Vous êtes sur la page 1sur 10

International Journal of Food Sciences and Nutrition,

September 2012; 63(6): 679–688

Attentional capture and understanding of nutrition labelling: a study


based on response times

GASTÓN ARES1,2, ANA GIMÉNEZ1,2, FERNANDA BRUZZONE1,2, LUCÍA ANTÚNEZ1,


ALEJANDRA SAPOLINSKI1, LETICIA VIDAL1, & ALEJANDRO MAICHE2
1
Sección Evaluación Sensorial, Departamento de Ciencia y Tecnologı́a de Alimentos, Facultad de Quı́mica, Universidad de la
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

República (UdelaR), Gral Flores 2124, CP 11800, Montevideo, Uruguay, and 2Facultad de Psicologı́a, Centro de Investigación
Básica en Psicologı́a (CIBPsi), Universidad de la República (UdelaR), Tristán Narvaja 1674, Montevideo, Uruguay

Abstract
The aim of the present work was to assess the influence of nutrition information format on attentional capture and consumers’
understanding. Sixteen labels of two products (yoghurt and pan bread) were designed following a four 2-level factors full
factorial design with the following variables: label background design, type of product, nutrition information format and traffic
light system. The labels were presented to 178 consumers, who were asked to decide whether the fat/sodium content of each
yoghurt/pan bread label was medium or low. Participant responses and reaction times were recorded. Results showed that type of
product, nutrition information format and traffic light system significantly affected the time needed by consumers to find the
For personal use only.

nutrition information and to classify the labels according to their content of a given nutrient. Meanwhile, consumers’
understanding of the labels was mostly affected by the content of the nutrient and the presence of the traffic light system.

Keywords: nutrition labelling, consumers’ research, reaction times, traffic light system

Introduction
Different strategies have been implemented worldwide According to the results from a qualitative study
to encourage people to make healthier choices, carried out in the European Union, the major
nutrition labelling being one of them. Considering criticisms to nutrition labelling were related to the
the increase in the consumption of pre-packaged foods difficulty in finding key information on the label
(Caraher et al. 1999), nutritional information on food (Directorate General for Health and Consumer
labels aims to provide information about the nutrient Protection 2005). Therefore, one of the major
content of products to help consumers make healthier challenges is to design nutrition labels which could
choices (Cowburn and Stockley 2005). be easily read and understood and which rapidly catch
The success of nutrition labelling depends on the consumers’ attention. For this reason, studying
ability of consumers to detect, read and understand attentional capture is a key step for the development
nutritional information (Mackison et al. 2010). of effective ways of communicating nutrition infor-
Despite the growing inclusion of nutrition information mation. In this context, the present work aims to study
on food labels some authors have reported that many how nutrition labelling format affects attentional
people do not often use nutrition information, or that, capture.
despite consumers report reading nutrition labelling, Solomon et al. (2002) defined attention as the
their actual use is considerably lower (Grunert and degree to which consumers focus on a stimuli within
Wills 2007). their range of exposure. According to Pieters and

Correspondence: G. Ares, Sección Evaluación Sensorial, Departamento de Ciencia y Tecnologı́a de Alimentos, Facultad de Quı́mica,
Universidad de la República (UdelaR), Gral Flores 2124, CP 11800 Montevideo, Uruguay. Tel: þ 598 2 9245735. Fax: þ 598 2 9241906.
E-mail: gares@fq.edu.uy

ISSN 0963-7486 print/ISSN 1465-3478 online q 2012 Informa UK, Ltd.


DOI: 10.3109/09637486.2011.652598
680 G. Ares et al.

Wedel (2004), consumers’ attention towards a The fact that nutrition information receives atten-
stimulus depends on the characteristics of the stimulus tion does not necessarily imply that consumers process
itself (bottom-up factors) and also on the expectations the information (van Herpen and van Trijp 2011).
about the product and consumers previous experi- After consumers perceive nutrition information, they
ences (top-down factors). The characteristics of the must understand it in order to make inferences about
stimulus may determine rapid and automatic atten- the content of a certain nutrient or the healthiness of
tional capture even when the consumer is not the product as a whole. In this sense, some authors
searching for it (Wolfe 1998). This type of attentional have stated that consumers do not usually know which
capture depends on the size, colour and the pop- nutrients are important (Hurt 2002) or find it difficult
up characteristics of the element from the background to determine if the concentration of a specific nutrient
in which it is included (Pieters and Wedel 2004). In the is low, medium or high (Black and Rayner 1992).
case of nutrition labelling, the format under which Carrillo et al. (2011) reported that nutritional knowl-
the information is displayed could have a large edge has a great influence on consumers’ perception
influence on consumers’ attention. and understanding of food labelling. Consumers with
From experimental studies on attentional capture, low nutritional knowledge considered the usual
it is known that attentional processes driven by nutritional labels too technical and reported to rarely
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

bottom-up factors, such as size and shape, determine take them into account when deciding their food
perceptual salience in advertisements (Janiszewski purchases.
1998). From this, it could be expected that According to several studies, the use of coloured
organization of the information plays a crucial factor systems such as the traffic light system or Guideline
in consumers’ attentional capture. Tables are usually Daily Amounts (GDA)-based systems makes the
considered the best organizers of information and interpretation of nutrition information easier due to
usually they simplify even the most complex layouts, a decrease in information processing (Grunert and
being one of the most recommended features in Wills 2007). In line with this, van Herpen and van
documents when information needs to be clearly Trijp (2011) reported that the use of traffic light
presented (Microsoft at work 2011). Usually, subjects system enhances healthy product choice, which
For personal use only.

