Content uploaded by Marlene Schwartz
Author content
All content in this area was uploaded by Marlene Schwartz on Sep 27, 2014
Content may be subject to copyright.
Public Health Nutrition: page 1 of 10 doi:10.1017/S1368980011003065
Purchases of ready-to-eat cereals vary across US household
sociodemographic categories according to nutritional value
and advertising targets
Katia Castetbon
1,2,
*, Jennifer L Harris
2
and Marlene B Schwartz
2
1
Usen, Institut de veille sanitaire/Universite
´Paris 13, 74 rue Marcel Cachin, F-93017 Bobigny Cedex, France:
2
Rudd Center for Food Policy and Obesity, Yale University, New Haven, CT, USA
Submitted 18 February 2011: Accepted 25 October 2011
Abstract
Objective: To describe ready-to-eat (RTE) cereal purchases in 2008 in the USA
according to cereal nutritional quality and marketing strategy and household
sociodemographic characteristics.
Design: Cross-sectional study of purchases in one year. Each type of cereal was
assigned to one of four nutrition quality categories (based on Nutrient Profile
Index, NPI) and one of four advertising categories based on television exposure
and analysis of packaging (child-targeted, family-targeted, adult-targeted and no
television advertising). Medians and distributions of purchase indicators were
calculated for the cereal categories and the distributions were compared across
sociodemographic groups.
Setting: RTE cereals (n249) with complete label and nutritional content.
Subjects: RTE cereal purchases according to household sociodemographic
characteristics obtained from Nielsen Homescan, a nationally representative panel
of households.
Results: Purchases of RTE cereals were highest in households with one or more
child and lowest in African-American and Asian households, as well as those
earning ,$US 30 000 per annum. The lowest-quality products were purchased by
four times as many households as the highest-quality cereals, but loyalty to these
products was lower. Purchases of cereals by households with children and in
African-American and Hispanic households increased as cereal nutritional quality
declined. Compared with non-advertised products, advertised child-targeted
cereals were purchased thirteen times more frequently; family-targeted brand
purchases were ten times higher; and adult-targeted cereals were purchased four
times more frequently.
Conclusions: Our findings suggest that improving the nutritional quality of
RTE cereals with advertising targeted to children could also lead to increased
consumption of healthier products by young people.
Keywords
Food advertising
Nutritional values
Purchases
Ready-to-eat cereals
Sociodemographic categories
Breakfast is an important part of a healthful diet, because it
facilitates energy balance over the day and typically supplies
major nutrients
(1)
. Consuming breakfast (compared with
skipping breakfast) has been associated with weight gain
prevention
(2)
and positive health status in general
(1)
.Young
people, especially teenagers, are at high risk of skipping
breakfast
(3–5)
, with one out of five 9–13-year-olds and a
third of 14–17-year-olds regularly skipping breakfast in the
USA in 1999–2006
(5)
.
Ready-to-eat (RTE) cereals are one common choice for
breakfast. RTE cereal consumption increased through the
mid-1990s in the USA
(4)
; a third of children (9–13 years
of age) and a quarter of adolescents (14–18 years) ate
RTE cereals for breakfast in the first half of the 2000s
(5)
.
RTE cereals provide carbohydrates including fibre and
micronutrients (due to enrichment widely practised since
the 1970s), but low fat content
(1)
. They have therefore
been recommended as components of a healthful
breakfast, together with dairy products (milk is usually
consumed with RTE cereals) and fruit (fresh or juice)
(6)
.
The research documenting the health benefits of RTE
cereals typically compares this choice with skipping
breakfast altogether
(7)
or with high-fat breakfast options
such as those including fried eggs, cheese, bacon or
sausage
(8)
. As shown in a controlled study
(9)
, the benefit
of RTE cereal consumption was sizeable only when
accompanied by nutritional education, suggesting that the
positive impact is probably due more to other healthy
SPublic Health Nutrition
*Corresponding author: Email katia.castetbon@univ-paris13.fr rThe Authors 2011
behaviours associated with such food choices than to the
nutritional quality of the products consumed.
One feature of RTE cereals is their wide variation in
nutritional composition across brands; specifically, some
products contain sizeable amounts of added sugars and
salt
(10)
. Child-targeted RTE cereals are of special concern
(11)
because they are more energy-dense and contain higher
amounts of sugars compared with their adult-targeted
counterparts. Previous studies have found that RTE cereals
of low nutritional quality continue to be advertised exten-
sively to children
(10)
despite cereal companies’ pledges to
reduce marketing of unhealthy products to children
(12)
.
Importantly, during the 2003–2007 period, exposure to RTE
cereal advertising decreased among 2–5-year-old children
(223 %), but remained relatively constant among older
children (23%) and adolescents (20?8%)
(13)
. Different
advertising exposure trends were also observed by cereal
manufacturer. For example, both General Mills and Kellogg
participate in the children’s food marketing pledges, but
older children’s exposure to General Mills advertising went
down by 10 % during that period whereas exposure to
Kellogg advertising went up by 7 %. Similarly, exposure
to advertising by younger children went down by 30 %
for General Mills products, but by just 11 % for Kellogg
products. Manufacturers also committed to reduce the
sugar and Na contents of cereals marketed to children
(10)
.
However, as of early 2009, the average nutritional quality of
children’s cereals improved by only 2–5 % v. formulations
in 2007
(10)
. In addition, how these small improvements
influence RTE cereal purchases and intake, especially in
children and adolescents, remains to be understood
(14)
.
