0% added healthiness : the persuasive power of front-of-package food claims and the role of scepticism amongst Dutch consumers

(1)

‘0% added Healthiness’

The Persuasive Power of Front-of-Package Food Claims and the Role of

Scepticism amongst Dutch Consumers

Author: Lucia Kabia

Student ID: 10003022

Master’s Thesis

Graduate School of Communication

Master’s programme Communication Science

Persuasive Communication Track

Supervisor: Dr. Paul E. Ketelaar

24-06-2016

(2)

i

Voorwoord

In 2010 zette ik voor het eerst voet op wetenschappelijke bodem, van de Universiteit van Amsterdam. Mijn intrede in de Geesteswetenschappen – Media & Cultuur welteverstaan – liet mij kennismaken met het gedachtegoed van grote denkers als Jean Baudrillard, Stuart Hall en Marshall McLuhan. Ik leerde om dieper in te gaan op filosofische denkwijzen en betekenislagen, om vraagtekens te zetten bij oppervlakkigheden, en heb dusdanig een kritische blik en

ontledende schrijfstijl ontwikkeld. Wat tijdens mijn bachelor al opviel, was dat mediatheorieën die zich richtten op communicatie en sociaalwetenschappelijke kwesties in het bijzonder mijn interesse wekten (e.g. Hall’s encoding-decoding, McLuhan’s ‘the medium is the message’). Na het behalen van mijn Bachelor of Arts heb ik er aldus voor gekozen om mij niet verder te ontwikkelen in de diepte, maar juist in de breedte. Zodoende ben ik – na het volgen van een schakelprogramma statistiek en methodiek, en een aantal maanden stage te hebben gelopen in marketing en publiciteit bij een filmdistributeur – uiteindelijk terecht gekomen bij de master Persuasive Communication Science.

In een jaar heb ik ontzettend veel geleerd. Waar ik doorgaans vlekkeloos door mijn bachelor ben gevlogen, werd ik tijdens mijn master – in een heel nieuw veld met nieuwe theorieën, methoden en analysetechnieken – behoorlijk op de proef gesteld. Er werd van mij verwacht dat ik (statistische) verbanden kon leggen tussen mensen en hun gedrag, en deze vervolgens kort en bondig kon rapporteren, zonder enige poespas. ‘Less is more’ was bij mij echter nooit aan de orde geweest (ter indicatie: mijn bachelor scriptie bestond na flink schrappen desalniettemin uit 16.000 woorden, met een richtlijn van 10.000 woorden). Tijdens mijn master had ik dan ook vaak moeite om mijn (ik geef toe, soms ietwat ver gaande) gedachtenkronkels in toom te houden wanneer ik papers moest schrijven met een limiet van 4 pagina’s. Keer op keer kreeg ik te horen over mijn interpretaties: ‘dit kun je zomaar niet stellen’, of ‘kom eens meteen to

(3)

ii

the point’. Uiteindelijk ben ik er toch redelijk in geslaagd om mijzelf duidelijker en bondiger uit

te drukken, zonder dat dit ten koste gaat is gegaan van inhoud. Daarnaast heb ik ontzettend veel verschillende facetten van de persuasieve communicatie meegekregen, die ik toen ik aan deze studie begon niet voor mogelijk zag. Zo ben ik positief verrast door het vak health

communication, waarbij ik veel heb geleerd over het opzetten en analyseren van

gezondheids-interventies, en heb ik door het schrijven van deze scriptie tevens uitgebreide kennis opgedaan over food marketing. Ik ben erg dankbaar voor alles wat ik heb geleerd en de goede begeleiding van alle docenten die mij constant intellectueel hebben uitgedaagd en mijn prestatiedrang bleven voeden. In het bijzonder wil ik mijn begeleider Paul Ketelaar bedanken, die mij gedurende mijn hele scriptieperiode heeft voorzien van razendsnelle en waardevolle feedback.

Door alles wat ik in relatief korte tijd heb geleerd voel ik mij voor het eerst echt

klaargestoomd voor het leven na mijn studie. Ik heb talloze interessante persuasieve technieken en theorieën geleerd waarvan ik zeker ben dat ik ze tijdens mijn loopbaan in de praktijk zal brengen. Ik ben buitengewoon trots op wat ik nu al heb bereikt en kijk voornamelijk terug op mijn studietijd als een zelfontwikkeling van ongekende omvang – met een paar kleine

zenuwinzinkingen, maar vooral heel veel ups, creativiteit, en opgedane kennis. Deze scriptie, die mij de titel Master of Science heeft opgeleverd, is de ultieme kroon op al mijn harde werk. De afsluiter van een zeer belangrijk hoofdstuk uit mijn leven, mijn studententijd, waarin ik ben gegroeid tot wie ik vandaag de dag ben. Met enige spanning maar ook heel veel zin zal ik echter aan een nieuw hoofdstuk beginnen, een carrière in de marketing en communicatie. The future

starts today!

(4)

iii

Abstract

Objective: The average supermarket nowadays has become a battlefield of products with

health-related symbols, logos and claims, all fighting for the attention of the consumer. These heuristics often prime ‘healthy thoughts’ in consumers’ minds, thereby presenting products as healthier than they actually are. This research investigated the persuasive power of front-of-package food claims, and whether scepticism could alter the way in which consumers are influenced by claims when assessing a product’s healthiness, and if that in turn has an effect on their purchase intentions.

Method: An online experiment (N = 182) with a 3 (Claim type: Objective claims vs. Subjective

claims vs. No claims (Control)) X 2 (Level of scepticism towards claims: High scepticism vs. Low scepticism) between-subjects factorial design was conducted. Results: Contrary to expectations, results showed that consumers who were exposed to objective or subjective claims did not express higher ratings of perceived healthiness or purchase intentions than consumers who were exposed to no claims at all, and moreover, scepticism towards claims did not play a moderating role in this relationship. However, highly sceptical consumers expressed significantly lower ratings of perceived healthiness and purchase intentions than less sceptical consumers – regardless of whether they were exposed to claims or not. This indicates that consumers were not influenced by claims when making inferences about the product they were exposed to, but rather by whether or not they were sceptical towards the product itself. Implications: The persuasive power of food claims diminishes when a claim is mismatched by the product type it appears on, and moreover when consumers take a sceptical stance towards the product itself. Therefore, food manufacturers are in any case advised to use a tailored marketing strategy approach for each of their products, which assesses which claims suit best for which product type.

Keywords: food marketing, objective and subjective claims, FOP information, nutrition, healthiness, purchase intentions, consumer scepticism

(5)

iv

Introduction

For the first time in human history, obesity is causing a bigger health crisis than malnutrition (Adams, 2012). A global landmark study which was published in 2012 in The Lancet reported alarming results about diseases linked to obesity such as diabetes type II, heart disease and cancer, which are killing triple the number of people who die from being underfed (Adams, 2012). Moreover, The World Health Organisation predicted that, within a couple of years, obesity-related diseases will be the leading causes of death in all countries of the world, even in the poorest (Berreby, 2013). And although it seems like people want to eat healthier and improve their diets – especially in the Western countries (Voedingscentrum, 2011; Pomeranz, 2013; Graham & Mohr, 2014) – recent studies have shown that no country has successfully reduced the obesity problem so far (Lee, 2016), thereby raising questions about underlying causes for this threatening global health problem.

