The effect of sugar on consumers’ product evaluations

(1)

Amsterdam Business School

Master Business Administration

Marketing track

The effect of sugar on consumers’ product evaluations

MSc Thesis by:

Bas Spanjaard

10646809

Topic: The sugar movement Supervisor: Dr. J.Y. Guyt Amsterdam, June 21, 2018

(2)

Statement of originality

This document is written by Student Bas Spanjaard 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 supervision of completion of the work, not for the contents.

(3)

Abstract

As the global awareness of the detrimental effects of sugar on health is rising, the research gap on sugar content in consumer products is surprising. An increasing body of research links sugar with diet-related diseases. The increasing importance of consumer awareness regarding daily sugar intake is illustrated by the FDA updating the nutrition and supplements facts labels, including “Added sugars” in grams and as percent daily values. This thesis examines the relationship between sugar content in consumer products, private labels and national brands, and product evaluations. More specifically, it takes perceived healthiness, tastiness and quality into account when defining this relationship. It is hypothesized that sugar content has a negative effect on perceived quality and perceived health but a positive effect on perceived taste. Also, it is hypothesized that national brands influence the

relationship between sugar content and perceived product evaluation stronger than private labels. Hypotheses are tested in a 2x2 factorial MANCOVA (N = 660). The results showed sugar content to have a significant negative effect on perceived healthiness, but a positive significant effect on perceived tastiness and quality. Furthermore, only a moderating effect of brand type on the relationship between sugar content and perceived healthiness was found. This research tries to aid brand and retail managers in how to deal with sugar content in products, the shifting demand of consumers, as well as the role private labels and national brands have in the ‘war on sugar’.

(4)

1. Introduction

Consumers worldwide are becoming increasingly aware of the need to maintain a healthy weight to prevent diet-related diseases, such as obesity, diabetes and heart conditions (Euromonitor, 2014). An increasing body of scientific research links sugar with such diet-related diseases like obesity, an increased risk of chronic diseases, cardiovascular diseases and dental caries (Louie, Moshtaghian, Rangan, Flood, & Gill, 2016). Some go even further, arguing that sugar is an addictive “poison” that causes a host of degenerative ailments, including cancer, even in thin people (“How sugar became Public Enemy No. 1”, 2017). But still, people in a majority of the countries researched by Euromonitor (2014) consume more sugar than what the World Health Organization recommends for daily intake. With this growing body of research on sugar in mind, the urge of choosing sugar-free products over sugar added products becomes increasingly important for consumers. Consequences of health problems are not only social, in the form of a reduced quality of life, but also economic, in the form of health care costs (Ma, Ailawadi, & Grewal, 2013). To put it in a monetary

context, in 2008 obesity cost the U.S. an estimated U.S. €120 billion, while the UK estimated a bill of roughly €69 billion.

A high level of free sugars intake is of serious concern. Free sugars are defined as sugars including added sugars and naturally present sugars in honey, syrups and fruit juices. A high level of free sugars intake is associated with poor dietary quality, obesity and an increased risk of chronic diseases, such as cardiovascular diseases, cancers and diabetes (World Health Organization, 2015). Chronic diseases are yearly responsible for 40 million (70%) of the world’s deaths. Each year, 15 million people die from a chronic disease while between the age of 30 and 69 (World Health Organization, 2017). Food plays a major role in human health (Van Kreijl & Knaap, 2004). Approximately 10 percent of the total annual

(7)

deaths in the Netherlands, can be attributed to an unhealthy dietary composition. Obesity is responsible for 5 percent of the fatalities (Van Kreijl & Knaap, 2004).

Surprisingly, despite the new recommendations, added sugars intake continue to be above the recommended level of 10% of the total energy intake. The added sugars intake among US children and adults, despite a decline from 2003 to 2012, has significantly increased between 1977 and 2012 (Powell, Smith-Taillie, & Popkin, 2016). It is important to note that there is no need for added sugars in a diet as they do not serve any physiological purpose (Gunnars, 2017).

Addressing these sugar-related issues, the U.S. Food and Drug Administration (FDA) responded to the growing research on nutrition science by updating the nutrition and supplements facts labels, required on most packaged goods, in 2016. This update included “Added sugars” in grams and as percent daily value on the label (FDA Food Labeling: Revision of the Nutrition and Supplement Facts Labels, 2016) and stresses the importance of consumer awareness regarding sugar intake. Additionally, the WHO (2015) added two recommendations regarding free sugar intake for adults and children. Firstly, the WHO strongly recommends reducing the intake of free sugars to less than 10% of total energy intake. Secondly, the WHO suggests a further reduction of the intake of free sugars to below 5% of total energy intake for additional health benefits.

Next to this increasing informing of consumers globally on the detrimental effects of sugar in consumer products, another interesting change in the retail landscape has taken place over the last decades. More and more retail stores have been carrying products with their own label. Where private labels were seen as poor cousins to the manufacturer brands in the past, manufacturers of branded products have been taken aback by the unexpected and continued increase in private label share since the 1970s (Kumar & Steenkamp, 2007). The authors take Germany, Europe’s largest economy and the third-largest economy in the world, as an

(8)

example; private label share in Germany increased over the last three decades from 12 percent to 34 percent. More recent research of Nielsen (2018) showed a further increase of private label market share in Germany from 35 percent in 2014 to 36 percent in 2016.

In the last decade, as the value for money continuum (from premium to budget private label) is being stretched and retailers are innovating quickly to meet shoppers’ expectations, private labels have received a growing attention by marketing scholars. Consumers today are connected at all times and have access to endless information. As a result, their expectations are changing and they are shopping differently. Many now see private label brands as being equal to or substitutional for multinational brands (Nielsen, 2018). The research of Nielsen (2018) also found a performance reversal among private label and manufacturer branded products in the United Sates in 2017. While manufacturers of all sizes saw flat or positive performance in the fourth quarter of 2016, private labels took the lead a year later, growing at 2%.

With the current importance of consumer awareness regarding sugar intake, this research aims to get more insights into how sugar affects consumers’ product evaluations. In a

research by Reuters (2016), it showed that 58% of American adults want to cut back on their sugar intake. However, despite these good intentions, Bandy (2014) and other authors argue there is not yet such a thing as an anti-sugar consumer. Consumers in both emerging and developed markets still consume considerably more sugar than is recommended by the FDA. This research will investigate if the actual behavior of consumers’ is in line with the growing amount of research on the detrimental effects of sugar on the body. Consumer understanding of nutrition and health claims is a key aspect of recommendations like those from the WHO.

The research question of this thesis will be: “How does sugar content in consumer products affect consumers’ product evaluations and how is this relationship influenced by using either private labels or national brands?”. Sub-questions to this main research question

(9)

are: “What is the effect of sugar content on perceived taste, perceived quality, perceived health and product evaluation?” and “How do private labels and national brands moderate the relationship between sugar content and product evaluation?”. By providing an answer to these questions, this study will provide an increased understanding to managers on sugar content in consumer-packaged goods and the role both private labels and national brands have in the ‘war on sugar’. By helping to make consumers aware of their sugar intake and the amount of sugar in consumer-packaged goods, I try to reduce consumers’ sugar intake and help manufactures respond to the shifting demands of consumers. This contributes to

healthier eating patterns, more satisfied consumers and potentially less diet-related diseases. The lack of research on sugar content and product perceptions is surprising given the growth of research on sugar-related health problems. This study goes beyond the existing literature by examining this gap.

