mandatory calorie disclosure: a comprehensive analysis of

MANDATORY CALORIE DISCLOSURE:
A COMPREHENSIVE ANALYSIS OF THE EFFECT ON
CONSUMERS AND RETAILERS
Natalina Zlatevska
The University of Technology Sydney
Nico Neumann
The University of South Australia
Chris Dubelaar
Deakin University
ABSTRACT
In 2017 restaurants in the United States will need to provide calorie information on their
menus as part of the Patient Protection and Affordable Care Act. The estimated cost of
complying with the new legislation is in excess of $300 million. In the present research, we
examine the efficacy of the legislation. Specifically, we explore the effect that it will have on
changing both consumer and retailer behavior. Three studies are included in this research: A
meta-analysis of 152 studies examining the effect of calorie disclosure on calories selected, a
meta-analysis of 45 studies examining the effect of calorie disclosure on calories offered by
retailers, and a natural field experiment in the United States comparing restaurants that have
adjusted their menus voluntarily to disclose calories with those that have not. Across these
three studies we reveal a small but significant and unequivocal calorie disclosure effect for
menu labels.
Keywords: Calorie Labels, Mandatory Disclosure, Meta-Analysis
With Americans spending almost half (46%) of their food dollars on meals purchased outside
of the home (NRA, 2002), and with food away from home accounting for 34% of an
individual’s total calories (Lin et al. 1999), helping consumers make informed decisions
about the nutritional content of food purchased in quick and full-service restaurants is a
logical next step in the fight against obesity. In response to the call to provide consumers with
more nutritional information at the point of purchase, legislation (as part of the Patient
Protection and Affordable Care Act) was passed in 2010, requiring restaurants in the United
States to include calorie information on their menus.
Supporters of the legislation argue that by equipping consumers with caloric information
when they are often unaware, or underestimate the nutritional content of the food they are
eating, will encourage them to make considered and possibly healthier product selections.
Ideally, disclosing caloric information will encourage consumers to select food items
containing less calories (Burton et al. 2006; Burton and Kees 2012).
The new legislation applies to quick and some full-service retail establishments. Retailers
have until May 2017 to comply with the imposed guidelines. The initial cost of implementing
the proposed menu changes is estimated to exceed $350 million for food retailers, with an
ongoing cost of compliance of $44 million (FDA 2011).
Because of the large mandatory cost imposed on retailers (VanEpps, Roberto, Park,
Economos, and Belich 2016), there is strong interest in whether the benefits of the proposed
legislation will outweigh the expenses and required efforts for restaurants to implement menu
labels. Moreover, a few obesity-related policy changes have actually been implemented in the
United States within the last 10 years, further increasing the focus of the public on the
success of this policy.
However, while significant pressure exists to show that the new policy will successfully
improve consumer well-being and to justify the extra financial burdens for retailers, prior
academic studies investigating the effect of calorie labelling initiatives have provided mixed
results. For instance, Long et al. (2015) present summary data that reveal that disclosure of
calories is correlated with people opting for less calories, whereas other studies (e.g.,
Schwartz et al. 2011) suggest that calorie disclosure does not affect food choices. Thus, a
key, outstanding question is, will mandatory calorie disclosure be successful in changing
consumer behavior?
In the present research, we examine the efficacy of the legislation. To do this, first, we
reexamine extant research exploring the effect of calorie labelling initiatives. When
reviewing previous efforts on synthesizing the existing literature on menu labeling initiatives,
we find that this research has significant methodological shortcomings. In particular, many
existing reviews are not of a meta-analytic nature (that is, they are qualitative, conceptual
reviews), and those that are, suffer from potential biases. The biases include lack of control
for moderating variables in the meta-analysis and strong limitation in synthesized studies
resulting from the small sample sizes (6–38 studies). These two problems with prior studies
make it difficult to come to conclusions about general effects.
Second, to shed light on the likely overall calorie disclosure effect, we present a metaanalytic approach using multilevel modelling techniques. This meta-analysis method
accounts for the potential sources of bias mentioned above and relies on 152 synthesized
cases (representing 1,666,598 observations). In particular, our meta-analysis accounts for
various sources of heterogeneity by including critical control variables into the model and by
capturing dependencies imposed by the nested structure of experiments from the same
authors, thus further reducing bias in estimates (Neumann and Böckenholt 2014). Our
findings based on this robust estimation indicate a small but significant and unequivocal
calorie disclosure effect for menu labels.
Third, we note that prior research solely focuses on consumers’ reaction to new labels and
less on actions by, and implications of, supply-side dynamics. However, retailers and their
menu adjustments play a key role in the ultimate success of any policy, independent of new
information disclosure. Obligatory incentives often drive the behavior of information
providers, sometimes for the better and sometimes for the worse (Lowenstein et al. 2014).
Without a comprehensive examination of the effect of calorie disclosure on both the
consumer and the retailer, determining whether the legislation will have substantial, broadbased effects is difficult. Following this rationale, we present two different analyses to
investigate likely supply-side adjustments to any new legislation. We perform a meta-analysis
of 45 studies (representing 27,250 observations) examining the calories offered by retailers
before and after menu changes began to be implemented in U.S. restaurants between 2006
and 2014. In a second analysis, we present data from a natural field experiment in the United
States by comparing restaurants that have adjusted their menus voluntarily to disclose
calories with those that have not. This comparison further supports the meta-analysis but also
allows controlling for portion sizes and specific energy and nutrition composition on calorie
consumption.
Finally, we conclude with a review and investigation of the total economic implications of
calorie disclosures for both retailers and consumers, including suggesting strategies for
improving the overall efficacy of different policies to tackle obesity.
THE EFFECT ON CONSUMERS
The implementation of mandatory calorie disclosure on menu boards at the point of purchase
is expected to have the same positive effects on consumer food choices as the Nutrition
Labeling and Education Act (NLEA) did in 1990. After the enforcement of the NLEA, 48%
of consumers reported changing their purchase decisions based on the nutrition labels
included on processed food packaging (Burton and Biswas 1993). However, although
momentum has continued to gather around menu labelling policies, and despite the
widespread support of consumers for the introduction of menu labelling (national polls show
that between 67% and 83% of people support calorie disclosure (Roberto, Marlene, Schwartz,
and Brownell 2009)), evidence supporting the efficacy of the initiative remains unclear.
Multiple studies, across many academic disciplines, have investigated the impact of calorie
disclosure on nutrition but have reached little consensus as to the overall effect the legislation
will have on consumers. For example, Bollinger et al. (2010), Roberto et al. (2010), and
Hammond et al. (2013) conduct experiments illustrating that calorie disclosure reduces
energy consumption. In contrast, Swartz et al. (2011) and Downs, Wisdom, and Lowenstein
(2015) perform field experiments and find no significant calorie reduction related to changes
in consumers’ food choices. The experimental results of Dumanovsky et al. (2011) and Girz
et al. (2012) even suggest an increase in calories consumed following disclosure.
However, not only individual studies come to conflicting conclusions regarding effect
magnitudes. We identified nine review studies that synthesized experimental research on
calorie consumption measures (see Table 1). Similar to individual studies, some of the
systematic reviews report that disclosure is effective in reducing the caloric content of chosen
meals. For example, Littlewood et al. (2015, p. 1) conclude that the results of their review
show a “statistically significant effect” of menu labelling, whereas Fernandes et al. (2016, p.
534) argue that “calorie labeling in menus is not effective to promote healthier food choices.”
<<<INSERT TABLE 1 HERE>>>
What could explain the contradictions in research findings on the influence of calorie
disclosure? After reviewing the nine summary studies, we make several key observations that
seem to provide plausible explanations about the mixed results among the existing systematic
reviews. First, six of the nine studies in Table 1 (Harnack and French 2008; Swartz et al.
2011; Burton and Kees 2012; Lazareva 2015; Fernandes et al. 2016; Van Epps et al. 2016)
represent qualitative reviews where the research team summarized experimental data under
the lens of several key criteria, often subjectively grouped by two to three raters. In contrast
to a quantitative meta-analysis, such narrative review can be biased by the views of the raters
or the selection of studies (Rosenthal and DiMateo 2001). In addition, qualitative groupings
and analyses suffer from lack of transparency and assessment standards, such as the use of
established effect sizes that account for study precision or statistical methods that are
deployed to neutrally determine final conclusions.
Moreover, the three remaining quantitative reviews appear to be subject to two possible
methodological flaws: (a) strong sampling constraints leading to a small sample of studies
and (b) lack of control mechanisms for moderating variables.
First, we find that the three review studies that represent traditional meta-analyses in Table 1
were based on low sample sizes with between 12 and 19 included studies (from a population
of several hundred available articles). Any outcomes based on such small number of studies
could be biased because of sampling variance or a very restrictive sampling framework
(DerSimonian and Laird 1986). When reviewing the three studies more closely, we find that
the selection criteria of these works have been limited in terms of regions, publication status,
years of publication, and labelling methods (see Table 2). These sampling limitations reduced
the pool of synthesized studies and most likely created an unrepresentative subsample of all
available studies.