Perhaps one should take a walk through the average supermarket, which has nowadays become a battlefield of products with health-related symbols, logos and claims, all fighting for the attention of the consumer. While front-of-package (FOP) information is supposed to help consumers in making fast, but well-informed food choices, manufacturers are increasingly causing a serious information asymmetry (Verbeke, 2005; Van Trijp & Van der Lans, 2007). They do so by luring consumers with prominently placed claims and symbols associated with health, while condemning other less healthy (yet factual) aspects of a food product to the

Nutrition Facts panel on the backside of the packaging – which is a mandatory list of ingredients, additives, allergens, and nutritional values (Van den Berg, Van Dooren, Peters, Van der Vossen-Wijmenga, & Werkman, 2011). Consumers who do not take a sceptical stance towards FOP food claims, or those who lack time, abilities and motivation to thoroughly inspect all nutrition

(8)

2 misled by manufacturers presenting their products as healthier than they actually are (Andrews, Burton, & Netemeyer, 2000; Mariotti, Kalonji, Huneau, & Margaritis, 2010; Pomeranz, 2013). However, in the current Western capitalistic society, food companies must use claims and other eye-catching FOP information in order to set their brand apart from competitors, which at the same time enables consumers to differentiate between products when standing in front of an overflowing shelve in the supermarket (Vieira, de Castro Alcantara, do Prado, Pinto, & de Rezende, 2015). Moreover, with an increasing demand for healthful products and the rise of so-called ‘functional foods’ – i.e. foods that contain special technologically developed health-promoting ingredients (Siró, Kápolna, Kápolna, & Lugasi, 2008) – manufacturers increased their use of nutritional and health claims.

The current research investigates the persuasive power of FOP health-related food claims. Yet, first thing to note is that not every claim or label can be put on a product without any

constraints. In the Netherlands, food manufacturers are bound to rules imposed by the European Union, following Regulation (EC) no 1924/2006. All FOP claims that have not been approved by the EU Commission (EC) – after scientific review by the European Food Safety Authority

(EFSA) – will be banned (Mariotti et al., 2010). According to this legislation, regulated claims are divided into three types, that is; nutritional claims (NCs), health claims (HCs), and reduction of disease risk claims (RDRCs). NCs state the presence or absence of certain substances in foods (e.g. ‘no added sugars’, ‘fat-free’, ‘rich in fibre’, etc.), whereas HCs are based on a relationship between an ingredient and a health benefit from that ingredient (e.g. ‘Vitamin C increases iron absorption’) (Van Buul & Brouns, 2015). RDRCs imply that ‘the consumption of a food

category, a food, or one of its constituents significantly reduces a risk factor in the development of a human disease’ (Mariotti et al., 2010). Next to regulated claims, there are also claims and

(9)

3 buzz words which are based on feelings and emotions rather than verifiable facts, and therefore they can be freely used without restrictions. Most of these words however are linked to ‘health’ by food manufacturers (Northup, 2014). Examples of commonly used unregulated claims are; ‘fresh’, ‘traditional’, and ‘pure’ – and in the Netherlands also bourgondisch (i.e. ‘hearty’) and ambachtelijk (i.e. ‘made with craftsmanship’) (Pell, 2004; ‘’Loze Kreten’’, 2015).

NCs and unregulated claims are most easily made because – unlike HCs and RDRCs – they require no or limited rules or scientific evidence of any causal relationship, making them very attractive and commonly used by food marketers (Van Buul & Brouns, 2015). Hence, the current research will solely focus on these types of claims. Since NCs are based on verifiable facts – that is, objective information – these types of claims will from now on be referred to as

objective claims. Unregulated claims are more susceptible to individual interpretation and thus

these claims will be referred to as subjective claims (Ford, Smith, & Swasy, 1990).

In previous studies, both objective and subjective food claims appeared to be positively linked to consumers’ perceived healthiness of products carrying such claims (Van Trijp & Van der Lans, 2007; Burton, Andrews, & Netemeyer, 2000; Graham & Mohr, 2014; Northup, 2014). Moreover, it is proved that personal characteristics such as gender and age influence how consumers evaluate food claims (Van Wezemael, Caputo, Nayga, Chryssochoidis, & Verbeke 2014; Lähteenmäki, 2013). Yet there is another trait expected to affect claim evaluations, but of which empirical evidence is scarce, that is, scepticism. In the current research, scepticism is defined as ‘the general tendency of consumers to disbelieve nutrition information’ (Petrovici, Fearne, Nayga Jr., & Drolias, 2012).Scepticism stems from Persuasion Knowledge, meaning that ‘consumers develop beliefs about advertising and promotional tactics that advertisers use to persuade them. Over time, consumers may become more sceptical of such tactics and thus they

(10)

4 may become more reactant to them’ (Friestad & Wright, 1994). This reactance can for example be expressed in negative purchase intentions, as products carrying claims or logos which are recognized as marketing tactics will likely be avoided (Elving & Steenhuis, 2014). Consumer programs such as Keuringsdienst van Waarde, Radar and Kassa, and independent organisations like Foodwatch and De Consumentenbond are known to reveal misleading food marketing cases, as they attempt to increase awareness and scepticism amongst Dutch consumers towards claims, labels and logos on food and beverages sold at supermarkets.

Moreover, it is argued that ‘scepticism provides consumers with a ‘’healthy’’ viewpoint from which to make product evaluations’ (Mohr, Eroǧlu, & Ellen, 1998), which makes it an important construct to examine. Yet, no empirical research has been conducted on the role of scepticism on perceived healthiness and purchase intentions of Dutch consumers, when they are exposed to objective claims compared to subjective claims. However, it is important to discover if scepticism can alter the way in which consumers are influenced by FOP food claims when assessing a product’s healthiness, as misleading claims can have a major negative impact on public health when consumers who are low in scepticism perceive products with such claims as healthier than they might actually be (Mariotti et al., 2010).

Hence, the current research will examine the gap that was already emphasized in 1998 by Mohr et al., as they argued that ‘consumers may be aware that marketers and advertisers may attempt to mislead them, but it is not clear whether this scepticism substantially alters the way in which consumers evaluate individual cases (i.e. FOP food claims) or reduces the likelihood of them being misled’. It is expected that scepticism towards a FOP food claim – especially at a high level – diminishes the positive effect of that claim on perceived healthiness and purchase intentions (Mohr et al., 1998). However, research is needed in order to establish such effects.

(11)

5 Therefore, this study aims to answer the following question: What is the role of Scepticism on

Perceived Healthiness & Purchase Intentions of Dutch consumers when they are exposed to

products with Objective Claims, compared to products with Subjective Claims?

Theoretical Background

Elaboration Likelihood Model

Running errands is for most consumers a routinely, low-involvement process, causing them to not consciously reflect about every item they put into their shopping basket (Hoyer, 1984). Empirical findings of consumer research explain that consumers believe search costs are relatively high – especially for routinely purchased products – because the risks of buying a dissatisfying low-involvement product (which could be easily replaced the next time one visits a supermarket) do not outweigh the costs of time loss due to extensive searching (Ford et al., 1990; Hoyer, 1984). Therefore, most consumers are not motivated to process all nutrition information that can be found on the front and backside of food and beverage packaging when doing

shopping (Ford et al., 1990). Hence, many consumers make use of the most easily accessible information, which is generally found at the front of packaging (Keller et al., 1997).