This introduction will be followed by a literature review, in which the existing literature is used to explore consumers’ perceptions of sugar, consumers’ product evaluations, private labels and national brands. Additionally, a conceptual framework is included, which consists of two hypotheses based on the current literature. Next, the research method is explained. The research will consist of an online experiment (N = 660) and the results are analyzed and presented in the results section. Finally, the main insights of this thesis are, summarized discussed and suggestions for future research are given.

(10)

2. Literature review and conceptual model

2.1 Added sugars and consumer perceptions of sugar

Nutrition claims, in particular sugar claims in food labelling such as “no added sugars”, highlight the beneficial nutritional aspects of food and drinks by providing information about energy or nutrient composition. They are a useful tool for communicating nutrition

information to consumers and encouraging healthy eating patterns (Patterson, Sadler, & Cooper, 2012). Healthy eating patterns are crucial for maintaining a healthy weight and preventing diet-related diseases. An increasing body of scientific research links sugar with diet-related diseases like obesity, an increased risk of chronic diseases, cardiovascular diseases and dental caries (Louie et al., 2016). Diet related diseases such as obesity have become a global pandemic. In 2005, at least 400 million people were obese (Body Mass Index > 30). Furthermore, the percentage of obese adults in the U.S. increased from 15% in the late 1970s to 36% in 2010 (Hu, 2013). In parallel with the increase in obesity, the prevalence of type 2 diabetes has nearly doubled in the US, from 5.3% during 1976-1980 to 11.3% in 2010. The number of people worldwide with type 2 diabetes is projected to reach over 550 million by 2030 (Hu, 2013).

Contradictory to health recommendations of the WHO (2015), Louie et al. (2016) found that a large proportion of the Australian youth are consuming alarming and excessive amounts of energy from added sugars, with over 80% of their participants reporting diets which exceeded the 10% cut-off set in the latest WHO guidelines. Supporting these

statements, Powell et al. (2016) found that the added sugars intake among both U.S. children and adults increased from 1977 to 2012. The mean adjusted added sugars intake among children increased from 275 kcal/day (roughly 69 grams) in 1977 to 387 kcal/day (roughly 99 grams) in 2003 and then declined to 326 kcal/day (roughly 82 grams). For adults, the mean adjusted added sugars intake increased from 228 kcal/day (roughly 57 grams) in 1977 to

(11)

341/kcal day (roughly 85 grams) in 2003 and then declined to 308 kcal/day (roughly 77 grams) in 2012 (Powell et al., 2016). This shows that the WHO recommendations are significantly exceeded and consumers are still consuming excessive amounts of sugar on a daily basis. Louie et al. (2016) found that compared with U.S. children and adolescents, Australian children and adolescents have substantially lower intakes of absolute added sugars but are still exceeding added sugar intake recommendations. This not only shows, in line with the Euromonitor International (2014) report, that added sugar intakes vary greatly among countries, but that added sugar intakes exceed the recommended intake for a majority of countries.

As the leading contributor to the excess of added sugars intake, excessive

consumption of sugar-sweetened beverages represents a risk for the population’s health (De Ruyter, Olthof, Seidell & Katan, 2012; Hu, 2013; Le Bodo, Paquette, Vallières, & Alméras, 2015; Louie et al., 2012; Wang, Bleich, & Gortmaker, 2008). Sugar-sweetened beverages are found to be the single largest source of added sugars and the largest source of energy intake in the US diet (Hu, 2013). A possible explanation for the excessive intake of sugar-sweetened beverages is because (liquid) sugars do not lead to a sense of satiety, so the consumption of other foods is not reduced, ultimately leading to a calorie surplus (De Ruyter et al., 2012). This is supported by the WHO (2015) and Gunnars (2017) who stated that there is no need for added sugars in a diet as they do not serve any physiological purpose. Schulze et al. (2004) found that in women, a larger consumption of sugar-sweetened soft drinks led to significantly increase in weight gain and an increased risk for developing type 2 diabetes. Hu (2013) even compared soft drink consumption trends to tobacco consumption trends, as both soft drinks and tobacco companies have a worldwide reach and aggressive marketing tactics designed to export unhealthy products to developing countries.

(12)

The decrease in consumption of added sugars over the past decade, as reported by Powell et al. (2016) has mainly come from a decrease in consumption of added sugars from beverages, while consumption of added sugars from foods show no significant change. This shows that the growing research on sugar-sweetened beverages already has had some effect on consumers’ consumption. Due to this decline in beverages, foods are becoming an increasingly important contributor of added sugars. Among children 2 to 18 years old, desserts, candy, and cereals were large contributors of added sugars (Powell et al., 2016).

An adequate solution to diet-related diseased would be the consumption of sugar-free beverages. This is supported by de Ruyter et al. (2012) who found, in an experiment on Dutch children, that weight, skinfold-thickness measurements, waist-to-height ratio and fat mass increased significantly less in a sugar-free beverage consuming group compared to a sugar holding beverage consuming group. However, contradictory to these health benefitting findings, a worldwide research by Technavio (2017) showed that sugar-free beverages still only hold a 35% market share in 2016.

Patterson et al. (2012) examined the consumer understanding of sugar claims on food and drink products. They found that participants’ (all women) associations with sugar

included sweetness, energy, nice taste, fattening, dental health, hyperactivity and diabetes. Beliefs about sugar were for example that white sugar is refined and bad for you, brown sugar is more natural and healthier for you, sugar is acceptable in moderation and that sugar in fruit is acceptable but too much is converted in fat. When introduced with reduced sugar claims, the authors found that this was familiar to all participants and reactions were initially positive. This is supporting earlier research of Lähteenmäki et al. (2010) who stated that health claims can be expected to increase perceived healthiness of products.

Furthermore, Patterson et al. (2012) found that negativity was expressed for the expressed taste and the replacement ingredients. This is also supporting earlier research by

(13)

Lähteenmäki et al. (2010), who stated that the impact of health claims can be negative when consumers are approached with claims containing ingredients and benefits they have not been exposed to before. This however is not expected for reduced sugar claims, parallel with the initial positive findings of Patterson et al. (2012). When questioned about the level of sugar reduction, the authors found a lack of awareness of any guidelines. Products with a claim for no added sugars were generally preferred above those with a reduced sugar claim as the process of not adding sugar was considered more ‘natural’ than taking something out. For no added sugar products, Patterson et al. (2012) also found that all participants expected there to be some form of sugars in the products, and all expected sweeteners to be added in.

Next to the influence of sugar claims on food and drink products, other influences are important to assess consumers’ product evaluations. Perceived healthiness is one of the factors that is important for the attitude towards food and drinks (Provencher, Polivy, & Herman, 2009; Ares and Gámbaro, 2007; Lähteenmäki et al., 2010). Next to perceived healthiness, these authors also found that both perceived product quality and perceived product tastiness are factors influencing consumers’ product evaluation. With the growing body of research on the detrimental effects of sugar on consumers’ health in mind, this research will seek for an understanding if and how these updated perceptions on sugar influence product evaluations.

