ArticlePDF Available

Abstract and Figures

Food marketing is facing increasing challenges in using portion size (e.g., “supersizing”) as a marketing tool. Marketers use portion size to attract customers and encourage purchase, but social agencies are expressing concern that larger portion sizes encourage greater consumption. This in turn raises concerns about excessive consumption and obesity. This paper addresses two questions that are central to this debate: How much effect does portion size have on consumption, and are there limits to this effect? A meta-analytic review revealed that, for a doubling of portion size, consumption increases by 35% on average. However, the effect has limits. An extended analysis showed that the effect of portion size is curvilinear: as portions become increasingly larger, the effect diminishes. In addition, although the portion-size effect is widespread and robust across a range of individual and environmental factors, the analysis showed that the portion-size effect is weaker among children, women, and overweight individuals, for non-snack food items and in contexts in which more attention is being given to the food being eaten.
Content may be subject to copyright.
Journal of Marketing
Vol. 78 (May 2014), 140 –154
© 2014, American Marketing Association
ISSN: 0022-2429 (print), 1547-7185 (electronic)
Natalina Zlatevska, Chris Dubelaar, & Stephen S. Holden
Sizing Up the Effect of Portion Size
on Consumption: A Meta-Analytic
Food marketing is facing increasing challenges in using portion size (e.g., “supersizing”) as a marketing tool.
Marketers have used portion size to attract customers and encourage purchase, but social agencies are
expressing concern that larger portion sizes encourage greater consumption, which can cause excessive
consumption and obesity. This article addresses two questions that are central to this debate: (1) How much effect
does portion size have on consumption? and (2) Are there limits to this effect? A meta-analytic review reveals that,
for a doubling of portion size, consumption increases by 35% on average. However, the effect has limits. An
extended analysis shows that the effect of portion size is curvilinear: as portions become increasingly larger, the
effect diminishes. In addition, although the portion-size effect is widespread and robust across a range of individual
and environmental factors, the analysis shows that it is weaker among children, women, and overweight
individuals, as well as for nonsnack food items and in contexts in which more attention is given to the food being
Keywords: portion size, food marketing, consumption norm, unit bias, obesity, supersizing
Natalina Zlatevska is Assistant Professor (e-mail: nzlatevs@bond. edu.
au), Chris Dubelaar is Professor (e-mail:, and
Stephen S. Holden is Associate Professor (e-mail: sholden@bond. edu.
au), Faculty of Business, Bond University. The authors thank Lauren
Block, Pierre Chandon, Peter Danaher, Wagner Kamakura, Harmen
Oppewal, Gulasekaran (Sekar) Rajaguru, Mike Reid, and three anony-
mous JM reviewers for their helpful comments on previous versions of this
article. Alexander Chernev served as area editor for this ar ticle.
The maxim “bigger is better” seems to characterize
both customer preferences and marketing efforts in
many domains, particularly those of food and drink
(Dubois, Rucker, and Galinsky 2012). By offering dispro-
portionately large increases in portion sizes for “a few extra
cents,” marketers win customers and profits (Dobson and
Gerstner 2010). The widespread adoption of this strategy is
reflected in growing portion sizes. For example, Coca-Cola
bottles have grown from the original 6.5 oz. (192 ml) bottle
to modern single-serve bottles of 10 oz. (300 ml), 16 oz.
(500 ml), and 20 oz. (600 ml). Moreover, artistic depictions
of the biblical Last Supper show that the quantities por-
trayed in the main dish, the amount of bread, and the size of
the plate have increased over the past 1,000 years (Wansink
and Wansink 2010).
Whether marketers have been responding to customer
preferences for larger-sized portions or whether they have
been shaping those preferences is not clear. What is clear is
that consumers eat and drink more from larger portion sizes
(Chandon and Wansink 2011; Wansink 2004). This has
become a cause for concern, and marketers are increasingly
challenged to consider the effects of portion size on con-
sumption (Dobson and Gerstner 2010; Haws and Winterich
2013; National Alliance for Nutrition and Activity 2002).
Morgan Spurlock famously drew public attention to mar-
keters’ use of portion size in his 2004 documentary, Super
Size Me. With worldwide attention on the “obesity epi-
demic” (Moore 2007; World Health Organization 2003,
2013), marketing and nutrition literature streams and gov-
ernment agencies have all identified food portion sizes as a
potential contributing factor to rising obesity rates (Centers
for Disease Control and Prevention 2004; Chandon and
Wansink 2011; Rolls 2003; Steenhuis and Vermeer 2009;
Young and Nestle 2002). New York City’s effort to restrict
the sale of soft drinks of 16 oz. or more is evidence that
social marketers and public health authorities are respond-
ing to the threat (Saul 2012). Although this attempt was
resisted and ultimately overturned in court (Hughes 2013),
the signal remains the same: social marketers and public
health agencies regard portion size as a problem.
However, no one seems to have adequately quantified the
size of the portion-size issue. We expect that an increase in
portion size is linked with an increase in consumption, and
we aim to show this linkage using standard meta-analytic
techniques. However, we considered that it would be more
useful to express how much consumption has changed as a
function of increasing portion size. With this idea in mind,
we developed an elasticity measure expressing consump-
tion change as a function of portion-size change based on a
linear regression analysis. Finally, we anticipate that the
portion-size effect is limited or constrained under certain
circumstances. First, we anticipate that the portion-size
effect on consumption will diminish for very large portion
sizes so that, eventually, increasing portion size will cease
to have any effect on consumption. Second, we expect that
Sizing Up the Effect of Portion Size on Consumption / 141
the portion-size effect will be moderated or constrained
under at least some individual and environmental condi-
tions. Before we proceed to quantifying the portion-size
effect and exploring its limits, we first aim to clarify the
concept of portion size and, in particular, to distinguish it
from similar but distinct size-related manipulations.
A Disambiguation of Portion Size
In collecting studies that ostensibly related to manipulations
of portion size, we found a confusing array of manipula-
tions for portion size, serving size, package size, and so
forth. We begin by setting aside the term “serving size”
because it is a normative concept referring to a recommen-
dation of the amount of food to consume (e.g., Mohr, Licht-
enstein, and Janiszewski 2012). Typically, recommended
serving sizes appear within standardized food information
labels. Portion size is a descriptive concept referring to the
quantity of food contained in a portion. It is usually indi-
cated as the weight or volume of the contents of the pack-
age.1Thus, one portion size may contain more (or less) than
one recommended individual serving size.
Next, we distinguish portion size and package size. This
distinction is sometimes acknowledged by marketers
through a notice advising consumers that the contents in the
package may settle during shipment. Thus, the size of the
container in which food or drink is served (be it a package,
plate, or cup) is a factor that can be manipulated indepen-
dently of portion size (e.g., Wansink 1996). In some studies,
portion size and container size are confounded: a small por-
tion is served on or in a small container and a large portion
is served on or in a large container (e.g., Rolls, Roe, Krall et
al. 2004; Van Kleef, Shimizu, and Wansink 2011). Although
crossing portion size with container size is conceptually
possible, it can be problematic because the physical limita-
tions of putting a large portion into a small container can
lead to an unbalanced design (e.g., Marchiori, Corneille,
and Klein 2012). Container size is itself a conceptually mul-
tifaceted variable that can be broken down into container
diameter, typically described as “plate size” (e.g., Koh and
Pliner 2009; Rolls et al. 2007), and container volume (e.g.,
Stroeble, Ogden, and Hill 2009; Van Kleef, Shimizu, and
Wansink 2011; Wansink 1996). Another variable related to
the container is the “perceived size,” which is influenced by
the container shape; for example, a tall, thin glass appears
to hold a larger quantity of drink than a short, squat glass
(Wansink and Van Ittersum 2003, 2005). Other studies have
focused on the size of the utensils used to serve or consume
the food while the portion size (as we have defined it)
remains fixed (e.g., Mishra, Mishra, and Masters 2012).
Another size variable researchers have examined is a
difficult-to-define quality that we label “granularity.” This
variable refers to whether a portion has a fine granularity
comprising many small parts or a coarse granularity com-
prising a few large parts. One form of granularity is parti-
tioning, which is the manipulation of the number and size
of packages that make up the portion size. For example, Do
Vale, Pieters, and Zeelenberg (2008, Study 2) presented
each participant in their study with either two 200 g pack-
ages of chips in one condition or nine 45 g packages of
chips in another. The portion size was roughly equivalent,
but the granularity was coarse in the first instance and fine
in the second. Although the researchers referred to “large”
and “small” package formats, respectively, we note that this
labeling overlooks that package size is necessarily con-
founded with number: a few large packages versus multiple
small packages. Another form of granularity relates to the
size of the food morsels in the portion, or many small food
morsels (e.g., mini cookies) versus fewer, larger food
morsels (e.g., regular cookies) (see, e.g., Scott et al. 2008).
Morsels and partitions are distinct and can be crossed or
confounded. For example, in a study by Scott et al. (2008),
the “small food configuration” comprised multiple small
packages of many mini M&M’s and the “large” configura-
tion comprised a few large packages of fewer, regular-size
In summary, there are several related but conceptually
distinct manipulations of size. All can be manipulated inde-
pendently and can therefore be crossed or confounded with
portion size and one another. Although these other size
manipulations are worthy of research, this article focuses on
portion size and its effect on consumption.
To conduct a meta-analysis to answer the questions about
the size and extent of the portion-size effect, we initially
looked for relevant papers through a search of ABI/
INFORM, ProQuest Digital Dissertations, Business Source
Premier, Web of Science, and other databases using key-
words related to the size or amount of food offered. Some
specific search terms were “portion size,” “serving size,”
and “unit bias,” used as a synonym for and potential expla-
nation of the portion-size effect (Geier, Rozin, and Doros
2006). We also manually searched through the following
journals and conference proceedings: Journal of Marketing,
Journal of Marketing Research, Journal of Consumer
Research, Journal of Consumer Psychology, Journal of
Public Policy & Marketing, Annual Review of Nutrition,
American Journal of Clinical Nutrition, Body and Society,
British Journal of Sociology, Social Science and Medicine,
Appetite, International Journal of Obesity, Advances in
Consumer Research, American Marketing Association pro-
ceedings, and Obesity Society abstract supplements. When
we found an article, we examined the references to identify
further studies. In addition, we used Web of Science, Sco-
pus, and Google Scholar to search the citations of included
articles. The approach we used is consistent with several
authors’ recommendations (Hunter and Schmidt 1990;
Rosenthal 1979). In an effort to counteract the file-drawer
problem often associated with a meta-analysis, we placed a
call on ELMAR (Electronic List for Marketing Academics
and Researchers) for working papers. Finally, we sent e-
mails to researchers in the domain asking for published and
unpublished works. We received two articles from different
1Some researchers, particularly food and nutrition scientists,
may report portion sizes in kilojoules or calories. This can be con-
fusing because recommended serving size is often reported as
kilojoules or calories on nutrition information labels.
authors as a result of our calls for unpublished research, but
neither of them manipulated portion size as we define it.
The criteria for inclusion in the meta-analysis for quan-
tifying the portion-size effect required that (1) the indepen-
dent variable manipulated portion size of food or drink and
(2) the dependent variable included an interpretable mea-
sure of consumption. Studies varied widely in terms of how
consumption was measured. Most reported actual consump-
tion, but there was considerable variation in the measures
used, including grams, ounces, kilojoules, calories, and
even percentages of a basic meal size (e.g., Levitsky and
Youn 2004). Furthermore, some studies reported intended
consumption; for example, “The subject was asked how
much of the product she would use in this situation”
(Wansink 1996, p. 3). Others measured perceived consump-
tion; for example, “Participants were asked to estimate how
many crackers they believed they consumed” (Wansink,
Payne, and Shimizu 2011, p. 1098). Accordingly, we report
on the portion-size effect for actual, intended, and perceived
consumption separately.