According to the Elaboration Likelihood Model (Cacioppo & Petty 1986), people process persuasive messages through either one of two routes of information processing, being a central or a peripheral one. Central route processing means that people engage in careful, cognitive consideration and evaluation of information, whereas peripheral route processing is characterized by a more emotional and aesthetic way of making judgements about the quality of a product or message (Samson & Voyer, 2012; Belch & Belch, 2015). Since most consumers only look at claims on products for a very short time – that is, less than 4 seconds – they will not engage in extensive information processing (Van Buul & Brouns, 2015). Moreover, the Nutrition Facts

(12)

6 panel at the backside of products can be hard to understand by the average consumer (Northup, 2014; Graham & Mohr, 2014), and even if consumers understand the Nutrition Facts panel properly, they are inclined to truncate their information search to the front of a package, as was found in a study of Ford and colleagues (1990). Thus, during shopping – when there is little motivation and ability to process – consumers’ thoughts take the peripheral route, making them depend on heuristics in order to make choices and inferences. These heuristics can for example be someone’s favourite trusted brand or the colour or shape of a product’s packaging, but it can also be FOP claims, both objective and subjective (Keller et al., 1997; Verbeke, 2005; Sütterlin & Siegrist, 2015).

Consumer Positivity Biases Explaining Healthiness Ratings

A reason why consumers perceive objective and subjective claims on food packaging as sufficient information sources to make quality judgements or purchase decisions, is that these claims activate already-stored knowledge or concepts. Priming is the triggering of a concept (e.g. ‘vitamin C’), which is then available to implicitly influence later thoughts and behaviours

(Northup, 2014). These concepts are stored in memory as nodes, being highly connected with other concepts (e.g. ‘healthy’). When a node is primed by a stimulus (e.g. a fruit juice package carrying a claim), the node becomes activated and easily accessible in memory. If the trigger (i.e. prime) is strong enough, the activation of the first node can be spread to the highly connected other concepts, making those easily accessible as well (Northup, 2014). Food marketers have successfully linked concepts like ‘fresh’ and ‘vitamin C’ to ‘healthy’, making those concepts activated from within consumers’ memory when they are exposed to those trigger words (Northup, 2014). As a consequence, priming can induce a positivity bias, causing consumers to overrate products due to the sheer presence of a FOP food claim. This effect is also called a

(13)

7

mere-label effect (Van Trijp & Van der Lans, 2007).

Another consumer bias found in previous studies is a so-called health halo effect, which causes consumers to make positive inferences about the overall healthiness of a product, based on the presence of a health-related claim about one particular nutrient or ingredient (Andrews et al., 1998; Roe, Levy, & Derby, 1999). In other words, one single ‘healthy’ aspect of a product is generalised to other aspects of the product not mentioned in the claim (Van Trijp & Van der Lans, 2007). In line with previous studies on the impact of FOP food claims, in which results showed that products labelled with both objective and subjective claims were perceived as healthier than products without such claims (Ford et al., 1990; Van Trijp & Van der Lans, 2007; Aschemann-Witzel & Hamm, 2010; Northup, 2014), it is hypothesized that:

H1: Products carrying FOP food claims (i.e. objective and subjective) will induce

higher ratings of perceived healthiness than products without any FOP food claims.

Theory of Reasoned Action to Predict Food Choice

According to attitude models such as the Theory of Reasoned Action (TRA), people’s behaviour and decisions are driven by whatever information happens to be available (Ajzen & Fishbein, 1980). At the core of this theory lays the notion that people’s intention to perform a certain behaviour is a latent variable that is determined by attitudes towards performing that behaviour and subjective norms, that is, perceived social pressure to carry out the behaviour in question (Petrovici, Ritson, & Ness, 2004). Although several scientists used to perceive social norms as weak predictors of the consumption of low-involvement goods (Shepherd, Sparks, Bellier, & Raats, 1992; Petrovici et al., 2004), social pressure likely became more salient with the emergence of (healthy) food trends rapidly spreading through social media (Helm, 2015). In the field of food choice studies, the TRA has been widely used to predict consumption

(14)

8 of and attitudes towards food and beverages (Shepherd, 1988; Petrovici et al., 2004; Burton et al., 2000). Consumer attitudes towards a certain food or beverage – which again influence intent dependent variables (e.g. purchase intentions) – are generally affected by product-related beliefs and evaluations (Shepherd et al., 1992). Such a product-related belief can be an affective reaction to food, such as perceived tastiness, or it can be a cognitive reaction such as pricing, nutritional value or perceived healthiness (Trendel & Werle, 2015). Both reactions can occur automatic and implicitly when one is exposed to a food product, and more often affective and cognitive

reactions are negatively correlated, in a sense that for example chocolate can be perceived as very tasty but at the same time very unhealthy. Reason why affective and cognitive reactions occur automatic is because they both work through priming (Trendel & Werle, 2015). As was previously discussed, objective and subjective claims referred to in the current research are presumed to prime health-related thoughts in consumers’ minds and therefore it is expected that their purchase intentions will be more strongly predicted by cognitive automatic reactions – that is, perceived healthiness – instead of affective reactions.

Moreover, previous research has found that perceived healthiness can play a mediating role in the relationship between exposure to a product with health-related FOP claims or logos and attitudes and purchase intentions for that product (Burton et al., 2000; Graham and Mohr, 2014). Next to this, a study conducted by Aschemann-Witzel & Hamm (2010) found that consumers who were more likely to choose food products labelled with FOP claims, also perceived these products as healthier. Therefore, it is expected that:

H2: Perceived healthiness will have a positive mediating effect on the relationship

between exposure to FOP food claims (i.e. objective and subjective) and purchase intentions, such that the healthier a product is rated – due to the claims the product is

(15)

9 carrying – the higher consumers’ purchase intentions will be.

The Role of Scepticism in Evaluations of Products Carrying FOP Claims

It is expected that in our current (Western) society people will become more and more sceptical towards processed and packaged foods (Elving & Steenhuis, 2014) due to, amongst other things, an increase in consumer programs that discuss food-related topics, and heavy use of social media which has a great contribution in creating public attention for food scandals and other food-related issues (KRO-NCRV, 2016; Raymond & Flood, 2014). Moreover, a focus group research conducted by the U.S. Food and Drug Administration (FDA) indicated that consumers were highly sceptical towards health and nutritional claims because they perceived these as attempts of manufacturers to sell more of their products (Keller et al., 1997), and likewise results emerged from the FDA Health and Diet Survey – which was conducted that same year – where 79% of the respondents strongly or somewhat agreed with the statement ‘’Claims on food labels are more like advertising than anything else’’ (Roe et al., 1999). Hence, it is known that consumers share sceptical thoughts about claims on food and beverages, but no empirical evidence is present about whether scepticism also alters the way in which consumers make inferences about a product’s healthiness when they are exposed to health-related FOP claims, and whether this in turn influences their purchase intentions. It is consequently hypothesized that:

H3a: Scepticism towards claims will have a moderating effect on the relationship between

exposure to FOP claims (i.e. objective and subjective) and perceived healthiness and purchase intentions, such that highly sceptical consumers will show lower ratings of perceived healthiness and lower purchase intentions than less sceptical consumers when they are exposed to FOP claims.

