Master Thesis
Informational Health and Taste Claims in Food
Advertising for (Un)Healthy Brand Images: Can they
Create a Healthy = Tasty Intuition?
Study: MSc Business Administration – Marketing Track
Institution: Amsterdam Business School / University of Amsterdam Course: Master Thesis
Thesis Supe rvisor: L.T. Moratis Student: Menno Visser
Student numbe r: 10833331 Due date: 29-01-2016
2
Statement of Originality
This document is written by Student Menno Visser who declares to take full responsibility for
the contents of this document. I declare that the text and the work presented in this document
is original and that no sources other than those mentioned in the text and its references have
been used in creating it. The Faculty of Economics and Business is responsible solely for the
3
Table of Contents
1. Abstract ... 5 2. Introduction ... 6 2.1 Background ... 6 2.2 Research gab ... 72.3 Research question and sub-questions ... 8
2.4 Academic contribution... 8
2.5 Practical contribution ... 9
2.6 Thesis structure ... 10
3. Literature review ... 11
3.1 Food Marketing and health ... 11
3.2 The Unhealthy = Tasty Intuition... 12
3.3 Informational stimuli and the UTI ... 14
3.4 Brand image and brand extensions ... 16
3.5 Conceptual model ... 19
4. Research method ... 20
4.1 Food brand advertisement selection ... 20
4.2 Pretest ... 21
4.3 Fictive brand development ... 22
4.4 Measure development ... 23
4.5 Data collection procedure ... 24
4.6 Sample ... 26 5. Data Analysis ... 27 5.1 Descriptive Statistics ... 27 5.2 Reliability Analysis ... 28 5.3 Correlation analysis ... 30 5.4 Hypothesis Testing ... 33
5.4.1 Perceived health existing brands ... 36
5.4.2 Perceived health fictive brands... 37
5.4.3 Perceived taste existing brands... 39
5.4.4 Perceived taste fictive brands ... 40
5.4.5 Combined effect of perceived health and taste ... 42
6. Discussion ... 46
4
6.1.1 A belief in the Unhealthy = Tasty Intuition (H1) ... 47
6.1.2 The effect of informational HN R and taste claims (H2 and H3) ... 49
6.1.3 Differences in brand image (H4) ... 50
6.2 Academic implications ... 51 6.3 Practical implications... 52 6.4 Limitations ... 53 6.5 Future research... 55 7. Conclusion... 57 8. References ... 59 9. Appendix ... 65
9.1 Stimuli used in IAT ... 65
9.2 Designed food ads and vignettes ... 66
9.2.1 Existing brand ads ... 66
9.2.2 Fictive brand ads... 68
9.2.3 Vignettes fictive brands ... 70
5
1. Abstract
Extant literature emphasizes that the amount of people that are considered as overweight or
obese is growing rapidly, while in contrast a trend occurs in which consumers eat healthier
food products. This still suggest that consumers associate unhealthy food with tastiness and,
therefore, have a belief in the Unhealthy = Tasty Intuition. However, prior research states that
it is possible to transform this belief in the UTI into a belief in the Healthy = Tasty Intuition
when informational health and taste claims are used in food ads. Nevertheless, these studies
pay little attention to ads that emphasize informational health and taste claims simultaneously
and ignore the influencing role of the brand image. Through an experiment with four
experimental groups (n = 189), this study examined the joint effect of informational health
and taste claims on perceived levels of health and taste for (un)healthy brand images in food
advertising. Participants were exposed to food ads in which three factors were included: (1)
informational health and taste claims were either present or absent, (2) the parent brand image
was either unhealthy or healthy, and (3) the ads were designed for both existing as well as
fictive brands. However, the results of this study revealed that consumers associate healthy
food with tastiness and thus already have a belief in the HTI. Furthermore, participants were
indifferent about the perceived health and taste of the products or evaluated the food products
in the ads without informational claims as healthier and tastier. Therefore, the main
conclusion of this research is that it is better not to use informational HNR and taste claims in
food advertising. However, when food companies still decide to use these claims in their ads,
it is only valuable for an existing brand with a healthy image or a new (fictive) brand with an
unhealthy image. These results contribute to existing literature and provides evidence for food
marketers that informational HNR and taste claims, which they already use in advertising,
6
2. Introduction
2.1 Background
Recently, a trend emerged in which consumers shifted their food intake towards fresh, natural
and minimally processed foods. Consumers are continuously confronted with healthy food
products and recipes of friends, relatives, as well as food companies as soon as they watch
TV, browse the internet, check social media, read a magazine, or listen to the radio (N ielsen,
2015). Despite this increased attention for healthier food habits, a contradictory problem
occurred. That is, the percentage of people who are considered as overweight or obese is
growing rapidly (N g et al., 2014).
Specifically, in the last 30 years the prevalence of overweight and obesity rose by
27.5% for adults and 47.1% for children. Moreover, the total amount of individuals who were
considered as overweight or obese was 2.1 billion in 2013 (N g et al., 2014). As a result public
policy makers had to develop strategies including: social-change marketing campaigns, health
promotion programs, and health education. These strategies need to directly affect behaviors
in order to increase the motivation to make healthy lifestyle choices. Nevertheless, inducing
such behavioral change will be a difficult task and many campaigns fall short of expectation
(Hill, Wyatt, Reed, & Peters, 2003; Marteau, Hollands, & Fletcher, 2012; Swinburn et al.,
2011).
One cause for the growing rate of overweight and obes ity is that consumers have the
tendency to over consume unhealthy food because they expect that unhealthy food will taste
better (Raghunathan, Naylor, & Hoyer, 2006). In other words, food that is perceived as
unhealthy is associated as having a better taste while food that is perceived as healthy is
associated as less tasty. Raghunathan et al. (2006) refer to this inverse relationship between
7
In addition, the UTI is in line with the general theory of food decision making.
Overconsumption of unhealthy food occurs because food decisions are primary based on taste
(Tepper & Trail, 1998). More recent literature have extended this view and found that food
selection is guided by (sensory) intrinsic properties (e.g. taste, aroma, and fat content) as well
as extrinsic properties (e.g. packaging, advertising and claims). Both properties can either be
health related or health unrelated. In particular, intrinsic and extrinsic properties have a
different impact in the food decision process (Hoppert et al., 2014; Mai & Hoffmann, 2012).
Although consumers are aware they need to take health into account when choosing
food, they are attracted by the intrinsic product attributes such as the flavor of the food.
However, extrinsic product attributes that for example message reduced calories or other
kinds of health cues may form the decisive factor in their food choice (Hoppert, Mai, Zahn,
Hoffmann, & Rohm, 2012; Hoppert et al., 2014). The latter effect is of great importance in
food marketing. Because marketers have the ability to change the extrinsic properties, while
they regularly do not have the ability to change the product itself (Hoppert et al., 2012). That
is why knowledge is needed on how informational stimuli provided by advertisement and
packaging influence food decision making.
