- Email: [email protected]
Adolescents’ perception of the healthiness of snacks
Food Quality and Preference 50 (2016) 94–101
Contents lists available at ScienceDirect
Food Quality and Preference journal homepage: www.elsevier.com/locate/foodqual
Adolescents’ perception of the healthiness of snacks Tamara Bucher a,b,⇑, Clare Collins b, Sabine Diem a, Michael Siegrist a a
Consumer Behavior, Institute of Food, Nutrition and Health (IFNH), ETH Zürich, Universitätstrasse 22, 8092 Zurich, Switzerland Nutrition and Dietetics, School of Health Sciences, Faculty of Health and Medicine, Priority Research Centre in Physical Activity and Nutrition, The University of Newcastle, University Drive, Newcastle, Callaghan, NSW 2300, Australia b
a r t i c l e
i n f o
Article history: Received 29 May 2015 Received in revised form 12 January 2016 Accepted 2 February 2016 Available online 3 February 2016 Keywords: Snacks Adolescents Health perception Portion size Hierarchical cluster analysis Multidimensional scaling Snacking Sorting task
a b s t r a c t Changes in snacking habits in developed countries are a growing cause for concern, since foods and beverages commonly consumed as snacks, tend to be both energy dense and nutrient poor. Adolescents are characterised by frequent snack consumption. Therefore, promoting more healthful snack choices to adolescents is important for optimising nutrient intake and lowering the risk of chronic disease. The ability to evaluate the healthiness of snacks is essential to making healthy choices. Previous research has shown that health claims can influence consumers’ perceptions of food products. However, little is yet known about consumers’ perceptions of how nutritious or healthy specific foods or beverages are. This knowledge is important for planning successful interventions and designing healthy snacks that will also appeal to population groups with a higher dietary risk, including adolescents. The aim was to investigate how adolescents evaluate the healthiness of snacks currently available for consumption in school environments. Seventy-five adolescents participated in a sorting task and evaluated the healthiness of 37 representative snacks. The data were analysed using hierarchical multiple regression and cluster analysis. The sugar (b = .51, P < .001), fruit (b = .49, P < .001), total fat (b = .41, P = .002) and nut content (b = .35, P = .002) were significant predictors of snacks’ perceived healthiness. The findings of this study are important for tailoring future interventions to promote healthy eating and setting priorities for nutrition education. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Overweight and obesity are directly responsible for at least 2.8 million deaths worldwide each year (WHO, 2014), while the health care costs stemming from poor dietary patterns account for more than 3.6% of the gross national product in developed countries (Popkin, Kim, Rusev, Du, & Zizza, 2006). In Switzerland, more than 40% of all adults and about 20% of children are overweight (Federal Statistical Office, 2012). Increases in the portion sizes of products (Young & Nestle, 2007) and changes in eating patterns, including more frequent snacking, have been identified as contributing to the obesity epidemic (Hill & Peters, 1998; Young & Nestle, 2002). Several studies show that portions sizes, especially those of energy-dense, nutrient-poor (EDNP) foods and snacks, have become larger in recent decades (Nielsen & Popkin, 2003; Steenhuis, Leeuwis, & Vermeer, 2010; Young & Nestle, 2003). ⇑ Corresponding author at: Priority Research Centre in Physical Activity and Nutrition, The University of Newcastle, University Drive, Newcastle, Callaghan, NSW 2300, Australia. E-mail address: [email protected] (T. Bucher). http://dx.doi.org/10.1016/j.foodqual.2016.02.001 0950-3293/Ó 2016 Elsevier Ltd. All rights reserved.