(16)

10 Theory of Economics of Information Explaining Higher Scepticism for Subjective Claims According to the Theory of Economics of Information, consumers show differences in scepticism depending on the ease and costs of evaluating the veracity of a claim (Stigler, 1961; Ford et al., 1990). Since objective information (e.g. fat percentage) is more easily verified than subjective information (e.g. ‘pure’), it is argued that consumers have a preference for objective claims over subjective claims (Ford et al., 1990; Andrews et al., 1998). In consonance, the U.S. Federal Trade Commission assumes that consumers are more wary of subjective claims because they are more ‘puffery’ (Ford et al., 1990). Ford et al. (1990) supported this assumption with empirical findings proving that respondents were more sceptical towards subjective claims than towards objective claims, and Burton and colleagues (2000) found that consumers felt that there was a conscious attempt to mislead them through the use of a general, subjective claim

(‘Healthy’), but these feelings did not occur for a specific, objective claim (‘No cholesterol’). Above mentioned findings indicate that subjective claims are likely to be subject to heightened scrutiny of sceptical consumers because of their lack of verifiability and superfluous appearance, to an extent that these types of claims might even evoke contrariwise effects on consumer evaluations (Andrews et al., 1998, 2000; Aschemann-Witzel & Hamm, 2010). A so-called boomerang effect – i.e. an unintended consequence of an attempt to persuade, resulting in the adoption of an opposing position instead (Jenkins & Dragojevic, 2011) – amongst highly sceptical consumers is expected to emerge, in such a way that these consumers perceive products with subjective claims as less healthy than they might actually are, which consequently can result in resistance towards purchasing products carrying such claims. It is therefore hypothesized that:

H3b: Scepticism will induce a moderating boomerang effect on the relationship between

(17)

11 intentions, such that consumers who are highly sceptical will rate food products carrying such claims as even less healthy than food products without such claims.

Figure 1. Conceptual Model of Hypothesized Interactions.

Method

Design

In order to reveal any causal effects as outlined in figure 1, an experiment with a 3 (Claim type: Objective claims vs. Subjective claims vs. No claims (Control)) X 2 (Level of scepticism towards FOP claims: High scepticism vs. Low scepticism) between-subjects factorial design was conducted. The experiment was embedded in an online survey format, which was constructed in Qualtrics. Dutch consumers were randomly assigned to one of three conditions, in which they were exposed to an image of the front of a package of mango-orange juice carrying either objective claims, subjective claims or no claims at all.

Sample

Participants needed to be at least 18 years old, understand and read Dutch, and visit a supermarket on a regular basis. These requirements were all stated in the recruitment text. Data was collected through convenience sampling, together with snowball sampling. Participants were recruited through social media, respondent websites, e-mail and personal addressing. When the

(18)

12 experiment was closed after two weeks of data collection, 220 Dutch-speaking consumers (Mage

= 35, SD = 16.3, 75% women) finished the survey.

However, since differences between the three conditions were hypothesized due to the manipulation of claim type, it was important to exclude participants from the sample who stated having seen any claims that belonged to a condition other than the condition they were assigned to. Furthermore, another criteria for inclusion in the final sample was that participants in the experimental conditions reported having seen at least two of three claims that matched their condition. Thus, following the responses on the manipulation check, 23 participants from the control condition were excluded, next to 7 participants from the objective claims condition and 8 participants from the subjective claims condition (total number of drop-outs N = 38), leaving a final sample of N = 182 (Mage = 34, SD = 15.6, 75.3% women) of which 50.5% was or had been

a university student. The final sample sizes of the three conditions were NControl = 61, NObjective =

59, and NSubjective = 62.

Product Choice

In the search for proper stimulus material, the following requirements were taken into account; the food product had to be familiar amongst Dutch consumers and the presence of both subjective and objective FOP claims needed to be realistic, in order to reduce experimental artificiality. Moreover, it was expected that scepticism reduces consumers to be misled, and therefore it was desired that the product had a somewhat misleading character – that is, FOP claims on such products can make them look healthy, but in fact the product has been proven to be not so healthy at all – in order to detect greater differences between high and low sceptical consumers on ratings of perceived healthiness. Furthermore, the product should not induce any

(19)

13 healthiness at baseline (Andrews et al., 2000) – since these effects do not allow for much

variance on the measure of perceived healthiness, which was important to properly test the hypotheses of the current research.

Although previous studies on the effects of FOP food claims commonly used yoghurt as a stimulus (Andrews et al., 1998, 2000; Ford et al., 1990; Keller et al. 1997; Roe et al., 1999; Lähteenmäki et al., 2010), this product was proven to cause a ceiling effect given its omnipresent healthy image (Van Trijp & Van der Lans, 2007). A recent survey conducted by the Health Ministry of the Dutch State Government found that information on meat products, sodas and fruit juices were perceived as most misleading according to Dutch consumers (Dijkzeul, Stutje, & Kuipéri, 2015). Since sodas were expected to cause a floor effect and meat products typically do not carry both subjective and objective claims, the chosen stimulus for this research was a non-branded1_{cardboard package for mango-orange juice, because it matched all previously} listed requirements (figure 2).

Pre-tests

Claim Choice. First, a small qualitative pre-test (N = 7) was conducted in which

participants were shown 26 cards with FOP food claims2 – 12 nutritional claims (objective), 12 subjective claims, and 2 logos (i.e. the ‘Healthier Choice’ and ‘Conscious Choice’ checkmarks)3 – and their top of mind thoughts about all claims together were asked. Most notable was the fact

1_{The stimulus was non-branded in order to control for brand likeability in measuring ratings of purchase intentions.} 2_{A list of all claims used can be found in Appendix A.}

3_{The Gezondere keuze (i.e. ‘Healthier choice’) checkmark with a green circle and the Bewuste keuze (i.e. ‘Conscious}

choice’) checkmark with a blue circle are nowadays present on almost every product in the average Dutch supermarket. The logos are developed by an association (i.e. Stichting Ik Kies Bewust) that was founded by big manufacturers such as

Unilever and FrieslandCampina, and they are intended to assist consumers in making healthier choices in an easy and

quick manner (Polderman, 2012). The checkmarks will only be placed on a company’s product when a certain amount of money is paid for it (Polderman, 2012). However, they are officially approved by the European Commission, and all products carrying the logos have been scientifically tested on containing less saturated fat, salt or sugar or more fibre compared to other products in the same product category (‘’Vinkje’’, n.d.) and thus they are considered as objective claims in the scope of the current research.

(20)

14 that all participants mentioned the word ‘healthy’, thereby confirming the mere-label effect that the presence of a claim automatically primes ‘healthy thoughts’. When participants took a closer look at the claims, half of them (N = 4) expressed more sceptical thoughts such as ‘oplichterij’ (i.e. ‘scam’) or ‘marketing’. Next, participants were asked to group the claims into at least two categories and the researcher asked them on behalf of which criteria they made their

distinction(s). In general, participants distinguished between claims which were perceived as ‘vague’ and ‘based on feelings’ on the one hand, and claims which were identified as more ‘factual’, ’verifiable’, and ‘content-related’ on the other hand. From each claim type (objective and subjective), three claims that were grouped together in all cases were selected, for the reason that they were most likely to appear on the front of a package of fruit juice.