Our search for research published through December
2013 returned 52 articles, which we broke down into 211
identifiable studies. However, we could not calculate effect
sizes for 17 of the studies collected because 13 did not indi-
cate the level of significance in the original article, and 4
studies were field based and did not provide the sample
size. We excluded an additional 6 studies that examined
nonfood consumption. Of the remaining 188 separate stud-
ies, 84 captured size manipulations other than portion size
and were thus excluded, leaving us with 104 studies that
captured the portion-size effect as we have defined it (for a
detailed list of these studies, see Table 1). Of these 104
studies, 23 seemed to confound portion size with container
size in their manipulation (marked in Table 1 with a super-
script “a” in the “Data Identifier” column). In terms of the
dependent measure, actual consumption was reported in 88
studies, intended consumption in 13, and perceived con-
sumption in 3.
The first effect-size metric we examined was the stan-
dardized difference in means expressed as Cohen’s d
(Cohen 1988). The mean difference reflects how much
more was consumed from the larger portion size than from
the smaller or “control” portion size. Thus, a positive value
for Cohen’s d reflects the expected portion-size effect, with
a larger mean difference reflecting a larger effect. We
adopted a random-effects perspective: we assumed the true
effect size to vary from one study to the next and that the
studies represented a random sample of effect sizes (Hunter
and Schmidt 1990).
We calculated Rosenthal’s fail-safe N (Rosenthal 1979)
for our study to be 1,554, this being “the number of [null
effect] studies that would need to be added to a meta-analysis
to reduce an overall statistically significant observed result
to non-significance” (Rosenberg 2005, p. 464). This num-
ber comfortably exceeds Rosenthal’s (1991) recommenda-
tion that, for a robust meta-analysis, the fail-safe N should
exceed 5k + 10, which is 530 (k = 104) in the current study.
We also produced a funnel plot showing portion-size effects
mapped against standard errors (see Figure 1). Most studies
appeared within the funnel, as is typically expected, but
there seems to be a skew such that studies with larger stan-
dard errors have larger effects. This could be interpreted as
a possible publication bias (strong effects from “small”
studies), but it is more likely that the variation in standard
errors reflects different strength manipulations of portion
size (Sterne et al. 2011). For example, two points (to the
lower right outside the triangle) represented different condi-
tions in an 11-day study (Rolls, Roe, and Meengs 2007): the
larger standard errors arise because of the much-larger
quantities of food being served. We examine the strength of
the portion-size manipulation at length subsequently.
We directly established the heterogeneity of effect sizes
through the I2index (Higgins and Thompson 2002; Huedo-
Medina et al. 2006). The I2index is calculated with the for-
mula 100 ¥(Q – df)/Q, where Cochran’s Q is as defined by
Hunter and Schmidt (1990). The I2index quantifies hetero-
geneity as low (25%), medium (50%), or high (75%) (Hig-
gins and Thompson 2002). We report the observed hetero-
geneity of each calculated effect size in the following
Results and Discussion
Quantifying the Size of the Portion-Size Effect
Increasing portion size reliably increased consumption (d =
.45, k = 88, I2= 65%). The top bar in Figure 2 shows the aver-
age size of the portion-size effect on actual consumption and
the 95% confidence limits around the estimate. Figure 2 also
shows that there was a significant but smaller effect of portion
size on intended consumption (d = .18, k = 13, I2= 41%)
and on perceived consumption (d = .38, k = 3, I2= 0%).
We note that a great deal of heterogeneity was observed
in the effect size for portion size on actual consumption (I2=
65%). This relatively high level of heterogeneity is likely
due to the treatment of all portion-size manipulations as
equal, which is simply not the case. The size of the portion-
size effect is likely to depend on the degree of change in
portion size, an observation we made previously, in line
with Sterne et al. (2011), as a potential explanation for the
observed asymmetry in the funnel plot. This then highlights
the limitations of the meta-analytic effect size, which treats
all portion-size manipulations as constant. As Chernev,
Bockeholt, and Goodman (2010) note, meta-analytic mean
effect sizes are not easily interpreted beyond simple com-
parisons of “control” versus “treatment.” Thus, the standard
meta-analysis makes no allowance for the size of the
change in portion size: a 50%, 100%, and 200% change in
portion size are all treated the same, whereas a measure of
effect scaled on the degree to which portion size has
changed will address the high heterogeneity. Importantly, it
will also provide a more practical and useful measure of the
portion-size effect.
To estimate how much actual consumption changes as a
function of the change in portion size, we developed scalar
measures for both portion size and consumption. We
recorded the proportional change in portion size as the
change in portion size relative to the smaller portion (see
Equation 1) and expressed the proportional change in con-
142 / Journal of Marketing, May 2014
Sizing Up the Effect of Portion Size on Consumption / 143
Summary of Studies Used in the Portion-Size Meta-Analysis
Portion Served Amount Consumed Moderators (Independent Variables)
Cohen’s Snack Food
Article ID Data Identifierc “Small” “Large” “Small” “Large” d Aged Gendere BMIf Foodg Focusg
Burger et al. (2011) 1 (Intended) 1 serve 2 serves — .04 1 0 0 1
Burger, Fisher, and 2 Blindfoldedb 410.g 820.g 266.13 g 322.58 g .21 1 0 0 0
Johnson (2011)
3 Not blindfoldedb 410.g 820.g 283.87 g 383.87 g .21 1 0 0 1
Diliberti et al. (2004) 4 b 248.g 377.g 234.4 8g 335.6 g .59 1 0 1
Fisher (2007) 5 2–3 years, other serveb 200.g 400.g 93.66 g 102.11 g .14 0 1 0 1
6 2–3 years, self serveb 200.g 400.g 93.66 g 89.44 g .14 0 1 0 1
7 5–6 years, other serveb 250.g 500.g 157.04 g 204.23 g .13 0 1 0 1
8 5–6 years, self serveb 250.g 500.g 157.04 g 169.72 g .13 0 1 0 1
9 8–9 years, other serveb 450.g 900.g 254.23 g 286.62 g .12 0 1 0 1
10 8–9 years, self serveb 450.g 900.g 254.23 g 267.61 g .12 0 1 0 1
Fisher, Arreola, et al. (2007) 11 Children, chickenb 152.07 g 304.14 g 110.33 g 147.52 g .45 0 0 0 1
12 Children, crackersb 40.06 g 80.12 g 20.35 g 24.89 g .11 0 0 1 1
13 Children, cerealb 40.g 80.g 27.g 40.75 g .45 0 0 0 1
14 Children, juice 240.23 mL 480.46 mL 172.34 g 172.34 g .11 0 0 0 1
15 Children, macaroni and cheeseb 300.g 600.g 149.67 g 158.28 g .11 0 0 0 1
16 Mothers, cerealb 80.g 160.g 50.75 g 54.5 g .11 1 0 1 0 1
17 Mothers, chickenb 200.g 400.g 161.27 g 212.14 g .26 1 0 1 0 1
18 Mothers, crackersb 60.g 120.g 45.67 g 53.46 g .11 1 0 1 1 1
19 Mothers, juice 336 mL 672 mL 255.32 g 357.45 g .45 1 0 1 0 1
20 Mothers, macaroni and cheeseb 400.g 800.g 240.4 g 280.79 g .26 1 0 1 0 1
21 Mothers, riceb 200.g 400.g 136.25 g 160.g .11 1 0 1 0 1
Fisher, Liu, et al. (2007) 22 High energy densityb 250.g 500.g 156.g 211.g .69 0 0 0 1
23 Reference energy densityb 250.g 500.g 160.g 209.g .61 0 0 0 1
Fisher, Rolls, and Birch (2003) 24 Children < 4 years of age 125.g 250.g 99.g 104.g .30 0 0 1
25 Children > 4 years of age 175.g 350.g 110.g 140.g .47 0 0 1
Flood, Roe, and Rolls (2006) 26 Womenb 360.g 540.g 300.g 331.g .49 1 0 0 1 1
27 Menb 360.g 540.g 320.g 403.g 1.13 1 1 0 1 1
Hermans et al. (2011) 28 — 388.1 9g 524.98 g 1.72 1 0 0 0 1
Jeffery et al. (2007) 29 767.kcal 1,528.kcal 687 kcal 1,019.kcal 1.14 1 0 1 0 1
Kral et al. (2009) 30 Applesauceb 122.g 244.g 90.g 129.1 g .54 0 0 1 1
31 Broccolib 75.g 150.g 24.g 25.g .05 0 0 0 1
32 Carrotsb 75.g 150.g 19.g 20.g .08 0 0 0 1
144 / Journal of Marketing, May 2014
Portion Served Amount Consumed
Moderators (Independent Variables)
Cohen’s Snack Food
Article ID Data Identifierc “Small” “Large” “Small” “Large” d Aged Gendere BMIf Foodg Focusg
Kral, Roe, and Rolls (2004) 33 High energy density, S–Mb 500.g 700.g 339.8 g 359.4 g .29 1 0 0 0 1
34 Low energy density, S–Mb 500.g 700.g 357.9 g 416.8 g .52 1 0 0 0 1
35 High energy density, M–Lb 700.g 900.g 359.g 392.9 g .17 1 0 0 0 1
36 Low energy density, M–Lb 700.g 900.g 416.8 g 424.2 g .06 1 0 0 0 1
Levitsky and Youn (2004) 37 Pasta, S–M 100% 125% 350.g 430.g .60 1 0 0 1
38 Bread sticks, S–M 100% 125% 11.g 14.g .60 1 0 0 1
39 Ice cream, S–M 100% 125% 59.g 85.g .60 1 0 1 1
40 Vegetable soup, S–M 100% 125% 130.g 160.g .60 1 0 0 1
41 Pasta, M–L 125% 150% 430.g 460.g .60 1 0 0 1
42 Bread sticks, M–L 125% 150% 14.g 16.g .60 1 0 0 1
43 Vegetable soup, M–L 125% 150% 160.g 190.g .60 1 0 0 1
44 Ice cream, M–L 125% 150% 85.g 97.g .60 1 0 1 1
Looney and Raynor (2011) 45 Study 1 150.g 300.g 84.2 kcal 99.kcal .07 0 1 1
Marchiori, Corneille, and 46 (Medium portion size)/ 200.g 600.g 30.4 g 59.8 g .62 1 0 1 0
Klein (2012) (small container size)
vs. (large portion size)/
(large container size)ab
Raynor and Wing (2007) 47 Study 1b 813.3 g 1629.4 g 521.01 g 932.86 g 1.05 1 0 1 1
Rolls, Engell, and Birch (2000) 48 3.6 years, S–Mb 150.g 263.g 44.8 g 54.6 g .18 0 0 0 0 1
49 3.6 years, M–Lb 263.g 376.g 54.6 g 39.6 g −.30 0 0 0 0 1
50 5.0 years, S–Mb 225.g 338.g 100.7 g 122.7 g .36 0 0 0 0 1
51 5.0 years, M–Lb 338.g 450.g 76.7 g 100.7 g .27 0 0 0 0 1
Rolls, Morris, and Roe (2002) 52 Plate, S–Mb 500.g 625.g 340 g 374.g .40 1 0 0 1
53 Serving dish, S–Mb 500.g 625.g 330 g 374.g .42 1 0 0 1
54 Plate, M–Lb 625.g 750.g 374 g 410.g .33 1 0 0 1
55 Serving dish, M–Lb 625.g 750.g 374 g 390.g .35 1 0 0 1
56 Plate, L–XLb 750.g 1,000.g 410 g 446.g .40 1 0 0 1
57 Serving dish, L–XLb 750.g 1,000.g 390 g 410.g .42 1 0 0 1
Rolls, Roe, Kral, et al. (2004) 58 Women, S–Mab 28.g 42.g 25.48 g 34.21 g .62 1 0 0 1 0
59 Men, S–M ab 28.g 42.g 26.19 g 39.13 g .73 1 1 0 1 0
60 Women, M–Lab 42.g 85.g 34.21 g 50.2 g .62 1 0 0 1 0
61 Men, M–Lab 42.g 85.g 39.13 g 61.05 g .73 1 1 0 1 0
62 Women, L–XLab 85.g 128.g 50.2 g 54.61 g .15 1 0 0 1 0
63 Men, L–XLab 85.g 128.g 61.05 g 81.84 g .73 1 1 0 1 0
64 Women, XL–XXLab 128.g 170.g 54.61 g 59.04 g .15 1 0 0 1 0
65 Men, XL–XXLab 128.g 170.g 81.84 g 83.64 g .33 1 1 0 1 0
Sizing Up the Effect of Portion Size on Consumption / 145
Portion Served Amount Consumed
Moderators (Independent Variables)
Cohen’s Snack Food
Article ID Data Identifierc “Small” “Large” “Small” “Large” d Aged Gendere BMIf Foodg Focusg