Recent observations on RTE cereals intake are also needed.
Studying RTE cereal purchases can contribute to our
understanding of the relationship between nutritional
quality, marketing and product intake
(15)
. Databases of
individual dietary intake assessments (such as 24 h
recalls) contain little or no information about brands
of manufactured products consumed, and generic foods
do not take into account variations in nutritional content
from one brand to another. Therefore, such surveys
do not enable analyses of specific nutrient variations
based on the exact products eaten
(16)
. Despite their
limitations
(17)
, databases designed to provide economic
information can be used to conduct detailed analyses of
purchase behaviours by finely classifying each RTE cereal
product according to its specific nutritional content and
comparing product purchases with advertising exposure.
Knowing how RTE cereal purchases vary according to
their nutritional content and advertising exposure could
lead to public health actions aimed at improving the
quality of products actually chosen by the population.
In addition, it is hypothesized that purchases vary across
sociodemographic categories of the population in relation
to the nutritional content and/or the advertising volume
and that different population categories may be prone to
‘respond’ differently to these factors. Improving access to
good-quality RTE cereals for subgroups of the population
so identified would also further public health action.
The objective of the present study was to describe
purchases of RTE cereal products in the US population
overall and across sociodemographic categories according
to (i) cereal nutritional content and (ii) product advertising,
including target market (child, family and adult) and
whether the product was advertised on television.
Methods
The Nielsen Homescan data (now ‘National
Consumer Panel’)
Panel recruitment and data collection
Nielsen recruits panel members, 18 years and older, living
in all US states and interested in participating in the
collection of data on their product purchases through its
dedicated website complementary to a random recruit-
ment. Participants are further selected on the basis of
sociodemographic characteristics they provide. Each time
they shop, they register the date, store name and location
of their purchases. They use a hand-held scanner to
register the barcode of all goods purchased along with
the quantity and whether the purchase was made with a
promotion or manufacturer’s coupons (specifying its
amount). They send the data once weekly through the
dedicated and secured website.
Based on the barcodes, Nielsen identifies the detailed
product and brand information, as well as weekly average
price when purchases are made at stores registered in the
store-level data (‘Scan Track’). Otherwise, volunteers
manually enter the price paid using a hand-held scanner.
Nielsen verifies this information according to the median
of prices in the area and corrects outlier entries (i.e. those
outside 99 % thresholds of acceptable actual prices). It
typically uses the median price that is most appropriate
for the store where the purchase has been made. The data
set used in the present analysis included RTE cereal
purchases made in 2008 by households in the Homescan
panel that provided any purchase registration for at
least 10 months during the year and purchased any RTE
cereal (n57 171).
Nielsen purchase indicators
Calibration based on the national census for household
size, household income, female head age, race, Hispanic
origin, male and female head education, head of house
occupation, presence of children and Nielsen county
size is carried out to provide national estimations for
sociodemographic characteristics and product purchases.
Based on the sociodemographic characteristics of
Homescan panel households, Nielsen provides estimates
for a nationally representative sample of households
and by sociodemographic category. The data set includes
the following measures for each RTE cereal in their
SPublic Health Nutrition
2 K Castetbon et al.
database: ‘Total spending’, the total dollars spent during
the year by the entire estimated population group; ‘Item
buyers’, the estimated total number of households who
bought the product during the year; ‘Buying rates’, the
mean annual spending per household on the cereal (in
dollars); ‘Purchase frequency’, the mean number of times
the product was bought during the year per household;
‘Purchase size’, the average amount spent per purchase
(in dollars); ‘Purchase cycle’, the average number of days
between two purchases; ‘Loyalty’, the average share of
spending for a given RTE cereal as a percentage of all
spending for RTE cereals during the year by households;
‘Dollars purchased on deal’, the share of total spending
when the product was on promotion at the store (in %);
and ‘Dollars with manufacturer coupon’, the share of total
spending using coupons supplied by manufacturers (in %).
Nielsen also computes an aggregated ‘volume index’
by sociodemographic category. The volume index
describes the extent to which the share of purchases of
a given RTE cereal, by households in a specific socio-
demographic category, compare with those that would
be expected given that category’s share of the total
population. The volume index is calculated by dividing
the percentage of RTE cereal spending in the socio-
demographic group by the percentage of the group in the
national population according to the US Census. A
volume index greater than 100 indicates that the group
purchases a higher-than-average amount of a given cereal;
whereas an index below 100 indicates that it purchases a
lower-than-average amount.
Nutrient content database and nutrient profiling
index model score computation
The detailed list of RTE cereals in the Homescan data set
included products purchased by a minimum of seventy-
five households during 2008, other products being
grouped into a summarizing category. This RTE cereal
list was then merged with a database of RTE cereals,
including nutrient content and advertising, developed for
a comprehensive analysis of the RTE cereal industry in
2008
(10)
. When the nutrient content was missing for a
given RTE cereal, it was completed by using manufacturer
websites and checking the nutrition facts on packaging in
the supermarket. This step was required for very few
cereals (,5 %), therefore it is unlikely to have biased the
final estimations. The nutrient data set included serving
size supplied by manufacturers and the content of energy
(kcal/serving), saturated fat (g/serving), sugar (g/serving),
fibre (g/serving) and Na (mg/serving). The nutrient content
database contained 573 registered RTE cereals.