Second, in addition to the restricted sample sizes, the three existing meta-analyses did not
account for moderating variables when estimating the average effect size. The three reviews
present qualitative and subgroup analyses to investigate the impact of study characteristics.
The subgroup analysis only allows investigating one variable at a time, and conducting
multiple tests raises the risk of false-positive results because of chance alone (Yusuf et al.
1991).
<<<INSERT TABLE 2 HERE>>>
Given the concerns regarding the methodological choices of the existing reviews, the
generalizability of the findings should be questioned. To address the shortcomings of the
existing quantitative reviews and to shed light on the overall effect of calorie disclosure on
consumer food choices, we remove the restrictions above and perform a comprehensive
meta-analysis of 152 experimental comparisons. Furthermore, to gain a better understanding
of heterogeneity across the different studies, we carry out a metaregression on different study
characteristics as well as a multilevel modelling estimation. Metaregression is an extension to
subgroup analysis, simultaneously allowing accounting for the effects of both continuous and
categorical moderators (Thompson 2002).
SYUDY 1: THE EFFECT OF CALORIE DISCLOSURE ON CALORIE SELECTION BY
CONSUMERS
Meta-Analysis Method
Studies relevant for the meta-analysis were initially identified through a search of
ABI/Inform, ProQuest Digital Dissertations, Business Source Premier, Web of Science,
PsychInfo, Scopus, Google Scholar, and other databases using the following keywords: menu
labeling, restaurant labeling, calories on menu, calorie disclosure, nutritional information on
menu, Patient Protection and Affordable Care Act. The references in papers found in our
search were also examined to identify further studies. In particular, a number of qualitative
conceptual reviews (Harnack and French 2008; Swartz et al. 2011; Burton and Kees 2012;
Lazareva 2015; Fernandes et al. 2016; VanEpps et al. 2016) and three quantitative metaanalytic reviews (Sinclair et al. 2014; Littlewood et al. 2015; Long et al. 2015) were
identified during the search. The references of these papers were consulted to identify more
studies. The search was not restricted to any particular years of publication, country of data
collection, or languages.
Intervention
The primary intervention examined in the analysis was the disclosure of calorie information
on a retail restaurant menu. Studies that manipulated calorie (or kilojoule, e.g., Morley et al.
2013) disclosure along with another intervention (e.g., calories plus traffic light symbols
(Hammond et al. 2013)—calories plus energy expenditure (Platkin et al 2014) and calories
plus additional nutrients (Burton et al. 2006))—were also included in the analysis. Studies
that manipulated only a symbol and not calories (e.g., heart healthy stickers on menu items)
were excluded from the analysis (Freedman and Connors 2011; Levin 1996; Sharma et al.
2011; Vyth et al 2010). Burton et al. (2015) was also excluded from the analysis, even though
it was identified during the search, because it explored the nutrition facts panel rather than
calorie disclosure on restaurant menus. For the purpose of the analysis, retail restaurants were
defined as quick-service (e.g., Yamamoto et al. 2005) cafeterias (e.g., Holmes et al. 2013) or
fine-dining establishments (e.g., Fotouhinia-Yepes 2013).
Outcome
To be eligible for inclusion, studies were required to report the amount of calories selected
(or purchased, e.g., Krieger et al. 2013) by consumers. One study was excluded because it
provided information about the proportion of items selected from the menu, rather than the
amount of calories selected (Davis-Chervin et al. 1985). Likewise, both Driskell et al. (2008)
and Hwang and Lorenzen (2008) were excluded from the analysis because their outcome
variables were not of interest.1
Data extraction
All necessary information was extracted from the published articles, protocols, and
commentaries related to each study. In some cases, where raw data were not available,
assumptions and calculations were made from the figures included in the articles or the
explanation of the results in text. Two studies (Mayer et al. 1987; Webb et al. 2011)
examining the effect of calorie disclosure on menus could not be included because of lack of
data even though they fit the eligibility criteria.
A total of 152 studies reported in 44 papers representing 1,666,598 observations were
included in the meta-analysis examining the effect of calorie disclosure on calories selected
(see Appendix 1 for a summary of included studies).
Effect-size measure
1
Use or nonuse of nutrition labels and willingness to pay more for healthier food items,
respectively
For the effect-size measure (ES) of our meta-analysis, we computed the raw mean differences
for the calorie selection before and after the treatment:
(1)
ES = Xafter − Xbefore ,
where X is the average outcome in calorie selection of the respective experimental condition.
Using the raw means instead of standardized measures (which divide the raw means by a
study’s standard deviation) provides two advantages: (1) we can easily interpret the results
(possible calorie reductions) and (2) our effect size is 100% scale-free. That is, using a raw
metric allows us to strictly separate the impact of calorie disclosure on means from study-tostudy differences in standard deviation (Bond Jr., Wiitala, and Richard 2003).
Next, we computed the corresponding standard errors (SE) using the Comprehensive MetaAnalysis software (Borenstein et al. 2007).2 Here, we need to distinguish between betweenand within-subject designs for the SE estimations. For between designs, the formulas are as
follows (cite):
(2)
S2
SE = √nafter +
after
S2before
nbefore
,
where S denotes the sample standard deviation and n the sample sizes of the respective group.
For within designs, the formulas are as follows:
2
(3)
S
SE = √ diff
,
n
where Sdiff denotes the standard deviation and n the sample sizes of the pairs.
Graphical analysis
2
For some cases, only the confidence intervals or p-values were available for the reported studies (see Appendix
1). However, the meta-analysis software can still provide estimates for these cases.
To evaluate the effect magnitudes provided by the synthesized meta-analysis data, we
produced a funnel plot (Sterne and Egger 2001), which maps effect estimates from individual
studies against a measure of study precision (e.g., standard errors).
A visual inspection of the funnel plot of our data (see Figure 1) shows that the scatter
resembles a symmetrical inverted funnel with fairly evenly spread data points. This pattern
suggests that the risk of publication bias and systematic heterogeneity across studies is low.
Moreover, we find that the center of the funnel is in the negative region of the effect scale,
suggesting an average calorie reduction effect across the 152 observations.
<<<INSERT FIGURE 1 HERE>>>
Multilevel analysis
Meta-analyses (Glass 1976; Rosenthal 1978) combine the data from independent
studies and investigate this collection of estimates to answer hypotheses of interest. Often this
analysis involves more than one effect observation within the same study. Ignoring this
nested structure can lead to standard errors that are too low and therefore to an increase in
false-positive results (Neumann and Böckenholt 2014). To address these concerns, we
perform a multilevel mixed-effects meta-analysis, which accounts for the dependencies of
effect sizes in each study.
Control variables
Given that effect results in prior studies were mixed, we include several control variables in
our meta-analysis to account for potential study differences based on varying experimental
design characteristics. Specifically, we control for the labelling method, between- versus
within-subject designs, field versus lab tests, whether the food category was unhealthy or not,
and whether the menu refers to children only or not. In addition, we follow the suggestion of
Rosenthal and DiMatteo (2001) to include a control variable for the publication year of a
study. Specifically, we add to the model the time lag between the publication year of each
study and the first data point in our research synthesis. Table 3 provides a summary statistic
for all tested variables.
<<<INSERT TABLE 3 HERE>>>
Model and estimation
For our meta-analysis, we use the metafor package (Viechtbauer 2010) and obtain the
estimates through maximum likelihood estimation. The mixed-effects model estimate
consists of three levels: the first level encompasses the effect sizes, the second incorporates
the observations, and the third adds the studies that provide the observations. To avoid
multicollinearity, all independent variables in the model are grand-mean centered (see the
correlation matrix in Table 4).3 The independent variables are included on level 2 if they vary
within studies; otherwise on level 3. These specifications lead to the following meta-analysis
equations:
Levels 1 and 2:
(4)
ESij = β00j + β11 UNHEALTHYij + β12 LABELij + β13 CHILDRENij + rij,
Level 3:
(5)
β00j = γ000 + γ100 WITHIN j + γ200 FIELDj + γ300 YEARSj + uj,
where subscript i is used for the observations (i = 1, …, 152) and subscript j is used for the
studies (j = 1, …, 42). Moreover, rij represents the random observation effect (on level 2) and
u1j the random study effect (on level 3). The vectors uj and rij are specified to follow a
multivariate normal distribution with the dispersion matrices Τu and Τr.
3
Because our analysis is based on a linear model, we report phi coefficients for the binary variables instead of tetrachoric
correlations.
<<<INSERT TABLE 4 HERE>>>
Results
First, we compare the fit of the full model with all moderators with a reduced model that
contains no independent variables using the AIC and BIC criteria and the log-likelihood (see
Table 5). While the log-likelihood of the full model is slightly better than that of the reduced
model, both AIC and BIC are lower for the reduced model, which we therefore use to assess
the overall effect of calorie selection.