Further, larger portions consistently lead to an increased intake (Rolls, 2014; Rolls, Morris, & Roe, 2002; Steenhuis & Vermeer, 2009). Snacks are often sold in pre-determined portion sizes, which unconsciously suggests a norm for how much should be consumed (Wansink & van Ittersum, 2007, 2013). Portion sizes likely affect intake because individuals consistently consume the vast majority of what they serve themselves (Wansink & Johnson, 2015). Being mindful of consuming healthy portion sizes is therefore essential. Besides larger portion sizes, the increased frequency of snacking in Western societies is a significant cause of concern (Piernas & Popkin, 2010). While eating in the kitchen or dining room at home is associated with a lower BMI in children and adults (Wansink & van Kleef, 2014), adolescents are thought to be particularly prone to making poor choices in terms of nutrition, as they frequently snack and eat outside of the home (Larson et al., 2008; Rangan, Kwan, Flood, Louie, & Gill, 2011; Zizza, Siega-Riz, & Popkin, 2001). Currently, it is unclear whether a higher frequency of snacking promotes weight gain (Hampl, Heaton, & Taylor, 2003; Hartmann, Siegrist, & van der Horst, 2013; Johnson & Anderson, 2010). Adolescents have high energy and nutrition requirements due to the adolescent growth spurt, especially during puberty (EUFIC, 2006),
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
and on average they engage in higher levels of physical activity (EUFIC, 2006). Eating snacks between main meals can help to satisfy these energy and nutrient requirements (EUFIC, 2006). Snacking per se does not have a negative impact on dietary quality. Rather, it can offer an opportunity for making healthy, lower energy density food choices, which result in a wider variety of foods being included in the diet (Whybrow & Kirk, 1997). A recent study conducted by Hartmann et al. found that in Switzerland, snacking was positively associated with intake of sweets and savouries, but also of fruit (Hartmann et al., 2013). Cluster analysis revealed that high-frequency snack consumption occurs in the context of healthy, as well as unhealthy, dietary behaviours and lifestyle patterns (Hartmann et al., 2013). Therefore, public health interventions should promote nutritious snack consumption among groups at particular risk of poor dietary habits. Environmental exposure, motivation, and ability have been suggested as the three key determinants of healthy versus unhealthy food choices (Brug, 2008). Along with motivation, the ability to evaluate the healthiness of snacks when faced with choosing from a range of options is important for adolescents. To date, very little evidence is available concerning why particular foods or beverages are perceived as nutritious or healthy. Lessons about food and nutrition in school could help young people to gain the knowledge necessary to make informed choices about the meals and snacks they regularly consume (EUFIC, 2006). Nutrition education for school-aged children is important because dietary habits, which affect food preferences, energy consumption and nutrient intake, develop in childhood and particularly during adolescence (EUFIC, 2006). In order to provide sustainable nutrition education and to promote healthy snack choices, it is important to understand how adolescents perceive these foods and to determine which criteria shape their perceptions when deciding whether or not a snack is ‘healthy’. Research has shown that health claims can influence consumers’ perceptions of food products (Lahteenmaki, 2013), although little is known about why particular foods are perceived as healthy or unhealthy (Bucher, Müller, & Siegrist, 2015). A study involving adults found a preference for reduced fat products compared to the full fat versions, even though products with zero fat were not preferred to the low fat varieties (Visschers & Siegrist, 2010). Therefore, fat content may also be relevant to health perception. Indeed, recent studies have found that when judging the healthiness of food, women predominantly relied on fat and fibre content (Rizk & Treat, 2014, 2015). Others have found that being ‘natural’ is an important criterion for food choice, with food additives being considered unnatural, unhealthy, or even a health risk (Bearth, Cousin, & Siegrist, 2014). In a study of high school students, moderation, balance, and variety were important criteria for health (Croll, NeumarkSztainer, & Story, 2001), with the term ‘healthy food’ being associated with high consumption of specific food groups such as ‘fruit and vegetables’ or low consumption of energy-dense, nutrientpoor ‘junk’ food (Croll et al., 2001). Adolescents associated healthy eating with fruit, vegetables, carbohydrates and vitamins in a focus group study conducted with children aged 9- to 18-years-old (Fitzgerald, Heary, Nixon, & Kelly, 2010). A recent study found that sugar content ( ), fruit content (+), caffeine content ( ) and sweetener content ( ) were important criteria for both parents and children when evaluating the healthiness of soft drinks (Bucher & Siegrist, 2015). However, it is currently not known whether similar criteria are important for evaluating snack foods. It seems likely that for these foods, other criteria such as energy from fat or protein and portion size may be relevant, although it remains unclear whether consumers actually consider portion size when evaluating the healthiness of products, since the volumes
95
were kept constant in the previous study of soft drinks (Bucher & Siegrist, 2015). To date, there is no universal definition of the ‘healthfulness’ of a food. This is likely due to the numerous factors that need to be considered, including nutrient content, cooking method, and portion size. Lobstein and Davies (2009) attempted to develop a means to define the healthiness of foods using a nutrient profile method. They included key nutrients that have a negative impact on perceived healthiness (i.e. energy, saturated fat, sugars and sodium), as well as others that have a positive impact (i.e. fruit, vegetable, nut, fibre and protein content), and used this information to calculate an overall score (Lobstein & Davies, 2009). This method allows the direct comparison of the healthiness of different foods across categories. However, portion size is not considered within this approach. Therefore, the aim of the current study was to examine the criteria that adolescents aged between 12- and 16-years-old use to judge the healthiness of a variety of snacks available in their school environment and to compare their evaluations with nutrient profile scores.