Rolls, Roe, Meengs, 66 Women, S–Mb 275.g 367.g 214.g 245.g .39 1 0 0 0 1
et al. (2004)
67 Men, S–Mb 275.g 367.g 265.g 334.g 1.60 1 1 0 0 1
68 Women, M–Lb 367.g 458.g 245.g 249.g .04 1 0 0 0 1
69 Men, M–Lb 367.g 458.g 334.g 383.g .72 1 1 0 0 1
70 Women, L–XLb 458.g 550.g 249.g 278.g .30 1 0 0 0 1
71 Men, L–XLb 458.g 550.g 383.g 415.g .37 1 1 0 0 1
Rolls, Roe, and 72 Women, S–M 100% 150% 4,400.kcal 5,000.kcal .22 1 0 0 0 1
Meengs (2006a)
73 Men, S–M 100% 150% 6,000.kcal 7,000.kcal 1.05 1 1 0 0 1
74 Women, M–L 150% 200% 5,000.kcal 5,400.kcal 1.05 1 0 0 0 1
75 Men, M–L 150% 200% 7,000.kcal 7,500.kcal 1.05 1 1 0 0 1
Rolls, Roe, and 76 100% energy density 3,060.g 4,080.g 2,017.g 2,279.g .79 1 0 0 0 1
Meengs (2006b)
77 75% energy density 3,060.g 4,080.g 1,981.g 2,251.g .73 1 0 0 0 1
Rolls, Roe, and 78 Men 2,135.g 3,154.g 1,918.g 2,215.g 1.58 1 1 0 0 1
Meengs (2007)
79 Women 1708.54 g 2500.g 1,439.g 1713.g 2.09 1 0 0 0 1
Scheibehenne, Todd, and 80 Eating in darkb 481.g 706.g 462.g 627.g .87 1 0 0 1
Wansink (2010)
81 Eating in dark (perceived) 481.g 706.g 496.g 576.g .48 1 0 0 1
82 Eating in lightb 451.g 636.g 432.g 525.g .63 1 0 0 1
83 Eating in light (perceived) 451.g 636.g 416.g 504.g .31 1 0 0 1
Spill et al. (2010) 84 Carrots, S–Mb 30.g 60.g 24.7 g 36.2 g .59 0 0 1
85 Carrots, M–Lb 60.g 90.g 36.2 g 38.1 g .27 0 0 1
Van Kleef, Shimuzu, and 86 Apple pieb 40.g 200.g 29.7 g 60.2 g .67 1 0 1 1
Wansink (2013)
87 Chocolateb 10.g 100.g 6.5 g 8.3 g .35 1 0 1 1
88 Potato chipsb 10.g 80.g 6.1 g 10.4 g .67 1 0 1 1
Wansink (1994) 89 Diet Pepsi (intended)a — — 231.mL 334.mL .52 1 1
90 Water (intended)a — 352.mL 317.mL .05 1 0
146 / Journal of Marketing, May 2014
Portion Served Amount Consumed
Moderators (Independent Variables)
Cohen’s Snack Food
Article ID Data Identifierc “Small” “Large” “Small” “Large” d Aged Gendere BMIf Foodg Focusg
Wansink (1996) 91 Study 2, bottled water 1,000.mL 2,000.mL 355.mL 410.mL .65 1 0 0 1
92 Study 2, tap water (intended)a 1,000.mL 2,000.mL 376.mL 387.mL .17 1 0 0 1
93 Study 3, regular price oil 472.mL 944.mL 105.mL 137.mL .28 1 0 1
94 Study 3, sale price oil 472.mL 944.mL 139.mL 141.mL .02 1 0 1
95 Study 4, Creamette spaghetti, 675.g 1,350.g 234.g 331.g .38 1 0 0 1
S–M (intended)a
96 Study 4, Creamette spaghetti, 1,350.g 2,025.g 331.g 321.g −.19 1 0 0 1
M–L (intended)a
97 Study 4, Crisco oil, 472.mL 944.mL 99.mL 134.mL .38 1 0 0 1
S–M (intended)a
98 Study 4, Crisco oil, 944.mL 1,416.mL 134.mL 124.mL −.19 1 0 0 1
M–L (intended)a
99 Study 4, M&M’s, 114.g 228.g 63.g 103.g .38 1 0 1 1
S–M (intended)a
100 Study 4, M&M’s, 228.g 342.g 103.g 122.g .19 1 0 1 1
M–L (intended)a
Wansink and Kim (2005) 101 Fresh popcornab 120.g 240.g 58.9 g 85.6 g 1.45 1 1 0
102 Stale popcornab 120.g 240.g 38.g 50.8 g .85 1 1 0
Wansink, Painter, and 103 b 510.3 g — 240.97 g 416.73 g .73 1 0 0 1
North (2005)
104 (Perceived) 510.3 g 190.71 g 198.18 g .28 1 0 0 1
aThese studies confounded container size with portion size.
bStudies in which portion served and amount consumed were codable in grams and were used in a test for curvilinearity.
cIndependent variable (+ dependent variable if not actual consumption).
d0 = study with participants aged 15 years and younger, and 1 = study with participants older than 15 years.
e0 = female participants only, and 1 = male participants only.
f0 = participants’ BMIs were £25, and 1 = participants’ BMIs were >25.
g0 = no, and 1 = yes.
Notes: Article indicates the source; ID indicates the identification number assigned to each observation in the meta-analysis; Data Identifier provides a combination of key independent and depen-
dent variables enabling the reader to identify the exact data used for each observation (line) in this table; Portion Served shows the “small” and “large” portion sizes (where available);
Amount Consumed is shown for the “small” and “large” portions, respectively; Cohen’s d is the measure of the effect size; and Moderators shows the value for each study on five modera-
tors (with a missing value used to show if the study was not codable).
Sizing Up the Effect of Portion Size on Consumption / 147
sumption as the change in consumption relative to the
amount consumed from the smaller portion (see Equation 2).
(1) DS/SS, and
(2) DC/CS,
DS = change in portion size (larger portion size – smaller
portion size),
SS= smaller portion size,
DC = change in consumption (amount eaten from larger
portion – amount eaten from smaller portion), and
CS= consumption from smaller portion size
The effect of portion size was then estimated by the
coefficient resulting from regressing the change in con-
sumption (Equation 2) on the change in portion size (Equa-
tion 1) for k = 86 studies.2Because many of the articles
used in the analysis provided multiple studies (see Table 1),
and some of those articles used between-subjects designs
and some used within-subject designs, we needed to imple-
ment a multilevel model to account for the fact that the
observations were not all independent. Furthermore,
because there is an infinite number of combinations of por-
tion size possible for both large and small portions, we
treated portion size as a random factor in the model, which
enabled us to extrapolate to the population at large from our
sample. This resulted in the following multilevel model:
DC/CSij = B0j + B1j ¥DS/SSij + rij,
B0j = g00 + g01 ¥Designj+ g02 ¥Article(Design)j+ u0j, and
B1j = g10 + g11 ¥Designj+ g12 ¥Article(Design)j+ u1j.
Because all articles used either a within-subject or a
between-subjects design, each article from which we
obtained multiple studies is nested within design. The term
Article is a series of 26 dummy variables to reflect that
there are 27 articles from which we draw our data; thus, gi2
is a coefficient reflecting the effect of each paper on the
intercept and slope of the line. We first estimated the base-
line model (intercept only) and then estimated the model
with both intercept and slopes. The fit (given by –2 ¥log
likelihood [–2LL]) for the intercept-only model is –23.02 and
uses 1 degree of freedom. The intercept and slopes model
we proposed uses 30 degrees of freedom and has a –2LL of
–121.03, for a net change of 98.02 (distributed as c2) for 29
degrees of freedom (p< .001), indicating that our proposed
model is a significant improvement on the baseline model.
In this model, the intercept is nonsignificant, as we
expected; the gammas for study design and 24 of the 26
dummy codes for articles are also nonsignificant. We
obtained significant effects for the Fisher (2007) and Raynor
and Wing (2007) articles only (–.21 and .39, respectively).
Across all food types and portion sizes (k = 86), our
multilevel model shows that when portion size is doubled
(i.e., when DS/SS= 1), the amount that respondents con-
sume increases by 35% on average (B = .35, t = 3.33, p<
.01, R2= .89). We note that it may be useful to interpret this
coefficient as the portion-size elasticity of consumption.
Although these results show that the portion-size effect is
substantial, it is smaller than we would expect if consump-
tion were guided by the portion size, as suggested by the
notion of a consumption norm whereby people eat a fixed
proportion of what they are served. If people were to follow
such a rule or heuristic, we would expect a coefficient of
100%. The most common version of this explanation is that
people tend to eat everything on their plate because of a
norm or “unit bias” perhaps established by parental instruc-
tions received during childhood (e.g., Birch et al. 1987; Fay
Notes: The funnel plot displays each observation (k = 104) as a
function of the effect size (Cohen’s d) and the standard error.
The angular lines mark the 95% confidence limits, with the
expectation that most studies will fall within these lines.
Asymmetry in the distribution of observations may suggest
the operation of a bias.
0 1 2 3–3 –2 –1
Cohen’s d
Standard Error
Funnel Plot of Cohen’s d by Standard Error
2We could not include 2 of the 88 studies that examined actual
consumption because they did not include portion sizes for both
large and small portions (i.e., Hermans et al. 2011; Wansink,
Painter, and North 2005). We consider neither “intended” nor
“perceived” consumption in further analyses.
Forest Plot of the Portion-Size Effect
Notes: The forest plot displays average effect sizes (shown as a
number in the box below the bar on the right hand side) and
their respective 95% CIs (indicated by the extremities of the
bar) for three dependent variables: actual, intended, and
perceived consumption. The numbers within the parenthe-
ses beneath each dependent variable label show the num-
ber of studies on which the effect-size estimate is based (k)
and the heterogeneity of the estimate (I2).
.2 .4 .6 .8–.8 –.6 –.4 –.2 0
(88, 65%)
(13, 41%)
(3, 0%)
et al. 2011; Geier, Rozin, and Doros 2006; Wansink 2004;
Wansink, Painter, and North 2005). However, a more gen-
eral version is that people eat a fixed percentage of the por-
tion served. Some are “plate cleaners” (Burger, Fisher, and
Johnson 2011), some always leave 10% because it is con-
sidered polite to do so, and some eat only 50% of the por-
tion because they are “on a diet.” If consumers followed
any of these rules, the expectation would be that “the per-
centage change in [portion] size would change consumption
by the same percentage” (Geier, Rozin, and Doros 2006, p.
522). However, our analyses show that a 100% change in
portion size leads to a 35% change in consumption, sug-
gesting that if such consumption norms are operating, they
are not driving the portion-size effect for everyone, or they
represent an otherwise incomplete explanation.
The Limits on the Portion-Size Effect
The coefficient of 35% may reflect the average size of the
portion-size effect across consumers: some people are fol-
lowing a consumption norm, but others are not. However,
another possibility is that the coefficient represents an aver-
age across different degrees of portion-size change. That is,
the consumption norm may operate for most people but
only up to some point, at which it begins to break down.
Ultimately, the portion-size effect must be reduced and
even eliminated; otherwise, the outcome would be the same
as for the character Mr. Creosote in the Monty Python film
The Meaning of Life, who explodes as a result of eating too
much. Accordingly, we would expect the doubling of a very
small portion to have greater effect than the doubling of a
very large portion.