The Nutrient Profile Index (NPI) score
(18)
was calcu-
lated for each product. The NPI is based on a model
developed for the Food Standards Agency (FSA) in the
UK and validated to reflect food quality assessments
by nutritionists
(19,20)
. The model takes into account both
‘positive’ (i.e. to encourage) and ‘negative’ (i.e. to limit)
nutrients together. Thus it provides a more nuanced
evaluation of foods’ nutritional quality based on their
entire nutrient composition. The original model was
launched to identify products that are healthy and can be
advertised to children on television in the UK; other
applications, with modified calculations, are currently
being evaluated for labelling products, for instance in
Australia and New Zealand (http://www.foodstandards.
gov.au/). Briefly, the model provides one score for indi-
vidual products based on points for components that
should be limited in the diet (energy, saturated fat, simple
sugars and Na; ‘A’ points) relative to points for components
considered favourable for a healthy diet (fruits, vegetables
and nuts, NSP fibre or AOAC (Association of Official
Analytical Chemists) fibre and protein; ‘C’ points). Points
are assigned based on nutrients in 100g of the product.
The overall score is then calculated by subtracting the
C points (from 0 to 15 points maximum) from the A points
(from 0 to 35 points maximum)
(18)
. In the original model, a
solid food is considered as ‘less healthy’ if it scores 4 or
more points. The initial model is difficult to interpret as
higher scores indicate lower nutritional quality, and the
range of scores falls between 215 and 35. Therefore, we
modified the original model calculation as follows
(10)
.
>The score was transformed as ‘(223NP score) 170’.
The new NP index (NPI) therefore falls between 0
[(22335 points) 170] and 100 [(223215 points) 1
70], with a higher NPI indicating better nutritional
content. An initial NP score lower than 4 points is
considered by FSA as a threshold for identifying
‘healthful products’. The corresponding threshold for
the NPI is therefore higher than 62 points.
>Since no information about AOAC and NSP fibre was
available separately, only the calculation for AOAC
fibre was used
(18)
.
Targeted advertising exposure
The present study categorized RTE cereals using the
same method as the previous analysis that also examined
advertising and packaging for 277 RTE cereals and
classified them as ‘child-targeted’, ‘family-targeted’ or ‘adult-
targeted’
(10)
. Cereals were classified as child-targeted if
they advertised directly to children in 2008. Cereals were
classified as family-targeted if their packaging or marketing
copy indicated that they were appropriate to feed children
and/or families, but the researchers found no evidence
of marketing directed to children. All other cereals were
classified as adult-targeted. Products not advertised on
television at all in 2008 were categorized separately as ‘no
advertising’ in our analyses.
Statistical analyses
Based on the NPI describing the nutrient content of RTE
cereals, four categories of nutrient quality were created:
(i) very poor (,40 points); (ii) poor (40–49 points);
SPublic Health Nutrition
Ready-to-eat cereal purchases and advertising 3
(iii) fair (50–62 points); and (iv) good (.62 points). Given
that individual data at the household level were not
available for the Homescan panel, distributions (median
and 25th–75th percentiles) are used to describe purchase
indicators (number of buyers, purchase frequency, etc.)
and volume indices across the NPI advertising target
categories. Medians were preferred due to small sample
sizes in some subgroups. For volume indices, box-and-
whisker plots are also used to illustrate distributions across
target-advertising categories for some sociodemographic
groups. Nielsen recommends that a volume index less than
80 indicates ‘under-purchasing’ of the product by a given
sociodemographic group, and that a volume index higher
than 120 indicates that the RTE cereal is purchased in
amounts sizeably higher than expected given the share
of the group in the overall population. Indeed, Nielsen
considers a difference of 20 % as meaningful given the
standard errors usually observed for panel volume esti-
mates. Sociodemographic characteristics analysed here
were presence of any child at home (and if yes, the age
category of the child), household size, race/ethnicity,
female head-of-household education, income and geo-
graphic region. Statistical comparisons across categories
were carried out using ANOVA and trend tests when
appropriate. Statistical analyses were conducted using the
STATA statistical software package version 10?0(2007;Stata
Corporation, College Station, TX, USA).
Results
The Homescan data set included purchases of 290 different
RTE cereals. Of these cereals, five consisted of ‘Ralston
Food products’ (comprised of thirty-five various store
brands) and one observation summarized private label
product (‘CTL-BR’) purchases for which nutrition data
were not available; in 2008, these two categories repre-
sented 10?4 % of total purchases in dollars. In addition,
thirteen items consisted of RTE cereal in mixed cereal
packs (1?1% of the total purchases), nutritional content
was not available for an additional eighteen cereals (mostly
because they had been discontinued; 0?9 % of the total
purchases) and four items were excluded because they
were not RTE cereals (cereal straws or wheat germ; 0?5%
of the total purchases in 2008). The 249 RTE cereals
included in the final analysis represented 87?1 % of total
purchases in US dollars in 2008.
Ready-to-eat cereal purchases across household
sociodemographic categories
Based on medians (25th–75th percentiles) of volume
index (‘All’ column, Table 1), purchases of RTE cereals
were lower than average (volume index median ,80) in
households with one member, African-American and
Asian households, and households earning ,$US 30 000
per annum. Purchases were also low (volume index
median ,90) in households in which the female head of
household was not a high-school graduate and those
without children in the home. In contrast, RTE cereal
purchases were higher than average (volume index
median .120) in households with five or more members
and those with at least one child of any age. In addition,
median volume indices increased regularly with the
number of household members, female head-of-household
education and income (Table 1).