2.2 Research gab
In the age of healthier food habits, two factors need considerable attention. That is, the food
needs to be healthy while also the taste needs to be good. This resulted that food companies
started to promote their products with informational claims. The most frequent used are health
and nutrition-related (HNR) claims and taste claims (Choi, Paek, & Whitehill K ing, 2012;
Choi & Springston, 2014). Extant literature focused on informational stimuli with either a
HNR or taste claim. However, they ignore the joint effect of HNR and taste claims (Choi et
al., 2012). Furthermore, prior research states that little attention has been paid on the effect of
8
Sandell, 2012). Therefore, the objective of this paper is twofold. First, it will examine if the
UTI vanishes and a healthy = tasty intuition (HTI) can be created when health and taste
claims are emphasized simultaneously in food advertising. Second, it will explore if this effect
differs for a healthy parent brand image compared to a n unhealthy parent brand image.
2.3 Research question and sub-questions
The main research question can be established as follows:
What is the joint effect of informational HNR and taste claims on perceived levels of health and taste for (un)healthy parent brand images in food advertising?
The following sub-questions will help answering the research question:
What is the role of health in food marketing?
What is the Unhealthy = Tasty intuition and how does it influence food decision making? What is the role of informational claims and what is the effect if they are co ntradicting? What is the effect of informational claims on perceived levels of health and taste?
What is the role of the parent brand image in the consumers’ evaluation of food products? 2.4 Academic contribution
As stated before, consumers have a belief in the UTI (Raghunathan et al., 2006). However,
other studies describe that this belief can be reduced. More specifically, an example stimulus
that reduces this belief is by using informational HN R and taste claims for food products.
Prior studies have used this stimulus in order to find the effect how consumers perceived the
health and taste of food products (Cheong & K im, 2011; Choi & Springston, 2014; Mai &
Hoffmann, 2015; Wansink & Chandon, 2006). However, these studies used either an
informational HNR or taste claim. In other words, no study used informational HNR and taste
9
literature is how consumers perceive the health and taste of food products when informational
HNR and taste claims are used simultaneously in advertising. Furthermore, as stated by
Paasovaara et al. (2012) there is little evidence if the effect on perceived health and taste will
be different for a brand with a healthy image compared to an unhealthy brand image. Hence,
the second contribution of this research is to examine if differences exist in the perceived
health and taste of consumers when a brand with a healthy image versus an unhealthy image
use informational claims in their advertising. In sum, the empirical question remains how
consumers will react to ads of (un)healthy food brands that emphasize both HNR and taste
appeals.
2.5 Practical contribution
In reaction to the trend of healthier food consumption, companies in the food industry
complemented their product portfolio with healthier products (Wilcox, Vallen, Block, &
Fitzsimons, 2009). These new products are often promoted among consumers with
advertising. In many cases these ads include both informational HNR and taste claims. Good
examples include published ads in food magazines or commercials on the television.
However, it remains unclear how consumers react to informational claims in food advertising.
Therefore, additional insights for marketing practitioners are required. This research could
provide evidence for food marketers if it is profitable to emphasize both informational health
and taste claims in food advertising or whether it is a waste of money to make such
advertisements. Furthermore, this study also makes a societal contribution. More specifically,
it will give insights how consumers respond when contradicting claims are used in food
advertising. Additionally, it gives an understanding if this could persuade consumers to
increase their motivation to make healthier lifestyle choices. Hence, this study can provide
10
successful health promotion program in order to reduce the amount of people that are
considered as overweight or obese.
2.6 Thesis structure
This paper starts with concise but focused review of the current literature on how food decision making is influenced by the UTI and about the effect of informational claims for food products in this context. Then the quantitative research methodology of this study will be explained in more detail. Hereafter, the findings that result from the data will be provided. Finally the paper will discuss the consequences of the research and the academic and practical implications are addressed.
11
3. Literature review
In this section of the paper an extensive review of the current literature will be given. First,
the importance of health in food marketing will be emphasized. Second, the UTI will be
described in more detailed and how this influences food decision making. Third, the effect of
informational claims on food products will be investigated. Fourth, the influential role of the
brand image will be examined. Finally, a conceptual framework that guides this study will be
introduced.
3.1 Food Marketing and health
As consumption changes into more healthy but tasty food, it also affects how food companies
conduct their marketing (Parker, 2003). Specifically, health issues have become more and
more important in food marketing over the last couple of years. Food marketers noticed that
this could be a very profitable source for many companies and they started to stress a healthy
lifestyle (Luomala, Paasovaara, & Lehtola, 2006). Correspondingly, various retailers and
brands in the food industry introduced healthier options to supplement their product portfolio
besides the less healthy product offerings (Wilcox et al., 2009).
Furthermore, food companies have also increased their attention in sustainability and
added more sustainable products to their portfolio, such as organic food. This also contributed
towards a more healthier product offering, since research indicate that consumers assume that
organic food is healthier (Hoogland, de Boer, & Boersema, 2007; Lee, Shimizu, K niffin, &
Wansink, 2013). A good example for this is the Albert Heijn, a large Dutch supermarket
chain, which added more than 1,000 organic products to their portfolio since last year.
However, from a consumer perspective, selecting food is a mundane activity which
can occur several times a day. Nevertheless, this decision making process also involves
satisfying different goals, such as : health, taste and price (Finkelstein & Fishbach, 2010).
12
what to eat (Wilcox et al., 2009). In order to satisfy the contradicting goals of tasty food
intake or healthy nutrition they need to compromise. According to Mai & Hoffmann (2015)
this is due to the fact that consumers believe that unhealthy food tastes better. Therefore, food
companies and policy makers should help the consumers to reduce their UTI and to market
healthier food more effectively.
3.2 The Unhealthy = Tasty Intuition
In general there is a belief in the UTI (Raghunathan et al., 2006). That is, consumers have the
intuition that food with a good taste is incompatible with a high degree of healthiness.
Raghunathan et al. (2006) propose that there are two underlying sources for the UTI. First, it
is generated internally on the basis that consumers believe that healthy food cannot go
together with a good taste. Second, an external source operates besides the interna l one
through repeated exposures by mass media or personal communication.
The majority of food consumption research supports the contradiction of health and
taste. The rationale for this is that food decision making is mainly based on the taste of food
(Tepper & Trail, 1998). Furthermore, other studies found that consumers expect food to taste
better and enjoy the taste more when they are faced with a taste claim for unhealthy food
compared to a claim for healthy food (Luomala et al., 2015; Raghunathan et al., 2006).