Analytically, we expected the portion-size effect to be
curvilinear. To test this idea, we first examined eight articles
included in the meta-analysis that reported three or more
levels of portion size and consumption (Kral, Roe, and
Rolls 2004; Levitsky and Youn 2004; Rolls, Engell, and
Birch 2000; Rolls, Morris, and Roe 2002; Rolls, Roe, Krall
et al. 2004; Rolls, Roe, and Meengs 2006a; Rolls, Roe,
Meengs et al. 2004; Spill et al. 2010) (for details, see Table
1). Each of these articles offered multiple portion-size
changes from differing bases. For example, Rolls, Roe,
Krall et al. (2004), using five portion sizes of chips (28 g
[S], 42 g [M], 85 g [L], 128 g [XL], 170 g [XXL]), yielded
four separate portion-size effects for men and women (i.e.,
eight studies): small to medium (S–M), medium to large
(M–L), large to extra-large (L–XL), and extra-large to
extra-extra-large (XL–XXL). Combining all these articles
(k = 42) and coding the smallest portion size as “small,” the
next as “medium,” and so forth, we found that the effect
size became smaller for successively larger portion-size
comparisons (as shown subsequently). This result is encour-
aging, considering that the size of the “small” portion var-
ied across the eight articles:
•Small to medium: dS–M = .56, k = 17, I2= 48%
•Medium to large: dM–L = .42, k = 17, I2 = 57%
•Large to extra-large: dL–XL = .37, k = 6, I2= 0%
•Extra-large to extra-extra-large: dXL–XXL = .23, k = 2, I2= 0%
In a separate effort to search for curvilinearity in the
portion-size effect in our data set, we examined all studies
in which both portion size and consumption could be coded
in grams (k = 71).3We also eliminated studies (k = 6) in
which consumption reflected a sum across multiple eating
occasions (Rolls, Roe, and Meengs 2006a, 2007) because
they are not comparable with single-eating-occasion fig-
ures. This left us with 65 studies for this analysis (identified
by a superscript “b” in the “Data Identifier” column of
Table 1). We broke down each study into separate observa-
tions: one observation of grams served and grams con-
sumed from the smaller portion, and another observation
from the larger portion. The resulting data set comprised
109 observations4showing amount consumed as a function
of amount served (as plotted in Figure 3) for a range of por-
tion sizes (10 g to 1,629.40 g) and a range of foods and
drinks. Fitting a curve to all 109 observations, we found
that the portion-size effect was curvilinear (see Figure 3;
consumption = .81 ¥portion size 3.65 ¥10−4 ¥[portion
148 / Journal of Marketing, May 2014
3For this analysis, studies in which portion sizes were reported
in volumes were converted to grams on the basis of a measure of
density (grams/unit volume), if available, and studies in which
portion size was reported in kilojoules (or calories) were con-
verted on the basis of measures of energy density (kilojoules/
gram), if available. We could not convert studies in which portions
consisted of food representing mixed energy densities (e.g., a meal
comprising different foods).
4The 65 studies yielded only 109 separate observations because
some articles provided studies at multiple overlapping levels (e.g.,
S–M, M–L), as Table 1 shows. In such cases, we entered only one
observation for each portion size. In addition, the grams served for
the “larger” portion in Wansink, Painter, and North’s (2005) bot-
tomless soup bowl study were not fixed, which led us to exclude
this observation; however, we retained the “smaller” portion
Notes: The figure shows a plot of 109 observations of the average
amount consumed by portion size. Modeling consumption as
a quadratic function of portion size returned the following
equation: Consumption = .81 ¥portion size 3.65 ¥10−4 ¥
(portion size)2, R2= .74.
0 1000200 400 600 800
Portion Size (g)
Amount Consumed (g)
Amount Consumed as a Function of Amount
Sizing Up the Effect of Portion Size on Consumption / 149
size]2; all coefficients significant at p< .001, R2= .74).5
Together with our earlier subanalysis, the results confirm
that the portion-size effect is likely limited for extremely
large portions.
We reason that the portion-size effect operates because
consumers rely on the external cue of portion size as a
guide to consumption rather than internal cues such as sati-
ation, at least for smaller portions. That is, they are sensitive
to portion size but insensitive to internal cues. We note that
this reasoning may seem difficult to reconcile with research
showing that consumer judgments of quantity are insensi-
tive to, and tend to underestimate, increases in portion size
(Chandon 2009; Chandon and Ordabayeva 2009; Van Itter-
sum and Wansink 2012). The resolution is, we believe, that
consumers may be using a heuristic—a consumption norm
(i.e., eat a certain percentage of the portion)—while simul-
taneously failing to perceive how much larger a portion size
may be. Therefore, the insensitivity of quantity judgment
and the operation of a consumption norm jointly support the
portion-size effect. Importantly, our data show that there is
a limit to this effect as portion sizes increase, which we
speculate may reflect internal cues such as satiation becom-
ing more salient. Satiation is a complex construct (for a
review, see De Graaf et al. 2004), but our purpose here is to
advance a relatively straightforward explanation as to why
a portion-size effect based on insensitivity to quantity and
the use of consumption norms is unsustainable and ulti-
mately diminishes or disappears.
Although our subanalyses show nonlinearity in the
portion-size effect, we acknowledge that such results are
based on a substantially reduced data set; however, we are
encouraged that evidence of nonlinearity emerges despite
the wide variety of food types represented. Although we
have shown that there is a limit to the portion-size effect at
larger portion sizes, we believe that much more work and
many more data points are required to explore more com-
pletely the curvilinearity of the portion-size effect.
In addition to the limits imposed on the portion-size
effect by physical limits and satiation, social agencies con-
cerned about containing the portion-size effect are likely to
be very interested in understanding the individual and envi-
ronmental factors that attenuate or eliminate the effect.
Despite the extensive number of studies in the literature
examining the portion-size effect, there are few consistent
theories describing any clear boundary conditions. Rather,
most studies tend to show that the effect is robust across a
wide range of contexts. For example, the portion-size effect
operates across a range of individual characteristics: both
men and women show the effect (Rolls, Roe, and Meengs
2007), it is unaffected by body mass index (BMI) (Fisher,
Arreola, et al. 2007), and it is evident even among those
with interest and/or expertise in nutrition (Chandon and
Wansink 2007; Tangari et al. 2010; Wansink, Van Ittersum,
and Painter 2006). Researchers have also shown that the
effect operates across a surprisingly wide range of environ-
mental conditions. For example, studies have shown that it
operates even for less palatable foods (Wansink and Kim
2005) and in situations in which the subjects are literally
blind to the portion size. Wansink, Painter, and North
(2005) show that people ate significantly more from a “bot-
tomless bowl” of soup that was continually and surrepti-
tiously refilled (thereby disguising portion size) than those
eating from the same size soup bowl that was not being
refilled. Researchers have also documented the portion-size
effect in a “dark restaurant” where customers were served
by blind people and ate their meals in complete darkness
(Scheibehenne, Todd, and Wansink 2010). The only factor
that seems to have consistently reduced the portion-size
effect is age: several studies have suggested that portion
size does not strongly affect consumption by younger chil-
dren (Birch et al. 1987; Fisher 2007; Fisher, Rolls, and
Birch 2003; McConahy et al. 2002; Rolls, Engell, and Birch
2000). However, young adults show a portion-size effect
similar to that of adults (Levitsky and Youn 2004; Looney
and Raynor 2011).
Notwithstanding the bulk of research showing few mod-
erators of the effect, we considered it worthwhile to exam-
ine a limited set of individual factors (age, gender, and
BMI) and environmental factors (snack food and food
focus) that might be identified as having some potential
influence on the portion-size effect. We examined age
because previous research has suggested that it is a modera-
tor of the effect, so we coded studies as comprising respon-
dents who were £15 or >15 years of age. Although there is
no evidence that gender moderates the portion-size effect,
many studies have reported the results for men and women
separately, thereby enabling the testing of gender as a mod-
erator. At least one study of BMI indicates that it has no
influence on the portion-size effect (Fisher, Arreola, et al.
2007); nevertheless, the possibility that higher-BMI people
are more susceptible to the portion-size effect warrants
examination in view of research suggesting that they are
less able to monitor internal cues of satiation and rely more
on external consumption cues (Wansink, Payne, and Chan-
don 2007). Therefore, we coded and recorded as separate
studies, when possible, the results for those samples with an
average BMI > 25 versus those with a BMI £25.
In terms of environmental factors, we examined snack
food and food focus. Although there is no evidence that
snack foods are more subject to a portion-size effect than
nonsnack foods, there has been a tendency in the criticism
of supersizing to conflate the portion-size effect with
unhealthy snack foods. In effect, if the portion-size effect
works equally for unhealthy foods and healthy ones, this
presents an opportunity for social marketers to supersize
healthy foods. Accordingly, we coded studies as using
snack food (e.g., M&M’s) versus nonsnack foods (e.g.,
pasta, water) when possible. Finally, Wansink (2010a, b)
5We estimated this as a simple regression of consumption on
portion size on the k = 109 data points. Because of the noninde-
pendence of our data, we also ran a model with portion size as a
random coefficient, which returned a similar result and fit (R2=
.65). However, this function showed a decline for portion sizes
beyond 800 g, which seems highly unlikely. In addition, we ran a
log-log model, which also provided a good fit (R2= .77), but a
plot of this function showed that it was virtually linear for the
range of portion sizes across which it was estimated. Altogether,
these results confirm our expectation that the portion-size effect is
suggests that mindlessness is a major factor contributing to
overeating and the portion-size effect. If true, greater atten-
tion to the food during the consumption episode might limit
the portion-size effect to some extent. Therefore, in an
effort to capture some degree of the mindless–mindful
dimension, we coded studies as having a food focus (partici-
pants knew their food consumption was being monitored)
or no food focus (participants were not aware that their food
consumption was being monitored). In the “no food focus”
condition, the researcher’s interest in food consumption was
disguised by presenting the food incidentally and, in some
instances, through a cover story. We reasoned that when there
was no food focus, consumption would be more mindless.
To assess moderation effects, we returned to our elastic-
ity model and again used a random coefficients model to
examine for an interaction of each moderator (e.g., age)
with portion-size change (i.e., DS/SS) (see Table 2). Note
that we could not code all studies on all moderators (see
missing values in Table 1), and only a limited crossing of
moderators was available, meaning that moderators may be
confounded with one another and with other unspecified
variables. For example, all the studies identified as high
BMI (>25) comprised adult female respondents only. Thus,
we had no observations for high-BMI men or children. To
maximize the number of available observations for each
moderator, we used the simpler linear model showing
change in consumption as a function of change in portion
size; however, we note that the results should be treated
with caution. We offer our analysis to encourage and assist
further research endeavors in this important and underex-
amined area.
The results for the moderator analysis confirm that the
portion size effect is robust and is observed across all condi-
tions examined (see Table 2). Nonetheless, the results sug-
gest that the size of the portion-size effect varies for differ-
ent individual characteristics. Adult consumption increased
by 39% for a doubling of portion size but increased by only
20% for children. We found that men responded to a dou-
bling of portion size with a 52% increase in consumption
and women responded with a 27% increase in consumption,
a result we did not expect that calls for more investigation.
Overweight people (BMI > 25) responded less (18%) to a
doubling in portion size than did those with a BMI of 25 or
less (34%). This result was also unexpected because it chal-
lenges research suggesting that overweight people are less
sensitive to satiation and more sensitive to external cues
(Wansink, Payne, and Chandon 2007). In terms of environ-
mental factors, people responded more to a doubling of por-
tion size of snack foods (37% increase in consumption)
than to nonsnack foods (27%). When respondents were told
that the study was about food, they responded less to a dou-
bling of portion size (26%) than when there was no food
focus (45%). This finding is consistent with the notion that
mindfulness might help overcome overconsumption and the
portion-size effect (Wansink 2010a, b).