<<<INSERT TABLE 5 HERE>>>
According to the reduced meta-analysis model, we find a significant main effect for selected
calories after disclosure across our data set of 152 observations. On average, consumers
reduce their energy by about 30 calories (γ00 = 25.7, p < .001). Despite the multilevel
structure, which accounts for unobserved within-study and author heterogeneity, we find a
significant Q statistic (Q151 = 588.2, p < .001). The Q statistic is one measure for variability in
effect estimates that is due to heterogeneity rather than sampling error (Higgins and
Thompson 2002). Hence, a substantial degree of heterogeneity for our data still seems to
exist, which is most likely related to consumer, product, or study characteristics that we could
not test in our meta-analysis.
For completeness, we also report the results of the full model in Table 6. As implied by the
model comparison tests (which favor the reduced model), none of the moderating variables
have a significant impact on calorie reduction decisions across our 152 observations.
<<<INSERT TABLE 6 HERE>>>
In particular, our meta-analysis suggests no statistically significant differences in effect sizes
based on different labelling techniques, sample characteristics (children versus mixed
groups), or experimental designs. Regarding the latter, neither within- versus between-subject
designs nor field versus lab studies affect the magnitude of effects significantly. In addition,
we find no significant trend pattern in reported effect sizes over the years.
STUDY 2: THE EFFECT OF CALORIE DISCLOSURE ON CALORIES OFFERRED BY
RETAILERS
Controversy surrounds the impending mandatory implementation of the legislation. Unlike
taxes and bans, the legislation has widespread consumer support; however, support from
retailers is mixed. Many retailers are actively fighting the implementation of the legislation,
with some asking for exemptions because of fear of the increasing cost of compliance and a
possible negative effect on sales (Roberto et al. 2009). For example, grocery stores and pizza
retailers argue that they should be exempt from calorie labelling because of the complexity
and cost of implementing the scheme on their menus given the customizable nature of their
product offerings (Roberto et al. 2009).
In light of this controversy, we also sought to establish the effect of disclosure on retailers.
Advocates of the new legislation suggest that in addition to calorie disclosure motivating
consumers to purchase and ultimately consume less energy, the legislation could also
encourage the restaurant industry to reformulate their product offerings so that they also
contain less energy (VanEpps et al. 2016). However, as research into the effects of nutrition
labels on food products following the implementation of the NLEA reveals, firm responses
are not always clear. For example, in some cases, the implementation of the NLEA reduced
brand nutritional quality (Moorman et al. 2012).
In study 2, we examine the effect of calorie disclosure on a change in the amount of calories
offered by retailers using a meta-analytic approach.
Meta-Analysis Method
The search strategy adopted in study 2 was the same as that utilized in study 1. The key
difference in the meta-analytic methodology adopted between the two studies was the
primary outcome examined. To be eligible for inclusion in study 2, studies were required to
report the amount of calories offered by retailers both before and after the retailers had
disclosed caloric information on their menus. A total of 45 studies reported in 4 papers
representing 27,250 observations were included in the meta-analysis examining the effect of
calorie disclosure on calories provided by retailers between 2006 and 2014 (see Appendix 2
for a summary of included studies).
Graphical analysis
A visual inspection of the funnel plot of our data (see Figure 2) shows that the scatter
resembles a symmetrical inverted funnel with fairly evenly spread data points.
<<<INSERT FIGURE 2 HERE>>>
Multilevel meta-analysis
We again perform a multilevel mixed-effects meta-analysis to account for the
dependencies of effect sizes in our data. In addition, similar to study 1, we include several
control variables in our model to account for potential study differences based on varying
characteristics. For our analysis of calorie adjustments, we control for between- versus
within-subject designs (variable WITHIN), whether the food category was unhealthy or not
(variable UNHEALTHY) and whether the menu refers to children only or not (variable
CHILDREN; see Appendix 2 for classifications). In addition, we account for the time interval
between the calorie adjustments (how many years before and after, variable DIFFERENCE).
Table 7 provides a summary statistic for all tested variables. We compare the full
metaregression model with a reduced meta-analysis model (without covariates).
<<<INSERT TABLE 7 HERE>>>
Model and estimation
The new mixed-effects model we estimate consists of four levels: the first level
encompasses the effect sizes, the second incorporates the observations, the third controls for
shared-base conditions,4 and the fourth adds the studies that provide the observations. All
independent variables in the model are grand-mean centered (see the correlation matrix in
Table 8).5 The independent variables are included on level 2 if they vary within studies;
otherwise on level 4. These specifications lead to the following meta-analysis equations:
Levels 1 and 2:
(6)
ESijk = α0jik + β1 UNHEALTHYijk + β2 CHILDRENijk + rijk,
Level 3:
(7)
α0j = β0ji + uij,
Level 4:
(8)
β0ji = γ0 + γ1WITHIN j + γ2 DIFFERENCEj + vj,
where subscript i is used for the observations (i = 1, …, 45), subscript j for the share
conditions (i = 1, …, 35), and subscript j for the studies (j = 1, …, 4). Moreover, rijk
represents the random observation effect (on level 2), uij the random condition effect (on level
3), and uj the random study effect (level 4). The vectors rijk, uij, and uj are specified to follow
a multivariate normal distribution with the dispersion matrices Τu, Τr, and Τv.
4
Some of our experimental comparisons may allow more than one effect size. To account for the dependencies for these
shared bases, we follow Neumann, Böckenholt, and Sinha (2016) and add an extra level to the model.
5
Because of the strong correlation between the variables DIFFERENCE and WITHIN, we also ran the analyses with three
moderators only (dropping one of the variables, respectively). The results are identical, clearly favoring the reduced model.
<<<INSERT TABLE 8 HERE>>>
Results
Similar to the first meta-analysis, the model fit criteria suggest that the reduced meta-analysis
model better fits the data (see Tables 9 and 10). According to this model, average retailers
reduce their offerings by about 10 calories (γ00 = 10.3, p < .02). As implied by the model
comparison tests (which favored a reduced model), none of the moderating variables have a
significant impact on calorie reduction decisions across our 45 observations.
<<<INSERT TABLES 9 AND 10 HERE>>>
Although study 2 provides evidence of a reduction in calories offered by retailers, another
form of firm response could occur as a reaction to the imposed legislation: a reduction in the
amount of food served (Berman and Lavizzo-Mourey 2008; Young and Nestle 2007). In the
following study, using a quasi-experiment, we explore the effect of calorie disclosure on both
the calories provided by retailers and the serving size offered to consumers.
STUDY 3: QUASI EXPERIMENT
THE EFFECT OF CALORIE DISCLOSURE ON CALORIES PROVIDED AND SERVING
SIZE
To further explore retailer response, we examine changes in the products offered by retailers
before and after the 2008 period, where retailers began to voluntarily disclose calories on
menus (Farley, Caffarelli, Bassett, Silver, and Frieden 2009). In addition to determining
whether food has been reformulated in terms of its calorie content, the quasi-experimental
data we synthesized enable us to examine changes in the serving size offered. One method of
changing the total calorie content of food is by altering the size of the portion provided
(Zlatevska and Holden 2017).
Method
We test the effect of calorie disclosure on the reformulation of food products offered
by retailers using a quasi-experimental design, similar to that of Moorman et al. (2012). The
quasi experiment examines both calories provided and food serving size before and after the
introduction of the voluntary calorie labelling of food retailers in 2008. The study examined
nutritional data from 2003 through 2016 from eight chain restaurants. Retailers were eligible
for inclusion in the study if they were ranked among the top 50 quick-service restaurants in
2010 (https://www.qsrmagazine.com/reports/2010-qsr-50), served entrées (were not coffee
shops), and met criteria regarding location and data availability. The location criteria included
the retailer having 20 or more locations nationally. Five restaurant chains had voluntarily
disclosed the calorie content of the food products on their menu prior to 2015, and three
restaurants had not. These three restaurants served as a comparison, control condition.
Nutritional data came from an archive of Internet data collected from publicly accessible
Web pages beginning in 2003. The archive shows chain restaurant company websites exactly
as they appeared in previous years. Dates were printed on each Internet menu; thus, we were
able to verify the year the nutritional information represented. Nutrient analysis was limited
to calories and serving size. Only the same menu items that were available both before 2010
and after 2015 were eligible for inclusion. A total of 204 menu items met the eligibility
criteria from menus of 5 total cases and 3 control restaurants during the period from 2003
through 2016.
Results
A paired comparison of means revealed that restaurants who disclosed calories on their
menus (case restaurants) significantly decreased the calorie content (t = 3.63, p <.001) and
serving size (t = 6.69, p < .001) of their product offerings. In comparison, we did not identify
such significant changes in restaurants that did not disclose calories on their menus (control
restaurants) in terms of calorie change (t = 1.11, p = .27) or serving size (t = .46, p = .64;
Table 1).
<<<INSERT TABLE 11 HERE>>>
We also calculated the percentage reduction in calories provided,
(9) ((CaloriesPre - CaloriesPost)/ CaloriesPre)*100,
and the serving size,
(10) ((ServingSizePre – ServingSizePost)/ ServingSizePre)*100.