Difference between Claim Types. A second pre-test (N = 67) was employed in order to

test for significant differences between the two experimental conditions. The pre-test was embedded in an online survey format, constructed in Qualtrics. Participants were randomly assigned to either the objective (N = 35) or subjective claims condition (N = 32) in which they were exposed to an image of a package of mango-orange juice carrying the claims selected from the qualitative pre-test described above (figure 2a & b). Next, participants were asked to rate the claims on the displayed packaging on four self-devised 7-point items with endpoints ‘Totally not based on regulations/Totally based on regulations’, ‘Very vague/Very clear’, ‘Totally based on feelings/Totally based on facts’, and ‘Very incredible/Very credible’.

A principal components analysis with oblique rotation (direct oblimin) was employed and the four items loaded on one factor with eigenvalue > 1 (EV = 2.37) explaining 59.3% of

variance. A Cronbach’s Alpha reliability test proved that the scale was reliable (α = .79) after item 1 (with endpoints ‘Totally not based on regulations/Totally based on regulations’) was

(21)

15 eliminated. The three remaining items were computed into an overall claim evaluation scale and an independent samples t-test was employed in order to compare the mean scores of the two experimental groups. Results showed that evaluations of the objective claims (M = 4.06, SD = 1.29) were significantly different from those of the subjective claims (M = 2.86, SD = 1.28) t (65) = 3.80, p < .001, as subjective claims were perceived as more vague, based on feelings and incredible than objective claims. Therefore, the manipulations turned out to be successful and analyses with two different experimental conditions in the main study proved to be sensible.

Measurement

For the current research, all answers – with the exception of demographics and

manipulation check – were measured on 7-point Likert scales, with the most negative response anchored left and the most positive response anchored right. An overview of the original and translated items from the multi-item scales used in this research can be found in Appendix B.

(22)

16

Dependent variable

Purchase Intentions were measured by one single item (i.e. ‘How likely would you be to

purchase the product, given the information shown on this package?’) that was adopted from a three-item scale from Keller et al. (1997). Reason for choosing just one single item from the original scale was that the other alternatives from the scale would become too similar when translated into Dutch (‘likely’ and ‘probable’ both mean ‘waarschijnlijk’ in Dutch), which in turn can cause irritation when participants experience difficulties in discriminating between three very similar-looking items (Petrescu, 2013). Moreover, three items to measure the same construct can lead to respondent fatigue, since multi-item measures compose an unnecessarily lengthier questionnaire (Bergkvist, 2015). Also, purchase intention is a doubly construct, that is, a construct with a simple, clear object (e.g. fruit juice) and a single and single-meaning attribute (e.g. purchasing), meaning that concepts such as purchase intentions are concrete and

unequivocally understood by participants, and therefore a single-item scale is proved to be sufficient and valid (Rossiter, 2002; 2010; Alexandrov, 2010). Besides, measuring purchase intentions with a single-item scale is very common in food-related studies (Andrews et al., 1998; Roe et al., 1999; Burton et al., 2000; Van Trijp & Van der Lans, 2007; Graham & Mohr, 2014).

Mediator Variable

Perceived Healthiness was measured by a single-item scale adopted from Werle, Trendel,

& Ardito (2013) which asks participants to ‘Rate the healthiness of the stimuli product’. The item was translated into Dutch as follows: ‘Hoe gezond denkt u dat dit product is?’ (1 = ‘Heel

erg ongezond; 7 = ‘Heel erg gezond’). Again, single-item scales are very common to measure

perceived healthiness in food studies and proved to be reliable (Andrews et al., 1998, 2000; Roe et al., 1999; Northup, 2014; Sütterlin & Siegrist, 2015).

(23)

17

Moderator Variable

Scepticism towards FOP claims was measured with a 9-item Consumer Scepticism

towards Advertising Scale, which is developed and tested by Obermiller & Spangenberg (1998). Examples of items are ‘We can depend on getting the truth in most advertising’ and

‘Advertising’s aim is to inform the consumer’ (1 = ‘Totally disagree’; 7 = ‘Totally agree’). Items were translated into Dutch and for each item the word ‘advertising’ was replaced by ‘claims on the front of packaging’ (i.e. ‘claims op de voorkant van verpakkingen’). The new scepticism scale proved to be very reliable (α = .89).

Covariates

Perceived Tastiness was measured with a single-item scale adopted from the same study

of Werle, Trendel, & Ardito (2013) as the perceived healthiness measure used in the current research, since both measures proved to generate significant results in the study of Werle and colleagues. Hence, participants were asked to rate the stimuli product on a scale from 1 = ‘Totally untasty’ (i.e. ‘Helemaal niet lekker’) to 7 = ‘very tasty’ (i.e. ‘Heel erg lekker’).

Persuasion Knowledge was measured with four items from Bearden, Hardesty & Rose’s

6-item Persuasion Knowledge Scale (2001). The first two items from Bearden et al.’s scale were merged into one item (i.e. ‘’I know when an offer is ‘too good to be true’’ and ‘I can tell when an offer has strings attached’) (1 = ‘Totally disagree’; 7 = ‘Totally agree’), since these two items have very similar meanings in Dutch. One item was removed due to close similarity with another question – i.e. both questions were related to recognizing sales gimmicks. The new 4-item

persuasion knowledge scale proved to be very reliable (α = .84).

Demographics were collected at the end of the questionnaire in order to obtain more

(24)

18 Besides, gender, age, and level of education proved to be important demographics in previous food claim studies (Roe et al., 1999; Ascheman-Witzel & Hamm, 2010).

Manipulation checks

During pretesting, significant mean differences between evaluations of objective and subjective claims were revealed. In order to reaffirm these results and test whether the

manipulation also worked in the main study, the claim rating task from the pre-test was repeated. Again, the objective claims (M = 3.68, SD = 1.30) were evaluated significantly different from subjective claims (M = 2.65, SD = 1.07), t (119) = 4.80, p < .001.

A second manipulation check was adopted from Roe et al. (1999) and asked whether participants observed any claims at the package of mango-orange juice they were exposed to (i.e. ‘’Heeft u enige claims (afzonderlijk van de tekst ‘Mango-Sinaasappelsap’) gezien op de eerder getoonde verpakking?’’). Cut-outs of all claims from both conditions and an ‘I did not see any of

these claims’-option were then showed, and participants were asked to select all claims that they

could remember having seen. Participants in the experimental condition who reported having seen less than two claims that matched their condition or any claims that mismatched their condition were eliminated, as well as participants in the control condition who selected any other option than ‘I did not see any of these claims’.

Procedure

After accepting the informed consent – in which the goal of the research and ethical information was outlined (Appendix C) – participants were randomly assigned to one of three conditions (i.e. the objective, subjective or control condition). Next, participants were exposed to their treatment-related image of a package of mango-orange juice (figure 2) and product

(25)

19 purchase intentions. Presenting the stimuli at the beginning of the questionnaire ‘warms up’ the participants, as it is more enjoyable to rate images than filling out a static lengthy questionnaire. After exposure to the stimuli, the moderator variable scepticism towards FOP claims was measured, followed by the highly related concept of persuasion knowledge. Scepticism towards FOP claims was established after exposure to the stimuli because it was expected to serve too much as a prime for the manipulation when measured beforehand, which would induce a negative bias in participants’ product evaluations (i.e. the mediating variable perceived healthiness and the dependent variable purchase intentions).