General Discussion
Our research makes four significant contributions. The first
is to clarify the conceptualization of portion size and related
size manipulations and their effects on consumption. We
have defined an array of distinct size-related constructs:
portion size (quantity of food in the portion), container size
(diameter and volume), container shape (perceived size),
utensil size, and granularity (size of partitions and food
morsels). Our research highlights the need for researchers
to define size manipulations more clearly. For example, the
tendency for marketing researchers to examine the effects
of “small” and “large” granules (e.g., partitions, food
morsels) on consumption consistently overlooks that granu-
larity manipulates both number and size simultaneously.
Any explanation attributed to granule size could equally be
attributed to granule number because the two are necessar-
ily confounded.
Our second key contribution is to quantify the portion-
size effect. The meta-analysis revealed that the portion-size
effect was a medium-sized effect (d = .45). This fits the
standard method of establishing effect sizes, but it does not
offer much useful insight for understanding how much por-
tion size affects consumption. Through a more in-depth
analysis, we establish that a doubling of portion size leads
150 / Journal of Marketing, May 2014
Results for Moderators of the Portion-Size Effect (Random Coefficients Model)
Model Fit
Moderator Model Estimate t-Value p-Value R2 –2LL
Age DS/SS .20 4.73 <.001 .89 –72.93
Adults ¥DS/SS .19 3.56 <.01
Gender DS/SS .27 8.19 <.001 .64 –65.36
Males ¥DS/SS .25 2.60 .01
BMI DS/SS .18 3.30 <.01 .88 –67.38
BMI £25 ¥DS/SS .16 2.51 .02
Snack food DS/SS .37 8.01 <.001 .79 –64.59
Not snacks ¥DS/SS –.11 –1.89 .07
Food focus DS/SS .26 9.20 <.001 .85 –67.56
No food focus ¥DS/SS .19 2.76 .01
Notes: Moderator indicates the variable being examined, Model indicates both a term for the portion-size effect (DS/SS) and a term for the
interaction of the moderator with the portion-size effect, Estimate is the regression coefficient, t-value and p-value provide the tests of
significance for each coefficient, and Model Fit provides measures of the fit of the model to the data.
Sizing Up the Effect of Portion Size on Consumption / 151
to a 35% increase in consumption on average across a range
of food types and contexts.
Our third contribution is to show that the portion-size
effect is curvilinear. In two separate subanalyses of our
data, we show that the increase in consumption with
increasing portion size dropped off as the portion sizes
grew larger. For smaller portion sizes, we believe that con-
sumption norms may be largely driving the portion-size
effect; however, the portion-size effect is eliminated for
larger portion sizes, perhaps because of an increasing
salience of and reliance on internal cues.
Our fourth contribution is to explore conditions under
which the portion-size effect is reduced. Our analyses sug-
gest that the portion-size effect was attenuated for children,
women, people with high BMIs, nonsnack foods, and eating
with a food focus. These findings suggest that the modera-
tors of the portion-size effect provide fruitful avenues for
further research.
Overall, our results imply that increasing portion size
leads to real and important increases in consumption. With-
out any corresponding increase in energy output, we would
expect the increase in consumption to contribute to signifi-
cant weight gain. Indeed, studies confirm that offering large
portion sizes delivered over multiple sessions and meals
does result in weight gain (Jeffery et al. 2007). Conversely,
and as the portion-size effect predicts, longer-term studies
have found that reduced portion sizes successfully reduced
consumption (Freedman and Brochado 2010).
The implication for governments and social agencies
concerned about obesity is that portion size may be an
important contributing factor and that efforts to limit por-
tion size may prove helpful (Goldman and Patton 2012).
However, our finding that overweight people respond less
to the portion-size effect than others raises the concern that
downsizing may do little to help those who are already
Another of our results suggests that current parental
interventions in childhood may be unhelpful. The finding
that children’s consumption is apparently less affected by
portion size than that of adults suggests that learning, accul-
turation, and adaptation play a role (e.g., Birch et al. 1987).
Simplistic rules such as “eat everything on your plate” cre-
ate norms that may contribute to the problem (Wansink,
Payne, and Werle 2008). Parents demanding that children
eat their greens before a dessert may also be “training” chil-
dren to override the internal cues that may naturally serve to
limit the operation of the portion-size effect.
Future Directions for Research
Our research offers multiple rich pathways for exploring
portion size and related size manipulations. More work is
required to tease apart the operation of the process through
which the portion-size effect occurs. On the one hand, we
fear that the quantification of the portion-size effect in our
research may be underestimated because many researchers,
keen to avoid “floor” effects, tend to choose generous small
portion sizes in their studies, so the observed portion-size
effects may be relatively constrained. On the other hand, the
fears raised about the portion-size effects may be overstated
because our research shows there are limits to the effect. We
have speculated that internal cues such as satiation will ulti-
mately limit the effect, and we have empirically shown that
there is a decline in the effect for larger portions. We
believe that scholars need to pay more attention to explor-
ing and delineating the nonlinear nature of the portion-size
We have also explored various moderators of the effect,
but we acknowledge that our work is tentative. However,
what we offer should encourage more systematic investiga-
tion of the boundary conditions—in particular, when do
people switch off their consumption norm, and when do
they stop eating? We believe that there is much to be
learned by investigating people’s thoughts and beliefs
around cessation of eating. Scholars might use protocol
research to understand whether people are monitoring their
hunger, their satiation, or both, and how and whether these
factors interact with external factors such as portion size.
We have speculated that the portion-size effect may be
mediated by a consumption norm—a rule about what pro-
portion of a portion one “ought” to eat. If true, this norm,
like many heuristics, may operate beneath awareness and
may be difficult to eliminate.
Finally, we note that our own research was hindered by
the widely varying approaches adopted by researchers in
examining and reporting portion-size effects. In particular,
studies used many different metrics for reporting portion
size and consumption, making comparisons extremely diffi-
cult or impossible. We addressed this limitation in part by
offering a clear conceptualization of the various types of size
manipulations. We hope that future researchers will consider
reporting manipulations and measures in weight to allow
for easier comparison of the absolute size of the portion-
size effect in specific studies and future meta-analyses.
Marketing Implications
The clear implication of our research is that supersizing
comes with a catch. Larger portions may facilitate sales, but
they also increase consumption. A doubling of portion size
leads to a 35% increase in consumption on average—poten-
tially more when doubling smaller portions and potentially
less when doubling larger portions. Encouragingly for the
social researchers concerned about the contribution of por-
tion size to obesity, our research shows that the portion-size
effect has limits, although the tolerance of public agencies
will surely be tested even before consumers “burn out”
(Ma, Ailawadi, and Grewal 2013; Wansink 2012). To stave
off the inevitable attack from governments and social agen-
cies concerned about obesity, the obvious but unpalatable
truth is that portion sizes should be restricted or reduced
(e.g., Wansink and Van Ittersum 2007). Our results suggest
that reducing portion sizes will reduce consumption and,
given the curvilinear relationship we found, with potentially
greater effect in the domain of smaller portions. Although
marketers may be reluctant to give up the apparent prof-
itability of larger and established sizes (Jain 2012), some
encouraging research has suggested that consumers may be
willing to choose a smaller portion option—even if not dis-
counted or if charged a premium on the price per unit
(Schwartz et al. 2012; Wansink 2012). That is, at least some
customers may pay a premium for a smaller portion. More-
over, public concern about obesity may make this segment
increasingly important. Other possibilities include working
with rather than against the portion-size effect. For exam-
ple, the portion-size effect is robust and works for snack
foods and healthier foods (albeit to a lesser extent). There-
fore, a dual approach for addressing social concerns about
overeating and obesity would be to encourage supersizing
of healthy foods while limiting such practices among
unhealthy foods.
Finally, the results from the food focus moderator are
promising, suggesting that “mindfulness” may help miti-
gate the portion-size effect, as Wansink and Sobal (2007;
see also Wansink 2010a, b) suggest. Although our results do
not show the portion-size effect completely eliminated for
food focus, nor indeed for any of the moderators, they do
suggest that encouraging people to be more conscious while
consuming food could be useful. Given the finding that
children are less affected by the portion-size effect, we
speculate that mindfulness essentially undoes people’s
training to “eat everything” they are served in a single por-
tion and instead inclines them to tune in to internal cues.
It is clear that consumers do use portion size as a guide
to consumption, leaving marketers with an obligation to be
mindful of the size of the portions that they market. Our
research shows that the effect is robust and observed across
many conditions, but there is at least some promise that
several factors may help reduce the effect. Overall, we offer
marketers more understanding of the portion-size effect so
that they may be better equipped to address how portion
size might be used in marketing.
152 / Journal of Marketing, May 2014
sumption Self-Regulation,” Journal of Consumer Research, 35
(3), 380–90.
Dobson, Paul W. and Eitan Gerstner (2010), “For a Few Cents
More: Why Supersize Unhealthy Food?” Marketing Science,
29 (4), 770–78.
Dubois, David, Derek D. Rucker, and Adam D. Galinsky (2012),
“Super Size Me: Product Size as a Signal of Status,” Journal of
Consumer Research, 38 (6), 1047–62.
Fay, Stephanie H., Danielle Ferriday, Elanor C. Hinton, Nicholas
G. Shakeshaft, Peter J. Rogers, Jeffrey M. Brunstrom, et al.
(2011), “What Determines Real-World Meal Size? Evidence
for Pre-Meal Planning,” Appetite, 56 (2), 284–89.
Fisher, Jennifer O. (2007), “Effects of Age on Children’s Intake of
Large and Self-Selected Food Portions,” Obesity, 15 (2), 403–
—, Angeles Arreola, Leann L. Birch, and Barbara J. Rolls
(2007), “Portion Size Effects on Daily Energy Intake in Low-
Income Hispanic and African American Children and Their
Mothers,” American Journal of Clinical Nutrition, 86 (6),
—, Yan Liu, Leann L. Birch, and Barbara J. Rolls (2007),
“Effects of Portion Size and Energy Density on Young Chil-
dren’s Intake at a Meal,” American Journal of Clinical Nutri-
tion, 86 (1), 174–79.
—, Barbara J. Rolls, and Leann L. Birch (2003), “Children’s
Bite Size and Intake of an Entree Are Greater with Large Por-
tions Than with Age-Appropriate or Self-Selected Portions,”
American Journal of Clinical Nutrition, 77 (5), 1164–70.
Flood, Julie E., Liane S. Roe, and Barbara J. Rolls (2006), “The
Effect of Increased Beverage Portion Size on Energy Intake at
a Meal,” Journal of the American Dietetic Association, 106
(12), 1984–90.
Freedman, Marjorie R. and Carolina Brochado (2010), “Reducing
Portion Size Reduces Food Intake and Plate Waste,” Obesity,
18 (9), 1864–66.
Geier, Andrew B., Paul Rozin, and Gheorghe Doros (2006), “Unit
Bias: A New Heuristic That Helps Explain the Effect of Portion
Size on Food Intake,” Psychological Science, 17 (6), 521–25.
Goldman, Henry and Leslie Patton (2012), “NYC Health Panel
Backs Bloomberg Ban on Super-Size Sodas,”,
(September 13), (accessed February 21, 2014), [available at
Haws, Kelly L. and Karen P. Winterich (2013), “When Value
Trumps Health in a Supersized World,” Journal of Marketing,
77 (May), 48–64.
Birch, Leann L., Linda McPhee, B.C. Shoba, Lois Steinberg, and
Ruth Krehbiel (1987), “‘Clean Up Your Plate’: Effects of Child
Feeding Practices on the Conditioning of Meal Size,” Learning
and Motivation, 18 (3), 301–317.
Burger, Kyle S., Marc A. Cornier, Jan Ingebrigtsen, and Susan L.
Johnson (2011), “Assessing Food Appeal and Desire to Eat:
The Effects of Portion Size and Energy Density,” International
Journal of Behavioral Nutrition and Physical Activity, 8 (Sep-
tember), 101–110.