A simple linear regression was calculated to predict calorie change based on the change in
serving size. A significant regression was found (F(1,202) = 108.41, p <.001), with an R 2 of
.59 (Figure 3). Thus, we find that serving size decreases with calorie reduction.
<<<INSERT FIGURE 3 HERE>>>
Discussion
The quasi experiment shows that, when restaurants choose to reveal calorie information, they
also choose to adjust the size and, possibly, energy density of the offerings. While this quasi
experiment raises a number of questions around self-selection bias and motivation, it also
highlights that restaurants can, and do, make changes in both directions. The results from this
simple quasi experiment highlight that the net effects of relative energy intake resulting from
mandatory disclosure on consumers depend also on supply-side dynamics and not only on
consumers’ nutritional choices.
General Discussion
The central focus of the present research was to investigate the effect of mandatory calorie
disclosure on menu labels on consumer choices and retail practice. We first surveyed the
existing literature and review articles that have provided conflicting results regarding the
success of calorie disclosures in affecting nutrition choices to date. We argue that the mixed
results on the overall effect may be attributed to the limited sample size (12–19) of the
existing quantitative studies and the post hoc separate (subgroup) analyses of moderating
characteristics that may inflate false-positive.
To offer a more robust effect estimation, we presented a multilevel modelling metaregression
based on 152 experimental comparisons, which accounts for various dependencies and
sources of heterogeneity. Interestingly, we find that adding moderator variables to a metaanalysis that already accounts for the nested structure of the data (e.g., several experiments
from the same author) does not increase explanatory power.
Our multilevel random-effects meta-analysis suggests that there is a small but significant
negative effect on total calories chosen or consumed when consumers are provided with
information regarding calories in their meals. While this effect may appear not strong at first
(average reduction of about 30 calories per meal occasion), the reduced number of calories,
with approximately one-third of meals eaten outside the home, can add up to a noticeable
effect over time.6
6
Hill, Wyatt, Reed, and Peters (2003) suggest that an increase in people’s energy intake of as little as 50–100 kcal per day is
In addition to the meta-analysis on the impact of calorie disclosure on consumers’ nutrition
behavior, our work is the first to investigate supply-side adjustments as a result of the policy.
To do so, we also performed a meta-analysis of 45 comparisons of calorie content/energy
differences between restaurants that did and did not introduce voluntarily menu labels in the
United States between 2006 and 2014. According to this analysis, we find that retailers also
respond to mandatory disclosure of calorie information in a significant manner, with a
reduction of 10 calories on average per food item on their menu.
To further investigate the retailers’ role with regard to policy reactions, we presented retailer
data from a quasi-experiment that compares calories provided as well as food serving size
before and after the introduction of the voluntary calorie labelling. This comparison indicates
again that restaurants that disclosed calories on their menus significantly decreased their
calorie content. However, we also find that the serving size was reduced for restaurants,
suggesting that the energy density was not reduced or reduced only very slightly.
Hence, both our meta-analytic and quasi-experimental results illustrate that when restaurants
are required to release information on the calorie content of their foods, they can reduce the
offered calorie count as a result. This new finding tells us that the net effect on consumers is
expected to be larger than that shown in just the consumer studies as the real impact will be a
combination of the two effects.
Implications
The findings of our research have relevant implications for public policy. According to our
analyses, the legislation will create behavioral changes at both the demand and the supply
side. In other words, consumers are likely to make lower calorie choices and restaurants are
enough to account for the rise in obesity.
likely to opt for reformulating their menus to reduce the number of calories offered. This
means that regulators and lawmakers should also monitor the menu adjustments of
restaurants to estimate the net effect of the new legislation on public welfare.
Research exploring the general implications of targeted transparency in behavioral economics
and finance suggests that mandatory disclosure of information through legislation does not
affect the recipients of the information much (Lowenstein 2014). However, in line with
Lowenstein (2014), our findings imply that retailers adjust their menus following new
policies too. However, these adjustments may be a double-edged sword, with retailers opting
to decrease portion size in addition to the reformulation of the food provided.
Limitations and Research Directions
Our meta-analysis indicated unexplained heterogeneity that was not captured by the
moderators we tested. Future research should explore additional characteristics that may
explain differences in the effect size. From a policy perspective, consumer characteristics and
menu features may be particularly valuable.
Moreover, our findings on the supply-side reactions toward policy changes raise several
questions. For example, are retailers’ portion size reductions driven by the legislation or other
phenomena? Do people choose different lower-calorie food products that also happen to have
been reduced by the restaurant, resulting in a net decrease in consumption that is greater than
the sum of the parts? The interplay of calorie reduction and hidden portion decreases (with
potential energy increases per weight) is a fruitful area of public policy research that requires
further inquiry.
References
Berman M, and Lavizzo-Mourey R, 2008, ‘Obesity Prevention in the Information Age:
Caloric Information at the Point of Purchase’. The Journal of the American Medical
Association 300(4), pp. 433-5
Bollinger,
B.,
Leslie,
P.
and
Sorensen,
A.,
2011.
Calorie
posting
in
chain
restaurants. American Economic Journal: Economic Policy, 3(1), pp.91-128.
Bond Jr F, Wiitala WL, and Richard D. 2003, Meta-analysis of raw mean differences.
Psychological Methods. 2003 Dec 8(4), pp. 406-18.
Borenstein, M., Hedges, L. and Rothstein, H., 2007. Meta-analysis: Fixed effect vs. random
effects. Meta-analysis. com.
Burton, S., Cook, L.A., Howlett, E. and Newman, C.L., 2015. Broken halos and shattered
horns: overcoming the biasing effects of prior expectations through objective information
disclosure. Journal of the Academy of Marketing Science, 43(2), pp.240-256.
Burton, S., Creyer, E.H., Kees, J. and Huggins, K., 2006. Attacking the obesity epidemic: the
potential health benefits of providing nutrition information in restaurants. American journal
of public health, 96(9), pp.1669-1675.
Burton S, and Biswas, A. 1993, Preliminary Assessment of Changes in Labels Required by
the Nutrition Labeling and Education Act of 1990. The Journal of Consumer Affairs 27(1)
Summer 1993, pp. 127–144
Burton S, and Kees J. 2012, Flies in the Ointment? Addressing Potential Impediments to
Population-Based Health Benefits of Restaurant Menu Labeling Initiatives. Journal of Public
Policy & Marketing: Fall 2012, Vol. 31(2), pp. 232-239.
Davis-Chervin D, Rogers T, and Clark, M. 1985, Influencing food selection with point-ofchoice nutrition information. Journal of Nutrition Education and Behaviour. 17(1), pp. 18–22
DerSimonian R, and Laird N. 1986, Meta-analysis in clinical trials. Control Clin Trials. Sep;
7(3), pp. 177-88.
Downs JS, Wisdom J, and Loewenstein G. 2015, Helping Consumers Use Nutrition
Information:
Effects
of
Format
and
Presentation. American
Journal
of
Health
Economics, 1(3), pp. 326-344.
Driskell MM, Dyment S, Mauriello L, Castle P, and Sherman K. 2008, Relationships among
multiple behaviors for childhood and adolescent obesity prevention. Prev Med. 2008; 46, pp.
209–215.
Dumanovsky T, Huang CY, Nonas CA, Matte TD, Bassett, MT, and Silver, LD. 2011,
Changes in energy content of lunchtime purchases from fast food restaurants after
introduction of calorie labelling: cross sectional customer surveys. British Medical
Journal 2011 Jul 26, pp. 343.
Farley TA, Caffarelli A, Bassett, MT, Silver L, and Frieden TR. 2009 New York City’s Fight
over Calorie Labelling. Health Affairs 28(6), pp. w1089-w1109.
Fernandes, A.C., Oliveira, R.C., Proença, R.P., Curioni, C.C., Rodrigues, V.M. and Fiates,
G.M., 2016. Influence of menu labeling on food choices in real-life settings: a systematic
review. Nutrition reviews, 74(8), pp.534-548.
Fotouhinia-Yepes M. 2013, Menu Calorie Labelling in a Fine Dining Restaurant: Will it
Make a Difference? Journal of Quality Assurance in Hospitality & Tourism 14(3), pp. 281293.
Freedman MR, and Connors R. 2010, Point-of-purchase nutrition information influences
food-purchasing behaviors of college students: a pilot study. J Am Diet Assoc. 2010 Aug
110(8), pp. 1222-6.
Girz, L., Polivy, J., Herman, C.P. and Lee, H., 2012. The effects of calorie information on
food selection and intake. International Journal of Obesity, 36(10), pp.1340-1345.
Glass, G.V., 1976. Primary, Secondary, and Meta-Analysis of Research 1. Educational
researcher, 5(10), pp.3-8.
Hammond, D., Goodman, S., Hanning, R. and Daniel, S., 2013. A randomized trial of calorie
labeling on menus. Preventive medicine, 57(6), pp.860-866.