2.2 Consumers’ product evaluations

As literature on sugar-free products and consumer perception is lacking, this paragraph will further explore the components of consumers’ product evaluations on sugar and sugar-free products. Bandy (2014) argues there is not yet such a thing as an anti-sugar consumer. She shows that many populations consume more sugar than is recommended, reduced sugar products represent only a very small portion of total sales and the average global consumer is

(14)

expected to purchase 3 grams of sugar a day more in 2019 compared to 2014. To a certain extend this overconsumption of added sugars can be accredited to aggressive marketing practices (Le Bodo et al., 2015; Hu, 2013). Le Bodo et al. (2009) found that, in the United States alone, carbonated beverage companies’ marketing expenditures targeting youth (mainly teens-directed) reached $395 million, which accounted for 22% of all food

categories. Le Bodo et al. (2005) also stated that advertisements are usually appreciated by youth and have been shown to influence their diet-related behaviors. Even though such advertising practices are very interesting and influential, this research will solely focus on consumers’ perceptions of products and how these are influenced by sugar content and the type of brand (national brand or private label).

Throughout the consumer buying process, from pre-purchase consideration to purchase, consumers evaluate products. Most consumer research considers product evaluation to be a goal-directed process, meaning consumers evaluate products with a particular situation in mind. Evaluative criteria emerge from this process. Evaluative criteria are the relevant set of product characteristics and product performance levels associated with each characteristic (Gardial, Clemons, Woodruff, Schumann, & Burns, 1994).

An essential element of product evaluation is the perceived quality. Quality has been recognized by leading marketing strategy authors as a core concept in building customer value and satisfaction. Even the best marketing department in the world cannot sell products which are poorly made or which fail to meet anyone’s need. Quality is an important tool for companies in order to create a sustainable competitive advantage (Ophuis & Van Trijp, 1995). Furthermore, the quality of products offered by a retailer influence customer retention (Grewal, Krishnan, Baker, & Borin, 1998).

Another factor influencing product evaluations is the perceived taste. Taste is suggested to be the most important quality attribute of food (Ophuis & Van Trijp, 1995).

(15)

Raghunathan, Naylor and Hoyer (2006) found that for unhealthy products, perceived tastiness is rated higher. They found that consumers overconsume foods that they perceive as

unhealthy (containing more fat) because they assume that such food tastes better. This is supported by Verbeke (2006) who found that consumers are not willing to compromise on taste for health. The connection between health labels and negative taste expectation might be a reason for consumers not to buy products with health claims.

The perceived healthiness of products is also an important factor for consumers’ product evaluations, as health has been named the most significant trend and innovation driver in the global food and drinks market (Lähteenmäki et al., 2010). However, health-related claims have resulted in higher ratings of perceived healthiness, but the increase has been small or moderate at the best (Lähteenmäki et al., 2010).

Hence, hypothesis 1 is as follows:

Hypothesis 1: A high sugar content in consumer products has a negative effect on perceived quality, perceived health and product evaluation but a positive effect on perceived taste

2.3 Private labels versus national brands

As the research on the detrimental effects of sugar in consumer products is increasing, another interesting and market changing development has taken place in the retail industry over the last decades. Private labels in the consumer-packaged goods industry have

experienced a worldwide surge in availability and market share in recent years. Private labels are brands owned by a retailer or distributor, sold exclusively in their own stores (Kumar & Steenkamp, 2007). Where private labels were primarily targeted to the poor in the past, purchasing private label products of (supposedly) comparable quality for a much lower price is increasingly considered as “smart” shopping (Kumar & Steenkamp, 2007). As Steenkamp,

(16)

van Heerde and Geyskens (2010) found, objective tests showed that there is often little quality difference between private labels and national brands. Also, the perceived quality difference is on average relatively small. This is supported by research of Abril and Sanchez (2016) who found that private labels initially emerged as low quality, low price brands, but consumers have significantly improved their price/quality ratio perceptions. However, Bao, Sheng, Bao and Stewart (2011) found that even though private label sales have expanded exponentially in recent years, consumers still perceive private labels to have a lower quality than national brands. This lower perception of private labels is found even though the objective quality is at par or even higher for some retailers (Bao et al., 2011).

Kumar and Steenkamp (2007) also found that in 2007, private labels already

accounted for one of every five items sold every day in U.S. supermarkets, drug chains, and the mass merchandisers, with the market share in Europe even being larger. More recent research by Abril and Sanchez (2016) showed that private labels are a notable concern for consumer-packaged goods companies, having reached a 20% market share in the United States and 35% in Europe. Nielsen (2018) information across more than 60 countries showed that private label products continue to gain share across all major geographies. The relentless store expansion by retailers over the past decade has given shoppers more access to private label and to brands. In recent years, e-commerce has given brands another way to reach the consumer.

Abril and Sanchez (2016) stated that consumers react differently to marketing mix efforts for manufacturer brands than for private labels. The risk that consumers perceive when choosing private labels is also found to be higher when consumers choose private labels over national brands. It is therefore concluded that the reasons for choosing a manufacturer brand versus a private label likely differ. Extant literature has shown differences in consumer behavior towards private labels and national brands. The importance of the price gap, product

(17)

innovation, non-price strategies (e.g. promotions), advertising and new product introductions are all factors that have been researched in previous studies on the choice between private labels and national brands (Abril & Sanchez 2016). Brand reputation has attracted much importance in the marketing literature as well. The impact of the brand on new product success is found to be involved for 10-15% of the success of a new product (Gielens & Steenkamp, 2007). Steenkamp et al. (2010) furthermore find that consumers still show a higher willingness-to-pay for national brands, compared to private labels.

As can be concluded from the extant literature, even though the gap of perceptions between private labels and national brands is narrowing, national brands are often seen as the superior product. While the objective quality between private labels and manufacturer brands have equalized in recent years, consumer perceptions have not yet caught up to the most recent developments in the retail industry. When taking the effect of sugar on product evaluations in mind, it is therefore expected that national brands have a stronger impact on the relationship between the possible factor sugar content and product evaluation.

Hence, hypothesis 2 is proposed as follows:

Hypothesis 2: The effect of sugar content on product evaluation is stronger for national brands than for private labels

Figure 1. Conceptual Model

Private Labels / National Brands

H1

H2

(18)

3. Methodology

The goal of this thesis is to come with new insights in the effects of added sugar on

consumers’ product evaluations. Furthermore, the moderating role of both private labels and national brands is researched. In the previous sections of this thesis the extant literature relevant for the research problem was discussed. The next section will discuss the research design used to answer the research question. Furthermore, both the experiment design and survey design, as well as the different conditions, will be clarified. Thereafter, in the second paragraph, the research procedure is explained. The variables product evaluation, sugar content and brand type are discussed in the third paragraph. Predictions will be stated in the fourth paragraph. The scales are provided to measure the constructs of the theoretical framework.