Purchase indicators according to ready-to-eat
cereals’ nutrient quality
RTE cereals were distributed across NPI categories as
follows: very poor (,40), n46 (18?5 %); poor (40–49),
n69 (27?8 %); fair (50–62), n89 (35?7 %); and good
(.62), n45 (18?1 %). Most RTE cereals contain little or no
saturated fat, protein or vegetables/fruits/nuts per 100 g;
the variation between cereal NPI scores is mostly due to
differences in sugar, fibre and Na content. Indeed, the
higher the NPI range, the lower the sugar content and the
higher the fibre content (Table 2). Variations in Na are not
sizeable in the lowest three NPI categories, but the Na
median is much lower in the ‘good’ quality products. In
addition, slightly higher energy content was observed as
the NPI score decreased.
Purchase indicators varied significantly across nutrient
content categories, with the exception of purchase fre-
quency (P50?10; Table 3). With improvement in NPI,
the number of buyers decreased (P50?01) but loyalty
increased dramatically (P,0?001). Buying rates per annum
andpurchasesize(P50?002) increased with nutritional
quality; whereas the purchase cycle (i.e. time between
two purchases) decreased (P,0?001). Finally, share of
purchases using coupons or promotions was lowest in the
extreme nutrient profile categories (NPI ,40 and NPI .62)
and highest in the two intermediate categories (Table 3).
In each household sociodemographic group, volume
index varied across nutrient content categories (Table 1).
The RTE cereal volume index decreased when nutritional
quality increased in households with at least one child,
while the opposite occurred in households without
children. A similar pattern was also observed for household
size (volume index increased in households with one or
two members and decreased in households with three
or more members) and according to female head-of-
household education level (volume index decreased with
nutritional quality in the first three education categories, but
increased for households with a college graduate female
head of household). Moreover, variations across nutritional
content categories differed according to race/ethnicity:
the volume index increased with nutritional quality in
Caucasian households, but decreased in Hispanic and
African-American households. Asian households showed a
different pattern with an increasing volume index up to the
fair (50–62) NPI category and then a decrease. RTE cereal
volume indices also increased significantly with nutritional
SPublic Health Nutrition
4 K Castetbon et al.
quality in the highest household income category ($$US
100 000 per annum) and in the East and West regions. By
contrast, volume indices decreased significantly with NPI
category in households earning $US 30 000–39 999 and
$US 50 000–59 999 per annum, as well as in the South
and Central regions.
SPublic Health Nutrition
Table 2 Energy and nutrient contents (median and 25th–75th percentiles) of ready-to-eat (RTE) cereals according to nutritional quality (NPI
category)-and target-advertising category-
-
, Homescan data, Nielsen, 2008
Energy (kJ/100 g) Energy (kcal/100 g) Sugar (g/100 g) Na (mg/100 g) Fibre (g/100 g)
nMedian 25th–75th Median 25th–75th Median 25th–75th Median 25th–75th Median 25th–75th
Nutritional quality (NPI range)-
Very poor (,40) 46 7004 6883–7234 1674 1645–1705 41?435?5–44?4 180 150–220 3?3 0–3?6
Poor (40–49) 69 6732 6485–7004 1609 1550–1674 32?117?0–25?0 180 150–230 3?73?3–7?1
Fair (50–62) 89 6485 6046–6686 1550 1445–1598 20?717?0–25?0 200 150–230 8?66?7–12?8
Good (.62) 45 6058 5933–6418 1448 1418–1534 13?31?5–19?3 10 0–120 10?99?1–13?6
Target-advertising category-
-
Child-targeted 17 6732 6565–7234 1609 1569–1705 37?532?3–41?4 180 140–180 3?73?4–6?7
Family-targeted 17 6494 6251–6778 1552 1494–1620 20?717?6–30?0 190 140–200 7?46?2–10?2
Adult-targeted 13 6276 5385–6565 1500 1287–1569 19?412?9–24?5 180 120–230 9?46?9–25?0
No advertising 202 6602 6301–7004 1578 1506–1674 23?716?7–33?3 170 130–220 6?73?3–10?9
NPI, Nutrient Profiling Index.
-For definition of NPI, see the Methods section and Rayner et al.
(18)
.
-
-
For definition of target-advertising categories, see the Methods section.