Participants subsequently continued to the manipulation checks section of the questionnaire. The first manipulation check was solely included for the two experimental

conditions, as this was a claim rating task. The second manipulation check, at which participants had to report which claims they could remember having seen, was accordingly asked to all participants. The questionnaire closed with the collection of demographics. Demographics were deliberately asked at the end of the questionnaire in order to avoid stereotype threat (Dobronte, 2013), meaning that respondents could feel stereotyped because of their level of income or education, which makes them try to perform excessively better than they normally would do.

Results

Before the main analyses, one-way ANOVAs with post hoc tests were employed as randomization check with all covariates, the moderator variable scepticism towards FOP claims and all demographics – with the exception of gender, for which a chi-test was conducted because of its dichotomous character – in order to check for unequal distributions of participants between the three conditions. The results of these analyses showed that the conditions did not differ on all demographic measures, as well as on the covariates persuasion knowledge and perceived

(26)

20 tastiness, and the moderator variable scepticism towards FOP claims.

Testing Hypothesis 1

H1 held that consumers who are exposed to food products with FOP claims (i.e. objective or subjective) show higher ratings of perceived healthiness than consumers who are exposed to products without any claims (i.e. control group). In order to test this hypothesis, the three groups were dummy coded into the control condition (N = 61) and the two experimental conditions grouped into one variable (N = 121). An independent samples t-test4_{then revealed that the mean} perceived healthiness scores of participants who were exposed to claims (M = 3.62, SD = 1.40) did not significantly differ from the mean perceived healthiness scores of participants who were not exposed to any claims (M = 3.51, SD = 1.25), t (180) = -.53, p = .599, indicating that H1 was rejected. In order to check for any individual effects, a test of variance was repeated with the two experimental groups taken separately. A one-way ANOVA5_{with post hoc test nevertheless} indicated that neither the objective claim condition (M = 3.51, SD = 1.33) nor the subjective claim condition (M =3.73, SD = 1.46) differed significantly from the control group condition (M

= 3.51, SD = 1.25) or from each other on perceived healthiness scores F (2, 179) = .53, p = .589.

Testing Hypothesis 2

H2 held that perceived healthiness has a positive mediating effect on the relationship between exposure to objective and subjective claims (IV) and purchase intentions (DV). A

4_{An independent samples t-test was chosen because the manipulation of the categorical independent variable was divided}

into two different groups with different people – that is, an experimental condition and a control condition – with only one continuous outcome variable. Moreover, scores on the dependent variable perceived healthiness were normally distributed with values of skewness and kurtosis close to 0 (skewness = .15, kurtosis = -.44) and the sample sizes for both conditions were > 30, making the t-test a reliable tool to assess mean differences between groups (Field, 2013).

5_{An analysis of variance was chosen because the manipulation of the categorical independent variable was now divided}

into three groups/conditions. ANOVA is a way of comparing the ratio of systematic variance to unsystematic variance in an experimental study, and in this way it can tell whether the experimental manipulation was generally successful. Hence, it does not provide specific information about which groups are affected and therefore, a post hoc test needed to be conducted in order to assess any reciprocal mean differences between the conditions (Field, 2013).

(27)

21 PROCESS macro mediation analysis (model 4) with a control versus experimental group dummy as IV6_{, purchase intentions as DV and perceived healthiness as mediator variable revealed that} there was no mediating effect of perceived healthiness scores on purchase intentions when participants were exposed to claims versus no claims (effect = .04) 95% CI [-0.11, 0.19]. Moreover, the mediation analysis was repeated for the two experimental conditions separately, coded as dummy variable, in order to detect any condition-specific effects, but again no

significant mediation effect was found (effect = .07) 95% CI [-0.09, 0.32]. Therefore, H2 was rejected.

Additional Analysis: Including Perceived Tastiness

However, perceived healthiness appeared to significantly interact with purchase

intentions (b = .33, b* = .27, t = 3.80, p < .001), yet only explaining 7% of variance in purchase intentions (Adj. R2_{= .07). Therefore, perceived tastiness was added to the model, since this}

product-related belief is in theory also able to predict purchase intentions (Trendel & Werle, 2015). A multiple regression analysis was employed7 with perceived healthiness and perceived tastiness as predictors of purchase intentions. Perceived tastiness significantly correlated with both perceived healthiness (r = .23, p = .001) and purchase intentions (r = .57, p < .001) and also proved to be a significant predictor of purchase intentions, as displayed in table 1. Moreover, the explained variance in purchase intentions significantly raised to 33.4% (Adj. R2 = .33).

6_{The PROCESS macro tool for SPSS makes use of logistic regression-based path analyses in order to reveal indirect and}

direct mediating, moderating and moderated mediation interactions. However, PROCESS macro analyses lose their predictive power when the independent variable is not dichotomous and therefore, all multi-categorical independent variables for the current research were first dummy coded into two groups instead of three (Hayes & Preacher, 2014).

7_{Multiple regression analyses seek to find a linear combination of predictors that correlate maximally with an outcome}

(28)

22 Table 1

Multiple Regression with Perceived Healthiness and Perceived Tastiness as Predictors of Purchase Intentions. b SE β t Sig. Perceived healthiness .18 .07 .15 2.40 .017 Perceived tastiness .59 .07 .53 8.52 .000

Note. Dependent variable: Purchase intentions (standardized). N = 182.

Testing Hypothesis 3a & 3b

H3a held that scepticism towards FOP claims has a moderating effect on the relationship between exposure to claims (IV) and perceived healthiness (mediator) and purchase intentions (DV). A PROCESS macro moderated mediation analysis (model 7) with the control versus experimental group dummy variable as IV, purchase intentions as DV, perceived healthiness as mediator variable and scepticism towards FOP claims as moderator variable revealed that there was no moderated mediation effect when participants were exposed to claims versus no claims (index = .02) 95% CI [-0.10, 0.20]. The analysis was repeated – yet with solely the experimental conditions dummy coded as IV – but again no significant moderated mediation effect was found (index = .03) 95% CI [-0.19, 0.22] and thus H3a was rejected. Since persuasion knowledge and scepticism are in theory related concepts8 (Friestad & Wright, 1994), the analyses were also repeated with persuasion knowledge as moderator variable instead of scepticism towards FOP claims. Nevertheless, no significant outcomes were found (index claims vs. no claims = .12) 95% CI [-0.02, 0.31]; (index objective claims vs. subjective claims = .11) 95% CI [-0.04, 0.39].

8_{In the current research, persuasion knowledge appeared to be marginally significantly correlated with scepticism (r =}

-.15, p = .038). The relationship is negative due to the fact that the statements used to measure these two constructs were

framed in the opposite direction in the questionnaire, indicating that low scores on the scepticism scale meant being highly sceptic whereas low scores on the persuasion knowledge scales meant having little persuasion knowledge.

(29)

23 H3b held that scepticism towards FOP claims induces a moderating boomerang effect on the relationship between exposure to subjective claims and ratings of perceived healthiness and purchase intentions. The PROCESS macro moderated mediation analysis (model 7) – as it was conducted in order to test H3a – was repeated, yet with a dummy coded subjective claim

condition versus the control condition as IV. Again no significant moderated mediation effect of scepticism was found that would in turn reveal a boomerang effect for the subjective claims condition (index = .02) 95% CI [-0.09, 0.20]. Therefore, H3b was rejected.