can understand the meaning of a scene in a few glances suggests that this system helps consumers to interpret
(Hollingworth and Henderson 1998, 1999), whereas the provided information.
text requires more eye fixations to be comprehended. When evaluating the healthiness of a food product
Thus, it could be assumed that more attention per unit consumers might not only take into account the
surface is required for text (linear format) than for nutrition information, since food labels communicate
scene perception (panel format). So, if panel and information using different types of signs which
linear formats are compared, it could be expected that help consumers to identify and categorize the product
panel format might enable consumers to more easily (Opperud 2004; Smith et al. 2010). In particular, the
find nutrition information. images used in the graphic design of the labels have a
In an attempt to increase consumers’ attention to great impact on consumers’ expectations and percep-
nutrition labelling, the provision of front-of-pack tion of food products (Ares et al. 2010), which could
nutrition labelling, in addition to complete nutrition even override the effect of nutrition information.
information, is being used progressively more (Cow- Furthermore, the graphic design of the labels could
burn and Stockley 2005). The traffic light system is an also affect the ability of nutrition information to catch
example of this trend. The nutritional signpost consumers’ attention by affecting its pop-up from the
proposed by the Food Standards Agency summarizes background.
key information by rating four nutrients (sugars, fat, Thus, consumers’ attention and understanding of
saturated fat and salt) as high, medium or low, using nutrition labelling determine, together with other
the traffic light scheme (red amber and green) (Food personal characteristics, whether the consumer will
Standards Agency 2007). The use of this simpler take into account nutritional information when buying
front-of-pack information could increase consumers’ the product or not. Thus, reducing the difficulty of
attention towards nutrition information and make information search and understanding for consumers
the access to information easier, which could could make it more likely that the information
lead consumers to further incorporate nutrition provided will be actually seen and used by the
information in their decision process (Jones and consumers when making their everyday food
Richardson 2007). Consumers might need to look at purchases.
the front-of-pack information for a shorter period of Most studies are based on self-reported retro-
time compared to detailed nutrition labelling in order spective behaviour, which can lead to considerable
to get the information provided and be able to over-reporting with regard to behaviours that are
compare products. According to van Herpen and van considered to be socially desirable (Podsakoff et al.
Trijp (2011), nutrition tables receive less attention 2003). In this context, Bialkova and van Trijp (2010)
than traffic light labels, particularly when consumers used a visual search procedure to evaluate the
face time constraints. attentional capture of front-of-pack nutritional
Attentional capture and understanding of nutrition labelling: a study based on response times 681

information. These authors asked participants to groups of similar size were considered: 54% of the
detect whether a nutrition label was present or not participants between 18 and 34 years old and 46%
and reported that attentional capture was faster and older than 35 years. All consumers had normal or
more accurate when front-of-pack labels were mono- corrected-to-normal vision and full colour vision.
chromatic rather than polychromatic. The main Consumers were recruited from public places, based
advantage of this approach is that it provides more on their interest and availability to participate in the
reliable information than other commonly used study and their frequency consumption of yoghurt and
methodologies such as self-report measures. However, pan bread (at least once a week).
it only focuses on consumers’ attention but it tells
nothing about the access to information by the
consumers. Considering that efforts in interpreting
nutrition information might also determine consu- Stimuli
mers’ willingness to search for it in real-life situations,
the second part of the process also deserves to be Two target products were considered: yoghurt and pan
studied. bread. These products were selected considering that
The aim of the present study was to evaluate the they are frequently consumed in Uruguay and that
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

influence of the format of nutrition labels on they differ in their perceived healthiness (Ares et al.
consumers’ attention and processing of nutrition 2008). For each product, consumers’ attention and
information, and to study whether this influence understanding were focussed on the content of a
depends on the graphic design of the labels, using a specific nutrient: fat for yoghurt and sodium for pan
methodological approach based on response times bread.
(RTs). For each product, 16 labels were designed following
a four 2-level factors full factorial experimental design.
Examples of two labels for each product are shown in
Materials and methods Figure 1.
In order to study the influence of nutrition
Participants
information format on consumers’ attention and
For personal use only.