—, Jennifer O. Fisher, and Susan L. Johnson (2011), “Mecha-
nisms Behind the Portion Size Effect: Visibility and Bite Size,”
Obesity, 19 (3), 546–51.
Centers for Disease Control and Prevention (2004), “Trends in
Intake of Energy and Macronutrients: United States 1971
2000,” Morbidity and Mortality Weekly Report, 53 (4), 80–82.
Chandon, Pierre (2009), “Estimating Food Quantity,” in Sensory
Marketing: Research on the Sensuality of Products, Aradhna
Krishna, ed. New York: Routledge, 323–43.
and Nailya Ordabayeva (2009), “Supersize in One Dimen-
sion, Downsize in Three Dimensions: Effects of Spatial
Dimensionality on Size Perceptions and Preferences,” Journal
of Marketing Research, 46 (December), 739–53.
and Brian Wansink (2007), “Is Obesity Caused by Calorie
Underestimation? A Psychophysical Model of Meal Size Esti-
mation,” Journal of Marketing Research, 44 (February), 84–
and (2011), “Is Food Marketing Making Us Fat? A
Multi-Disciplinary Review,” Foundations and Trends in Mar-
keting, 5 (3), 113–96.
Chernev, Alexander, Ulf Bockenholt, and Joseph Goodman
(2010), “Commentary on Scheibehenne, Greifeneder, and
Todd: Choice Overload: Is There Anything to It?” Journal of
Consumer Research, 37 (3), 426–28.
Cohen, Jacob (1988), Statistical Power Analysis for the Behav-
ioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.
De Graaf, Cees, Wendy A. Blom, Paul A. Smeets, Annette Stafleu,
and Henk F. Hendriks (2004), “Biomarkers of Satiation and
Satiety,” American Journal of Clinical Nutrition, 79 (6), 946–
Diliberti, Nicole, Peter L. Bordi, Martha T. Conklin, Liane S. Roe,
and Barbara J. Rolls (2004), “Increased Portion Size Leads to
Increased Energy Intake in a Restaurant Meal,” Obesity
Research, 12 (3), 562–68.
Do Vale, Rita C., Rik Pieters, and Marcel Zeelenberg (2008), “Fly-
ing Under the Radar: Perverse Package Size Effects on Con-
Sizing Up the Effect of Portion Size on Consumption / 153
Hermans, Roel C.J., Junilla K. Larsen, Peter Herman, and Rutger
C.M.E. Engels (2011), “How Much Should I Eat? Situational
Norms Affect Young Women’s Food Intake During Meal
Time,” British Journal of Nutrition, 107 (4), 588–94.
Higgins, Julian P.T. and Simon G. Thompson (2002), “Quantify-
ing Heterogeneity in a Meta-Analysis,” Statistics in Medicine,
21 (11), 1539–58.
Huedo-Medina, Tania B., Julio Sánchez-Meca, Fulgencio Marín-
Martínez, and Juan Botella (2006), “Assessing Heterogeneity
in Meta-Analysis: Q Statistic Or I2Index?” Psychological
Methods, 11 (2), 193–206.
Hughes, Mark (2013), “New York Judge Overturns Supersized
Fizzy Drinks Ban,” The Telegraph, (March 11), (accessed Feb-
ruary 21, 2014), [available at news/
Hunter, John E. and Frank L. Schmidt (1990), Methods of Meta-
Analysis: Correcting Error and Bias in Research Findings.
Newbury Park, CA: Sage Publications.
Jain, Sanjay (2012), “Marketing of Vice Goods: A Strategic
Analysis of the Package Size Decision,” Marketing Science, 31
(1), 36–51.
Jeffery, Robert W., Sarah Rydell, Caroline L. Dunn, Lisa J. Har-
nack, Allen S. Levine, Paul R. Pentel, et al. (2007), “Effects of
Portion Size on Chronic Energy Intake,” International Journal
of Behavioral Nutrition and Physical Activity, 4 (1), 27–32.
Koh, Jiaqi and Patricia Pliner (2009), “The Effects of Degree of
Acquaintance, Plate Size, and Sharing on Food Intake,”
Appetite, 52 (3), 595–602.
Kral, Tanja V.E., A.C. Kabay, Liane S. Roe, and Barbara J. Rolls
(2009), “Effects of Increasing the Portion Size of Fruit and
Vegetable Side Dishes at a Meal on Children’s Intake Regula-
tion,” Appetite, 52 (3), 842.
—, Liane S. Roe, and Barbara J. Rolls (2004), “Combined
Effects of Energy Density and Portion Size on Energy Intake in
Women,” American Journal of Clinical Nutrition, 79 (6), 962–68.
Levitsky, David A. and Trisha Youn (2004), “The More Food
Young Adults Are Served, the More They Overeat,” Journal of
Nutrition, 134 (10), 2546–50.
Looney, Shannon M. and Hollie A. Raynor (2011), “Impact of
Portion Size and Energy Density on Snack Intake in Preschool-
Aged Children,” Journal of the American Dietetic Association,
111 (3), 414–18.
Ma, Yu, Kusum L. Ailawadi, and Dhruv Grewal (2013), “Soda
Versus Cereal and Sugar Versus Fat: Drivers of Healthful Food
Intake and the Impact of Diabetes Diagnosis,” Journal of Mar-
keting, 77 (May), 101–120.
Marchiori, David, Olivier Corneille, and Olivier Klein (2012),
“Container Size Influences Snack Food Intake Independently
of Portion Size,” Appetite, 58 (3), 814–17.
McConahy, Kristen L., Helen Smiciklas-Wright, Leann L. Birch,
Diane C. Mitchell, and Mary F. Picciano (2002), “Food Por-
tions Are Positively Related to Energy Intake and Body Weight
in Early Childhood,” Journal of Pediatrics, 140 (3), 340–47.
Mishra, Arul, Himanshu Mishra, and Tamara M. Masters (2012),
“The Influence of Bite Size on Quantity of Food Consumed: A
Field Study,” Journal of Consumer Research, 38 (5), 791–95.
Mohr, Gina S., Donald R. Lichtenstein, and Chris Janiszewski
(2012), “The Effect of Marketer-Suggested Serving Size on
Consumer Responses: The Unintended Consequences of Con-
sumer Attention to Calorie Information,” Journal of Marketing,
76 (January), 59–75.
Moore, Elizabeth S. (2007), “Perspectives on Food Marketing and
Childhood Obesity: Introduction to the Special Section,” Jour-
nal of Public Policy & Marketing, 26 (Fall), 157–61.
National Alliance for Nutrition and Activity (2002), From Wallet
to Waistline: The Hidden Costs of Super Sizing. Washington,
DC: National Alliance Institute for Nutrition and Activity.
Raynor, Hollie A. and Rena R. Wing (2007), “Package Unit Size
and Amount of Food: Do Both Influence Intake?” Obesity, 15
(9), 2311–19.
Rolls, Barbara J. (2003), “The Supersizing of America: Portion
Size and the Obesity Epidemic,” Nutrition Today, 38 (2), 42–
—, Dianne Engell, and Leann L. Birch (2000), “Serving Por-
tion Size Influences 5-Year-Old but Not 3-Year-Old Children’s
Food Intakes,” Journal of the American Dietetic Association,
100 (2), 232–34.
—, Erin L. Morris, and Liane S. Roe (2002), “Portion Size of
Food Affects Energy Intake in Normal-Weight and Overweight
Men and Women,” American Journal of Clinical Nutrition, 76
(6), 1207–1213.
—, Liane S. Roe, Kitti H. Halverson, and Jennifer S. Meengs
(2007), “Using a Smaller Plate Did Not Reduce Energy Intake
at Meals,” Appetite, 49 (3), 652–60.
—, ——, Tanja V.E. Kral, Jennifer S. Meengs, and Denise E.
Wall (2004), “Increasing the Portion Size of a Packaged Snack
Increases Energy Intake in Men and Women,” Appetite, 42 (1),
—, ——, and Jennifer S. Meengs (2006a), “Larger Portion
Sizes Lead to a Sustained Increase in Energy Intake Over 2
Days,” Journal of the American Dietetic Association, 106 (4),
—, ——, and — (2006b), “Reductions in Portion Size
and Energy Density of Foods Are Additive and Lead to Sus-
tained Decreases in Energy Intake,” American Journal of Clin-
ical Nutrition, 83 (1), 11–17.
—, ——, and — (2007), “The Effect of Large Portion
Sizes on Energy Intake Is Sustained for 11 Days,” Obesity, 15
(6), 1535–43.
—, —, —, and Denise E. Wall (2004), “Increasing the
Portion Size of a Sandwich Increases Energy Intake,” Journal
of the American Dietetic Association, 104 (3), 367–72.
Rosenberg, Michael S. (2005), “The File-Drawer Problem Revis-
ited: A General Weighted Method for Calculating Fail-Safe
Numbers in Meta-Analysis,” Evolution, 59 (2), 464–68.
Rosenthal, Robert (1979), “The ‘File Drawer Problem’ and the
Tolerance for Null Results,” Psychological Bulletin, 86 (3),
— (1991), Meta-Analytic Procedures for Social Research.
Newbury Park, CA: Sage Publications.
Saul, Michael H. (2012), “NYC Board of Health Passes ‘Soda
Ban,’” The Wall Street Journal, (September 13), (accessed
February 21, 2014), [available at
Scheibehenne, Benjamin, Peter M. Todd, and Brian Wansink
(2010), “Dining in the Dark: The Importance of Visual Cues
for Food Consumption and Satiety,” Appetite, 55 (3), 710–13.
Schwartz, Janet, Jason Riis, Brian Elbel, and Dan Ariely (2012),
“Inviting Consumers to Downsize Fast-Food Portions Signifi-
cantly Reduces Calorie Consumption,” Health Affairs, 31 (2),
Scott, Maura L., Stephen M. Nowlis, Naomi Mandel, and Andrea
C. Morales (2008), “The Effects of Reduced Food Size and
Package Size on the Consumption Behavior of Restrained and
Unrestrained Eaters,” Journal of Consumer Research, 35 (3),
Spill, Maureen K., Leann L. Birch, Liane S. Roe, and Barbara J.
Rolls (2010), “Eating Vegetables First: The Use of Portion Size
to Increase Vegetable Intake in Preschool Children,” American
Journal of Clinical Nutrition, 91 (5), 1237–43.
Steenhuis, Ingrid H.M. and Willemijn M. Vermeer (2009), “Por-
tion Size: Review and Framework for Interventions,” Interna-
tional Journal of Behavioral Nutrition and Physical Activity, 6
(August), 58–67.
Sterne, Jonathan A.C., Alex J. Sutton, John P.A. Ioannidis, Norma
Terrin, David R. Jones, Joseph Lau, et al. (2011), “Recommen-
dations for Examining and Interpreting Funnel Plot Asymmetry
in Meta-Analyses of Randomised Controlled Trials,” British
Medical Journal, 343 (July), d4002.
Stroeble, Nanette, Lorraine G. Ogden, and James O. Hill (2009),
“Do Calorie-Controlled Portion Sizes of Snacks Reduce
Energy Intake?” Appetite, 52 (3), 793–96.
Tangari, Andrea H., Scot Burton, Elizabeth Howlett, Yoon-Na
Cho, and Anastasia Thyroff (2010), “Weighing in on Fast Food
Consumption: The Effects of Meal and Calorie Disclosures on
Consumer Fast Food Evaluations,Journal of Consumer
Affairs, 44 (3), 431–62.
Van Ittersum, Koert and Brian Wansink (2012), “Plate Size and
Color Suggestibility: The Delboeuf Illusion’s Bias on Serving
and Eating Behavior,” Journal of Consumer Research, 39 (2),
Van Kleef, Ellen, Mitsuru Shimizu, and Brian Wansink (2011),
“Serving Bowl Selection Biases the Amount of Food Served,”
Journal of Nutrition Education and Behavior, 44 (1), 66–70.