However, there is also a study which suggest that health and taste can go along with
the consumption of food. It is seems that in some extent consumers think that in order to stay
healthy, life should not deny pleasure. In this case tasty food is seen as justified relaxation
(Luomala et al., 2006). Based on the intrinsic and extrinsic product properties as described by
Mai & Hoffmann (2012), other studies state that a belief in the UTI can be reduced by
emotional ability training (K idwell, Hasford, & Hardesty, 2015); adding health- unrelated
attributes to healthy products that elicit compensation effects for the loss in taste (Mai, Zahn,
13
Trendel, & Ardito, 2013); the extent in which individuals differ in motivation, for example:
taste lovers vs. nutrition fact seekers or dieters vs. non-dieters (Irmak, Vallen, & Robinson,
2011; Mai & Hoffmann, 2012); and most interesting food brands that emphasize both HNR
and taste claims, whereby taste should be the dominant factor (Raghunathan et al., 2006).
Additionally, there are also other concepts that are closely related to the UTI. Prior
studies divided healthy and unhealthy food into virtues and vices (Chernev & Gal, 2010;
Chernev, 2011; Wertenbroch, 1998). Virtues are food products that are considered as healthy
(e.g. fruit and cottage cheese), vices in contrast, are food products considered as unhealthy
(e.g. chocolate and French fries). According to Wertenbroch (1998), vices have immediate
benefits but delayed costs, for example: chocolate has a delicious taste which is experienced
right away, but its potential cost, such as weight gain, will be encountered later on. Virtues, on
the other hand, have immediate costs and delayed benefits, for instance: many people do not
like the taste of cottage cheese, but the product has long term health benefit due to the low fat
and high levels of protein. In a similar vein, Luomala et al. (2015) state that consumers have
inclusive and exclusive health and taste perceptions of food. Whereas, an inclusive perception
of health and taste means that healthy is perceived as tasty, and unhealthy as untasty. In
contrast, if health and taste are exclusive, healthy is considered as untasty, and unhealthy as
tasty. In sum, based on the preceding discussion it can be proposed that a belief in the UTI
exists in the mind of the consumer. Therefore, the following hypothesis is formulated.
H1. Despite the fact that consumers consume healthier food, in general they still have a belief in the Unhealthy = Tasty Intuition.
However, it should be noted that it is possible to transform a belief in the UTI into a
HTI. There are several actions food marketers can undertake in order to create this, which can
14
previously this paper will focus on the extrinsic attributes. In particular, the application of
informational claims that emphasize both health and taste. The next section will examine this
more extensively.
3.3 Informational stimuli and the UTI
As already mentioned, food decision making is based on intrinsic and extrinsic product
attributes. Intrinsic product attributes includes things as: taste, aroma, texture and fat content
(Mai & Hoffmann, 2012). As food marketers cannot always change the product itself, they do
have the possibility to make changes in the extrinsic attributes. These extrinsic product
attributes embrace: price, branding, production method, packaging, advertising, informational
claims, nutrition information, or certification (Mai & Hoffmann, 2012b; Mai et al., 2014). The
aim of this paper is to examine if a HTI can be created by using informational claims that
emphasize both health and taste used in food advertisement.
In practice it can be seen that in addition to healthier product portfolios, food
companies already use of informational stimuli in the form of HNR and taste claims in
advertising (Choi et al., 2012; Choi & Springston, 2014). In general, HNR claims can be
subdivided into three categories. The first category is health claims: ‘’these describe the
relation between food, a food component, or dietary supplement ingredient, and the reduction
of the risk of a disease or health-related condition’’. The second group is nutrient content
claims: ‘’these describe the percentage of a nutrient in a product relative to the daily value of
the product, which means that there needs to be a certain amount of a nutrient provided in a
single serving of the particular food item’’. The final category is structure/function claims:
‘’which describe the effect of a dietary supplement on the structure or function of the body’’ (Nan, Briones, Shen, Jiang, & Zhang, 2013, p. 264). These three categories are still quite
broad and vague in order to use them in food advertising. Therefore, Nan et al. (2013)
15
common used. They distinguishes three different types of nutrient content claims. First,
‘healthy claims’ give a general claim that the food is ‘healthy’. Second, ‘nutrient content claims’ describe a specific nutrient component in food products (e.g. low in fat, saturated fat
and calories). Third, ‘general nutrition claims’ imply the healthiness of food by using
unspecific words (e.g. wholesome and nutritious).
The latter types described by Nan et al. (2013) are in line with other studies that also
identified three distinct ways in communicating health, namely: functional claims, process
claims, and health symbols (Anker, Sandøe, Kamin, & Kappel, 2011; Chrysochou & Grunert,
2014). Functional claims imply that a food product contains a particular nutritional
component and, therefore it helps in maintaining a good health (e.g. contains O mega3 or low
in saturated fat). In contrast process claims, suggest that products in a certain product category
are healthier due to the production process compared to similar products in the same product
category (e.g. organic). Finally, health symbols add an emotional or experiential dimension to
food products by showing abstract concepts such as: depiction of physical activity, slim
bodies, and fruits, that gives a symbolic health related association to consumers. It should be
noted that these types can be placed on a continuum. In other words, with functional claims
the health message is directly conveyed and with health symbols the health message is
inferred. This paper will use the health and taste claims as described by Anker et al. (2011)
and Chrysochou & Grunert (2014)
Although consumers are skeptical of HNR claims, they still believe most nutrition
information on product packaging (Wansink & Huckabee, 2005). Results from experiments
conducted by Choi et al. (2012) and Choi & Springston (2014) show that when HNR claims
are used in advertising, consumers evaluate the product as healthier but as less tasty than ads
with taste claims. In contrast, consumers perceive food products as more tasty and less
16
However, using these claims independently of each other cannot overcome consumers’ UTI.
Therefore, it is suggested that by stressing health and taste claims together in food advertising
could break the UTI (Cheong & K im, 2011; Choi & Springston, 2014; Mai & Hoffmann,
2015; Raghunathan et al., 2006; Wansink & Chandon, 2006). This preceding discussion leads
to the following hypotheses.
H2. Using informational claims on food products that emphasize both health and taste have a positive effect on the perceived health.
H3. Using informational claims on food products that emphasize both health and taste have a positive effect on the perceived taste.
3.4 Brand image and brand extensions
As mentioned previously, food brands are extending their product portfolios with healthier
options (Wilcox et al., 2009). Brands in the food industry can be seen as having either a
healthy or unhealthy image. A food brand has a healthy image when the corporate
organization intentionally attempts to communicate the value of health implicitly and/or
explicitly, while this is linked to a product of the corporate organization and consumers are
able to associate the products/company with this value (Chrysochou, 2010). Despite of the
current image, brands with both healthy and unhealthy images have extended their product
portfolio with healthier products. Examples of these brand extensions are Coca Cola Life, a
rather unhealthy brand image; and Zonnatura vegetable spread that can be categorized as a
healthy brand image.