Additional Analysis: Comparing High vs. Low Sceptical Consumers

However, participants who were low in scepticism towards FOP claims showed

significantly higher ratings of perceived healthiness (MLow = 4.11, SD = 1.24; MHigh = 3.54, SD

= 1.34) t (180) = 2.71, p = .007 and purchase intentions (MLow = 3.68, SD = 1.58; MHigh = 2.81,

SD = 1.57) t (180) = 2.97, p = .003 than participants who were highly sceptical towards claims,

as was revealed when scepticism was dummy coded into a high versus low scepticism variable – based on a split of mean scores on the scepticism scale9 – and an independent samples t-test was employed with perceived healthiness and purchase intentions as DVs and scepticism (NHigh =

145; NLow = 37) as IV. However, both effects were rather small, that is ƞ2 = .04 for perceived

healthiness and η2_{= .05 for purchase intentions and moreover, all participants were taken} together regardless of whether they were exposed to claims or not.

9_{As the sample generally scored very high on scepticism towards claims (M = 2.36, SD = 1.00 on a 7-point scale with 1 =}

very sceptical and 7 = not very sceptical), a median split in order to divide participants into a high and low sceptical group was not sensible, since participants who scored above the median (2.22) would be considered as low sceptical, despite the fact that their scores were very close to 1 = very sceptical. Therefore, the sample was divided based on participants’ mean scores on the scepticism scale. 4 = neutral was more considered as being rather less than highly sceptical and therefore the breakpoint was set at 3, meaning that all participants with a mean score of 0-3 were considered as highly sceptical towards claims and the rest as being low in scepticism.

(30)

24

Conclusion

The goal of the current research was to explore the role of scepticism towards FOP food claims on ratings of perceived healthiness and purchase intentions of Dutch consumers when they were exposed to food products with objective claims, compared to food products with subjective claims. Analyses revealed that products carrying either objective or subjective claims did not induce higher ratings of perceived healthiness than products without any claims, and that perceived healthiness did not have a mediating effect on the relationship between exposure to these claims and purchase intentions. Moreover, scepticism towards FOP claims did not turn out to play a moderating role in the relationship between exposure to claims – both objective and subjective – and perceived healthiness and purchase intentions, nor did scepticism induce a boomerang effect on the relationship between exposure to subjective claims and ratings of perceived healthiness and purchase intentions amongst highly sceptical consumers.

Discussion

Theoretical Reflection

The direct positive effect of being exposed to FOP food claims compared to no claims on perceived healthiness as it was predicted in H1 was not found, indicating that participants were neither influenced by objective, nor by subjective claims when they rated the healthiness of the stimulus product. This result refutes the mere-label effect, meaning that consumers overrate products due to the sheer presence of a claim (Van Trijp & Van der Lans, 2007). Next to this, the claims did not automatically prime ‘healthy thoughts’ as was argued by Northup (2014).

Moreover, results from previous studies in which food products labelled with FOP claims were perceived as healthier than products without such claims (Ford et al., 1990; Van Trijp & Van der Lans, 2007; Aschemann-Witzel & Hamm, 2010) were not reproduced in the current research.

(31)

25 Contrary to what Burton et al. (2000) found, perceived healthiness also did not mediate the relationship between exposure to FOP claims and purchase intentions, as was formulated in H2. Perceived healthiness proved to be a significant predictor of purchase intentions, thereby confirming the Theory of Reasoned Action which holds the notion that product-related beliefs (e.g. perceived healthiness) influence intent dependent variables (e.g. purchase intentions) (Ajzen & Fishbein, 1980; Shepherd et al., 1992). However, exposure to FOP claims – both objective and subjective – did not have an influence on perceived healthiness, indicating that participants did not use the claims as heuristic cue to assess the quality of products (Sütterlin & Siegrist, 2015). Scepticism towards FOP claims did not turn out to have a moderating effect on the relationship between exposure to objective or subjective claims and perceived healthiness and purchase intentions (H3a). Besides, there was no boomerang effect revealed in which highly sceptical participants in the subjective claims condition rated the product as significantly less healthy than participants in the no claims condition or in the objective claims condition (H3b). The fact that participants did not show higher ratings of scepticism towards the subjective claims is not in line with the Theory of Economics of Information, stating that consumers show

differences in scepticism depending on the ease and costs of evaluating the veracity of a claim (Stigler, 1961), as well as results found by Ford et al. (1990), demonstrating that consumers were more sceptical towards subjective claims than they were towards objective claims. The non-significant results found in the current research could be due to the fact that in all conditions participants were for the greater part very sceptical towards FOP claims, allowing little variance between treatments. The overall high ratings of scepticism however are in line with results from studies from the U.S. FDA about consumers being generally very sceptical towards claims (Keller et al., 1997; Roe et al., 1999).

(32)

26 Along with the above mentioned results, some unexpected yet interesting interactions were revealed. First, perceived tastiness turned out to be a stronger predictor of purchase intentions than perceived healthiness, thereby refuting the idea that the claims used in this research would prime health-related thoughts which consequently would lead to purchase intentions being more strongly predicted by cognitive automatic reactions (i.e. perceived healthiness) instead of affective automatic reactions (Trendel & Werle, 2015). However, the finding is in line with arguments about taste expectations being the strongest driver of food choices in any case (Chandon & Wansink, 2012) and moreover, mean ratings of tastiness (M = 4.42, SD = 1.45) turned out to be higher than mean ratings of healthiness (M = 3.58, SD = 1.35), which could explain why participants were more driven by perceived tastiness instead of

healthiness when they made a (hypothetical) purchase decision.

Another significant outcome which was not hypothesized came to light when comparing participants who scored low on scepticism towards FOP claims (including those from the control group) to participants who scored high on this measure. The low scepticism group perceived the product as healthier and expressed higher purchase intentions than the high scepticism group, regardless of whether participants were exposed to claims or not. Although scepticism in the current research was very explicitly measured towards FOP claims, this finding indicates that scepticism towards claims is likely to be intertwined with general scepticism towards the food industry or towards the stimulus product (i.e. mango-orange juice) itself.

Limitations

All outcomes of the current research turned out to be condition-independent, meaning that participants did not base their judgements on the claims they were exposed to, but more on the product of its own. A main limitation is therefore product choice. It is sensible that fruit juice

(33)

27 already serves as a cue of being not so healthy, and due to the fact that this product has

repeatedly been given negative media attention (Consumentenbond, 2012; ‘’Sinaasappelsap net zo Slecht als Cola’’, 2015; ‘’Jamie Oliver Heeft Gelijk’’, 2015) consumers might already be sceptical towards it and thus they do not have to rely on claims in order to make inferences about the product’s healthiness.

Moreover, due to the fact that scepticism towards FOP claims was measured after exposure to the stimulus product instead of before, it is likely that participants did not base their responses on FOP claims in general – as they were instructed to do so – but solely on (claims in the context of) the product they were exposed to, that is, mango-orange juice. This assumption raises concerns about whether the scale that was used in the current research served as a true indicator of scepticism towards FOP claims in general, or that it was moreover a measure of scepticism towards the particular stimulus product, thereby offering an explanation for the fact that no significant effects of the manipulation (i.e. claims vs. no claims) were revealed, but nevertheless differences on ratings of perceived healthiness and purchase intentions between participants who scored high versus low on the scepticism scale occurred, regardless of whether they were indeed exposed to claims or not.