Table 1 Ready-to-eat (RTE) cereal volume index-(median and 25th–75th percentiles) according to nutritional quality (NPI category)-
-
,
Homescan data, Nielsen, 2008y
Nutrient quality (NPI range)
Very poor (,40) Poor (40–49) Fair (50–62) Good (.62) All
Median 25th–75th Median 25th–75th Median 25th–75th Median 25th–75th Median 25th–75th
Children at home
None ,18 years*** 43 38–57 76 47–92 97 80–112 111 93–127 85 56–106
Any ,6 years*** 192 165–216 152 113–204 100 66–162 77 53–132 136 80–186
Any of 6–12 years*** 239 202–279 152 116–224 99 72–133 63 40–126 132 81–206
Any of 13–17 years*** 229 202–262 141 111–199 101 77–127 71 43–125 125 86–196
Household size
1 member*** 27 19–36 45 30–63 68 54–84 90 67–105 56 33–79
2 members*** 50 41–64 84 60–110 109 86–130 127 96–139 94 63–123
3–4 members*** 151 135–158 129 107–144 106 89–128 89 65–110 117 92–142
$5 members*** 282 231–315 209 141–284 113 73–156 87 41–130 151 91–246
Race/ethnicity
Caucasian*** 101 96–106 105 99–110 108 100–111 113 107–118 106 99–112
African American*** 74 53–118 73 51–101 52 35–84 36 21–48 56 37–89
Asian 67 26–103 74 41–101 88 53–147 50 27–94 72 39–113
Hispanic** 122 98–139 100 72–125 97 66–128 74 49–106 98 69–129
Female head education
No high-school graduate** 111 73–161 94 68–122 69 37–99 80 55–102 83 52–114
High-school graduate*** 111 106–121 108 95–121 91 64–104 99 76–121 103 85–116
Some college*** 124 118–135 108 94–122 101 81–120 87 67–101 107 85–123
College graduate*** 103 76–119 117 94–131 129 107–159 121 93–141 118 94–140
Income ($US per annum)
,20 000 80 64–101 72 54–90 66 44–84 68 44–113 71 51–91
20 000–29 999 83 71–106 79 69–89 79 49–93 79 54–105 79 60–97
30 000–39 999* 107 85–131 103 88–116 86 65–105 87 61–109 94 72–114
40 000–49 999 112 95–126 100 83–111 89 74–116 93 74–121 100 82–119
50 000–69 999* 116 105–133 113 100–125 103 86–123 100 70–113 108 92–124
70 000–99 999 109 92–121 109 99–123 113 86–135 107 81–126 111 91–125
$100 000*** 86 55–114 107 85–125 123 103–149 126 87–167 112 86–139
Regionjj
East** 98 58–110 101 81–122 113 83–138 107 81–136 105 78–124
Central** 105 95–125 109 99–124 97 76–114 92 73–117 104 85–121
South 96 81–114 93 84–103 84 68–100 85 73–110 91 74–106
West* 107 85–127 101 78–112 104 79–141 111 80–125 104 79–125
NPI, Nutrient Profiling Index.
-Volume index 5ratio of RTE cereal spending in the sociodemographic group divided by percentage of the sociodemographic group in the US population.
-
-
For definition of NPI, see the Methods section and Rayner et al.
(18)
.
yComparison tests across NPI categories: *P,0?05, **P,0?01, ***P,0?001.
jjCensus definition.
Ready-to-eat cereal purchases and advertising 5
Purchase indicators according to whether the
product was advertised on television and by
target audience
Of the RTE cereals in the final analysis, forty-seven were
advertised on television in 2008. All child-targeted RTE
cereals with television advertising (n17) had an NPI score
in the very poor to poor range (,50). By contrast, eleven
of thirteen adult-targeted advertised cereals and twelve of
seventeen family-targeted cereals exhibited an NPI score
.50 (four and two cereals, respectively, scored .62).
RTE cereals that were not advertised on television in 2008
(n202) had the following distribution by NPI score:
17?4 % were very poor (,40), 27?7 % were poor (40–49),
35?6 % were fair (50–62) and 19?3 % were good (.62).
Nutrient content variations were observed by advertising
category (Table 2). Whereas the nutrient content of
no-advertised products fell between the three other
categories, the highest energy and sugar contents were
observed in child-targeted RTE cereals and the lowest
in the adult-targeted products. The Na content was
comparable across advertising categories.
Purchase indicators varied greatly by advertising target
(Table 4). Compared with cereals not advertised in 2008,
the median number of buyers for advertised child-
targeted RTE cereals was more than thirteen times higher,
buyers for advertised family-targeted products were ten
times higher and those for advertised adult-targeted
cereals were nearly four times higher. Purchase frequency
and dollar share using promotions or coupons were
also higher across the board for advertised compared
with not-advertised RTE cereals. Additionally, advertised
child-targeted RTE cereals exhibited unique purchase
SPublic Health Nutrition
Table 3 Ready-to-eat (RTE) cereal purchase indicators (median and 25th–75th percentiles) according to nutritional quality (NPI category)-,
Homescan data, Nielsen, 2008
Nutrient quality (NPI category)
Very poor (,40) Poor (40–49) Fair (50–62) Good (.62)
(n4–6) (n69) (n89) (n45)
Median 25th–75th Median 25th–75th Median 25th–75th Median 25th–75th
Number of buyers (31000) 2170 867–6093 1627 586–4967 731 333–2239 526 211–2029
Buying rates ($US)-
-
6?04?9–6?95?54?4–7?47?14?7–9?26?85?2–9?8
Purchase frequency (per annum) 1?81?5–2?11?71?4–2?01?71?4–2?11?91?4–2?3
Purchase size ($US) 3?25 3?03–3?69 3?36 3?0–3?75 3?88 3?54–4?26 3?83 3?24–4?21
Purchase cycle (d)y54 44–59 52 45–61 46 38–52 41 36–47
Loyaltyjj 5?34?2–5?95?13?6–7?26?74?4–8?66?95
?3–9?2
% on dealz18?46?9–32?329?416?6–35?629?220?4–36?620?911?8–29?8
% with coupons-- 2?50?7–7?24?01?8–10?46?11?3–11?31?70?4–5?4
NPI, Nutrient Profiling Index.
-For definition of NPI, see the Methods section and Rayner et al.
(18)
.
-
-
Mean annual spending per household on the cereal (in $US).
yAverage number of days between two purchases.
jjAverage share of spending for a given RTE cereal as a percentage of all spending for RTE cereals during the year by households.
zShare of total spending when the product was on promotion at the store.
--Share of total spending using coupons supplied by manufacturers.