A food company with an unhealthy image extending its portfolio with healthier
products can lead to customer confusion and difficulties to manage the image. However,
according to the theory of brand concept management – which is defined as the overall
17
simultaneously (Park, Jaworski, & Maclnnis, 1986). Although, firms should have one
dominant brand concept, which is the foundation for the brand image. It is proposed that
every brand image can be based on a single or several brand concepts that can tap into
functional and symbolic needs (Bhat & Reddy, 1998). According to Bhat & Reddy (1998) a
functional brand concept satisfy immediate and practical needs, while a symbolic brand
concept satisfy needs such as self-expression and status.
Yet, this solely focus the brand image of food companies, while the theory of brand
extensions also need to be considered. A favorable parent brand image facilitates the
acceptance of new products and this will transfer positive attitude from the parent brand to the
brand extension (Bhat & Reddy, 2001; Broniarczyk & Alba, 1994). It is also found that
perceived fit is the most dominant factor in explaining the success of brand extensions
(Dwivedi, Merrilees, & Sweeney, 2010). Therefore, it is crucial that brand extensions are
consistent with the brand concept (Park, Milberg, & Lawson, 1991). However, Monga & John
(2010) argue that many brands introduce successful new products in distant product
categories, which have few similarities with the parent brand. They refer to these brands as
being more elastic.
In particular, brands with a symbolic brand concept are considered as more elastic and
have the ability to extend in more dissimilar product categories, whereas a functional brand
concept is considered as less elastic (Park et al., 1991). Monga & John (2010) found that
brand elasticity is collectively determined by the brand concept and consumer styles of
thinking. The latter determinant – consumer styles of thinking – is subdivided into two
categories, that is: holistic thinkers who are more orientated on the whole context, including
the relationship between the parent brand and the brand extension; and analytic thinkers who
18
functional brand concepts extending into new distant product categories, while in this case
holistic thinkers generate favorable responses to the dissimilar brand extension.
Nevertheless, the gap in extension evaluation between analytic and holistic thinkers
can be reduced by introducing sub-brands in the brand architecture (Aaker & Joachimsthaler,
2000), providing more information about the distant brand extension, and matching the
message frame to the styles of thinking (Monga & John, 2010). Thus, it can be concluded that
food companies with both healthy and unhealthy brand images have the possibility to extent
their product portfolio with healthier products, while still got widely accepted by consumers,
even when this is into a new or distant product category.
However, as the health trend affects the entire marketing of food companies, it is not
enough to just introduce healthier options or to emphasize a healthy image. That is why
several food companies implemented the practice of health branding (Chrysochou, 2010).
Health branding is defined as: ‘’strategies that are adopted by firms to communicate the value
of health’’. Specifically, as argued by Chrysochou (2010), on the one hand the firm should
convey a healthy brand image and, on the other hand, it is a tool for social marketing
initiatives that aims for public health communication.
From the consumer perspective, findings generally confirm that consumers have
higher ad-related evaluation when the there is a match-up effect between the perceived
healthiness of the product and the informational claim that is used. In other words, consumers
preferred enhanced healthiness by HNR claims only for healthier food, while a taste claim
might be more appropriate for unhealthy products. Therefore, food marketers should be
cautious about highlighting mismatched claims (Choi et al., 2012; Choi & Springston, 2014).
This match-up effect might also be useful for the moderated effect of the parent brand image.
Consumers are more likely to perceive food as more tasty when a firm with a healthy image
19
consumers perceive food as healthier when nutritious ingredients are added by brands with a n
unhealthy image (Chrysochou, 2010). This leads to the following hypothesis.
H4. The brand image act as a moderator, so that a healthy brand image will lead to higher degrees of perceived health and taste compared to an unhealthy brand image.
3.5 Conceptual model
Based on the preceding hypotheses the following conceptual model can be designed.
H1. Despite the fact that consumers consume healthier food, in general they still have a belief in the Unhealthy = Tasty Intuition.
H2. Using informational claims for food products that emphasize both health and taste have a positive effect on the perceived health of the product.
H3. Using informational claims for food products that emphasize both health and taste have a positive effect on the perceived taste of the product.
H4. The brand image acts as a moderator, so that having a healthy brand image will lead to higher degrees of perceived health and taste compared to an unhealthy image.
Figure 1: Conceptual model
A belief in the Unhealthy = Tasty Intuition Healthy vs. unhealthy Parent Brand Image Perceived health Perceived taste Informational claims on food products H1 H2 H1 H3 H4
20
4. Research method
The research approach of this study was of explanatory nature in which an experiment was
conducted in order to test the proposed hypotheses. Data has been collected by a
cross-sectional internet- mediated questionnaire (Saunders, Lewis, & Thornhill, 2012). Participants
were approached digitally by e- mail, social media and face-to-face in the canteen of the
University of Amsterdam. This research design is generally used by stud ies which are closely
related, such as the study of Choi et al. (2012) and Raghunathan et al. (2006). The main
advantage of this design is that it enables to approach a large amount of respondents
(Saunders et al., 2012). In the end, the overall experimental design was a 2 (informational
claims: absent vs. present) x 2 (parent brand image: unhealthy vs. healthy) x 2 (brand: existing
vs. fictive) within subjects design, whereby 4 experimental groups have been created.
4.1 Food brand adve rtisement selection
In order to test the conceptual model, the most appropriate food brand ads needed to be
selected. Therefore, several steps were taken to select the ads that were most suitable for this
research. First, the different food categories that were known in the market have been
described. The second step was to search for food ads in food magazines. An ad was suitable
if a food brand was displayed, the product shown were a relatively new brand extension of the
parent brand, the ad already emphasized both informational HNR and taste claims, and within
each food category an ad was found of a brand with a healthy image as well as an unhealt hy
image. In the final step the most suitable food ads were selected and listed. In total 16 food
brands within eight food categories met the four prerequisites. The food categories were:
dairy, soft drinks, biscuits, juices, breakfast cereals, dressing and sauces, sandwich spreads,
and chips. Table 1 gives an overview of the selected food categories and the brands with a
21
Food c ateg ory He althy br and ad Unhe althy br and ad
Dairy Arla Skyr Yoghurt Ca mp ina Botergoud Halfvolle roo mboter
Soft drin ks Holy Soda Coca-Cola Zero
Biscuits Bolletje Goed Be zig Peijnenburg Ontbijtkoe k Zero Juices Healthy People Kokoswater Appelsientje Halfzoet Brea kfast cereals Quaker Co mp lete Melkunie Havermoutpap Dressing and sauces Kühne Dressing Re mia Feestsaus
Sandwich spreads Zonnatura Groentespread Hein z Sandwichspread Chips Rawlicious Boeren koolchips Lay’s Ovenbaked
Table 1: Selected food categories and brands
Hereafter three product categories were removed from the list, these categories
include: dressing and sauces, biscuits, and breakfast cereals. First, the category dressing and
sauces was removed because ‘Remia Feestsaus’ had an unclear HN R claim in the
advertisement. Secondly, the category biscuits have been removed since it is likely that
consumers cannot really discriminate between the brand images of the products ‘Bolletje Goed Bezig’ and ‘Peijnenburg Ontbijtkoek Zero’. Finally, breakfast cereals were removed because it is uncertain whether consumers associate the brand image of ‘Melkunie
Havermoutpap’ as unhealthy due to the prefabricated disposition of the product, compared the truly healthy brand image of ‘Quaker Complete’.