Harnack LJ, and French, SA. 2008, Effect of point-of-purchase calorie labeling on restaurant
and cafeteria food choices: A review of the literature. International Journal of Behavioral
Nutrition and Physical Activity 5(1), pp. 63.
Hedges LV, and Vevea JL. 1998, Fixed-and random-effects models in metaanalysis. Psychological methods, 3(4), pp. 486.
Higgins, J. and Thompson, S.G., 2002. Quantifying heterogeneity in a meta‐
analysis. Statistics in medicine, 21(11), pp.1539-1558.
Holmes, A.S., Serrano, E.L., Machin, J.E., Duetsch, T. and Davis, G.C., 2013. Effect of
different children’s menu labeling designs on family purchases. Appetite, 62, pp.198-202.
Hwang J, and Lorenzen CL. 2008, Effective nutrition labeling of restaurant menu and pricing
of healthy menu. Journal of Foodservice 19(5), pp. 270-276.
Krieger, J.W., Chan, N.L., Saelens, B.E., Ta, M.L., Solet, D. and Fleming, D.W., 2013. Menu
labeling regulations and calories purchased at chain restaurants. American journal of
preventive medicine, 44(6), pp.595-604.
Lazareva, Y., 2015. Can nutrition menu labelling positively influence consumer food
choices? A review of the literature. Surrey Undergraduate Research Journal, 1(1).
Levin, S., 1996. Pilot study of a cafeteria program relying primarily on symbols to promote
healthy choices. Journal of Nutrition Education, 28(5), pp.282-285.
Lin, B.H., Guthrie, J. and Frazao, E., 1999. Nutrient contribution of food away from
home. America’s eating habits: Changes and consequences, 750.
Littlewood, J.A., Lourenço, S., Iversen, C.L. and Hansen, G.L., 2016. Menu labelling is
effective in reducing energy ordered and consumed: a systematic review and meta-analysis of
recent studies. Public health nutrition, 19(12), pp.2106-2121.
Long, M.W., Tobias, D.K., Cradock, A.L., Batchelder, H. and Gortmaker, S.L., 2015.
Systematic review and meta-analysis of the impact of restaurant menu calorie
labeling. American journal of public health, 105(5), pp.e11-e24.
Loewenstein, G., Sunstein, C.R. and Golman, R., 2014. Disclosure: Psychology changes
everything. Annu. Rev. Econ., 6(1), pp.391-419.
Mayer, J.A., Brown, T.P., Heins, J.M. and Bishop, D.B., 1987. A multi-component
intervention for modifying food selections in a worksite cafeteria. Journal of Nutrition
Education, 19(6), pp.277-280.
Moorman, C., Ferraro, R. and Huber, J., 2012. Unintended nutrition consequences: firm
responses to the nutrition labeling and education act. Marketing Science, 31(5), pp.717-737.
Morley, B., Scully, M., Martin, J., Niven, P., Dixon, H. and Wakefield, M., 2013. What types
of nutrition menu labelling lead consumers to select less energy-dense fast food? An
experimental study. Appetite, 67, pp.8-15.
Neumann N. and Böckenholt U. 2014, A meta-analysis of loss aversion in product choice.
Journal of retailing 90(2), pp. 182-97.
Platkin, C., Yeh, M.C., Hirsch, K., Wiewel, E.W., Lin, C.Y., Tung, H.J. and Castellanos,
V.H., 2014. The effect of menu labeling with calories and exercise equivalents on food
selection and consumption. BMC obesity, 1(1), p.21.
Roberto CA, Schwartz MB, and Brownell KD. 2009, Rationale and Evidence for MenuLabeling Legislation. American Journal of Preventive Medicine 2009; 37(6) pp. 546-51.
Roberto, C.A., Larsen, P.D., Agnew, H., Baik, J. and Brownell, K.D., 2010. Evaluating the
impact of menu labeling on food choices and intake. American Journal of Public
Health, 100(2), pp.312-318.
Rosenthal R, and DiMatteo MR. 2001, Meta-analysis: recent developments in quantitative
methods for literature reviews. Annual Review of Psychology. 2001; (52), pp. 59-82.
Rosenthal R. 1978, Combining results of independent studies. Psychological Bulletin 85(1),
Jan 1978, pp. 185-193.
Schwartz, J., Riis, J., Elbel, B. and Ariely, D., 2012. Inviting consumers to downsize fastfood portions significantly reduces calorie consumption. Health Affairs, 31(2), pp.399-407.
Sharma, L.L., Teret, S.P. and Brownell, K.D., 2010. The food industry and self-regulation:
standards to promote success and to avoid public health failures. American Journal of Public
Health, 100(2), pp.240-246.
Sinclair, S.E., Cooper, M. and Mansfield, E.D., 2014. The influence of menu labeling on
calories selected or consumed: a systematic review and meta-analysis. Journal of the
Academy of Nutrition and Dietetics, 114(9), pp.1375-1388.
Sterne JA, and Egger M. 2001. Funnel plots for detecting bias in meta-analysis: guidelines on
choice of axis. Journal of Clinical Epidemiology. 2001 Oct; 54(10), pp. 1046-55.
Swartz JJ, Braxton D, and Viera AJ. 2011, Calorie menu labeling on quick-service restaurant
menus: an updated systematic review of the literature. International Journal of Behavioral
Nutrition and Physical Activity 2011 (8): pp. 135
Thompson 2002. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0
VanEpps, E.M., Downs, J.S. and Loewenstein, G., 2016. Calorie label formats: using
numeric and traffic light calorie labels to reduce lunch calories. Journal of Public Policy &
Marketing, 35(1), pp.26-36.
VanEpps EM, Roberto CA, Park S, Economos CD, and Bleich SN 2016, Restaurant menu
labeling policy: Review of evidence and controversies. Current Obesity Reports. 5(1) PP. 7280.
Viechtbauer 2010 Conducting Meta-Analyses in R with the metafor Package. Journal of
Statistical Software 36(3), pp. 1-48
Vyth El, Steenhuis IHM, Roodenburg AJC, Brug J, and Seidell JC. 2010, Front-of-pack
nutrition label stimulates healthier product development: a quantitative analysis. International
Journal of Behavioral Nutrition and Physical Activity 20107:65
Webb, K.L., Solomon, L.S., Sanders, J., Akiyama, C. and Crawford, P.B., 2011. Menu
labeling responsive to consumer concerns and shows promise for changing patron
purchases. Journal of Hunger & Environmental Nutrition, 6(2), pp.166-178.
Yamamoto, J.A., Yamamoto, J.B., Yamamoto, B.E. and Yamamoto, L.G., 2005. Adolescent
fast food and restaurant ordering behavior with and without calorie and fat content menu
information. Journal of Adolescent Health, 37(5), pp.397-402.
Young LR, and Nestle M. 2007, Portion sizes and obesity: responses of fast-food companies.
Journal of Public Health Policy. 2007 Jul; 28(2), pp. 238-48.
Yusuf S, Wittes J, Probstfield J, and Tyroler HA. 1991, Analysis and Interpretation of
Treatment Effects in Subgroups of Patients in Randomized Clinical Trials. Journal of the
American Medical Association. Jul 3; 266(1), pp. 93-8.
Zlatevska N. and Holden S. 2017, “Nudging the Weight off. Small Portions Big
Effects,” Working Paper.
TABLE 1
REVIEWS EXAMINING THE EFFECT OF CALORIE DISCLOSURE ON CALORIES
SELECTED
Review
K
Effect
Harnack, French (2008)
Qualitative
6
Mixed
Swartz et al. (2011)
Qualitative
7
No Effect
Burton, Kees (2012)
Qualitative
10
Mixed
Lazareva (2015)
Qualitative
16
Mixed
Fernandes et al. (2016)
Qualitative
38
No Effect
VanEpps et al. (2016)
Qualitative
16
Mixed
Sinclair et al. (2014)
Quantitative
17
No Effect (p = .35)
Littlewood et al. (2015)
Quantitative
12
Reduction Effect (p < .05)
Long et al. (2015)
Quantitative
19
Mixed (p = .02)
TABLE 2
SAMPLING LIMITATIONS OF PREVIOUS META-ANALYSES
Meta-analysis
Sinclair et al. (2014)
Littlewood et al. (2015)
Long et al. (2015)
Sampling limitations
Regions, participants, and
year of publication.
Details
Studies from the United States, Canada, Australia, and
New Zealand only. Participants must be “generally”
healthy. Studies from Jan. 1990 until March 2013 were
included.
Year of publication
Studies from 2012 to 2015 were included.
Publication status and label
type
Dissertations and non-peer-reviewed work excluded.
Experiments with menu labelling formats other than
those under (US) federal menu calorie labelling laws not
included.