One-hundred and seventy-eight people participated in understanding, the following independent variables
the study. Participants were 58% females and 42% were considered: label background (Background A vs.
males and ranged in age from 18 to 70. In order to Background B), type of product (regular vs. low fat
work with a consumer sample balanced in age, two for yoghurt and regular vs. low salt for pan bread),

Figure 1. Examples of two labels used in the study: (a) Backgrounds A and B for yoghurt, (b) Backgrounds A and B for pan bread.
682 G. Ares et al.

nutrition information format (panel vs. linear) and Procedure


traffic light system (with vs. without).
Participants completed two tasks: one for yoghurt and
Two products were considered in the present study
the other for pan bread. Half of the subjects completed
to investigate whether consumers were able to under-
the task for pan bread first, whereas the other half
stand when the fat/sodium content could be con-
performed the task for yoghurt first. Both tasks were
sidered low, particularly considering the increasing
separated by a 5-min break.
number of low-fat and low-salt products commercia-
On each task consumers were asked to classify the
lized in Uruguay. In order to avoid repetition of the
nutrient content of the label in ‘medium’ or ‘low’.
same nutrition information on the labels, small
They evaluated a total of 18 labels, the 16 labels of the
modifications were introduced for labels correspond-
experimental design plus 2 dummy labels. These latter
ing to the same type of product (regular vs. low fat).
labels were presented at the beginning of the task in
The sodium content of the pan bread labels ranged
order to familiarize participants with the methodology
from 23 to 76 mg per portion for low sodium products
and to show them the two ranges of each nutrient.
and from 195 to 276 mg per portion for regular
Thus, although consumers were not specifically
products. Meanwhile, the fat content of the yoghurt
explained what the terms ‘medium’ and ‘low’ meant,
labels ranged from 0.9 to 1.5 g per portion for low fat
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

when completing the trials they were aware of the


products and from 7.5 to 8.4 g per portion for regular
categories. This procedure avoided the increase in
yoghurts. These concentrations were selected accord-
correct responses by the familiarization with the
ing to the technical guidance proposed by the Food
categories during the trials.
Standards Agency (2007).
The labels were presented on a PC screen following
The levels of the variable nutrition information
complete block experimental design (William’s Latin
format were selected considering that in Uruguay
Square) using DirectRT software (Jarvis 2004). After
compulsory nutrition information (energy, total
seeing the labels consumers were asked to decide
carbohydrates, proteins, total fat, saturated fat, trans
whether the fat content of each yoghurt (or the sodium
fat, total fibre and sodium) could be displayed on food
content of each pan bread) was medium or low, by
labels under two formats: panel (a table) or linear (a
pressing the M (for ‘medio’, medium in Spanish) or B
For personal use only.

paragraph) (Ministerio de Salud Pública 2006). An


(for ‘bajo’, low in Spanish) key, respectively. Partici-
example of these formats is shown in the yoghurt labels
pants’ responses and RT were recorded using the
of Figure 1a. Moreover, considering that the front-of-
software. Participants were asked to respond as fast
pack information is not being used by food companies
and accurately as possible. For all trials, feedback
yet in Uruguay, it was interesting to study whether the
asking for faster responding was given if participants
inclusion of the traffic light system enhances con-
did not respond within 60 s, whereas consumers were
sumers’ attention and understanding of nutrition
shown a message on the screen asking them to take
information.
more time if they responded in less than 0.5 s. This last
Finally, in order to study whether consumers’
message was used to discourage participants to
attention and understanding of nutrition information
complete the task without taking enough time to find
depend on the graphic design of the label, two
the information on the label. It can be assumed that
backgrounds were considered for each product. The
the arrival of sensory information and motor answer
backgrounds were selected based on the results of
takes at least 0.2 s; therefore, if subjects take less than
previous qualitative consumer studies in order to have
0.5 s to answer, they are probably answering without
one background associated with a healthy product
processing the information.
(Background A) and one associated with different
sensory characteristics but not with health (Back-
ground B). The name of the brands on each label was Data analysis
selected to match consumers’ associations with the
Response times. The following analysis of variance
graphic design of the labels. The brand Vitalize was
(ANOVA) model was used to evaluate the influence of
used for the healthy yoghurt, whereas Láctico (lactic in
the experimental variables on consumers’ RTs for each
English) was selected for the yoghurt associated with
type of product:
sensory characteristics such as acidity.
Labels were designed using GIMP 2.6. The labels
included all the information that is compulsory for Response time ðRTÞ ¼ mean þ main effect for the
food labels in Uruguay. The influence of consumers’
knowledge of the nutrient content of their usual consumer ðset as randomÞ þ main effect for the
products was avoided by designing labels that were not
similar to products currently available in the Urugua- conjoint variables ðset as fixedÞ þ interaction effects
yan market. Moreover, the location of nutrition
information format and traffic light system was among the conjoint variables ðset as fixedÞ
different for each type of label to avoid consumers’
familiarization. þ random error:
Attentional capture and understanding of nutrition labelling: a study based on response times 683