3.1 Sample and design

For this experiment, data was collected through an online experiment to test the propositions (see Appendix 1). Data consists of 165 participants. Initially 198 participants started the experiment, however only 165 respondents finished and thus remained useful for analysis. As every respondent answered questions on four different categories, a total of 660 observations were recorded and used for the correlation and regression analyses (N = 660). Participants were recruited via personal contacting, social media, marketplace crowdsourcing and by e-mail to collect quantitative data through an online experiment, distributed via an online link. Firstly, participants were recruited through personal contacts of the author, a master student business administration at the University of Amsterdam, via Facebook. Simultaneously an email was send to all master students in business administration at the University of Amsterdam and Amazon’s Mechanical Turk (MTurk) program was used to create a more diverse subject pool. A total of 20 respondents were recruited via Amazon’s Mechanical

(19)

Turk. The online experiment was created with the program Qualtrics. In this program, all participants were forced to complete every question, to minimize missing data.

To investigate consumers’ product evaluation on sugar holding products versus sugar-free products and private label versus national branded products, an online experiment was conducted. The effect will be examined in a 2 (sugar level: sugar vs. zero sugar) x 2 (private label vs. national brand) between-subjects design. Therefore, there will be 4 conditions. Condition 1: a sugar-containing product from a private label. Condition 2: a sugar-containing product from a national brand. Condition 3: a sugar-free product from a private label and finally. Condition 4: a sugar free product from a national brand. Qualtrics was used to randomly allocate participants to the different conditions and product categories.

3.1.1 Product categories

Because this research wants to be as generalizable as possible, this research was conducted across four different product categories. The product categories chosen are often known for their level of sugar (more than 20 grams of sugar per 100 gram/ml in the regular products), but also hold sugar free alternatives. These categories are chosen as it is expected that the proposed contrast between sugar content and no sugar content will be best observable for categories high in sugar which include sugar free alternatives as well. It is expected that generalizing the effects will be easier when the results are based on effects found in multiple categories.

The categories selected were chosen upon availability in both Albert Heijn’s (The Netherlands) online grocery platform (www.ah.nl) and Walmart’s (United States) online grocery platform (www.walmart.com). The supermarkets were chosen due to their respective sizes in their countries, with Albert Heijn being the market leader in the Dutch retail market. For the private labels, both the private label of Albert Heijn (‘AH’) and Walmart (‘Sam’s

(20)

Choice’ and ‘Great Value’) were used. The intercontinental availability of categories was chosen so a wide variety of respondents would be familiar with the products and brands presented. To increase the external validity, real brands within these categories were used. To prevent preexisting attitudes biasing the examined effects, multiple categories were shown to respondents. Because not all chosen products had a ‘zero sugar’ version, fictional ‘zero sugar’ products were created in Photoshop by adding the phrase ‘zero sugar’ to the regular products. The four product categories chosen were: liquid beverages, breakfast cereals, ice cream and condiments. All products chosen showed names or text in English, except for the ice cream products. It was therefore chosen to show a national brand from the Netherlands (which contained text in Dutch) and the United States. Table 1 shows the different products used in the different conditions. The visual representations of the conditions are shown in Appendix 2.

(21)

Categories Conditions Liquid beverages Breakfast cereals

Ice creams Condiments

Condition 1

Sugar & private label

‘AH Classic Cola regular’ & ‘Sam’s Cola’

‘AH special flakes’ & ‘Great Value Corn Flakes cereal’ ‘AH Roomijs Vanille’ & ‘Great Value French Vanilla ice cream’ ‘AH Tomaten Ketchup’ & ‘Great Value Tomato Ketchup’ Condition 2 Sugar & national brand ‘Coca-Cola’ ‘Kellogg’s Special K Original’ ‘Hertog Vanille Roomijs’ & ‘Dreyer’s Vanilla Bean’ ‘Heinz Tomato Ketchup’ Condition 3

Zero sugar & private label

‘AH Classic Cola No Sugar’ & ‘Sam’s Cola Zero Calories’ ‘AH Special flakes no sugar’* & ‘Great Value Corn Flakes cereal zero sugar’* ‘AH Roomijs Vanille Suikervrij’* & ‘Great Value Zero Sugar French Vanilla ice cream’* ‘AH Tomaten Ketchup Zero Sugar’* & ‘Great Value Tomato Ketchup Zero Sugar’* Condition 4

Zero sugar & national brand ‘Coca-Cola Zero’ ‘Kellogg’s Special K Original Zero sugar’* ‘Hertog Vanille Roomijs Suikervrij’* & ‘Dreyer’s Zero Sugar Vanilla Bean’* ‘Heinz Tomato Ketchup No Sugar’*

Note, * Photoshop used to create the ‘no sugar’ version

Table 1. Product categories, conditions and products

3.2 Experiment procedure

Because the experiment has been conducted online, all the participants accessed the survey through an online link. When this link was accessed, a short introduction was shown.

Thereafter, the participants are shown the four different categories, in random order. For each category shown, a picture of one of the four conditions is shown at random to the

participants. After taking a close look at the pictures, the respondents were asked to answer the questions below the pictures. Participants are asked to rate these products on ‘perceived healthiness’, ‘perceived taste’ and ‘perceived quality’ against 7-point Likert scales, with different end poles per scale. As concluded from the literature review, these factors are

(22)

important for measuring consumers’ product evaluations. ‘Willingness to buy’ and

‘Willingness to recommend’ are added to the survey to examine if a high evaluation leads to a higher willingness to buy and higher willingness to recommend the product and how these differ for private labels and national brands. After having answered these questions for the four different product categories, participants are asked about their age, gender, nationality, awareness of sugar intake recommendations and some questions about their general health interest.

3.3 Measurements

3.3.1 Dependent Variable: Product evaluation

For this research, the variable product evaluation is measured through ‘perceived product healthiness, ‘perceived tastiness’ and ‘perceived quality’. By combining these three variables, the variable ‘product evaluation’ was created. The ‘perceived product healthiness’ was measured using a modified 3-item questionnaire. The first item, ‘How healthy is this product

for you?’ (+), was based on the work of Provencher, Polivy and Herman (2009) and Ares and

Gámbaro (2007). This item was scored against a 7-point Likert scale, ranging from (1) ‘very unhealthy’ to (7) ‘very healthy’. The second and third items are taken from a questionnaire developed by Lähteenmäki et al. (2010). An example item is ‘This product will reduce the

risk of cardiovascular diseases’ (+). The two items were scored against a 7-point Likert scale,

ranging from (1) ‘Not at all’ to (7) ‘Extremely’.

The ‘perceived tastiness’ was measured using a 2-item questionnaire (Raghunathan et al., 2006). An example item is ‘How tasty do you think this product will be?’ (+). The two items were scored against a 7-point Likert scale, ranging from (1) ‘Not at all’ to (7)

‘Extremely’. The original right end pole ‘very’ was changed to ‘extremely’ to keep end poles uniform throughout the survey.