Table 4 Ready-to-eat cereal (RTE) purchase indicators (median and 25th–75th percentiles) according to target-advertising category-,
Homescan data, Nielsen, 2008
Target-advertising category
Child-targeted Family-targeted Adult-targeted No advertising
(n17) (n17) (n13) (n202)
Median 25th–75th Median 25th–75th Median 25th–75th Median 25th–75th
Number of buyers ($1000) 11 165 8684–19 059 8646 1783–16 627 3354 2033–5779 869 331–2236
Buying rates ($US)-
-
6?45?2–7?97?75?6–10?17?95?8–9?86?34?6–8?2
Purchase frequency (per annum) 2?01?8–2?32?11?7–2?32?01?6–2?31?71?4–2?1
Purchase size ($US) 3?23?0–3?53?93?6–4?14?13?6–4?33?63?1–4?0
Purchase cycle (d)y60 54–63 49 44–56 44 42–47 46 38–54
Loyaltyjj 5?54?3–6?57?85?1–8?67?85?2–9?85?44
?0–7?6
% on dealz35?931?9–38?935?828?6–40?334?329?0–40?322?412?2–32?2
% with coupons-- 9?24?1–11?84?53?7–12?99?06?4–19?22?80?8–7?8
-For definition of target-advertising categories, see the Methods section.
-
-
Mean annual spending per household on the cereal (in $US).
yAverage number of days between two purchases.
jjAverage share of spending for a given RTE cereal as a percentage of all spending for RTE cereals during the year by households.
zShare of total spending when the product was on promotion at the store.
--Share of total spending using coupons supplied by manufacturers.
6 K Castetbon et al.
behaviours: their median purchase size was the lowest and
median purchase cycle was the highest of the advertised
categories. Finally, median loyalty for RTE advertised
cereals targeted to children was comparable to that of
not-advertised cereals, while advertised family- and adult-
targeted RTE cereals showed the highest loyalty.
In several population subgroups, RTE cereal volume
indices also varied according to whether products were
advertised or not and across target-advertising categories.
As would be expected, in households with at least
one child (Fig. 1a), volume indices were higher for child-
targeted advertised RTE cereals than for advertised cereals
targeted to families and adults (P,0?001). In contrast,
households without children purchased family- and
adult-targeted advertised RTE cereals more often (Fig. 1b).
Variations according to the number of household members
followed the same patterns (data not shown). Additionally,
Caucasian households (Fig. 2a) were less likely to purchase
child-targeted RTE cereals compared with family- and adult-
targeted RTE cereals (P,0?001), while African-American
households exhibited the opposite pattern (P,0?001;
Fig. 2b). Finally, in households in which the female head
of household had some college (P50?047), the volume
index (median (25th–75th percentile)) was higher for
advertised child-targeted RTE cereals (123 (116–128))
than for family-targeted (104 (96–117)), adult-targeted (90
(82–101)) and not-advertised RTE cereals (106 (84–124)).
For households in the remaining sociodemographic
categories, the volume index of RTE cereals did not
vary significantly according to whether the product was
advertised or the advertising target.
Discussion
Analyses of various aggregated indicators of RTE cereal
purchases highlight interesting variations across socio-
demographic categories in US households. RTE cereal
purchases varied according to nutritional value, television
advertising and target market. Nevertheless, the extent of
such variations was not the same across household
categories. Overall, these results illustrate the need for
improvement in the nutritional content of RTE cereals
advertised on television and targeted to children.
To our knowledge, such a detailed description of RTE
cereal-buying patterns across sociodemographic categories
has not been previously published. One study conducted in
1996 in Canada
(21)
showedthat,overall,RTEcerealswere
bought by fewer than half of households, and purchased
moreoftenwithincreasingeducationandincomeandwith
the presence of youth aged ,15 years at the dwelling. Our
observations based on purchase volume index are similar to
previous studies of US individual intake data. For adults
(22)
and children
(5)
, consumption of RTE cereals was lower in
African-American, Hispanic and low socio-economic status
households (based on education for adults and poverty-to-
income ratio for children in these studies) compared with
non-Hispanic white and high socio-economic status house-
holds. Lower purchases of RTE cereals could be markers of
low breakfast intake in these subgroups of the population as
RTE cereals are commonly consumed at breakfast. Given the
consequences of skipping breakfast on health
(1)
, this can be
considered a public health issue for which surveillance of
individual diet behaviours is necessary.
In the present study of 2008 purchases, US households
with at least one child bought relatively more RTE cereals,
which would be expected as many RTE cereals are
marketed as ‘kid products’. Nevertheless, households
with at least one child exhibited particularly elevated
purchases of RTE cereals of poor nutritional quality. This
can be related to the fact that while product advertising
was closely associated with buying patterns (purchases of
child-targeted advertised products were thirteen times
higher than not-advertised products), child-targeted pro-
ducts were also those with the poorest nutritional content
(i.e. in 2008 all child-targeted advertised products had
SPublic Health Nutrition
0
20
40
60
80
100
120
140
160
180
200
220
240
Child-
targeted
Family-
targeted
Adult-
targeted
No
advertising
Advertising category
Volume index
0
20
40
60
80
100
120
140
160
180
200
220
240
Child-
targeted
Family-
targeted
Adult-
targeted
No
advertising
Advertising category
Volume index
(a)
(b)
Fig. 1 Ready-to-eat (RTE) cereal volume index according to
target-advertising category in: (a) households including at least
one child; and (b) households without children (Homescan data,
Nielsen, 2008). Volume index 5ratio of RTE cereal spending
in the sociodemographic group divided by percentage of the
sociodemographic group in the US population; for definition of
target-advertising categories, see the Methods section
Ready-to-eat cereal purchases and advertising 7
an NPI ,50). Our observations together with previous
analyses
(23)
underline the potential impact of television
advertising on food-buying behaviours, especially for
advertising directed at children. Nevertheless, indicators
analysed here such as purchase frequency, time between
purchases and resulting loyalty suggest that buying beha-
viours for healthier products could increase with increased
advertising (adult-targeted products were bought four
times as often as products not advertised in 2008) and
potentially the use of promotions and coupons
(24)
.