When the effects were significant, honestly significant Results


differences were calculated using Tukey’s test at a
Response times
significance level of 5%.
Consumers’ ability to find and interpret nutrition
information was evaluated through the use of RTs. As
Percentage of correct answers. A logistic regression
given in Table I, the time needed by consumers to find
analysis was used to analyse and interpret the
nutrition information and classify yoghurt labels
percentage of correct answers of consumers’
according to their fat content significantly depended
classification of the evaluated food labels. This
on the type of product, the format under which
analysis enables evaluating the influence of many
nutrition information was displayed and the presence
different independent variables on dependent
of the traffic light system. Meanwhile, in the case of
variables that are binary (Freeman 1987). In this
pan bread labels RTs depended on the type of product,
case, logistic regression was used to study the influence
nutrition information format, traffic light system and
of the experimental variables (label background, type
the interaction between these two latter variables. It is
of product, nutrition labelling format, traffic light
interesting to highlight that the label background did
system and their interaction) on the probability of
not significantly affect consumers’ speed to find the
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

consumers’ correctly classifying the evaluated labels as


information and to classify the labels for any of the
medium or low in a given nutrient.
products considered (Table I).
Consumers’ responses were classified using a binary
As shown in Figure 2, RTs were shorter when
coding system; true if the participant correctly
consumers had to classify the salt content of pan bread
classified the label, and false otherwise.
labels than when they had to classify the fat content of
The model was specified as follows:
yoghurt labels.
Furthermore, for both yoghurt and pan bread, the
Probability of giving a correct answer classification of low fat/sodium products resulted in
higher RTs than regular products, as shown in
¼ a0 þ a1 label background Figure 2.
For personal use only.

Figure 3 shows the influence of nutrition infor-


þ a2 type of product mation and traffic light system on RTs. RTs were
shorter when nutrition information was displayed
þ a3 nutrition information format
using panel format and when the traffic light system
þ a4 traffic light system þ a5 label background was used. For yoghurt labels these variables did not
show a significant interaction, as given in Table I,
£ type of product þ a6 label background which indicates that the influence of traffic light system
on consumers’ attentional capture and information
£ nutrition information format processing did not depend on the format under which
the information was displayed. When the traffic light
þ a7 label background £ traffic light system system was incorporated to the labels, consumers
needed approximately 4 s less time to classify the fat
þ a8 nutrition information format content of yoghurt labels than when it was not present.
As for yoghurt labels, RTs for the classification of
£ traffic light system
pan bread labels were shorter when nutrition
þ a9 nutrition information format information was displayed under panel format and
when the traffic light system was used. However, there
£ type of product þ a10 type of product was a significant interaction between these two
variables (Table I). As shown in Figure 3, the influence
£ traffic light system: of the traffic light system was considerably larger when
nutrition information was displayed under linear
format than when panel format was used. RTs were
The statistical significance of the independent vari- reduced by 5 s approximately when the traffic light
ables was confirmed using the Chi-square statistic. system was included on pan bread labels that had the
A significance level of 5% was used as the criterion for nutrition information displayed under linear format,
statistical significance. whereas the reduction in RTs was only some
milliseconds more than 1.5 s when panel format was
considered.
Linear regression. A linear regression was carried out
between average RTs and the percentage of correct
Percentage of correct responses
responses for all the evaluated labels.
All statistical analyses were carried out using Consumers showed a relatively high understanding of
R language (R Development Core Team 2007). the information presented on the labels. They were
684 G. Ares et al.

Table I. Results of the ANOVA carried out on RTs for yoghurt and pan bread labels.

Yoghurt Pan bread

Variable F p-Value F p-Value

Label background 2.76 0.0967ns 0.04 0.8346ns


Type of product 4.49 0.0342* 10.75 0.0011**
Nutrition information format 46.9 ,0.0001*** 168.1 ,0.0001***
Traffic light system 112.9 ,0.0001*** 201.7 ,0.0001***
Label background £ type of product 0.48 0.4889ns 0.07 0.4650ns
Label background £ nutrition information format 0.52 0.4692ns 0.53 0.4650ns
Label background £ traffic light system 0.72 0.3964ns 0.04 0.8463ns
Type of product £ nutrition information format 1.15 0.2834ns 0.04 0.8485ns
Type of product £ traffic light system 0.0008 0.9771ns 1.03 0.3092ns
Nutrition information format £ traffic light system 1.06 0.3043ns 43.75 ,0.0001***

Note: Whereas ns indicates no significant effect for a confidence level of 95%; * Indicates significant effect for a confidence level of 95%;
** Indicates significant effect for a confidence level of 99%; *** Indicates significant effect for a confidence level of 99.9%
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

able to correctly classify 78.1% of the yoghurt labels content when the information was displayed under
according to their fat content, whereas they correctly linear format than when panel format was used.
classified 80.0% of the pan bread labels according to Regarding the influence of the traffic light system on
their sodium content. Although the percentage of consumers’ understanding of sodium content, as
correct answers was similar for both the products, the shown in Figure 5b it clearly depended on whether
influence of the evaluated variables on the labels nutrition information was presented under panel or
clearly differed (Table II). linear format. The traffic light system significantly
In the case of yoghurt labels, the percentage of increased the percentage of correct classifications
correct responses was significantly affected by the label regardless of how nutrition information was displayed.
background and the interaction between the label However, the increase in the percentage of correct
For personal use only.