(23)

The ‘perceived quality’ was measured using a 4-item questionnaire (Bao et al., 2011). This ‘quality perception’ scale was based on work of Grewal et al. (1998) and Keller and Aaker (1992). The quality perception scale used by Bao et al. (2011) showed excellent internal consistency (α = .95). Each item was scored against a 7-point Likert scale, ranging from (1) ‘Strongly disagree’ to (7) ‘Strongly agree’. An example item is ‘This product is of

high quality’ (+). One item, ‘This product is of very bad quality’ (-), needed to be recoded

(see Appendix 1) because it was negatively worded. The scale is therefore reversed and the range will change into (1) ‘Extremely’ to (7) ‘Not at all’.

3.3.2 Independent Variables: Sugar content and brand type

For both independent variables, dummy variables were created. First, a dummy variable which determined which type of sugar content was shown to the participants was created. This dummy variable had the values ‘0 = No sugar’ and ‘1 = Sugar’. Secondly, the moderating variable for the type of brand was created. Brand type had values ‘0 = Private label’ and ‘1 = National brand’.

3.3.3 Descriptive Variables: WTB, WTR and GHI

‘Willingness to buy’ and ‘Willingness to recommend’ are added to the survey to examine if a high evaluation leads to a higher willingness to buy and higher willingness to recommend the product (Sweeney & Soutar, 2001). Each item was scored against a 7-point Likert scale, ranging from (1) ‘Strongly disagree’ to (7) ‘Strongly agree’. An example item is ‘I would be

willing to buy this product’ (+).

Consumers ‘General Health Interest’ was based on research from Roininen, Lähteenmäki and Tuorila (1999), to assess consumers’ orientation towards eating healthy.

(24)

This short survey is used as a control variable and to see how healthy the respondents perceive themselves.

3.4 Analysis and prediction

To test the model presented in Figure 1, standard statistical procedures are used (Baron & Kenny, 1986). The data allows for many different types of analysis using SPSS. Descriptive analysis is used show some general statistics of the sample. The described multi-item scales will be tested for internal validity. Correlation will be looked for between all the variables, and the dependent variable (consumers’ product evaluation) is modeled as a function of the two independent variables (sugar content and brand type) in a 2x2 factorial MANCOVA.

For the correlation tested, a moderate to strong negative correlation between the independent variable ‘sugar content’ and the dependent variables ‘perceived product quality’, ‘perceived product healthiness’ is predicated. The correlation between ‘sugar content’ and ‘perceived product tastiness’ is predicted to be positive. Also, the correlation between ‘brand type’ and ‘sugar content’ is expected to be positive. Then hypothesis 1 stated a negative effect of high sugar on ‘perceived product quality’, ‘perceived product healthiness’ and ‘product evaluation’ but a positive effect on ‘perceived taste’. Hypothesis 2 stated national brands to have a stronger moderating effect than private labels, on the relationship between sugar content and product evaluation.

(25)

4. Results

The previous section of this thesis discussed the methodology of the research, including the research design. In the current section the results of the research and the analysis of these results will be discussed. First, the pre-test will be discussed. Then descriptive statistics will be discussed, followed by the second part which will discuss the reliability of the scales. Then correlation will be looked for between all the variables and at last the results from the 2x2 factorial MANCOVA analysis will be treated.

4.1 Pre-test and data preparation

To check for clarity of the conditions and the reliability of the scales, a pre-test was executed for a small group of people (N = 13). It was found that the recoding of counter-indicative items applied to some items of ‘perceived product quality’ (part of the dependent variable) and general health interest (part of the demographics section). Other scales needed to be adapted, for example ‘gender’ from (1) ‘Male’, (2) ‘Female’ to (0) ‘Male’, (1) ‘Female’. Also, the scales of the ‘Willingness to buy’ needed to be adapted due to a recode value error in Qualtrics. The scales (15) ‘strongly disagree’ to (21) ‘strongly agree’ were adapted to (1) ‘strongly disagree’ to (7) ‘strongly agree’. Furthermore, the original right end pole of the ‘perceived tastiness’ construct was changed. The right end pole ‘very’ was changed to ‘extremely’ to keep the end poles uniform throughout the survey.

The dataset generated by Qualtrics was remodeled in a second database in order to make sorting possible on the different conditions. The new variables ‘category’, ‘condition’, ‘sugar content’ and ‘brand type’ were created. ‘Category’ had values ‘1 = Liquid beverages’ to ‘4 = Condiments’. ‘Condition’ had values ‘1 = Private label & sugar’ to ‘4 = National brand & sugar free’. ‘Sugar content’ had values ‘0 = No sugar’ and ‘1 = Sugar’ and brand

(26)

type had values ‘0 = Private label’ and ‘1 = National brand’. In the new database, as every respondent evaluated 4 conditions, a total of 660 observations were recorded.

4.2 Descriptive statistics

Data consists of 165 participants. Initially 198 participants started the questionnaire, however only 165 respondents finished the survey completely and thus remained useful for analysis (N = 165). No missing data was found in these 165 answers. There were more female than male respondents (59.4% female). The respondents were aged between 19 and 80 years old. Most respondents were aged between 19 and 31 years old (80%). This high number is probably caused by the high wiliness of students to fill in the survey of another student and because of the similar age of the author. The largest group (70.3%) of respondents was Dutch (N = 116), 12.7% of respondents were from the United States (N = 21) and 4.2% was German (N = 7).

To ensure the validity of this study, it was important that participants were randomly assigned to the multiple conditions. This was automatically done by Qualtrics. Each

participant evaluated 4 product categories, thus resulting in 660 observations. Frequency checks showed condition 1 (‘private label & sugar’) contained 167 observations, condition 2 (‘national brand & sugar’) contained 160 observations, condition 3 (‘private label & zero sugar’) contained 166 observations and condition 4 (‘national brand & zero sugar’) contained 167 observations. Furthermore, the mean age of the participants did not significantly differ across the conditions. Finally, the gender division did significantly differ across conditions.

Regarding the sugar recommendations knowledge of the participants, 117 (70.9%) of the 165 participants did not know how many grams of sugar go in a sugar cube. Furthermore, 109 (66.1%) participants were not aware that the WHO recommends adults and children to reduce their daily intake of free sugars to less than 10%. However, most participants (27.9%) were more concerned about sugar than fat (23%) or carbs (20%). At last, the mean of the

(27)

general health interest survey (M = 4.6121, SD = 1.0607) showed that on average, the respondents identified themselves as moderately interested in eating healthy.

An independent t-test was conducted to compare willingness-to-pay in sugar and no sugar conditions, as well in private label and national brand conditions. First, it was found that there was not a significant difference between the scores for no sugar (M = 1.7715, SD = 0.9892) and sugar (M = 1.7938, SD = 0.9400); t(658) = -.297, p = .766. This indicates that sugar content does not have an effect on willingness-to-pay. Secondly, it was found that there was a significant difference between the scores for private label (M = 1.5653, SD = 0.8310) and national brand (M = 2.0038, SD = 1.0392); t(658) = -5.993, p = .001. These results suggest that brand type does have an effect on willingness to pay. Specifically, the results suggest that for national brands, consumers show a higher willingness to pay.