However, scepticism towards FOP claims as a dichotomous variable had to be analysed with the inclusion of participants in the control group who were not exposed to any claims at all, since the low scepticism group would otherwise become too small (N < 30) to perform proper statistical analyses of variance. Therefore, another limitation of the current research is that the sample was characterized as being highly sceptical (towards FOP claims) for the greater part. The high mean of scepticism could be explained by sampling error, since an overwhelming 90% of the sample was highly educated (i.e. highest level of education was either HBO or WO) and it

(34)

28 is proved that people who are highly educated are also more sceptical (Roe et al., 1999).

Implications & Further Research.

The first and most important thing to keep in mind when conducting research on effects of FOP claims on food and beverages, is that the effects of different claim types are inconsistent across different food categories and products (Andrews et al., 2000; Van Trijp & Van der Lans, 2007; Aschemann-Witzel & Hamm, 2010). In the current research, no significant effects were found of exposure to objective or subjective health-related food claims on ratings of perceived healthiness or purchase intentions. However, a product was used of which consumers most likely already formed opinions on, which could not be altered or influenced by the claims used in this study. Therefore, the current experimental research should be repeated with different products – preferably products which had not been repeatedly in the news or public debate due to their misleading true nutritional values.

In order to obtain more insight into the public food debate, opinions could be assessed by conducting a discourse analysis, for example by measuring electronic word-of-mouth (eWOM) – that is, any online positive or negative statement made by potential, actual, or former consumers about a product or company (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004). Further research should then investigate whether negative or positive eWOM has an effect on consumer attitudes and purchase intentions.Nutritional knowledge – likely obtained through eWOM or consumer organisations and television programs about food issues – should moreover be measured and taken into account as a moderator variable, since this construct is likely to be related to how consumers react to certain types of products, and the extent to which consumers are persuaded by FOP food claims (Andrews et al., 2000).

(35)

29 intentions were found for highly sceptical consumers compared to less sceptical consumers, and therefore it is advised that future research also takes scepticism into account. As an improvement of the current research, more participants who are less sceptical (towards claims) should

consequently be sampled in order to detect any statistically valid differences on product

evaluations between this group of consumers and consumers who are highly sceptical. In order to obtain a well-balanced sample including these two types of consumers, future research should adopt a quota sampling method (Bryman, 2012) by measuring scepticism as a binary personality trait at the beginning of the research and set an equal distribution of high and low sceptical participants as benchmark for the final sample.

Another important finding of the current research is that consumers’ purchase intentions are nowadays still more driven by taste expectations than by perceived values of healthiness. However, the overall low mean scores of perceived healthiness found in this research

demonstrate that the product used, that is, fruit juice, has lost the healthy image it once held (Bech-Larsen & Grunert, 2003) and that it is viewed nowadays more as a hedonic product – at least for highly sceptical consumers10. The low ratings of healthiness moreover revealed a floor effect, meaning that the product induced considerably low levels of perceived healthiness at baseline (Andrews et al., 2000). Future empirical research should therefore also investigate the effects (e.g. on purchase intentions) of taste claims, such as ‘Nu nog lekkerder!’ (i.e. ‘Now even tastier!’), since these types of claims are likely to be more suitable for hedonic products.

As a matter of fact, previous research from Choi et al. (2012) revealed that positive effects on consumer attitudes and claim believability were greatest when there was a match between claim type (taste claim vs. nutrient-content claim) and product type (product perceived

10_{Less sceptical participants showed slightly higher ratings of perceived healthiness than highly sceptical consumers in the}

(36)

30 as unhealthy vs. product perceived as healthy). However, no significant proof of such a match-up effect was found on purchase intentions, and moreover, the products used were already

unequivocally matched up as healthy (i.e. yoghurt) and unhealthy (i.e. chocolate chip cookie) in consumers’ minds – leaving room for further examination of the effects of taste claims versus health-related claims on purchase intentions or actual purchase of, for instance, more neutral products such as pasta (Aschemann-Witzel & Hamm, 2010).

From a practical perspective, food manufacturers are in any case advised to use a tailored marketing strategy approach for each of their products, which assesses which claims suit best for which product types (Aschemann-Witzel & Hamm, 2010). For fruit juice for example – which is nowadays perceived by Dutch consumers more as a hedonic product – it is likely that FOP taste claims will have a greater positive impact on purchase intentions than health-related claims. From a societal perspective, above mentioned implications combined with the non-significant claim effects found in the current research imply that health-related FOP food claims (both objective and subjective) do not match well with less healthy products. This is a very positive outcome for public health, indicating that consumers are less easily misled by claims intended to make a product look healthier, when in reality the product is not so healthy at all.

In conclusion, the persuasive power of FOP food claims seems to diminish when a claim is mismatched by the product type it appears on. Moreover, scepticism indeed provides Dutch consumers with a ‘healthy’ viewpoint from which they make product evaluations, as the highly sceptical consumers expressed low ratings of perceived healthiness for a hedonic product. And as it is predicted that consumers will become more and more sceptical, this given fact might imply a step forward for our current society – afflicted by obesity-related diseases – in reaching a turning point.

(37)

31

References

Print

Ajzen, I., & Fishbein, M. (1980). Understanding Attitudes and Predicting Social Behavior, Englewood Cliffs, New Jersey: Prentice Hall.

Alexandrov, A. (2010). Characteristics of single-item measures in Likert scale format. The

Electronic Journal of Business Research Methods, 8(1), 1-12.

Andrews, J. C., Netemeyer, R. G., & Burton, S. (1998). Consumer generalization of nutrient content claims in advertising. The Journal of Marketing, 62(4), 62-75.

Andrews, J. C., Burton, S., & Netemeyer, R. G. (2000). Are some comparative nutrition claims misleading? The role of nutrition knowledge, ad claim type and disclosure

conditions. Journal of Advertising, 29(3), 29-42.

Aschemann-Witzel, J., & Hamm, U. (2010). Do consumers prefer foods with nutrition and health claims? Results of a purchase simulation. Journal of Marketing Communications, 16(1-

2), 47-58.

Bearden, W. O., Hardesty, D. M., & Rose, R. L. (2001). Consumer self‐confidence: Refinements in conceptualization and measurement. Journal of Consumer Research, 28(1), 121-134. Bech-Larsen, T., & Grunert, K. G. (2003). The perceived healthiness of functional foods: A conjoint study of Danish, Finnish and American consumers' perception of functional foods. Appetite, 40(1), 9-14.

Belch, G. E., & Belch, M. A. (2015). Advertising and Promotion: An Integrated Marketing

Communications Perspective. (10th Global Ed.). New York: McGraw-Hill.

Bergkvist, L. (2015). Appropriate use of single-item measures is here to stay. Marketing

Letters, 26(3), 245-255.

Bryman, A. (2012). Social Research Methods. (4th Ed.). New York: Oxford University Press. Burton, S., Andrews, J. C., & Netemeyer, R. G. (2000). Nutrition ad claims and disclosures: Interaction and mediation effects for consumer evaluations of the brand and the ad. Marketing Letters, 11(3), 235-247.

Van Buul, V. J., & Brouns, F. J. (2015). Nutrition and health claims as marketing tools. Critical Reviews in Food Science and Nutrition, 55(11), 1552-1560.

Chandon, P., & Wansink, B. (2012). Does food marketing need to make us fat? A review and solutions. Nutrition Reviews, 70(10), 571-593.

0% added healthiness : the persuasive power of front-of-package food claims and the role of scepticism amongst Dutch consumers