Sociodemographic characteristics were also associated
with RTE cereal-buying patterns, in particular household
race/ethnicity and education of the female head of house-
hold. RTE cereal purchases made by white households were
rather homogeneous for all products (based on volume
index medians and 25th–75th percentiles). African-American
and Hispanic households under-purchased RTE cereals
overall, but they purchased more of the cereals with poorest
nutritional quality, especially the child-targeted advertised
RTE cereals. Similar buying patterns were found with
lower female head-of-household education, lower income
and southern region, and indicates that sociodemographic
groups for whom RTE cereal consumption is not the
norm in adults
(22)
maybemoresusceptibletoadvertising
directed towards their children and adolescents. Again,
these observations support the need to promote healthier
RTE cereals among population subgroups in which advertis-
ing seems to have a direct impact.
The present study has some limitations. First, we chose
to use a standardized nutrition score so that we could
classify products within the RTE cereal group according
to their nutritional quality. This does not tell us how
consuming these cereals influenced the entire diet over
a day
(18)
. Second, the requirements for participating in
the Homescan panel might produce biased estimations
despite the calibration of data on the census
(17)
. House-
holds must regularly register and transmit information
about their grocery store purchases which can be a time-
consuming process. Therefore, retired individuals may be
over-represented among one-member households relative
to young single adults. This potential self-selection bias
could explain why one-member households purchased
more RTE cereals of good nutritional quality. Further
research could examine the buying behaviours of young
SPublic Health Nutrition
0
20
40
60
80
100
120
140
160
180
200
220
240
Advertising category
Volume index
0
20
40
60
80
100
120
140
160
180
200
220
240
Advertising category
Volume index
0
20
40
60
80
100
120
140
160
180
200
220
240
Advertising category
Volume index
0
20
40
60
80
100
120
140
160
180
200
220
240
Advertising category
Volume index
Child-
targeted
Family-
targeted
Adult-
targeted
No
advertising
Child-
targeted
Family-
targeted
Adult-
targeted
No
advertising
Child-
targeted
Family-
targeted
Adult-
targeted
No
advertising
Child-
targeted
Family-
targeted
Adult-
targeted
No
advertising
(a) (b)
(c) (d)
Fig. 2 Ready-to-eat (RTE) cereal volume index according to target-advertising category in various race/ethnic groups: (a)
Caucasian; (b) African American; (c) Asian; and (d) Hispanic (Homescan data, Nielsen, 2008). Volume index 5ratio of RTE cereal
spending in the sociodemographic group divided by percentage of the sociodemographic group in the US population; for definition
of target-advertising categories, see the Methods section
8 K Castetbon et al.
single adults to evaluate whether they continue to pur-
chase the less-healthy cereals consumed when they were
younger and living with their parents. Third, these
aggregated data do not provide information about indi-
vidual household purchasing patterns nor individual
intake within households, especially by gender and age.
We assume that a product targeted to children purchased
by households with children is predominantly eaten
by children; however, parents may also consume these
cereals. Indeed, as previously mentioned, such house-
holds did not also purchase adult- and family-targeted
RTE cereals in higher quantities compared with other
households. Finally, our results are based on a cross-
sectional survey; therefore, we cannot make conclusions
about the direct impact of advertising on behaviours.
Experimental studies, such as naturalistic trials in ‘real
contexts’, are necessary to confirm a causal relationship.
In spite of these limitations, only these kinds of economic
databases enable researchers to attribute a precise nutri-
tional value to each product and to describe purchases
using various complementary indicators.
Conclusions
RTE cereal purchases are an interesting example to
understand the relationship between advertising, nutri-
tional quality and food purchase behaviours. Indeed, RTE
cereal advertising is widely targeted to children, but often
for products with nutritional quality that is far from
optimal. A great variety of products exists in the market
and advertising for high-quality products should be
encouraged given its probable effect on purchases. This is
relevant to improve diet, nutritional status and health
in the population categories which seem particularly
susceptible to food advertising according to our results
(e.g. households with children, African-American and
Hispanic households). Changes in food composition
toward lower contents of sugar and Na and higher fibre
content would be a complementary public health action.
The NPI model could be used to identify new product
formulations that would improve the overall nutritional
quality of different cereals. In addition, cost variance
analyses would help identify strategies to increase pur-
chases of healthy RTE cereals, which tend to be priced
higher than the less-healthy cereals. It would also be
interesting to conduct equivalent analyses for other
manufactured products containing sizeable amounts of
sugar, salt and/or fat.
Acknowledgements
The research was funded by the Robert Wood Johnson
Foundation and the Rudd Foundation. The authors declare
no conflict of interest. K.C. was a visiting researcher at
Rudd Center for Food Policy and Obesity at the time of
the study; she conceived the analysis design, carried out
analyses, interpreted results and drafted the manuscript.
J.L.H. contributed to the analysis design conception,
acquisition of data and interpretation of results, and
revised the manuscript. M.B.S. contributed to interpreta-
tion of results and revision of the paper. All authors
approved the final version of the paper for publication.