Another independent t-test was conducted to compare willingness-to-buy and willingness-to-recommend in sugar and no sugar conditions, as well in private label and national brand conditions. First, it was found that there was not a significant difference between the scores for no sugar (M = 3.9865, SD = 1.7937) and sugar (M = 4.2951, SD = 1.8183); t(658) = -2.195, p = .786. This indicates that sugar content does not have an effect on willingness to buy and willingness to recommend. Secondly, it was found that there was not a significant difference between the scores for private label (M = 3.7267, SD = 1.7670) and national brand (M = 4.5596, SD = 1.7607); t(658) = -6.065, p = .172. This suggests that brand type does not have an effect on willingness to buy and willingness to recommend.

Both Skewness and Kurtosis showed acceptable scores (Field, 2009) for the variables ‘perceived healthiness’, ‘perceived tastiness’, ‘perceived quality’, ‘product evaluation’, ‘willingness to pay’, ‘willingness to buy’ and ‘general health interest’ (see Appendix 3, A1). The most extreme Skewness score being 0.691 (SE = 0.95) and the most extreme score for Kurtosis being -1.100 (SE = .190).

(28)

4.3 Reliabilities and correlations

In Table 1, the means, standard deviation, and reliabilities (Cronbach’s Alpha) of the

variables are presented. Tavakol & Dennick (2011) suggested a minimum Cronbach alpha of 0.60 for reliable data and preferably a Cronbach’s alpha higher than 0.80. The variables consist out of items discussed in the methodology (see section 3) and are shown in Appendix 1. Reliability checks were run for ‘perceived product health’, ‘perceived product tastiness’ and ‘perceived product quality’, as well as for the short survey on respondents’ ‘general health interest’.

The results found that ‘perceived product health’ (α = .930), ‘perceived product tastiness’ (α = .962) and ‘perceived product quality’ (α = .786) all showed excellent

reliabilities. The Cronbach’s alpha of ‘perceived product quality’ would increase to 0.916 if one item was deleted, which was done accordingly to increase the reliability. Reason for this deletion is that this question is very similar, but negatively worded, as another question in the ‘perceived product quality’ questionnaire. Combining the original nine items for a total product evaluation score also showed excellent reliability (α = .877). However, one item of the ‘perceived product quality’ was again deleted to increase the Cronbach’s alpha to 0.912. The small survey on respondents’ general health interest also showed great reliability (α = .824). The mean of ‘perceived product health’ (M = 2.7263, SD = 1.1518) show that on average, the products were not perceived as healthy. The means of ‘perceived product tastiness’ (M = 4.2780, SD = 1.0549) and ‘perceived product quality’ (M = 4.3899, SD = 1.0170) indicate that the product tastiness and quality were perceived as moderately high.

As expected, sugar content was strongly significant and moderately negative correlated with perceived product healthiness (r = -.203, p < .01). According to the matrix there is an indication that sugar content has an effect on perceived product healthiness, which will be validated in the linear regression model. Sugar content was, as expected, also strongly

(29)

significant and moderately positive correlated with perceived product tastiness (r = .222, p < .01). This suggests that sugar content has an effect on perceived product tastiness, which again will be validated in the linear regression. A surprising finding was the significant and weakly positive correlation between sugar content and perceived product quality (r = .091, p < .05.

Brand type was found to be strongly significant and moderately to strongly positive correlated with perceived product health (r = .165, p < .01), perceived product tastiness (r = .168, p < .01) and perceived product quality (r = .441, p < .01). These outcomes might indicate that national brands score higher on product evaluations. Unexpectedly, sugar content and brand type were very weakly positive correlated, although this correlation was not found to be significant (r = -.012, p = .754).

(30)

T ab le 1. D es cr ip tive s tat ist ics M SD 1 1. G en d er a 0.59 0.493 2. A ge 28.92 11.235 .003 3. A w ar en es s of s u gar /WH O s tan d ar d s _b 0.3152 0.3705 -.046 4. Wi llin gn es s t o p ay 1.7825 .96449 .019 5. Wi llin gn es s t o b u y/ Wi llin gn es s t o r ec om m en d 4.1394 1.8111 -.029 6. G en er al H eal th I n te re st 4.6121 1.0583 .174** 7. S u gar c on te n t c .50 .50 .030 8. B ran d T yp e _d .50 .50 .011 9. P er ce ive d p rod u ct h eal th 2.7261 1.5638 -.086* 10. P er ve ive d p rod u ct t as tin es s 4.2780 1.6475 -.061 11. P er ce ive d p rod u ct q u al ity 4.3900 1.4334 -.076* 12. P rod u ct e val u at ion 3.7980 1.2060 -.095* N ot e, N = 660 * C or re la tion i s s igni fic ant a t t he 0.05 l eve l ( 2-ta ile d) . ** C or re la tion i s s igni fic ant a t t he 0.01 l eve l ( 2-ta ile d) . a 0 = M al e, 1 = F em al e b 0 = N o, 1 = Y es c 0 = N o s uga r, 1 = S uga r d 0 = P riva te l abe l, 1 = N at iona l br and D es cr ipt ive s and c or re lat ions be tw ee n t he v ar iabl es ( C ronbac h' s A lphas on di agonal ) 2 3 4 5 6 7 8 9 10 11 12 .080* -.021 -.059 -.101** -.051 .486** .158** .245** -.056 -.089* (.824) .055 .012 .012 .085* .040 -.051 .049 .227** .230** .010 -.012 -.025 .113** .338** .366** -.135** -.203** .165** (.930) -.145** -.049 .423** .745** -.085* .222** .168** .235** (.962) -.034 -.043 .399** .641** -.091* .091* .441** .427** .582** (.916) -.090* .010 .497** .752** -.133** .050 .322** .709** .788** .846** (.912)

(31)

4.4 Results

A 2x2 factorial MANCOVA (GLM) was conducted to examine the relationship between the independent variable (‘sugar content’) and the dependent variables (‘perceived product healthiness’, ‘perceived product tastiness’ and ‘perceived product quality’) and the moderating effect of ‘brand type’. Because this research wants to be as generalizable as possible, the between-subjects factor of the factorial 2x2 MANCOVA are the different conditions. Because it is a 2x2 design, post-hoc tests are not needed.

As a control variable, the mean score of the ‘General Health Interest’ survey was used. The unmanipulated ‘GHI’ score showed equal means in the different conditions (see Appendix 3, A2). Furthermore, the boxplots for the means per condition of ‘perceived healthiness’, ‘perceived tastiness’, ‘perceived quality’, ‘product evaluation’, ‘willingness to pay’ and ‘willingness to buy/recommend’ can be found in Appendix 3, A3 till A8.

The boxplots show that, as expected, the mean scores of ‘perceived healthiness’ (Appendix 3, A3) are highest for the zero sugar conditions, with the highest score in ‘national brand & zero sugar’. Also, as expected, the mean scores of ‘perceived tastiness’ (Appendix 3, A4) are highest for the sugar conditions. Interestingly, the mean scores of ‘perceived quality’ (Appendix 3, A5) and ‘product evaluation’ (Appendix 3, A6) are highest for the national brand conditions. The ‘willingness to pay’ (Appendix 3, A7) show that respondents are willing to pay more for national brands than for private labels. Also, respondents are found to be more willing to buy and recommend national brands (Appendix 3, A8). The multivariate results show that there was a statistically significant interaction effect of the control variable ‘General Health Interest’ on the combined dependent variables, Pillai’s Trace = .023, F(653) = 5.069, p = .002, η2_{= .023. The direction was found to be negative by the parameter}

estimates. A higher ‘General Health Interest’ score leads to a lower ‘perceived healthiness’ (-.187), ‘perceived tastiness’ (-.151) and ‘perceived quality’ (-.135).