4.2 Pretest
As listed in the previous paragraph a total of 5 food categories with 10 food brands were
selected, 5 with a healthy image and 5 with an unhealthy image. In order to ensure that the
two most appropriate food brands were selected, a pretest was conducted in the form of a
survey. This was necessary because the two food advertisements that were selected for the
experiment needed maximum variance in order to discriminate between a healthy a nd an
unhealthy brand image.
The survey consisted of one question in which participants needed to indicate whether
22
that completed the pretest was N = 22. Table 2 gives an overview of the descriptive statistics
of the selected food brands. The results show that Arla Skyr Yo ghurt has the highest mean
value (M = 5.14). This indicates that respondents perceive this as the healthiest product. In
contrast, Coca-Cola Zero has the lowest mean value (M = 2.59). This indicates that
respondents perceive this products as the most unhea lthy one. In order to test if the means of
Arla Skyr Yoghurt and Coca-Cola Zero were significant different from each other, a
Paired-Samples t-test was conducted. The results report a p-value of 0.000 < 0.001. Thus, it can be
concluded that there is a significant difference between the two means and there is maximum
variance. This means that it is suitable to use Arla Skyr Yoghurt as the product with a healthy
brand image and Coca-Cola Zero as the product with an unhealthy brand image in the
experiment.
M SD N
Arla Skyr Yoghurt 5.14 1.082 22
Ca mp ina Botergoud Halfvolle Roomboter 3.41 1.368 22
Holy Soda 4.38 1.161 22
Coca-Cola Zero 2.59 1.221 22
Healhy People Kokoswater 4.64 1.293 22
Appelsientje Halfzoet 3.45 1.335 22
Zonnatura Goentespread 5.00 1.512 22
Hein z Sandwichspread 3.32 1.393 22
Rawlicious Boeren koolchips 4.36 0.953 22
Lay’s Ovenbaked 3.14 1.246 22
Table 2: Descriptive statistics pretest
4.3 Fictive brand development
The same product categories as the existing brand s were chosen in order to create the fictive
brands. This was done because in this case the maximum variance between the healthy and
unhealthy images remained. Thus, for the healthy brand image a dairy product and for the
23
named Melkhof Yoghurt, whilst the fictive brand with an unhealthy image was called
Coca-Nade.
4.4 Measure development
The data in the experiment was collected through a cross-sectional survey. In this part the
scales that have been chosen to measure the different constructs in the survey will be
described. First the questionnaire asked respondents for their demographics, this includes:
gender (nominal variable), age (ratio variable), education level (ordinal variable), and income
(ordinal variable). For the other measures existing and validated variables were used.
The pretest survey consisted of one item from Werle et al. (2013). The item was
measured on a 7-point Likert scale ranging from 1 (Very Unhealthy) to 7 (Very Healthy). The
question asked is: ‘’Please indicate the extent to which you perceive the following food products as Healthy or Unhealthy.’’
An implicit belief in the Unhealthy = Tasty intuition was measured with an Implicit
Association Test (IAT) which was slightly adapted from Raghunathan et al. (2006).
Respondents were asked to categorize stimuli correctly, but also have to do this with some
speed. The stimuli shown come from four categories, which includes: 5 healthy food images,
5 unhealthy food images, 7 words associated with good taste, and 7 words associated with bad
taste. An overview of the stimuli can be found in appendix 1. Participants completed three
blocks of questions. The first was a block where they could practice and the two remainders
were critical test blocks. A more detailed overview of the blocks are shown in Table 3. The
practice block consisted of two trials that showed one image and one word. The two critical
blocks for analysis consisted of 16 trials that showed an equal amount of stimuli from all four
categories. The first critical test block were the congruent condition, in which participants
need to categorize unhealthy food images together with good taste words into one group and
24
second test block were the incongruent condition, in which participants need to categorize
healthy food images together with good taste words into one group and unhealthy food
images together with bad taste words into the other category. An implicit belief in the UTI
would be reflected in faster responses in the congruent condition compared to the incongruent
condition.
An explicit belief in the Unhealthy = Tasty Intuition was measured with three items
from Mai & Hoffmann (2015). This was measured on a 5-point Likert scale ranging from 1
(Strongly disagree) to 5 (strongly agree). An example of one item is: ‘’Things that are good
for me rarely taste good’’. The reported Cronbach’s Alpha (α) = 0.79.
Perceived Health was measured with three items from Mai & Hoffmann (2015). This was measured on a 7-point Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly
agree). An example of one item is: ‘’The product helps me stay fit’’. The reported Cronbach’s
Alpha (α) = 0.84.
Perceived Taste was measured with two items from Mai & Hoffmann (2015). This was measured on a 7-point Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree).
An example of one item is: ‘’The product is tasty’’. The reported Cronbach’s Alpha (α) = 0.93.
Block Number of
Trials
Functi on Ite ms assigned to left-key response
Ite ms Assigned to right-key Res ponse
1 2 Practice Words Images
2 16 Critica l test block Unhealthy food images + good taste words
Healthy food images + bad taste words 3 16 Critica l test block Healthy food images +
good taste words
Unhealthy food images + bad taste words
Table 3: Sequence of blocks in IAT
4.5 Data collection procedure
The research approach of this paper was of explanatory nature. In order to test the proposed
25
been collected by a cross-sectional survey. The survey was designed in Qualtrics and the data
were analyzed by using SPSS. The overall experimental design were a 2 (informational
claims: absent vs. present) x 2 (parent brand image: unhealthy vs. healthy) x 2 (brand: existing
vs. fictive) within subjects design, in which participant were randomly assigned to one of the
four groups. First participants were exposed to an existing brand ad. Hereafter they were
exposed to a fictive brand ad. Therefore 8 food ads and 2 vignettes were designed. The
designed food ads and vignettes can be found in appendix 2.
