Resource Economics
and regulations. Australia and New Zealand Health Policy, 5, 12. Review, 37, 2.
Swinburn, B. A., Jolley, D., Kremer, P. J., Salbe, A. D., & Ravussin,
E. (2006). Estimating the effects of energy imbalance on changes

123