(32)

The Box’s test of equality of covariance matrices is violated but due to the large and equal group sized it is assumed that MANOVA is robust enough for these variances. Based on research of Tabachnick and Fidell (2007), Pillai’s Trace test will be used to test the statistical significance of the differences between groups.

The Levene’s test of equality of error variances was found to hold for ‘perceived healthiness’ and to be violated by ‘perceived tastiness’ and ‘perceived quality’. Again, MANOVA is assumed to be robust enough due to the large sample size (Tabachnick & Fidell, 2007).

4.4.1 Main effects

Multivariate tests of the 2x2 factorial MANCOVA show that there was a statistically

significant effect of sugar content on the combined dependent variables, Pillai’s Trace = .125,

F(653) = 31.051, p < .001, η2_{= .125. Also, there was found to be a statistically significant}

effect of brand type on the combined dependent variables, Pillai’s Trace = .211, F(653) = 58.242, p < .001, η2_{= .211.}

For sugar content, the effect on each dependent variable separately shows to be significant. There was a statistically significant main effect of sugar for ‘perceived

healthiness’, F(1, 655) = 27.811, p < .001, η2_{= .041. A significant main effect of sugar for}

‘perceived tastiness’, F(1, 655) = 37.556, p < .001, η2_{= .054 was found as well. Also, a}

significant main effect of sugar for ‘perceived quality’ was found, F(1, 655) = 8.477, p = .004, η2_{= .013. The marginal means for ‘perceived healthiness’ were 2.417 (SE = .083) for}

sugar and 3.031 (SE = .082) for no sugar, a statistically significant mean difference of -.614, 95% CI [-.843, -.385], p < .001. The marginal means for ‘perceived tastiness’ were 4.661 (SE = .087) for sugar and 3.909 (SE = .086) for no sugar, a significant mean difference of .752, 95% CI [.511, .993], p < .001. The marginal means for ‘perceived quality’ were 4.542 (SE =

(33)

.071) for sugar and 4.253 (SE = .070) for no sugar, a significant mean difference of .289, 95% CI [.094, .484], p = .004.

Another interesting finding was the significant effect of brand type on each dependent variable separately. There was a statistically significant main effect of brand type for

‘perceived healthiness’, F(1, 655) = 19.101, p < .001, η2_{= .028. A significant main effect of}

brand type for ‘perceived tastiness’, F(1, 655) = 21.311, p < .001, η2_{= .032 was found as}

well. Also, a significant main effect of brand type for ‘perceived quality’ was found, F(1, 655) = 164.152, p < .001, η2_{= .200. The marginal means for ‘perceived healthiness’ were}

2.978 (SE = .083) for national brands and 2.470 (SE = .082) for private labels, a significant mean difference of .509, 95% CI [.280, .737], p < 0.01. The marginal means for ‘perceived tastiness’ were 4.568 (SE = .087) for national brands and 4.002 (SE = .086) for private labels, a significant mean difference of .566, 95% CI [.325, .807], p < 0.01. The marginal means for ‘perceived quality’ were 5.033 (SE = .070) for national brands and 3.762 (SE = .070), a significant mean difference of 1.271, 95% CI [1.076, 1.466], p < 0.01.

Thus, “H1: A high sugar content in consumer products has a negative effect on

perceived quality, perceived health and product evaluation but a positive effect on perceived taste” is partially supported. Sugar showed to have a significant negative effect on perceived

healthiness (p < .001), but a positive significant effect on perceived tastiness (p < .001) and quality (p = .004).

4.4.2 Moderating effect

Then, the interaction effects between the two independent variables are tested (Hypothesis 2). To test hypothesis 2, a 2x2 factorial MANCOVA was performed. Appendix 3, A9 shows the means and standard deviations of the dependent variables for the different conditions. The multivariate results show that there was a statistically significant interaction effect between

(34)

sugar content and brand type on the combined dependent variables, Pillai’s Trace = .019,

F(653) = 4.294, p = .005, η2_{= .019.}

On an individual level, there was a significant interaction effect between sugar content and brand type for ‘perceived healthiness’, F(1, 655) = 6.004, p = .015, η2_{= .009.}

However, there was no significant interaction effect between sugar content and brand type for ‘perceived tastiness’, F(1, 655) = 2.555, p = .110, η2_{= .004 and ‘perceived quality’ F(1, 655)}

= .895, p = .345, η2_{= .001. As such, a simple main effects analysis was conducted for}

‘perceived healthiness’. There was a significant difference between sugar and no sugar for private labels F(1,655) = 4.033, p = .045 and national brands F(1, 655) = 29.539, p < .001. Therefore, simple comparisons were run for the differences in the mean ‘perceived

healthiness’ score between brand types for sugar and no sugar. The means for perceived healthiness for private labels were 2.634 (SE = .116) for zero sugar and 2.305 (SE = .116) for sugar, a statistically significant mean difference of .329, 95% CI [.007, .651], p = .045. The means for perceived healthiness for national brands were 3.428 (SE = .116) for no sugar and 2.528 (SE = .118) for sugar, a statistically significant mean difference of .899, 95% CI [.574, 1.224], p < .001. See Appendix 3, A10 for a visual representation of this significant

interaction effect of sugar content and brand type on perceived healthiness.

Thus, “H2: The effect of sugar content on product evaluation is stronger for national

brands than for private labels” is only partially supported. When looking at the interaction

effect between brand type and sugar on an individual level, only a moderating effect of brand type on the relationship between sugar content and perceived healthiness is found (F(1, 655) = 6.004, p = .015, η2_{= .009). However, simple comparisons show that for the ‘perceived}

healthiness’, ‘perceived tastiness’ and ‘perceived quality’, statistically significant mean scores between sugar and no sugar are larger for national brands than for private labels.

(35)

5. Discussion and conclusion

This section gives a summary of the aforementioned results and explains what they suggest for the relationship between the variables. Furthermore, the implications, strengths,

limitations and suggestions for future research are discussed. This thesis is based on the assumption that consumers are updating their perceptions of sugar content in consumer products, due to the increasing awareness of and knowledge on sugar. An increasing body of scientific research links sugar with diet-related diseases like obesity, an increased risk of chronic diseases, cardiovascular diseases and dental caries (Louie et al., 2016; WHO, 2015). Also, the FDA (FDA Food Labeling: Revision of the Nutrition and Supplement Facts Labels, 2016) has responded to the growing research on nutrition science by updating the nutrition and supplements facts labels, including “added sugars” in grams and as percent daily values.