The survey was translated into Dutch to ensure a better understanding of the questions.
The technique that were used is back-translation. In other words, the survey is designed in
English and then translated into Dutch. Subsequently, an objective person translated the
survey back into English. Hereafter it was possible to compare the two version. After
comparison it was not necessary to make any adjustment.
At November 26th 2015 the survey was distributed and closed at December 11th 2015.
In total 250 unique respondents filled in the survey. Of the 250 survey’s, 61 were incomplete, resulting in 189 valuable surveys. The survey was divided into four parts, which is
summarized in Table 4. A more detailed view of the survey can be found in appendix 3. First,
an introduction was given and participants were asked for their demographics. In the second
part, participants completed the three blocks of the IAT. Hereafter, participants were asked to
answer the three items for an explicit belief in the UTI. In the last part, participants were
randomly assigned into one of the four experimental groups. First they were asked to study
the designed food advertisement of an existing brand comprehensively and, hereafter, to
answer the three items of perceived health and two items of perceived taste. Finally, they were
asked to read the vignette and to study the designed food ad of the fictive brand
comprehensively and, hereafter, also to answer the three items of perceived health and two
26
Surve y part
Conte nt Expose d to brand ad:
Group 1 Group 2 Group 3 Group 4
1 Introduction and demographics
No ad e xposure No ad e xposure No ad e xposure No ad e xposure
2 IAT No ad e xposure No ad e xposure No ad e xposure No ad e xposure 3 Exp lic it belief in
the UTI
No ad e xposure No ad e xposure No ad e xposure No ad e xposure
4 Perceived Health and Perceived Taste
Existing brand: healthy image, informat ional c laims present
Existing brand: healthy image, informat ional c laims absent
Existing brand: unhealthy image, informat ional c laims present
Existing brand: unhealthy image, informat ional c laims absent
Fictive brand: unhealthy image, informat ional c laims absent
Fictive brand: unhealthy image, informat ional c laims present
Fictive brand: healthy image, informat ional claims absent
Fictive brand: healthy image, informat ional claims present
Table 4: A complete overview of the survey flow
4.6 Sample
The population were consumers of food brands who live in the Netherlands. Since almost
everybody is a consumer, the potential population is very broad and diverse. The only
prerequisite that the respondents should meet is the ability to evaluate the designed
advertisement. A non-probability convenience sampling has been used, because the
population is broad and diverse. Participants were approached digitally through social- media
and e- mail. Furthermore, participants were also approached face-to-face in the canteen of the
27
5. Data Analysis
In this section insights will be provided about the descriptive data of the sample, the
correlations between the variables, and the proposed hypotheses will be tested.
5.1 Descriptive Statistics
The data set yielded 91 man (48.1%), 88 women (46.6%) and 10 respondents who did not
reveal their gender. The age of the respondents varied between 18 and 64 years old. On
average the respondents are 34 years old. Unfortunately, the variables educational background
and income were slightly skewed. More specifically, educational background had a slightly
negative skewness (-0.59) and income had a slight positive skewness (0.76). As indicated in
the research method, after participants completed the general questions of the survey, they
were randomly assigned to one of the four experimental groups. A detailed description of the
distribution of the sample data per experimental group is given in table 5.
Respondent Group
Sample size
Gender Age Education Income
N Men Women Mean
age BO MO MBO HBO WO < 20 20 – 30 30 – 40 40 – 50 50 – 60 60 > Group 1 47 24 22 34.61 0% 4.3% 31.9% 27.7% 36.2% 34% 12.8% 21.3% 21.3% 4.3% 6.4% Group 2 47 16 26 33.38 0% 8.5% 21.3% 29.8% 40.4% 40.4% 19.1% 14.9% 12.8% 6.4% 6.4% Group 3 47 24 20 34.10 2.1% 6.4% 21.3% 29.8% 40.4% 40.4% 21.3% 10.6% 10.6% 6.4% 10.6% Group 4 48 27 20 33.87 0% 8.3% 18.8% 37.5% 35.4% 41.7% 14.6% 16.7% 4.2% 8.3% 12.5%
Table 5: Descriptive data of sample
The next step was to investigate if there were significant differences between the
experimental groups. Participants were randomly assigned into the experimental groups in
order to avoid large discrepancies, but further statistical prove could confirm this. First, a
cross tabulation was designed and a Pearson Chi-Square test have been conducted to test if
there were significant differences between the gender distribution between the experimental
groups. The p-value of the Pearson Chi-Square test is 0.281 > 0.05. This implies that there
28
The same steps have been performed for the demographic variables age, education,
and income. However, for all variables more than 10% of the cells did not meet an expected
count of at least 5. For this reason it did not meet the prerequisites for a Person Chi-Squares
test and instead a Kruskal-Wallis test was performed. The Kruskal-Wallis tests showed the
following p-values: age 0.849 > 0.05, education 0.966 > 0.05, and income 0.856 > 0.05. This
indicates that there were no significant differences in all variables between the experimental
groups. Thus, it can be concluded that the distributions did not lead to a response bias.
Finally, an analysis was performed in order to handle missing data. Hence, apart from
the missing values in the control variables, the other variables contained a total of 15 missing
values. Group 1 (Existing brand: healthy image, informational claims present; and Fictive
brand: unhealthy image, informational claims absent) yielded 3 missing cases. Group 2
(Existing brand: healthy image, informational claims absent; and Fictive brand: unhealthy
image, informational claims present) yielded 4 missing cases. Group 3 (Existing brand:
unhealthy image, informational claims present; and Fictive brand: healthy image,
informational claims absent) yielded 8 missing values. Group 4 (Existing brand: unhealthy
image, informational claims absent; and Fictive brand: healthy image, informational claims
present) yielded 0 missing values. In all 15 cases it was possible to compute an estimated
value for analysis by using Hotdeck imputation (Myers, 2011). O ne prerequisite for Hotdeck
imputation is that the percentage of missing values should be lower than 10%. This was the
case for all the variables. Thus, all 15 values have been replaced by similar values in the
dataset that matches the missing values.
5.2 Reliability Analysis
Since there were no counter- indicative items in the dataset it was not necessity to recode any
of the items. Therefore, the consistency of the different scales that were used could be
29
As mentioned before, participants have been randomly divided over four different groups.
Participants were asked to answer the same questions, but the stimuli that were shown
differed in each group. Table 6 gives an overview of the Cronbach’s Alpha (α) per construct.