TABLE 3
STUDY 1: INCLUDED VARIABLES AND SUMMARY STATISTICS
Within
min
max
mean
var
Label
Field
Unhealthy
0
0
0
0
1
1
1
1
0.157895 0.289474 0.493421 0.57236842
0.133845 0.207041 0.251612 0.24638376
Children Years
0
1
1
42
0.092105 36.940789
0.084176 26.254749
TABLE 4
STUDY 1 CORRELATION MATRIX
Within
Label
Field
Unhealthy
Children
Years
Within
Label
Field
Unhealthy
Children
Years
1.00
–.24
.25
–.20
.24
–.52
–.24
1.00
–.36
–.10
–.20
.10
.25
–.36
1.00
–.12
.18
–.19
–.20
–.10
–.12
1.00
.23
.11
.24
–.20
.18
.23
1.00
–.06
–.52
.10
–.19
.11
–.06
1.00
TABLE 5
STUDY 1 MODEL COMPARISON
DF
Full model
Reduced model
9
3
LL
–857.8
–860.5
AIC
1733.6
1727.1
BIC
1760.8
1736.2
TABLE 6
RESULTS OF THE REDUCED META-ANALYSIS MODEL
Coefficient
Estimate
SE
P-value
Intercept
γ000
–33.2
8.0
<.001
Within
γ100
–11.7
23.5
.62
Field
γ200
15.3
16.4
.35
Years
γ300
–0.1
1.2
.92
Unhealthy
Label
β11
β12
–16.5
–15.7
12
11.3
.17
.16
Children
β13
16.1
30
.60
TABLE 7
STUDY 2 CORRELATION MATRIX
Within
Difference
Children
Unhealthy
Within
Difference Children Unhealthy
1.00
–.97
–.29
–.32
–.97
1.00
.13
.41
–.29
.13
1.00
–.42
–.32
.41
–.42
1.00
TABLE 8
STUDY DESCRIPTIVES
Within
Difference Children
Unhealthy
min
0
1
0
0
max
1
6
1
1
range
1
5
1
1
mean
.67
2.51
.18
.44
var
.23
4.76
.15
.25
TABLE 9
STUDY 1 MODEL COMPARISON
DF
Full model
Reduced model
8
4
LL
- 228.9
–229.4
AIC
473.7
466.9
BIC
488.2
474.1
TABLE 10
RESULTS OF THE REDUCED META-ANALYSIS MODEL
Coefficient
Intercept
γ000
Estimate
–10.3
SE
4.33
P-value
<.02
TABLE 11
CHANGES IN CALORIES PROVIDED AND SERVING SIZE
Calories Pre
Calories Post
Serving Size Pre
Serving Size Post
Case Restaurants
418.19
402.26
187.05 g
178.05 g
Control Restaurants
484.74
477.98
241.02 g
239.93 g
FIGURE 1
FUNNEL PLOT OF 152 OBSERVATIONS FOR CALORIE SELECTION
FIGURE 2
FUNNEL PLOT OF 45 OBSERVATIONS FOR CALORIES PROVIDED
FIGURE 3
REDUCTION IN CALORIES AS A FUNCTION OF REDUCTION IN SERVING SIZE
β = .86, p <.001
APPENDIX 1
15
21.85
800
815
22
within
field
NC
c
1
Aaron, Evans, Mela
1995
Low restraint
150
55.60
750
900
43
within
field
NC
c
1
Auchincloss et al.
2013
–113
63.24
1891
1778
648
between
field
NC
c
1
Balfour et al.
1996
Calories
–105
49.12
711
606
130
within
lab
NC
c
1
Calories
Study Design
Observation(n)
Mean Calories After
Raw SE
Study Descriptor
Healthy (Yes = 1, No = 0)
Label Method (c = calorie only, co = calorie +
other information)
High restraint
Lab or Field
1995
Study Name
Mean Calories Before
Aaron, Evans, Mela
Raw Difference in Means
Children (NC = Not Only Children in Sample, C =
Children Only in Sample)
STUDY 1: INCLUDED STUDIES
Bassett et al.
2008
–51.7
26.66
765.5
713.8
1805
between
field
NC
c
1
Bollinger et al
2010
–14.4
7.35
247
232.6
1510000
within
field
NC
c
0
Brissette et al.
2013
–59.6
30.06
947.7
888.1
1094
between
field
NC
c
0
Cantor et al.
2015
2008
–13
19.26
796
783
762
between
field
NC
c
0
Cantor et al.
2015
2013–14 post 2 and post 3
–4
5.93
839
835
3757
between
field
NC
c
0
Cantor et al.
2015
2013–14 post 3 and post 4
–31
45.96
835
804
3349
between
field
NC
c
0
Downs et al.
2013
No recommendation
14
25.26
334
348
561
between
field
NC
c
0
Downs et al.
2013
Per-meal or per-day calorie recommendation
41
23.68
369
410
561
between
field
NC
c
0
Dowray et al.
2013
Burger calories
–33.3
30.42
414.85
381.58
346
between
lab
NC
c
0
Dowray et al.
2013
Burger calories + miles walking
–78.6
25.80
414.85
336.24
346
between
lab
NC
co
0
Dowray et al.
2013
Burger calories + min walking
–41.9
26.83
414.85
372.93
346
between
lab
NC
co
0
Dowray et al.
2013
Dessert + miles walking
–5.48
27.48
74.55
69.07
112
between
lab
NC
co
0
Dowray et al.
2013
Dessert calories
9
32.11
74.55
83.55
112
between
lab
NC
c
0
Dowray et al.
2013
Dessert calories + min walking
1.94
31.73
74.55
76.49
112
between
lab
NC
co
0
Dowray et al.
2013
Drinks + miles walking
–28.5
27.77
146.55
118.05
192
between
lab
NC
co
0
Dowray et al.
2013
Drinks + min walking
–24.2
29.84
146.55
122.39
192
between
lab
NC
co
0
Dowray et al.
2013
Drinks calories
–29.3
27.87
146.55
117.3
192
between
lab
NC
c
0
Dowray et al.
2013
Salad calories
6.16
32.65
136.55
142.71
164
between
lab
NC
c
1
Dowray et al.
2013
Salad calories + miles walking
–27
27.91
136.55
109.59
164
between
lab
NC
co
1
Dowray et al.
2013
Salad calories + min walking
–22.5
31.15
136.55
114.05
164
between
lab
NC
co
1
Dowray et al.
2013
Side calories
–43.4
23.14
245.3
201.92
268
between
lab
NC
c
1
Dowray et al.
2013
Side calories + miles walking
–52
23.29
245.3
193.35
268
between
lab
NC
co
1
Dowray et al.
2013
Side calories + min walking
–15
21.84
245.3
230.29
268
between
lab
NC
co
1
Dumanovsky et al.
2011
Calories Au Bon Pain
–80
28.59
554.5
474.5
309
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Burger King
43.6
45.20
923.8
967.4
2298
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Domino's
–280
239.61
1309.1
1028.9
99
between
field
NC
c
0
Dumanovsky et al.
2011
Calories KFC
–58.9
33.08
926.7
867.8
990
between
field
NC
c
0
Dumanovsky et al.
2011
Calories McDonald’s
–44
23.93
829.2
785.2
5269
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Papa John’s
–51.4
115.38
622.8
571.4
408
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Pizza Hut
–96.2
100.69
1039
942.8
139
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Popeye’s
26
46.16
948.5
974.5
822
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Subway
133.1
27.16
749.2
882.3
3877
between
field
NC
c
1
Dumanovsky et al.
2011
Calories Taco Bell
34.5
67.50
773.2
807.7
588
between
field
NC
c
0
Dumanovsky et al.
2011
Calories Wendy's
–37.1
61.07
858
820.9
1001
between
field
NC
c
0
Elbel et al.
2009
Calorie New York
21
31.12
825
846
809
between
field
NC
c
0
Elbel et al.
2009
Calorie Newark
3
4.44
823
826
347
between
field
NC
c
0
Elbel et al.
2013
Female
–93
60.07
979
886
760
between
field
NC
c
0
Elbel et al.
2013
Male
–9
59.11
964
955
1169
between
field
NC
c
0
Elbel, Gyamfi, Kersh
Elbel, Gyamfi, Kersh
2011
2011
Calories female 0–12 New York
Calories female 0–12 Newark
61
134
68.74
123.98
480
511
541
645
65
13
between
between
field
field
C
C
c
c
0
0
Elbel, Gyamfi, Kersh
2011
Calories female 13+ New York
4
78.04
712
716
71
between
field
C
c
0
Elbel, Gyamfi, Kersh
2011
Calories female 13+ Newark
170
165.16
560
730
20
between
field
C
c
0
Elbel, Gyamfi, Kersh
2011
Calories male 0–12 New York
–96
77.24
622
526
62
between
field
C
c
0
Elbel, Gyamfi, Kersh
2011
Calories male 0–12 Newark
–128
98.06
621
493
20
between
field
C
c
0
Elbel, Gyamfi, Kersh
2011
Calories male 13+ New York
67
90.54
747
814
65
between
field
C
c
0
Elbel, Gyamfi, Kersh
2011
Calories male 13+ Newark
–83
169.93
878
795
17
between
field
C
c
0
Ellison, Lusk, Davis
2014
Calories + traffic light study 2
–67.8
26.16
740.82
673.07
645
between
field
NC
co
1
Ellison, Lusk, Davis
2014
Calories study 1
Ellison, Lusk, Davis
Ellison, Lusk, Davis
Ellison, Lusk, Davis
2014
2013
2013
Calories study 2
Calories + traffic light entree
Calories + traffic light extra calories
Ellison, Lusk, Davis
2013
Ellison, Lusk, Davis
5
8.97
622.44
627.44
2151
between
field
NC
c
1
–32.5
–129
60
27.69
190.48
88.60
740.82
663
103
708.36
534
163
603
92
92
between
between
between
field
field
field
NC
NC
NC
c
co
co
1
1
1
Calories entree
–15
22.15
663
648
92
between
field
NC
c
1
2013
Calories extra calories
66
97.46
103
169
92
between
field
NC
c
1
Finkelstein et al.