The research question guiding this study was: “How does sugar content in consumer products affect consumers’ product evaluations and how is this relationship influenced by using either private labels or national brands?”. Sub-questions to this main research question are: “What is the effect of sugar content on perceived taste, perceived quality, perceived health and product evaluation?” and “How do private labels and national brands moderate the relationship between sugar content and product evaluation?”.

5.1 Summary and key findings

This thesis aimed to find the effect that sugar content in consumer products has on perceived product healthiness, perceived product tastiness and perceived product quality. It also tried to determine whether brand type, being national brand or private label, acted as a moderator on this relationship between sugar content and perceived product evaluations. By providing an answer to these questions, this study aimed to provide an increased understanding on sugar

(36)

content in consumer-packaged goods and the role both private labels and national brands have in the ‘war on sugar’.

Contradictory to health recommendations of the WHO (2015), Louie et al. (2016) found that a large proportion of the Australian youth are consuming alarming and excessive amounts of energy from added sugars. Powell et al. (2016) found that the added sugars intake among both US children and adults increased from 1977 to 2012. This is supported by the findings of Euromonitor (2014), which found that people in a majority of the countries researched consume more sugar than what the World Health Organization recommends for daily intake. The update of the nutrition and supplements facts labels by the FDA in 2016, adding “Added sugars” in grams and as percent values, is perhaps the best illustration of the increased consumer awareness regarding sugar intake. Still, the lack of research on sugar content in consumer products is surprising, given the growth of research on sugar-related health problems. This study goes beyond the existing literature by examining this gap.

Therefore, the first hypothesis (H1) proposed that a high sugar content in consumer products would negatively influence perceived product quality and perceived product healthiness but positively influence perceived product taste. This hypothesis was partly confirmed. Sugar showed to have a significant negative effect on perceived healthiness (p < .001), but a positive significant effect on perceived tastiness (p < .001) and quality (p = .004). This is partly in line with research of Patterson et al. (2012) and Lähteenmäki et al. (2010), who found that health claims can be expected to increase perceived healthiness of products. Furthermore, the higher perceived tastiness of products high in sugar content is in line with earlier research stating consumers assume unhealthy products to taste better (Grewal et al., 1998; Raghunathan et al., 2006; Verbeke, 2006). Research on quality perceptions and sugar content is lacking, but it might be that the positive influence of sugar on perceived quality is possibly due to the correlation between tastiness and quality (see Table 1).

(37)

The second hypothesis (H2) proposed that brand type would enhance the relationship between sugar content and perceived product evaluation, being stronger for national brands than for private labels. This hypothesis was based on the assumption that even though the gap of perceptions between private labels and national brands is narrowing, national brands are often seen as the superior product. Steenkamp et al. (2010) found that consumers still show a higher willingness-to-pay for national brands, compared to private labels. The finding of significant differences in means of WTP between national brands and private labels in this study, confirms these findings of Steenkamp et al. (2010). Hypothesis 2 was only partially supported. When looking at the interaction effect between brand type and sugar on an individual level, only a moderating effect of brand type on the relationship between sugar content and perceived healthiness was found. However, simple comparisons showed that for the ‘perceived healthiness’, ‘perceived tastiness’ and ‘perceived quality’, statistically

significant mean scores between sugar and no sugar were larger for national brands than for private labels.

5.2 Implications, strengths, limitations and future research

There has been an increasing global awareness of the need to maintain a healthy weight to prevent diet-related diseases, such as obesity, diabetes and heart conditions and an increasing body of scientific research connecting sugar to these diet-related diseases. This makes sugar a topic of great interest for marketing scholars, retailers and consumers worldwide. This study tries try to reduce consumers’ sugar intake and help manufactures respond to the shifting demands of consumers. This contributes to healthier eating patterns, more satisfied

consumers and potentially less diet-related diseases. The lack of research on sugar content and product perceptions is surprising, given the growth of research on sugar-related health problems. This study goes beyond the existing literature by examining this gap. This study

(38)

provides an increased understanding to managers on sugar content in consumer-packaged goods and the role both private labels and national brands have in the ‘war on sugar’.

One limitation of this study might be the use of Walmart and Albert Heijn as private label brands. These grocery stores might be perceived as stores of high quality. The

difference might be bigger for other grocery stores, known for example for their low pricing strategy. Different retailers might be of interest for future research. However, as private labels are defined as any brand that is owned by retailers and sold only in its own outlets, it is

questionable if the findings on private label brands can ever be generalizable. The nature of private labels (being part of a retailer) might be diminishing the external validity. Also, due to a limited amount of time, only one type of product per category was used. An advice for future research is the use of more products per category. Also, the national brands used are well known and have dominant positions in the market. Even though through the use of multiple categories it was tried to generate a better view, such brands might still evoke existing attitudes among respondents.

Private labels are found to be struggling to gain consumer trust in Asia and the Middle East, where consumers are fiercely brand-loyal. A strength of this research is therefore the sample consisting of mainly American and European respondents. As the goal of this study was to generalize the results of sugar content in consumer products, another strength is the inclusion of multiple product categories. This research could be a start of research linking sugar content in consumer products with all types of marketing and retail related activities. The growing awareness of consumers on sugar and the detrimental effects of sugar on health should be more than enough reason to investigate the role of sugar content in consumer-packaged goods. Both managers and consumers should be informed how to deal with sugar content in products, to decrease yearly diet-related deaths and lower for example health care costs in the future.

(39)

Reference list

Abril, C., & Sanchez, J. (2016). Will they return? Getting private label consumers to come back: Price, promotion, and new product effects. Journal of Retailing and Consumer

Services, 31, 109-116.

Ares, G., & Gámbaro, A. (2007). Influence of gender, age and motives underlying food choice on perceived healthiness and willingness to try functional

foods. Appetite, 49(1), 148-158.

Bandy, L. (2014). Nutrition Trends: The sugar-free consumer [Powerpoint presentation]. Retrieved from http://go.euromonitor.com/rs/805-KOK-719/images/

NutritionTrendsSugarFreeConsumer.pdf

Bao, Y., Sheng, S., Bao, Y., & Stewart, D. (2011). Assessing quality perception of private labels: Intransient cues and consumer characteristics. Journal of consumer

Marketing, 28(6), 448-458.

Baron, R. M., & Kenny, D. A. (1986). The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations.

Journal of Personality and Social Psychology, 51(6), 1173-1182.

De Ruyter, J. C., Olthof, M. R., Seidell, J. C., & Katan, M. B. (2012). A trial of sugar-free or sugar-sweetened beverages and body weight in children. New England Journal of

Medicine, 367(15), 1397-1406.

Euromonitor International. (2014, November). The Sugar Backlash and its Effects on Global Consumer Markets.

FDA Food Labeling: Revision of the Nutrition and Supplement Facts Labels, 21 C.F.R. § 81.103 (2016).

The effect of sugar on consumers’ product evaluations

Amsterdam Business School

Master Business Administration

Marketing track

The effect of sugar on consumers’ product evaluations

MSc Thesis by:

Bas Spanjaard

10646809

Statement of originality

Abstract

Table of contents

1. Introduction

2. Literature review and conceptual model

3. Methodology

4. Results

5. Discussion and conclusion

Reference list