The Cronbach’s Alpha presents the estimator of the internal consistency and verify if all the items in each scale measure the same. As can be seen in table 6 most Cronbach’s Alpha
were above 0.7. This indicates that that the scales are good and that there is a high level of
internal consistency. It should be noted that for an explicit belief in the UTI the second item is
removed. This was done because the Cronbach’s Alpha (0.347 < 0.7) and the Corrected Item-Total Correlation (0.33 < 0.3) both were too low. As can be seen in table 6 there were three
items that do not meet the requirement of a Cronbach’s Alpha of 0.7, these are: Perceived
health when exposed to an unhealthy fictive brand without informational claims (Group 1),
Perceived health when composed to an unhealthy existing brand with informational claim
(Group 3), and Perceived health when exposed to an unhealthy existing brand without
informational claim (Group 4). Fortunately, all three variables do meet the threshold value
that can be considered as reasonable, which is 0.6. Furthermore, all items showed a Corrected
Item Total Correlation > 0.30 and the difference of the Cronbach’s Alpha if an item was deleted was not > 0.10. Thus, also these items can be used for further analysis and no item
30 Explicit belief UTI* Gr oup Expose d to advertisement of existing br and Perceive d health Perceive d taste Expose d to advertisement of ficti ve br and Perceive d health Perceive d taste α α α α α 0.739 Group 1 0.771 0.902 0.651 0.947 Group 2 0.792 0.904 0.773 0.978 Group 3 0.643 0.904 0.915 0.935 Group 4 0.663 0.957 0.841 0.983
Table 6: Reliability scores of all items. * Explicit belief UTI were asked before getting divided into a group
5.3 Correlation analysis
In this section a table with the means and standard deviations as well as a correlation matrix
have been created. Table 7 gives an overview of the former whilst table 8 shows the
correlations. For this analysis the independent variables (an implicit belief in the UTI and
explicit belief in the UTI), the dependent variables (perceived health and perceived taste), and
the control variables (gender, age, educational background, and income) are included. It
should be noted that the moderator variables (informational claims and brand image) are
included in these variables. This is the case because these stimuli are manipulated in the
31
Vari ables N of i te ms Mean St. De vi ation n
Implicit IAT (c ongruent) 16 4.725 7.726 189
Implicit IAT (inc ongrue nt) 16 2.178 1.087 189
Explicit IAT 2 2.050 0.811 189 Gr oup 1 HBICPH 3 4.560 1.036 47 HBICPT 2 2.957 1.041 47 FUZCPH 3 1.887 0.769 47 FUZCPT 2 3.670 1.606 47 Gr oup 2 HBZCPH 3 4.270 1.233 47 HBZCPT 2 3.149 1.108 47 FUICPH 3 2.482 1.161 47 FUICPT 2 4.170 1.364 47 Gr oup 3 UBICPH 3 2.227 1.123 47 UBICPT 2 3.511 1.627 47 FHZCPH 3 4.142 1.529 47 FHZCPT 2 2.947 1.230 47 Gr oup 4 UBZCPH 3 2.444 3.927 48 UBZCPT 2 3.927 1.857 48 FHICPH 3 4.757 1.093 48 FHICPT 2 2.740 0.951 48
Table 7: Mean and S t. Deviation of all items
It becomes clear from table 7 that both existing and fictive brands with a healthy image scores
higher on perceived health compared to an unhealthy brand image. O n t he other hand the
opposite is true, both existing and fictive brands with an unhealthy image scores higher on
perceived tasted compared to a healthy brand image. Another interesting result is that the
mean response latency of the congruent condition of the IAT is higher than the mean response
latency of the incongruent condition. In the next section a more detailed analyses will be
32 Vari able M SD 1 2 3 4 5 6 7 8 9 1. Gender 1.49 0.501 1 2. Age 34 12.828 -0.16* 1 3. Education 3.99 0.970 0.40 -0.51** 1 4. Income 2.56 1.662 -0.29** 0.57** -0.13 1 5. IAT congruent 4.73 7.726 -0.09 0.25** -0.17* 0.16* 1 6. IAT incongruent 2.18 1.087 -0.15* 0.52** -0.40** 0.25** 0.29** 1
7. Exp lic it IAT 2.05 0.811 -0.13 0.05 -0.15* -0.02 0.01 0.03 1 8. Perceived health 3.34 0.879 -0.03 0.14* -0.02 0.06 -0.09 -0.01 0.00 1 9. Perceived taste 3.36 0.963 0.00 0.13 -0.08 0.06 0.05 0.10 0.03 -0.19** 1
* Corre lation is significant at the 0.05 level (2-tailed). ** Corre lation is significant at the 0.01 level (2-tailed).
Table 8: Correlation matrix
Table 8 shows that among the control variables four significant correlations can be observed.
Although age is negatively correlated (r = -0.16) with gender, it can be concluded that there is
an absence of relation since the coefficient is between -0.2 and 0.2. Education is negatively
correlated (r = -0.51) with age. This phenomenon can be explained due to the fact that the
youngest part of the respondents consisted primarily of students with an academic
background. Apparently the group of elder participants had a more diverse educational
background. The income of the participants is negatively correlated (r = -0.29) with gender.
Gender has been coded as 1 being male and 2 being female. This indicates that males have a
higher income than females. Finally, income is also positively correlated (r = 0.57) with age.
This can be explained by the fact that the younger part of the respondents are mainly students
which in general have a low income. Furthermore, elder people usually have a higher income
compared to younger workers due to more work experience.
There are also a couple of significant correlations between the control variables and
33
significantly correlated with age (r = 0.25), education (r = -0.17), and income (r = 0.16).
Moreover, the incongruent condition of the IAT is significantly correlated to all control
variables. That is, gender (r = -0.15), age (r = 0.52), education (r = -0.4), and income (r =
0.16). The correlation coefficients of both the congruent and incongruent conditions with
respect to the control variables have the same direction. Hence, this indicates that the elder
group of the participants, which also have a higher income, have higher response latencies
than the younger participants. Furthermore, it also appears that a lower educational
background and being a male (only in the incongruent condition) also leads to higher response
times. Additionally, the explicit IAT is negatively correlated (r = -0.15) with education and
perceived health is positively correlated (r = 0.14) with age. However, no further inference
can be made because the coefficient are pretty close to zero, which indicates that there is an
absence of a relation.
Finally, also some significant correlations among the variables of the conceptual
model can be reported. The incongruent condition of the IAT is positively correlated (r =
0.29) with the congruent condition of the IAT. This is as expected since both conditions
together form the belief in the UTI of the participants. Lastly, perceived health is negatively
correlated (r = -0,19) with perceived taste. Also this is as expected, because as stated by
Raghunathan et al. (2006) food with a good taste is incompatible with a high degree of
healthiness.