2011
Calories period 1
5.7
2.91
1211.3
1217
23358.8
between
field
NC
c
1
Finkelstein et al.
2011
Calories period 2
3
1.53
1211.3
1214.3
22593.6
between
field
NC
c
1
Gerend
2008
Men
92
64.28
1052
1144
111
between
lab
NC
c
0
Gerend
2008
Women
–146
48.99
934
788
177
between
lab
NC
c
0
Girz et al
Girz et al
Girz et al
Girz et al
Girz et al
2012
2012
2012
2012
2012
Men restrained cal + framing study 2
Men restrained cal study 2
Men unrestrained cal + framing study 2
Men unrestrained cal study 2
Restrained pasta high cal study 1
169.8
123.8
–139
–88.1
–37.6
129.71
105.59
129.27
102.32
98.56
698.39
698.39
930.31
930.31
480.51
868.18
822.15
790.98
842.18
442.95
27
44
28
44
21
between
between
between
between
between
lab
lab
lab
lab
lab
NC
NC
NC
NC
NC
co
c
co
c
co
1
1
1
1
0
Girz et al
2012
Restrained pasta low cal study 1
–56.3
110.14
480.51
424.18
22
between
lab
NC
c
0
Girz et al
2012
Restrained salad high cal study 1
164.8
86.96
246.27
411.08
15
between
lab
NC
c
1
Girz et al
Girz et al
2012
2012
Restrained salad low cal study 1
Unrestrained pasta high cal study1
166.4
–176
75.59
64.67
246.27
514.58
412.68
338.7
22
38
between
between
lab
lab
NC
NC
c
c
1
0
Girz et al
2012
Unrestrained pasta low cal study 1
–92.6
62.34
514.58
422.02
42
between
lab
NC
c
0
Girz et al
2012
Unrestrained salad high cal study 1
205.3
83.83
304.2
509.48
17
between
lab
NC
c
1
Girz et al
2012
Unrestrained salad low cal study 1
–5.71
83.84
304.2
298.49
19
between
lab
NC
c
1
Girz et al
2012
Women restrained cal + framing study 2
–62
75.12
490.35
428.4
34
between
lab
NC
co
1
Girz et al
2012
Women restrained cal study 2
–26.6
81.30
490.35
463.78
46
between
lab
NC
c
1
Girz et al
2012
Women unrestrained cal + framing study 2
–113
82.67
509.27
396.74
37
between
lab
NC
co
1
Girz et al
2012
Women unrestrained cal study 2
–32.8
68.50
509.27
476.44
60
between
lab
NC
c
1
Hammond et al.
2013
Calories + traffic lights
–62.8
35.06
839.6
776.8
318
between
lab
NC
co
1
Hammond et al.
Hammond et al.
2013
2013
Calories 1
Calories shown in traffic lights
–95.4
–74.7
38.11
36.41
839.6
839.6
744.2
764.9
327
314
between
between
lab
lab
NC
NC
c
co
1
1
Harnack et al.
2008
Calorie
65.7
45.25
739
804.7
301
between
lab
NC
c
0
Harnack et al.
2008
Calorie + price
22
41.28
739
761
300
between
lab
NC
co
0
Hoefkens et al.
2011
Calories
1
24.14
597
598
224
within
lab
NC
c
1
Holmes e al.
Krieger et al.
Krieger et al.
Krieger et al.
2013
2013
2013
2013
Combos
Calories Burger 16–18 months post
Calories Burger 4–6 months post
Calories Coffee Chain 16–18 months post
–56
–12.8
–9.4
–22.2
28.52
71.59
72.21
30.92
279.36
904.7
904.7
154.3
223.32
891.9
895.3
132.1
453
1390
1393
1106
within
between
between
between
field
field
field
field
C
NC
NC
NC
c
c
c
c
1
0
0
0
Krieger et al.
2013
Calories Coffee Chain 4–6 months post
–10.6
39.78
154.3
143.7
803
between
field
NC
c
0
Krieger et al.
2013
Calories Sandwich 16–18 months post
–9.5
48.57
871.5
862
1495
between
field
NC
c
1
Krieger et al.
2013
Calories Sandwich 4–6 months post
35.3
47.20
871.5
906.8
1496
between
field
NC
c
1
Krieger et al.
2013
Calories Taco 16–18 months post
–113
55.30
979.6
866.7
1202
between
field
NC
c
0
Krieger et al.
2013
Calories Taco 4–6 months post
–8.6
87.71
979.6
971
1192
between
field
NC
c
0
Lee and Thompson
2016
Burger
–43.2
24.56
521.33
478.14
384
between
lab
NC
c
0
Lee and Thompson
2016
Burger co
–26.5
27.79
521.33
494.82
384
between
lab
NC
co
0
Lee and Thompson
2016
Dessert
–24.9
42.65
311.04
286.15
90
between
lab
NC
c
0
Lee and Thompson
2016
Dessert co
75.9
58.32
311.04
386.94
90
between
lab
NC
co
0
Lee and Thompson
2016
Drinks
–19.2
21.08
198.51
179.33
378
between
lab
NC
c
0
Lee and Thompson
2016
Drinks co
43.24
27.24
198.51
241.75
378
between
lab
NC
co
0
Lee and Thompson
2016
Salad
61.81
50.87
296.14
357.95
82
between
lab
NC
c
1
Lee and Thompson
2016
Salad co
1.66
51.55
296.14
297.8
82
between
lab
NC
co
1
Lee and Thompson
2016
Side
58.32
17.15
348.54
406.86
326
between
lab
NC
c
0
Lee and Thompson
2016
Side co
–12.6
14.93
348.54
335.98
326
between
lab
NC
co
0
Liu et al.
Liu et al.
2012
Calories
–84.1
23.03
1759.61
1675.52
208
between
lab
NC
c
1
2012
Colored calories
–305
24.19
1759.61
1454.55
203
between
lab
NC
co
1
Liu et al.
Lowe et al.
2012
2010
Mayer et al.
Milich et al.
Morley et al.
1987
1975
2013
Morley et al.
Rank-ordered calories
–154
–86.1
20.41
25.35
1759.61
656.09
1605.5
570
212
96
between
within
lab
field
NC
NC
c
c
1
1
Calories selected
Calorie (kj)
–7.34
–48
–117
8.35
24.42
59.49
461.56
507
1105.88
454.22
459
989
265
450
515
within
within
between
field
field
lab
NC
NC
NC
co
c
c
1
0
0
2013
Calorie (kj) + %DI
–91.1
134.93
1105.88
1014.81
519
between
lab
NC
co
0
Morley et al.
2013
Calorie (kj) + %DI + TL
–23.2
34.34
1105.88
1082.7
518
between
lab
NC
co
0
Morley et al.
2013
Calorie (kj) + TL
–120
60.83
1105.88
986.38
519
between
lab
NC
co
0
Pang and Hammond
Pang and Hammond
Pang and Hammond
2010
2010
2010
Calories
Calories + health statement
Calories + physical activity scale
–31.2
–35.2
–23
10.07
11.00
10.35
332.8
332.8
332.8
301.6
297.6
309.8
117
107
111
between
between
between
lab
lab
lab
NC
NC
NC
c
co
co
0
0
0
Platkin
2009
Lunch 1 calories
72.95
112.04
986.62
1059.57
42
between
lab
NC
c
0
Platkin
2009
Lunch 1 calories + EE
–146
122.00
986.62
840.79
42
between
lab
NC
co
0
Platkin
2009
Lunch 2 calories
–96.6
127.31
995.4
898.82
42
between
lab
NC
c
0
Platkin
2009
Lunch 2 calories + EE
–154
123.88
995.4
841.31
42
between
lab
NC
co
0
Prins et al.