background and the type of product and between the answers depended on the nutrition information
format, being higher for the linear format.
type of product and the traffic light system (Table II).
As shown in Figure 4a, the percentage of correct
responses was higher for Background B when regular
yoghurts were considered, whereas it was lower for Relationship between RTs and percentage of correct
low-fat yoghurts. responses
The influence of the traffic light system clearly RTs were significantly ( p ¼ 0.0198) negatively corre-
depended on the type of product considered. For low- lated with the percentage of correct answers for the
fat yoghurts, the percentage of correct responses classification of yoghurt and pan bread labels accord-
almost did not change when the traffic light system was ing to their nutrient content. Longer RTs tended to
used, whereas this signpost had a clear influence for correspond to a lower percentage of correct answers.
regular yoghurts. As shown in Figure 4b, the However, the correlation between these dependent
percentage of correct classifications for regular variables was low (2 0.41) and explained only 17% of
yoghurts increased from 77.3% to 85.0% when the the variance of the experimental data.
traffic light system was used. In the case of pan bread Coincidentally, when consumers incorrectly classi-
labels, results were markedly different from those of fied the labels their RT was longer than when they
yoghurt labels. As given in Table II, the percentage of
correct classifications of pan bread labels was 15000
significantly affected by the type of product, nutrition
information format, the presence of traffic light system 12153.7 a 12942.8 b
Response time (ms)

and the interactions between the type of product and


the nutrition information format and between the Yogurt
10000
nutrition information format and the traffic light Pan bread
system. As shown in Figure 5a, the percentage of 8110.7 b
consumers who correctly classified the sodium content 7213.9 a
of the pan bread labels was lower when it was low with
respect to the regular products. Meanwhile, nutrition 5000
Regular Low
information format only affected the percentage of
Type of product
correct responses when low sodium products were
considered. In this case, consumers correctly classified Figure 2. RT for the classification of yoghurt and pan bread labels
a larger number of labels according to their sodium as a function of their fat/salt content.
Attentional capture and understanding of nutrition labelling: a study based on response times 685

(a) 20000
on RTs was the same for correct and incorrect
responses.
Response time (ms) Linear
Panel 15994.2 d
15000
13060.3 c Discussion and conclusions
11653.5 b
10000 Methodological issues
9485.0 a
Nutrition labelling is one commonly used approach
that aims to help people to make healthier food choices
5000
With Without (Cowburn and Stockley 2005). In order to success-
Traffic light system fully convey nutrition information it is necessary to
design nutrition labelling in a way that is easy for
(b) 14000 consumers to detect, read, interpret and understand it
Linear 12283.6 c (Mackison et al. 2010). Considering that consumers
Response time (ms)

Panel invest limited time when making their food purchase


Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

decisions, they may not rely on detailed complex


9000 nutrition information but take into account simple
and easily accessible information (van Herpen and
6588.4 b 6926.6 b van Trijp 2011). Therefore, one of the key issues for
the success of nutrition labelling is its ability to catch
4850.6 a consumers’ attention and to provide information that
4000
With Without is easily found and interpreted.
Traffic light system In the present work, the time needed by consumers
to find nutrition information and to classify labels was
Figure 3. RT for the classification of labels with different used to estimate how easy it was for consumers to find
nutritions’ information format and with/without traffic light system and interpret information about the content of a
For personal use only.

for: (a) yoghurt and (b) pan bread.


specific nutrient (fat or sodium) in food labels. Thus,
longer RTs were related to more difficulty in finding
correctly classified them. For yoghurt labels the nutrition information and to a deeper processing once
difference between the RTs of correct and incorrect the information was found in order to reach a decision.
answers was 2329 ms, whereas for pan bread it was Meanwhile, consumers’ understanding of nutrition
2644 ms. However, it is interesting to highlight that the information displayed on food labels was assessed by
influence of the independent variables on RT was not asking them to indicate whether the fat content of
affected by the type of response, i.e. the conclusions yoghurts and the sodium content of pan bread were
regarding the influence of the independent variables medium or low.

Table II. Results of the logistic regression performed on the percentage of correct answers for yoghurt and pan bread labels.