, ——, and —— (2013), “Just a Bite: Considerably
Smaller Snack Portions Satisfy Delayed Hunger and Craving,”
Food Quality and Preference, 27 (1), 96–100.
Wansink, Brian (1994), “How and Why a Package’s Size Influ-
ences Usage Volume,” MSI-94-100, MSI Working Paper
— (1996), “Can Package Size Accelerate Usage Volume?”
Journal of Marketing, 60 (July), 1–14.
— (2004), “Environmental Factors That Increase the Food
Intake and Consumption Volume of Unknowing Consumers,”
Annual Review of Nutrition, 24, 455–79.
(2010a), “From Mindless Eating to Mindlessly Eating Bet-
ter,” Physiology and Behavior, 100 (5), 454–63.
— (2010b), Mindless Eating: Why We Eat More Than We
Think. New York: Bantam Books.
(2012), “Package Size, Portion Size, Serving Size, Market
Size: The Unconventional Case for Half-Size Servings,” Mar-
keting Science, 31 (1), 54–57.
— and Junyong Kim (2005), “Bad Popcorn in Big Buckets:
Portion Size Can Influence Intake as Much as Taste,” Journal
of Nutrition Education and Behavior, 37 (5), 242–45.
, James E. Painter, and Jill North (2005), “Bottomless
Bowls: Why Visual Cues of Portion Size May Influence
Intake,” Obesity, 13 (1), 93–100.
—, Collin R. Payne, and Pierre Chandon (2007), “Internal and
External Cues of Meal Cessation: The French Paradox
Redux?” Obesity, 15 (12), 2920–24.
—, ——, and Mitsuru Shimizu (2011), “The 100-Calorie
Semi-Solution: Sub-Packaging Most Reduces Intake Among
the Heaviest,” Obesity, 19 (5), 1098–1100.
—,—, and Carolina Werle (2008), “Consequences of
Belonging to the ‘Clean Plate Club,’” Archives of Pediatric
and Adolescent Medicine, 162 (10), 994–95.
— and Jeffery Sobal (2007), “Mindless Eating: The 200 Daily
Food Decisions We Overlook,” Environment and Behavior, 39
(1), 106–123.
— and Koert Van Ittersum (2003), “Bottoms Up! The Influ-
ence of Elongation on Pouring and Consumption Volume,”
Journal of Consumer Research, 30 (3), 455–63.
— and — (2005), “Shape of Glass and Amount of Alcohol
Poured: Comparative Study of Effect of Practice and Concen-
tration,” British Medical Journal, 331 (7531), 1512–14.
— and —— (2007), “Portion Size Me: Downsizing Our Con-
sumption Norms,” Journal of the American Dietetic Associa-
tion, 107 (7), 1103–1106.
—, —, and James E. Painter (2006), “Ice Cream Illusions:
Bowls, Spoons, and Self-Served Portion Sizes,” American
Journal of Preventative Medicine, 31 (3), 240–43.
— and Craig S. Wansink (2010), “The Largest Last Supper:
Depictions of Food Portions and Plate Size Increased over the
Millennium,” International Journal of Obesity, 34 (5), 943–44.
World Health Organization (2003), “Controlling the Global Obe-
sity Epidemic,” (accessed February 21, 2014), [available at
— (2013), “Ten Facts on Obesity,” (March), (accessed February
21, 2014), [available at
Young, Lisa R. and Marion Nestle (2002), “The Contribution of
Expanding Portion Sizes to the US Obesity Epidemic,” Ameri-
can Journal of Public Health, 92 (2), 246–49.
154 / Journal of Marketing, May 2014
... Perhaps the most extensively researched size nudge is the "portion-size effect" (PSE; [67]), in which the amount people eat is shaped by the amount they are given to eat in a sitting. One meta-analysis estimated that a 100% increase in portion size leads to a 35% increase in consumption (d + =.45, k = 88; [68]. Importantly, people fail to compensate later for excess consumption due to larger portion sizes. ...
... Importantly, people fail to compensate later for excess consumption due to larger portion sizes. Marchiori and colleagues [68] argued that the implicit activation of an exaggerated norm or reference point in situ underlies the PSE; this activated norm then serves as an anchor that people fail to adjust for (cf. ref [69]). ...
Full-text available
Interventions are effective in promoting health behavior change to the extent that (a) intervention strategies modify targets (i.e., mechanisms of action), and (b) modifying targets leads to changes in behavior. To complement taxonomies that characterize the variety of strategies used in behavioral interventions, we outline a new principle that specifies how strategies modify targets and thereby promote behavior change. We distinguish two dimensions of targets-value (positive vs. negative) and accessibility (activation level)-and show that intervention strategies operate either by altering the value of what people think, feel, or want (target change) or by heightening the accessibility of behavior-related thoughts, feelings, and goals (target activation). We review strategies designed to promote target activation and find that nudges, cue-reminders, goal priming, the question-behavior effect, and if-then planning are each effective in generating health behavior change, and that their effectiveness accrues from heightened accessibility of relevant targets. We also identify several other strategies that may operate, at least in part, via target activation (e.g., self-monitoring, message framing, anticipated regret inductions, habits). The Activation Vs. Change Principle (AVCP) offers a theoretically grounded and parsimonious means of distinguishing among intervention strategies. By focusing on how strategies modify targets, the AVCP can aid interventionists in deciding which intervention strategies to deploy and how to combine different strategies in behavioral trials. We outline a research agenda that could serve to further enhance the design and delivery of interventions to promote target activation.
... In general, the health expert's first-person narrative emphasizing the health risks of sugar decreased individuals' WTP for sugar-containing food, but did not modulate their WTP for sugar-free food. This supports earlier investigations on other healthy eating nudges (e.g., size enhancements), suggesting that interventions are more effective at reducing unhealthy eating than increasing healthy eating (12,56,57). This result is also in line with the notion of negativity bias (58,59). ...
Full-text available
Recent studies have revealed types of eating nudges that can steer consumers toward choosing healthier options. However, most of the previously studied interventions target individual decisions and are not directed to changing consumers’ underlying perception of unhealthy food. Here, we investigate how a healthy eating call—first-person narrative by a health expert—affects individuals’ willingness to pay (WTP) for sugar-free and sugar-containing food products. Participants performed two blocks of a bidding task, in which they had to bid on sweets labeled either as “sugar- free” or as “sugar-containing.” In-between the two blocks, half of the participants listened to a narrative by a dietary specialist emphasizing the health risks of sugar consumption, whereas the remaining participants listened to a control narrative irrelevant to food choices. We demonstrate that the health expert’s narrative decreased individuals’ WTP for sugar-containing food, but did not modulate their WTP for sugar- free food. Overall, our findings confirm that consumers may conform to healthy eating calls by rather devaluating unhealthy food products than by increasing the value of healthy ones. This paves the way for an avenue of innovative marketing strategies to support individuals in their food choices.
... These trends are concerning considering the evidence that increases in portion sizes are robustly linked to increases in energy consumption among children and adults (Hollands et al., 2015;Marteau et al., 2015;Zlatevska et al., 2014). ...
The average adult in the UK consumes 200-300 calories beyond their Guideline Daily Amount. For working adults, more than one-third of calories are consumed in the workplace, making this an important environment for intervention. This thesis makes a contribution to the academic literature, by improving our understanding of how and when offering lower-energy alternatives (‘swaps’) is effective, and to public health by refining an intervention which could be delivered in workplace canteens. Two scoping reviews were conducted (studies 1 & 2) and pointed towards the potential effectiveness of pre-ordering lunch and offering healthier swaps as strategies that may help to improve the healthfulness of food and drink choices. When offering lower energy swaps for snacks and non-alcoholic drinks, studies 3 (n=449) and 4 (n=3,481) recruited samples of UK adults in employment to test the effect of different messages on the acceptance of swaps in an experimental online canteen. The results indicated that messages focusing on the lower-energy content of swaps offered may be an effective and acceptable approach. When highlighting the energy content of swaps offered, increasing the interpretability of this information, by providing physical activity calorie equivalent information (PACE) (i.e., the number of minutes walking required to expend the energy contained) further increased the acceptance of snack and drink swaps offered. In study 5, an online version of a real-world canteen was developed and the intervention (prompts to swaps accompanied by a PACE message) was due to be tested in a real-world trial with the healthcare organisation Bupa. However, due to Covid-19, it was tested qualitatively with employees (n=30) of this organisation across the full lunch menu to provide insights about the factors perceived to influence swap acceptance and the acceptability of the intervention. Swap acceptance was facilitated by the provision of PACE information, and swap similarity in terms of taste, texture, and expected satiety as well as the perception that alternatives provided meaningful energy savings. Overall, the intervention was viewed as an acceptable approach to help reduce energy intake in the workplace. Following refinements to the intervention, Study 6 tested the effect of offering lower-energy swaps with and without PACE messages on the energy of hypothetical lunches pre-ordered with a representative online sample of working adults (n=2,150). Offering swaps with and without a PACE message was found to significantly reduce average energy ordered at lunch compared to when no swaps were offered, the PACE message was more acceptable, and there was no evidence of significant interactions between intervention efficacy and participant characteristics. Offering lower-energy swaps in the workplace when employees pre-order is an acceptable and promising intervention to reduce the energy of foods and drinks ordered. Future work should replicate this research in real-world settings.
... First, it should reduce portion size choices. Portion size positively predicts caloric intake (Zlatevska, Dubelaar, and Holden 2014), and interventions that reduce portion size choices reduce caloric intake (Schwartz et al. 2012). Second, the intervention should be easily implementable, particularly for online food-delivery systems. ...
Body Mass Indices and obesity rates are increasing worldwide, and one way to reduce caloric intake is to reduce portion size choice. In this research, the authors develop a behavioral intervention aimed at reducing portion size choices in the context of online food ordering. In eight experiments (including a field experiment), the authors show that the sequential presentation of two food images that move from partial to whole reduces hunger perceptions and portion size choices relative to all comparable sequences. This effect occurs because the partial-to-whole image presentation primes the concept of reaching fullness, which in turn reduces perceptions of hunger and portion size choices. The effect of image sequence on both hunger perceptions and portion size choices is mediated by the accessibility of the concept full and is attenuated when visualization of the dynamic sequence (partial-to-whole) is inhibited. The partial-to-whole sequence effect is observed even when the sequential images are unrelated to food, and is robust across languages, age groups, food type, and choice contexts. The brief, low-cost intervention can be implemented across several online food-ordering contexts (e.g., school cafeteria apps, diet apps) and has important public policy implications.
... This study also rules out product popularity, quality, novelty, and processing fluency as an alternative explanation. Specifically, previous research has pointed out that larger-sized products are often associated with higher popularity because these products are usually of greater value (Zlatevska et al., 2014); therefore, it is likely that products with higher popularity might be considered to be larger (Zhang et al., 2021). In addition, previous studies have also demonstrated that product quality might be associated with product size, because consumers often associate high quality with the high price that often appears at the same time as the large size (Pracejus et al., 2006). ...
Full-text available
This study demonstrates a visual phenomenon in online product presentation: product size perception is influenced by the depth of field of the presentation image. Depth of field refers to how blurry or sharp the background around the focused subject is a shallow depth‐of‐field image result in a clear focused subject and a blurry background, while in a deep depth‐of‐field image, both the subject and the background are clear. One eye‐tracking study, three behavioral experiments, and one field study show that a shallow (vs. deep) depth‐of‐field product presentation (i.e., a clear product with a blurry background) increases consumers' product size perceptions. This effect is mediated by the greater attention allocated to the product and is moderated by product familiarity. Specifically, when product familiarity is low, consumer attention mediates the significant effect of depth of field on product size perception. However, when product familiarity is high, the effects of depth of field and consumer attention decrease. The current research contributes to the previous research on product presentation and product size perception by investigating the effect of a novel factor, the depth of field, on consumers' estimations of product size. Overall, the findings encourage online retailers to carefully adapt the depth of field technique in their product presentation according to their objectives (e.g., attract consumers' interest vs. provide accurate information) and consumers' familiarity with their products.