2012
–163
69.03
678.39
515
56
between
field
NC
c
1
Pulos, Leng
2010
Restaurant A
–16.8
8.57
630
613.2
2796
within
field
NC
c
1
Pulos, Leng
2010
Restaurant B
–55.6
28.34
880
824.4
1320
within
field
NC
c
1
Pulos, Leng
2010
Restaurant C
–53.6
27.31
1620
1566.4
840
within
field
NC
c
1
Pulos, Leng
2010
Restaurant D
–20.6
10.51
710
689.4
5418
within
field
NC
c
1
Pulos, Leng
2010
Restaurant E
12.1
17.94
770
782.1
3108
within
field
NC
c
1
Pulos, Leng
2010
Restaurant F
–1
1.48
620
619
2922
within
field
NC
c
1
Roberto et al.
Roberto et al.
2010
2010
Calories
Calories + rec. daily intake information
–124
–203
157.09
153.51
1458.92
1458.92
1334.72
1256.37
191
199
between
between
field
field
NC
NC
c
co
1
1
Roseman et al.
2013
Intention to use labels
–27.9
30.04
709.9
682.04
239
between
lab
NC
c
0
Roseman et al.
2013
No intention to use labels
81.94
62.28
757.97
839.91
63
between
lab
NC
c
0
Schmitz, Fielding
1986
–71
26.47
638
567
823
within
field
NC
c
1
Schwartz et al.
2012
Study 2 entree baseline
19
19.21
536
555
1072
between
field
NC
co
1
Schwartz et al.
Schwartz et al.
Schwartz et al
2012
2012
2012
Study 2 entrée downsized
Study 2 side dish baseline
Study 2 side dish downsized
52
–6
19
23.53
7.81
13.44
508
484
437
560
478
456
1047
978
481
between
between
between
field
field
field
NC
NC
NC
co
co
co
1
1
1
Tandon et al.
Tandon et al.
2011
2011
Calories normal weight children
Calories normal weight parents
19
–88
50.83
134.69
782
780
801
692
55
24
within
within
field
field
C
NC
c
c
0
0
Tandon et al.
2011
Calories overweight/obese children
–56
107.97
936
880
20
within
field
C
c
0
Tandon et al.
2011
Calories overweight/obese parents
–111
188.89
848
737
42
within
field
NC
c
0
Temple et al.
2011
Calories female
–270
123.32
750
480
23
between
lab
NC
c
0
Temple et al.
2011
Calories male
–400
182.70
1010
610
23
between
lab
NC
c
0
Vanderlee, Hammond
2013
–128
38.78
563
435
1003
between
field
NC
c
1
Wei, Miao
2013
Calorie healthful
96.35
58.77
618.54
714.89
94
between
lab
NC
co
1
Wei, Miao
2013
Calorie unhealthful
–55.9
52.74
668.33
612.4
94
between
lab
NC
co
0
Wisdom, Downs,
Lowenstein
2010
Calories study 1 calorie label
–48.1
54.39
843.62
795.57
146
between
lab
NC
c
0
2010
Calories study 1 calories + recommended daily
intake
–85.5
54.28
843.62
758.11
146
between
lab
NC
co
0
–71.7
48.74
902.88
831.15
173
between
lab
NC
c
0
Wisdom, Downs,
Lowenstein
Wisdom, Downs,
Lowenstein
Wisdom, Downs,
Lowenstein
2010
2010
Calories study 2 calorie label
Calories study 2 calories + recommended daily
intake
–111
48.66
902.88
792.21
173
between
lab
NC
co
0
Yamamoto et al.
2005
Calories Denny’s
21
31.03
1031
1010
106
within
lab
C
c
0
Yamamoto et al.
2005
Calories McDonald’s
–45
22.70
933
888
106
within
lab
C
c
0
Yamamoto et al.
2005
Calories Panda Express
–37
18.66
874
837
106
within
lab
C
c
0
Yong et al.
2009
Calories
–12.4
4.60
646.5
634.1
28150
within
field
NC
c
1
353.9
361.7
384.1
239.4
200.6
197.2
411.6
517.9
486.9
823.5
779.1
515.8
650.4
684.8
416.2
335.6
418.4
545.2
617.8
563.4
995
453
453
835
589
589
1497
761
761
253
253
4341
1331
1331
279
279
352
352
1521
1521
within
within
within
within
within
within
within
within
within
within
within
within
within
within
within
within
within
within
within
within
NC
NC
NC
C
C
C
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
1
1
1
1
1
1
0
0
0
0
0
1
1
1
0
0
1
1
1
1
2012
2013
2012
2012
2012
2013
2012
2012
2013
2012
2013
2012
2012
2013
2012
2013
2012
2013
2012
2013
Date End
355.2
384.1
336.7
238.9
235.1
200.6
413.1
481.5
517.9
796.3
823.5
517.8
658.4
650.4
536.4
416.2
486.3
418.4
699.4
617.8
Date Start
Study Design
4.08
43.88
41.15
1.17
51.12
5.04
1.51
38.12
224.07
65.87
251.48
1.46
52.98
73.53
58.25
59.96
52.88
117.19
24.75
16.50
Healthy (Yes = 1, No = 0)
Observation(n)
–1.3
–22.4
47.4
0.5
–34.5
–3.4
–1.5
36.4
–31
27.2
–44.4
–2
–8
34.4
–120.2
–80.6
–67.9
126.8
–81.6
–54.4
Children (NC = Not Only
Children in Sample, C =
Children Only in Sample)
Mean Calories After
Study Descriptor
Appetizers
Appetizers 2013–2014
Appetizers 2012–2013
Children’s menu
Children’s menu 2012–2013
Children’s menu 2013–2014
Dessert
Dessert 2012–2013
Dessert 2013–2014
Main burger 2012–2013
Main burger 2013–2014
Main course
Main entree 2012–2013
Main entree 2013–2014
Main pizza 2012–2013
Main pizza 2013–2014
Main salad 2012–2013
Main salad 2013–2014
Main sandwich 2012–2013
Main sandwich 2013–2014
Mean Calories Before
2015
2016
2016
2015
2016
2016
2015
2016
2016
2016
2016
2015
2016
2016
2016
2016
2016
2016
2016
2016
Raw SE
Study Name
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Raw Difference in Means
Appendix 2
STUDY 2: INCLUDED STUDIES AND SUMMARY STATISTICS
2013
2014
2013
2013
2013
2014
2013
2013
2014
2013
2014
2013
2013
2014
2013
2014
2013
2014
2013
2014
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bleich, Wolfson, Jarlenski
Bruemmer et al.
Bruemmer et al.
Bruemmer et al.
Bruemmer et al.
Bruemmer et al.
Deierlein, Peat, Claudio
Deierlein, Peat, Claudio
Deierlein, Peat, Claudio
Deierlein, Peat, Claudio
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
Namba et al.
2016
2016
2015
2016
2016
2012
2012
2012
2012
2012
2015
2015
2015
2015
2013
2013
2013
2013
2013
2013
2013
2013
2013
2013
2013
Main soup 2012–2013
Main soup 2013–2014
Toppings/ingredients
Toppings/ingredients 2012–2013
Toppings/ingredients 2013–2014
Burger 2009–2010
Pizza 2009–2010
Sandwich 2009–1010
Sit Down 2009–1010
Tex-Mex 2009–2010
Children fast food 2010–2014 main
Children fast food 2010–2014 side
Children sit down 2010–2014 main
Children sit down 2010–2014 side
Adult a la carte entree restaurant A
Adult a la carte entree restaurant B
Adult a la carte entree restaurant C
Adult a la carte entree restaurant D
Adult a la carte entree restaurant E
Adult side dish restaurant A
Adult side dish restaurant B
Adult side dish restaurant C
Adult side dish restaurant D
Adult side dish restaurant E
Children a la carte
23.2
8.7
–0.5
28
–20.9
–9
–2
–23
–74
–33
9.6
–23.6
22.4
–45.9
–34
–126
11
94
72
–73
–7
–21
22
–44
–6
22.80
8.55
0.58
22.82
52.39
4.57
2.96
5.84
18.91
8.36
17.12
20.61
25.44
21.45
39.97
39.34
55.19
50.82
50.78
63.96
26.98
34.55
56.77
66.65
7.58
226.9
250.1
118.6
105.6
133.6
748
624
628
1044
690
268.3
150.7
381
180.2
525
415
272
503
379
462
136
185
351
307
261
250.1
258.8
118.1
133.6
112.7
739
622
605
970
657
277.9
127.1
403.4
134.3
491
289
283
597
451
389
129
164
373
263
255
210
210
2160
1514
1514
217
212
347
704
291
134
73
211
140
74
90
28
119
58
37
33
25
54
25
24
within
within
within
within
within
within
within
within
within
within
between
between
between
between
between
between
between
between
between
between
between
between
between
between
between
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
C
C
C
C
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
C
1
1
1
1
1
0
0
1
1
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
1
2012
2013
2012
2012
2013
2009
2009
2009
2009
2009
2010
2010
2010
2010
2005
2005
2005
2005
2005
2005
2005
2005
2005
2005
2005
2013
2014
2013
2014
2014
2010
2010
2010
2010
2010
2014
2014
2014
2014
2011
2011
2011
2011
2011
2011
2011
2011
2011
2011
2011