Product Variable Parameter estimate Standard deviation p-Value

Yoghurt Label background (label B vs. label A) 20.477 0.175 0.0064*


Type of product (regular vs. low) 20.051 0.179 0.7774ns
Nutrition information format (panel vs. linear) 0.245 0.183 0.1810ns
Traffic light system (with vs. without) 0.049 0.181 0.7853ns
Label background £ type of product 0.603 0.185 0.0011*
Label background £ nutrition information format 0.207 0.207 0.2613ns
Label background £ traffic light system 0.070 0.185 0.5465ns
Type of product £ nutrition information format 20.155 0.186 0.4058ns
Type of product £ traffic light system 0.380 0.186 0.0413**
Nutrition information format £ traffic light system 0.112 0.186 0.5465ns
Pan bread Label background (label 2 vs. label 1) 0.072 0.176 0.6819ns
Type of product (regular vs. low) 0.571 0.186 0.0021*
Nutrition information format (panel vs. linear) 0.417 0.183 0.0225**
Traffic light system (with vs. without) 0.902 0.191 ,0.0001***
Label background £ type of product 0.109 0.200 0.5865ns
Label background £ nutrition information format 20.373 0.197 0.0580ns
Label background £ traffic light system 0.030 0.197 0.8803ns
Type of product £ nutrition information format 0.419 0.204 0.0400**
Type of product £ traffic light system 0.087 0.205 0.6721ns
Nutrition information format £ traffic light system 21.248 0.198 ,0.0001***

Note: Whereas ns indicates no significant effect for a confidence level of 95%; * Indicates significant effect for a confidence level of 99%;
** Indicates significant effect for a confidence level of 95%; *** Indicates significant effect for a confidence level of 99.9%.
686 G. Ares et al.

(a) 90
the fat content of yoghurt labels was low or medium
Percentage of correct answers (%) Low fat 83
was more difficult than the decision on the sodium
79.2 Regular content of pan bread labels. A possible explanation for
80 78.4
this behaviour is that consumers needed more time to
71.9 classify the labels according to their fat content
70 because information was presented using three
different values (saturated, trans and total), whereas
for sodium only one value was presented on the labels.
60
Besides, it was clearly more difficult to decide that
the fat content of yoghurts and the sodium content of
50 pan bread were low than that to decide that it was
Background A Background B
medium (Figure 2). This suggests that the cognitive
Label background process needed to classify a product as low in a certain
(b) 90 nutrient was more difficult than that needed to classify
85 a regular product. According to these results, it
Percentage of correct answers (%)

Low fat
seemed that consumers did not have a clear
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

Regular
80 77.3
76.4 predetermined criterion to decide whether a product
73.9
was low or high in a certain nutrient. Instead, each
70 time consumers faced a label they had to select a limit
for classifying the product as low or medium in a
specific nutrient. Thus, consumers needed more time
60
to select the limit when the product was low. It might
be interesting to study whether these differences in the
50 time required for the classification depend on the
With Without
number and name of categories. Moreover, in the case
Traffic light system
of pan bread, consumers were clearly more accurate
For personal use only.

Figure 4. Percentage of correct answers for consumers’


when classifying medium than low sodium labels
classification of the fat content of yoghurt for: (a) labels with
different fat contents and backgrounds, (b) labels with different fat
(a)
contents and with/without traffic light system.
Percentage of correct answers (%)

Low sodium 87.4


90 86.1
The correlation between these two dependent Regular
variables might indicate whether RTs were more
80 76.7
related to the search for nutrition information or to
information processing. In the present work, RTs were 69.9
significantly negatively correlated with the percentage 70
of correct responses, which suggests that RTs were
related to processing time since consumers found 60
more difficult to interpret those labels for which they
required more time to find and process the infor- 50
mation. However, the correlation between RTs and Linear Panel
the percentage of correct responses was low and the Nutrition information format
influence of the evaluated independent variables did
(b)
not differ for consumers when considering correct and
Percentage of correct answers (%)

incorrect responses. This suggests that RTs were more 90 88.2 Linear
related to the difficulty in finding nutrition infor- Panel
81
mation than to information processing. 80 76.3
74.6
Therefore, in the present study RTs were related to
consumers’ search and interpretation of nutrition 70
information, whereas the percentage of correct
responses corresponded to a measure of their 60
understanding.
50
With Without
Influence of nutrition information format on consumers’ Traffic light system
attention and understanding
Figure 5. Percentage of correct answers when classifying the
Results showed that the type of product, nutrition sodium content of pan bread for: (a) labels with different nutritions’
information format and traffic light system signifi- information format and sodium content, (b) labels with different
cantly affected RTs (Table I). The decision of whether nutritions’ information format and with/without traffic light system.
Attentional capture and understanding of nutrition labelling: a study based on response times 687