5.4 Hypothesis Testing
In this section of the paper tests were performed in order to assess the hypotheses derived
from the proposed conceptual model.
H1. Despite the fact that consumers consume healthier food, in general they still have a belief in the Unhealthy = Tasty Intuition.
34
The first hypothesis will be tested in two ways. First, a Paired-Samples T Test was
performed. Therefore, the mean scores of both the congruent and incongruent condition in the
IAT were computed. As can be seen in table 9 the mean latency of the congruent condition
denoted 4725.40 milliseconds. In contrast, the mean latency of the incongruent condition was
2178.10 milliseconds. Both conditions were answered by the same sample and, therefore, a
matched pairs of observations originated. Hence, for testing the difference in the means, a
Paired-Samples t-test have been conducted. As already mentioned the mean response latency
of the incongruent condition is lo wer than the congruent condition. This difference is
significant with a p-value of 0.000 < 0.001. Thus, participants were able to respond
significantly faster when asked to group healthy foods with words describing tastiness, rather
than words describing lack of taste. This indicates that participants have a stronger implicit
association between healthy foods and tastiness compared to unhealthy foods and tastiness.
Mean n St. De vi ation
IAT congruent 4.725 189 7.725
IAT incongruent 2.178 189 1.087
Mean St. De vi ation St. Err or Mean t df Sig. (2-taile d) Pair IAT_ Con – IAT INCon 2.547 7.484 0.544 4.679 188 0.000
Table 9: Paired-S amples T Test of the implicit IAT
Second, the explicit belief in the UTI was analyzed. Therefore the two items of the
explicit IAT were averaged. As can be seen in table 10, the results indicate a mean of 2.050,
which is almost identical to the score ‘disagree’. The p-value that resulted from the O
ne-Sample T Test were 0.000 < 0.05. Thus, participants significantly disagree explicitly with the
belief that healthy food is considered as untasty.
n Mean St. De vi ation t df Sig. (2-taile d) Exp lic it IAT 189 2.050 0.811 34.738 188 0.000
35
After doing an analysis among the participants who do have an explicit belief in the
UTI (n = 34 with M ≥ 3), the results show that they do have an implicit belief in the HTI.
Participants had a mean response latency of 4307.1 milliseconds in the congruent condition
and 2215.4 milliseconds in the incongruent condition. The mean difference equals 2091.7
milliseconds which is statistically significant with a p-value of 0.000 < 0.001. Interestingly,
the mean difference of the latter case is lower than the mean difference of the general IAT
(2091.76 milliseconds < 2547.25 milliseconds). This result indicates that participants who
explicitly agree that unhealthy foods are more tasty, in contrast have a stronger implicit belief
in the HTI compared to participants that explicitly do not agree in the UTI. For this re ason, an
Independent-Samples t-test was performed in order to test if the difference in mean response
latency between those two group was statistically significant. As can be seen in table 11, the
p-value of Levene’s Test equals 0.579 > 0.10 and, therefore, equal variances can be assumed.
In this case there is a p-value of 0.696 > 0.05. Thus, it can be concluded that the mean
difference between those groups is not statistically significant.
n Mean St. De vi ation IAT ≥ 3 34 2.092 3.149
< 3 155 2.647 8.135
Levene’s Test for Equality of Vari ances
t-test for Equality of Means
F Sig. t df Sig. (2-tailed)
IAT Equal variances assumed Equal variances not
assumed
0.309 0.579 -0.391 187 0.696
-0.655 137.283 0.513
Table 11: Independent-S amples T Test for mean response latency differences between respondents who explicitly agree to a belief in the UTI compared to those who explicitly disagree
In sum, the results show that in general participants do not belief that unhealthy food is
36
significant stronger implicit belief that healthy foods are more tasty. In addition, although
some participants indicated that they do have an e xplicit belief in the UTI, the opposite seems
true. The results show that these participants also had an implicit belief in the HTI. This belief
was even stronger, even though it was statistically insignificant, compared to participants who
did not had an explicit belief in the UTI. Hence, this indicates that in general the respondents
also had an explicit belief in the HTI. It can thus be concluded that hypothesis 1 should be
rejected.
H2. Using informational claims for food products that emphasize both health and taste have a positive effect on the perceived health of the product.
H3. Using informational claims for food products that emphasize both health and taste have a positive effect on the perceived taste of the product.
5.4.1 Perceived health existing brands
In order to test Hypothesis 2, first a dummy variable was computed with 0 = informational
claims absent, and 1 = informational claims present. Hereafter, two Independent-Sample t-test
were performed in order to test if the mean scores were significantly different from each
other. In the first Independent-Samples t-test the healthy brand ads and in the second the
unhealthy brand ads have been analyzed. The results can be found in table 12 and 13.
Infor mational Clai ms Dummy n Mean St. De vi ation Healthy brand image Absent 47 4.270 1.233
Present 47 4.560 1.036
Levene’s Test for Equality of Vari ances
t-test for Equality of Means
F Sig. t df Sig. (2-tailed)
Healthy brand image Equal variances assumed Equal variances not assumed 1.895 0.172 -1.238 92 0.219 -1.238 89.345 0.219
37
Infor mational Clai ms Dummy n Mean St. De vi ation Unhealthy brand image Absent 48 2.444 1.184
Present 47 2.227 1.123
Levene’s Test for Equality of Vari ances
t-test for Equality of Means
F Sig. t df Sig. (2-tailed)
Unhealthy brand image Equal variances assumed Equal variances not assumed 0.151 0.698 0.918 93 0.361 0.919 92.909 0.361
Table 13: Independent-S amples T Test – Unhealthy brand image
In the case of the healthy brand ad the mean of perceived health is 4.270 without
informational claims, and with informational claims the mean is 4.560. This indicates that
participants somewhat agree that the y perceive Arla Skyr Yoghurt as healthy. Furthermore,
participants who were exposed to the ad with informational claims have a higher level of
perceived health than participants who were exposed to the ad without informational claims.
However, for the unhealthy brand ads the means were 2.444 in the without claim condition
and 2.227 in the ad with informational claims. This indicates that participants do not perceive
Coca Cola zero as healthy and no further inferences can be made.
Hence, for the healthy brand the statistics can be further analyzed. The results show
that equal variances can be assumed because the p-value of Levene’s Test is 0.172 > 0.10.
However, the p-value for the Independent-Samples t-test is statistically insignificant, because
the p-value is 0.219 > 0.05. Thus, for existing brands with both a healthy and unhealthy image
it can be concluded that consumers do not perceive the food product as healthier when
informational claims are used in advertisement.
5.4.2 Perceived health fictive brands
The same procedure was performed for the fictive brands which were designed for this