... Overall, the question arises of how well the consumer can estimate the amount of food when asked to report the quantity of foods consumed as multiples of portion size and whether there are differences between overweight or obese and normal weight in the portion size assessment of different foods. On one hand, individuals with BMI over 25 tend to be less responsive to larger food portions (Zlatevska et al., 2014); however, they are more likely to choose larger portions fueling further weight gain (Westerterp-Plantenga et al., 1996;Flegal and Troiano, 2000;Young andNestle, 2003, 2012). Accordingly, higher BMI is associated with selecting larger amounts of food in a setup where self-selection of portion size was permitted (Burger et al., 2007;Lewis et al., 2015). ...
Full-text available
Portion sizes of meals have been becoming progressively larger which contributes to the onset of obesity. So far, little research has been done on the influence of body weight on portion size preferences. Therefore, we assessed whether Body Mass Index (BMI), as well as other selected factors, contribute to the estimation of food portions weight and the subjective perception of portion sizes. Through online questionnaires, the participants were asked to estimate the weight of pictured foods in the first study. In the second study, the participants indicated how the depicted varying portion sizes of different meals relate to their actual consumed real-life portion sizes. A total of 725 and 436 individuals were included in the statistical analysis in the first and second study, respectively. BMI and gender had a small effect on the capacity to estimate the weight of foods. The main predictor for portion size choices was the factor gender with men estimating ideal portion sizes as larger than women. Further, age and hunger together with external and restrictive eating behaviors were among the deciding factors for portion size choices. As expected, externally motivated eaters chose bigger portions while restrictive individual smaller ones. Gender- and age-related differences in portion size preferences likely reflect distinct energy requirements. The individuals with a higher BMI do not differ strongly from other BMI groups in their portion-related preferences. Therefore, other factors such as meal frequency, snacking, or a lifestyle, may contribute more to the onset, development, and maintenance of overweight.
This study synthesizes the artificial intelligence literature into a Meta‐analytic framework based on the theory of reasoned action and the unified theory of acceptance and use of technology 2, and examines concrete relationships between the constructs of this framework. This meta‐review also performed a moderation analysis to investigate the possible reasons for inconsistent findings across studies. The findings suggest that three methodological moderators (sample type, gender dominance, and publication type), and one contextual moderator (level of country's technology advancement) lead to inconsistencies in the relationships between study constructs. Academically, this review synthesizes the artificial intelligence literature and resolves inconsistencies in the literature and also adds constructs to both theory of reasoned action and the unified theory of acceptance and use of technology 2. Practically, this meta‐analysis offers multiple implications for businesses interested in enhancing customer adoption of artificial intelligence. Especially, companies can increase customers' adoption of artificial intelligence by making it more user‐friendly, and advantageous and by adding pleasing features to it.
This contribution reviews the main normative and positive arguments that can used in the assessment of the costs and benefits of food marketing restrictions, focusing specifically on theoretical and empirical developments in the economics of advertising, consumer behaviour and industrial organization since the 70s.
Full-text available
Purpose Obesity among elderly consumers precipitates undesirable health outcomes. This study aims to investigate the effects of environmental cues on food intake of elderly consumers in an aged-care facility. Design/methodology/approach A longitudinal study conducted over 17 weeks in situ within an aged-care facility with 31 residents investigated how auditory (soothing music), olfactory (floral-scented candle) and visual (infographic on health benefits of the main meal component) cues influenced food intake quantity during a meal, while accounting for portion size effect (PSE). Findings Analysing the cross-sectional results of individual treatments and rounds did not reveal any consistent patterns in the influence of the three environmental cues. Longitudinal analyses, however, showed that the presence of auditory and olfactory cues significantly increased food intake, but the visual cue did not. Moreover, PSE was strong. Research limitations/implications Extending research into environmental factors from a commercial to a health-care setting, this study demonstrates how the presence of auditory and olfactory, but not cognitive cues, increased food intake behaviour among elderly consumers. It also shows that a cross-sectional approach to such studies would have yielded inconclusive or even misleading findings. Merely serving more would also lead to higher food intake amount. Practical implications Environmental factors should be a part of health-care providers’ arsenal to manage obesity. They are practical and relatively inexpensive to implement across different health-care settings. However, the same environmental factors would have opposite desired-effects with normal or underweight residents, and hence, aged-care facilities need to separate the dining experience (or mealtime) of obese and other residents. Quantity served should also be moderated to discourage overeating. Originality/value While studies into managing obesity, particularly among older adults, have mainly focused on techniques such as pharmacotherapy treatments with drugs, dietary management or even lifestyle change, less attention has been given to the influence of environmental cues. This study, executed in situ within an aged-care facility, provided evidence of the importance of considering the impact of environmental factors on food intake to help reduce obesity.
Full-text available
An analysis of consumers' Weblogs and two experiments address: (1) the differences in evaluations of menu items when consumers are versus are not provided with meal calorie information, and (2) their perception of calorie levels of different types of meals. Consumers provided their calorie estimates for specific meals offered by four different fast food restaurants, and an experiment assessed effects on consumer evaluations for calorie disclosures for actual items from two of these restaurants. Results show the complex relationship between consumer perceptions regarding the restaurants, the meals and the food items that can influence consumers' calorie estimates and evaluations of meals in restaurants.
Packaging influences usage behavior long after it has influenced purchase. Managers of consumer packaged goods and public policy officials have, therefore, questioned whether a package's size influences usage volume. Although often assumed, it has never been supported. Four laboratory studies and a final study in a Laundromat identify circumstances in which larger package sizes encourage greater use than do smaller package sizes. Unit cost is a key factor mediating this relationship. After noting useful implications for decisions regarding package size portfolios, sales promotions, and public policy, the author concludes by identifying other important but overlooked factors that increase usage volume and provide research opportunities.
Background: Portion size influences children's energy intakes at meals, but effects on daily intake are unknown. Objective: Effects of large portions on daily energy intake were tested in 5-y-old Hispanic and African American children from low-income families. Maternal food intake data were collected to evaluate familial susceptibility to portion size. Design: A within-subjects experimental design with reference and large portion sizes was used in a study of 59 low-income Hispanic and African American preschool-aged children and their mothers. The portion size of 3 entrées (lunch, dinner, and breakfast) and an afternoon snack served during a 24-h period were of a reference size in one condition and doubled in the other condition. Portion sizes of other foods and beverages did not vary across conditions. Weighed food intake, anthropometric measures, and self-reported data were obtained. Results: Doubling the portion size of several entrées and a snack served during a 24-h period increased energy intake from those foods by 23% (180 kcal) among children (P < 0.0001) and by 21% (270 kcal) among mothers (P < 0.0001). Child and maternal energy intakes from other foods for which portion size was not altered did not differ across conditions. Consequently, total energy intakes in the large-portion condition were 12% (P < 0.001) and 6% (P < 0.01) higher in children and mothers, respectively, than in the reference condition. Child and maternal intakes of the portion-manipulated foods were not correlated. Conclusions: Large portions may contribute to obesigenic dietary environments by promoting excess daily intakes among Hispanic and African American children.
Background: Increases in both the portion size and energy density of food have both been shown to increase energy intake, but the combined effects of such increases have not been investigated. Objective: The objective was to determine the combined effects of energy density and portion size on energy intake in women. Design: This study used a within-subjects design. Once a week for 6 wk, 39 women were served breakfast, lunch, and dinner ad libitum. The main entrée at lunch was formulated in 2 versions that varied in energy density (5.23 or 7.32 kJ/g), each of which was served in 3 different portion sizes (500, 700, or 900 g). The 2 versions were matched for macronutrient composition and palatability. Breakfast and dinner were standard meals. Results: Increases in portion size and energy density led to independent and additive increases in energy intake (P <0.0001). Subjects consumed 56% more energy (925 kJ) when served the largest portion of the higher energy-dense entrée than when served the smallest portion of the lower energy-dense entrée. Subjects did not compensate for the additional intake by eating less at the subsequent meal. Despite substantial differences in energy intake, no systematic differences in ratings of hunger and fullness across conditions were observed. Conclusions: The energy density and the portion size of a food act independently to affect energy intake. The findings indicate that large portions of foods with a high energy density may facilitate the overconsumption of energy.
During 1971-2000, the prevalence of obesity in the United States increased from 14.5% to 30.9%.¹ Unhealthy diets and sedentary behaviors have been identified as the primary causes of deaths attributable to obesity.² Evaluating trends in dietary intake is an important step in understanding the factors that contribute to the increase in obesity. To assess trends in intake of energy (i.e., kilocalories [kcals]), protein, carbohydrate, total fat, and saturated fat during 1971-2000, CDC analyzed data from four National Health and Nutrition Examination Surveys (NHANES): NHANES I (conducted during 1971-1974), NHANES II (1976-1980), NHANES III (1988-1994), and NHANES 1999-2000. This report summarizes the results of that analysis, which indicate that, during 1971-2000, mean energy intake in kcals increased, mean percentage of kcals from carbohydrate increased, and mean percentage of kcals from total fat and saturated fat decreased (Figures 1 and 2). An expert advisory committee appointed by the U.S. Department of Health and Human Services and the U.S. Department of Agriculture (USDA) is conducting a review of the Dietary Guidelines for Americans.³ Revised guidelines will be published in 2005.
Although we are just beginning to understand how environmental factors such as portion size affect eating behavior, the available data suggest that large portions of energy-dense foods are contributing to the obesity epidemic. Several possible strategies for adjusting portions to bring intake back in line with energy requirements are discussed. The continuing rise in the rates of obesity calls for urgent action.
Nutritional labels are mandatory on virtually all packaged food items sold in the United States. The nutritional information on these labels is reported on a "per-serving-size" basis. However, unbeknownst to many consumers, current Food and Drug Administration regulations allow manufacturers some discretion in setting serving sizes—a factor that the authors hypothesize has implications for consumer behavior. For example, adopting a smaller serving size allows marketers to reduce the reported calories, fat, sugar, and carbohydrates in a product serving, which in turn can influence the anticipated consequences of consumption. Three studies show that manipulating the serving size, and thus calories per serving, for equivalent consumption amounts influences the anticipated guilt of consumption, purchase intentions, and choice behavior. However, the results also show that individual difference and context variables, which heighten consumer attention to nutritional information in general, often focus attention on calorie information but not serving size. This leads to the counterintuitive finding that more nutritionally vigilant consumers are more heavily influenced by serving size manipulations. The authors discuss the managerial and public policy implications.
Despite the challenged contention that consumers serve more onto larger dinnerware, it remains unclear what would cause this and who might be most at risk. The results of five studies suggest that the neglected Delboeuf illusion may explain how the size of dinnerware creates two opposing biases that lead people to overserve on larger plates and bowls and underserve on smaller ones. A countercyclical sinus-shaped relationship is shown to exist between these serving biases and the relative gap between the edge of the food and the edge of the dinnerware. Although these serving biases are difficult to eliminate with attention and education, changing the color of one’s dinnerware or tablecloth may help attenuate them. By showing that the Delboeuf illusion offers a mechanistic explanation for how dinnerware size can bias serving and intake, we open new theoretical opportunities for linking illusions to eating behavior and suggest how simple changes in design can improve consumer welfare.
This research proposes that consumers’ preference for supersized food and drinks may have roots in the status-signaling value of larger options. An initial experiment found that consumers view larger-sized options within a set as having greater status. Because low-power consumers desire status, we manipulated power to test our core propositions. Whether induced in the lab or in the field, states of powerlessness led individuals to disproportionately choose larger food options from an assortment. Furthermore, this preference for larger-sized options was enhanced when consumption was public, reversed when the size-to-status relationship was negative (i.e., smaller was equated with greater status), and mediated by consumers’ need for status. This research demonstrates that choosing a product on the basis of its relative size allows consumers to signal status, illustrates the consequences of such a choice for consumers’ food consumption, and highlights the central role of a product category’s size-to-status relationship in driving consumer choice.