(Figure 5), which could be related to the lack of companies depends on the size of the label and the
nutritional knowledge. This is in agreement with the space available for nutrition information. However,
results from previous studies. Ares et al. (2008) considering that when making their choices consumers
reported that nutritional knowledge of Uruguayan might rely on information that is easily accessible, they
consumers was low, particularly in what refers to the might be more prone to read information presented
nutritional content of different food products. Besides, under panel form than when it is displayed under
Carrillo et al. (2011) reported that Spanish consumers linear format.
had a scarce knowledge of the nutritional effect of Regarding the use of the traffic light system, it
individual food components and that despite the fact clearly provided an improvement in attentional
that in general Spanish consumers were able to capture and consumers’ understanding of nutrition
understand food labels in a simple task, nutritional information for both products. The only difference in
knowledge had a large influence on the importance the results of both the products was that in the case of
given to nutritional information. Thus, informative yoghurt labels, the influence of the traffic light system
strategies seem necessary to increase consumers’ was independent of nutrition information format,
knowledge about the nutrient content of foods and whereas in the case of pan bread labels it was not. As
particularly of how to evaluate a food product based on shown in Figure 3, the influence of the traffic light
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

a given nutrient. system was larger when nutrition information was


An interesting finding of the present study is that displayed under linear format than when panel format
both yoghurt and pan bread label design did not affect was used. This could be explained considering that
consumers’ ability to find and interpret nutrition when the traffic light system was used consumers
information. This suggests that the graphic design of might have not read all the information displayed in
the labels did not affect consumers’ ability to find and linear format, requiring less time for the classification.
interpret nutrition information. However, for yoghurt As shown in Figures 3 – 5, the use of traffic light
the percentage of correct classifications was signifi- system caused a large reduction in RTs and a slight
cantly affected by the graphic design and non- increase in the percentage of correct classifications.
nutritional information of the labels (Table II). These results sum up to the published studies that
For personal use only.

According to the results, consumers seemed to report that multiple traffic light systems facilitate
associate labels designed with Background B with processing (van Herpen and van Trijp 2011). These
a regular product and not with a low-fat yoghurt and authors reported that despite the fact that consumers
therefore they tended to classify its fat content as from various European countries evaluated nutrition
medium. Apart from graphic design it is important to table format positively, this format received little
take into account that flavour information and product attention and did not stimulate healthy choices.
name did also vary between Background A and According to a study carried out by the Food
Background B, which could have also influenced Standards Agency (2005), among different formats,
consumers’ responses. This result indicates that traffic light systems yield the highest consumers’
consumers’ perception of the graphic design and understanding, measured in terms of their ability to
non-nutritional information of the labels significantly classify a product as high, medium or low in two given
affected their perception of the fat content of yoghurts. nutrients. Besides, traffic light labels and logos have
This is in agreement with previous research that shows been reported to enhance healthy food choices (van
that the images used on labels have a great impact on Herpen and van Trijp 2011).
consumers’ expectations and perception of food Considering that in the present study all the
products (Ares et al. 2010), which might even override information on the label was presented on a single
the effect of nutrition information. This area of plane, the influence of traffic light system when used
research has not received much attention yet and on the front of a package on consumers’ attention
should be further explored since consumers’ percep- could be even larger in a real-life setting. Therefore,
tion of the healthiness of food products might be displaying nutrition information in the front of the
clearly affected by the graphic or non-verbal signs of package could be an interesting alternative to increase
food labels. consumers’ awareness of nutrition information by
Furthermore, nutrition information format signifi- significantly reducing the effort needed for finding and
cantly affected the time required by consumers to interpreting this information on a food label.
classify the evaluated labels according to their content
of a given nutrient (Table I). It was clearly easier for
Limitations and implications for further research
consumers to find, read and interpret nutrition
information when it was presented using panel format The major limitation of the present work is that RT
(Figures 2 and 3). Despite the fact that both formats was related to both attention and cognitive processing.
are allowed by Uruguayan legislation, differences in Further research is needed to overcome this limitation
their ability to capture consumers’ attention and and to independently measure attention to and
their ease of interpretation should be considered. processing of nutrition information. The application
Nowadays, the use of linear or panel format by food of eye-tracking techniques could help to fill this
688 G. Ares et al.

gap since they could provide information on whether Carrillo E, Varela P, Fiszman S. 2011. Influence of nutritional
consumers take more time because they deeply process knowledge on the use and interpretation of food labels. J Food
Sci, doi: 10.1111/j.1750-3841.2011.02479.x (in press).
the information or because they cannot find the right Cowburn G, Stockley L. 2005. Consumer understanding and use of
information or have trouble in understanding the nutrition labelling: a systematic review. Public Health Nutr 8:
information presented to them. 21–28.
Another limitation of the present work is that some Directorate General for Health and Consumer Protection 2005.
information, such as flavour and product name, was The European consumers’ attitudes regarding product labelling
– qualitative study in 28 European countries. Versailles: Optem.
not controlled in the labels. This information could
Food Standards Agency 2005. Quantitative evaluation of alternative
have also influenced consumers’ perception of the fat food signposting concepts, www.food.gov.uk/multimedia/pdfs/
and salt content of the yoghurts and pan breads, signpostquanresearch.pdf. Accessed 28 June 2011.
respectively. Further research should be carried out to Food Standards Agency 2007. Front-of-pack traffic light signpost
specifically address this issue, studying how graphic labelling. Technical guidance, Issue 2. http://www.food.gov.uk/
multimedia/pdfs/frontofpackguidance2.pdf. Accessed 28 June
design, flavour, product name and other non-
2011.
nutritional information influence consumers’ percep- Freeman DH. 1987. Logistic regression Applied categorical data
tion of the nutrient content and healthiness of food analysis. New York: Marcel Dekker. p 237–274.
products. Grunert KG, Wills JM. 2007. A review of European research on
Int J Food Sci Nutr Downloaded from informahealthcare.com by University of Nebraska on 04/07/15