The authors are grateful to Vishnudas Sarda (Rudd Center)
for his help on data management.
References
1. Rampersaud GC, Pereira MA, Girard BL et al. (2005)
Breakfast habits, nutritional status, body weight, and
academic performance in children and adolescents. JAm
Diet Assoc 105, 743–760.
2. Timlin MT, Pereira MA, Story M et al. (2008) Breakfast
eating and weight change in a 5-year prospective analysis
of adolescents: Project EAT (Eating Among Teens).
Pediatrics 121, e638–e645.
3. Alexy U, Wicher M & Kersting M (2010) Breakfast trends in
children and adolescents: frequency and quality. Public
Health Nutr 13, 1795–1802.
4. Siega-Riz AM, Popkin BM & Carson T (1998) Trends in
breakfast consumption for children in the United States
from 1965–1991. Am J Clin Nutr 67, issue 4, 748S–756S.
5. Deshmukh-Taskar PR, Nicklas TA, O’Neil CE et al. (2010)
The relationship of breakfast skipping and type of break-
fast consumption with nutrient intake and weight status in
children and adolescents: the National Health and Nutrition
Examination Survey 1999–2006. J Am Diet Assoc 110,
869–878.
6. US Department of Agriculture (2010) MyPyramid for Kids
Web site. http://www.mypyramid.gov (accessed January
2011).
7. Albertson AM, Affenito SG, Bauserman R et al. (2009) The
relationship of ready-to-eat cereal consumption to nutrient
intake, blood lipids, and body mass index of children as they
age through adolescence. JAmDietAssoc109, 1557–1565.
8. Cho S, Dietrich M, Brown CJ et al. (2003) The effect of
breakfast type on total daily energy intake and body mass
index: results from the Third National Health and Nutrition
Examination Survey (NHANES III). J Am Coll Nutr 22,
296–302.
9. Rosado JL, del RA, Montemayor K et al. (2008) An
increase of cereal intake as an approach to weight reduction
in children is effective only when accompanied by nutrition
education: a randomized controlled trial. Nutr J 7,28.
10. Harris JL, Schwartz MB, Brownell KD et al. (2009) Cereal
FACTS: Evaluating the Nutrition Quality and Marketing of
Children’s Cereals. New Haven, CT: Rudd Center for Food
Policy & Obesity, Yale University; available at http://
www.cerealfacts.org/media/Cereal_FACTS_Report.pdf
11. Schwartz MB, Vartanian LR, Wharton CM et al. (2008)
Examining the nutritional quality of breakfast cereals
marketed to children. J Am Diet Assoc 108, 702–705.
12. Schwartz MB, Ross C, Harris JL et al. (2010) Breakfast cereal
industry pledges to self-regulate advertising to youth: will
they improve the marketing landscape? J Public Health
Policy 31, 59–73.
13. Powell LM, Szczypka G & Chaloupka FJ (2010) Trends in
exposure to television food advertisements among children
and adolescents in the United States. Arch Pediatr Adolesc
Med 164, 794–802.
14. World Health Organization (2006) Marketing of Food and
Non-alcoholic Beverages to Children. Report of a WHO
Forum and Technical Meeting, Oslo, Norway, 2–5 May 2006.
SPublic Health Nutrition
Ready-to-eat cereal purchases and advertising 9
Geneva: WHO; available at http://www.who.int/diet
physicalactivity/publications/Oslo%20meeting%20layout
%2027%20NOVEMBER.pdf
15. Larson NI, Story MT & Nelson MC (2009) Neighborhood
environments: disparities in access to healthy foods in the
US. Am J Prev Med 36, 74–81.
16. SlimaniN,DeharvengG,UnwinIet al. (2007) The EPIC
nutrient database project (ENDB): a first attempt to standardize
nutrient databases across the 10 European countries partici-
pating in the EPIC study. Eur J Clin Nutr 61, 1037–1056.
17. Einav L, Leibtag E & Nevo A (2008) On the Accuracy of
Nielsen Homescan Data. Economic Research Report no. 69.
Washington, DC: US Department of Agriculture, Economic
Research Service; available at http://www.ers.usda.gov/
Publications/ERR69/ERR69.pdf
18. Rayner M, Scarborough P, Lobstein T (2009) The UK Ofcom
Nutrient Profiling Model. http://www.dphpc.ox.ac.uk/bhfhprg/
publicationsandreports/publications/bhfhprgpublished/
nutrientprofilemodel (accessed January 2011).
19. Arambepola C, Scarborough P & Rayner M (2008)
Validating a nutrient profile model. Public Health Nutr
11, 371–378.
20. Scarborough P, Arambepola C, Kaur A et al. (2010) Should
nutrient profile models be ‘category specific’ or ‘across-the-
board’? A comparison of the two systems using diets of
British adults. Eur J Clin Nutr 64, 553–560.
21. Ricciuto L, Tarasuk V & Yatchew A (2006) Socio-demographic
influences on food purchasing among Canadian house-
holds. Eur J Clin Nutr 60, 778–790.
22. Siega-Riz AM, Popkin BM & Carson T (2000) Differences in
food patterns at breakfast by sociodemographic characteristics
among a nationally representative sample of adults in the
United States. Prev Med 30, 415–424.
23. Story M & French S (2004) Food advertising and marketing
directed at children and sdolescents in the US. Int J Behav
Nutr Phys Act 1,3.
24. Hawkes C (2009) Sales promotions and food consumption.
Nutr Rev 67, 333–342.
SPublic Health Nutrition
10 K Castetbon et al.