As stated by Bialkova and van Trijp (2010), display consumer response to nutrition information on food labels. J
size, colour, location and familiarity of front-of-pack Public Health 15:385–399.
Jarvis BG. 2004. DirectRT research software, version 2004
information impact consumers’ attention. However, in [Computer Program]. New York: Emprisoft.
the present study these variables were not considered Hollingworth A, Henderson JM. 1998. Does consistent scene
and should be included as independent variables in context facilitate object perception? J Exp Psychol Gen 127:
further research. Besides, other formats of front-of- 398 –415.
pack nutrition information should be studied since Hollingworth A, Henderson JM. 1999. Object identification is
isolated from scene semantic constraint: evidence from object
consumer liking and perception of these signposts
type and token discrimination. Acta Psychol 102:319–343.
might play a key role in their use of nutrition Hurt E. 2002. Nutritional labelling: European Union and United
information in their everyday life (Grunert and Wills Kingdom perspectives. Asia Pac J Clin Nutr 11:S77–S79.
For personal use only.

2007). Janiszewski C. 1998. The influence of display characteristics on


Research is also needed to compare results from the visual exploratory search behavior. J Consum Res 25:290–301.
Jones G, Richardson M. 2007. An objective examination of
present work from those performed under real-life
consumer perception of nutrition information based on healthi-
settings using observational studies. Moreover, con- ness ratings and eye movements. Public Health Nutr 10:
sidering that the influence of nutrition information 238 –244.
format depends on the product and the nutrient Mackison D, Wrieden WL, Anderson AS. 2010. Validity and
considered, it is necessary to consider a wider range of reliability testing of a short questionnaire to assess consumers’
use, understanding and perception of food labels. Eur J Clin Nutr
products and nutrients, in particular considering that
64:210–217.
the type of nutrient and product had a large influence Microsoft at work 2011. 12 tips for creating better docume-
on the results. Furthermore, consumers’ behaviour nts, http://www.microsoft.com/atwork/skills/documents.aspx.
when comparing labels instead of evaluating products Accessed on 23 October 2011.
in monadic sequence could provide interesting Ministerio de Salud Pública 2006. Decreto 117/06. Montevideo,
complementary information on the cognitive process Uruguay: Ministerio de Salud Pública.
Opperud A. 2004. Semiotic product analysis. In: McDonagh D,
involved in their evaluations. Hekkert P, Van Erp J, Gyim D, editors. Design and emotion.
London: Taylor and Francis. p 137–141.
Declaration of interest: The authors report no Pieters R, Wedel M. 2004. Attention capture and transfer in
conflicts of interest. The authors alone are responsible advertising. Brand, pictorial, and text-size effects. J Mark 68:
for the content and writing of the paper. 36–50.
Podsakoff PM, MacKenzie SB, Lee JY, Podsakoff NP. 2003.
Common method biases in behavioural research: a critical review
of the literature and recommended remedies. J Appl Psychol 88:
References
879 –903.
Ares G, Giménez A, Gámbaro A. 2008. Influence of nutritional R Development Core Team 2007. R: A language and environment
knowledge on perceived healthiness and willingness to try for statistical computing. ISBN 3-900051-07-0 Vienna, Austria:
functional foods. Appetite 51:663– 668. R Foundation for Statistical Computing.
Ares G, Piqueras-Fiszman B, Varela P, Morant Marco R, López AM, Smith V, Mogelvang-Hansen P, Hyldig G. 2010. Spin versus fair
Fiszman S. 2010. Food labels: do consumers perceive what speak in food labelling: a matter of taste? Food Qual Prefer 21:
semiotics want to convey? Food Qual Prefer 21:689–698. 1016–1025.
Bialkova S, van Trijp H. 2010. What determines consumer attention Solomon M, Bamossy G, Askegaard S. 2002. Consumer behaviour:
to nutrition labels? Food Qual Prefer 21:1042–1051. a European perspective. Edinburgh Gate: Pearson Education
Black A, Rayner M. 1992. Just read the label: understanding Ltd. p 36–49.
nutrition information in numeric verbal and graphic communi- van Herpen E, van Trijp HCM. 2011. Front-of-pack nutrition
cation. London: HMSO. labels. Their effect on attention and choices when consumers
Caraher M, Dixon P, Lang T, Carr-Hill R. 1999. The state of have varying goals and time constraints. Appetite 57:148–160.
cooking in England: the relationship of cooking skills to food Wolfe JM. 1998. Visual search. In: Pashler H, editor. Attention. East
choice. Br Food J 101:590–609. Sussex: Psychology Press.

Vous aimerez peut-être aussi