2. Methods 2.1. Sample characteristics and recruitment Seventy-five adolescents were recruited via flyers distributed at schools in the city of Zurich and surrounding areas. Written informed consent was obtained from all participants. The participants performed the snack-sorting task individually. Two subjects were excluded, one for not following the instructions and another due to incomplete data. Hence, the final analysis included 73 adolescents aged between 12- and 16-years-old (42 girls, mean age 14.2 ± 1.2 years, mean BMI 20.2 ± 2.3 kg/m2). There were no significant differences in BMI by gender. Three participants (4.1%) were in primary school, while the others (n = 70, 95.6%) attended secondary school, which in Switzerland is split into three levels or sections. More specifically, 60 adolescents (82.2%) attended the ‘Sekundarschule’ (medium level), one (1.4%) attended the ‘Realschule’ (lower level), and nine (12.3%) attended the gymnasium (higher level). Some 60 participants (82.2%) were Swiss, nine (12.3%) had another nationality, and four did not answer the relevant question (5.5% missing). All participants lived in the city or the surrounding suburbs of Zurich.
2.2. Selection of snacks Previous studies have shown that healthy children aged sevenyears-old and above are able to complete sorting tasks involving various beverages (Bucher & Siegrist, 2015), and that they can discriminate between healthy and unhealthy foods to some extent (Strachan & Pavie-Latour, 2008). We therefore expected that adolescents aged 12- to 16-years-old would be able to perform a similar task with snack foods. To select snacks regularly consumed by that age group, the snack supply of several schools was analysed. To this end, two cafeterias and two canteens in Zurich were visited and the manager of one school cafeteria was interviewed via telephone about their snack supply. In addition, students, teachers and staff were asked to list their schools’ available snacks or to provide photographs of vending machines and cafeterias. A list of snacks available at seven schools was collected, and a wide range of different healthy and unhealthy snack foods were selected for this study. For some snacks of a similar composition, a range of portion sizes were added to the selection in order to determine the influence of portion size on healthiness evaluations. For example, two
96
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
types of yoghurt with cereal, ‘Emmi Mixit Classic’ (200 g) and ‘Emmi Knusper Mix Classic’ (150 g), were included. 2.3. Calculation of nutrient profile scores As an objective measure of the healthiness of particular foods, we used the Ofcom Nutrient Profiling (NP) Model (Lobstein & Davies, 2009; Rayner, Peter Scarborough, & Tim Lobstein, 2009). This formula scores negative points for total energy, saturated fat, and sugar and sodium content, while it subtracts positive points for fruits, vegetables and nuts, and fibre and protein content (see Table 2 and Supplementary Table 1). A lower score indicates a healthier food product. The Ofcom NP scores were shown to strongly correlate with expert opinion and so the model is currently used to set boundaries for advertising bans on unhealthy food products for children (Rayner et al., 2009). The details that were used to calculate the NP scores for the snacks in this study are reported in Supplementary Table 1. 2.4. Experimental procedure To investigate the criteria that adolescents use to evaluate the healthiness of snacks, we conducted a snack-sorting task using a similar setting to a previous study that investigated perceptions of soft drinks (Bucher & Siegrist, 2015). Participants were invited to the lab individually and then instructed to arrange the 37 snacks in order along a three-metre line. The line ran from the most ‘unhealthy’, labelled on the left end (0 cm), to the most ‘healthy’, labelled on the right end of the line (300 cm), with tied rankings allowed if participants were unable to differentiate between two or more products. The 37 snacks were randomly arranged on a small table opposite the long table that held the three-metre line. Nutrient labels were present on most of the packaged foods and the adolescents were free to use them (or not) as they saw fit (see Supplementary Table 1). The study manager gave instructions from a neutral standing position so as to avoid influencing participants’ decisions either way. It was explained that the sorting task was based on their personal opinions and that there were no right or wrong answers. The participants were further instructed to think out loud and to explain the criteria they were using to organise the snacks, with the study manager taking notes of the participants’ comments. To record all the relevant criteria in a timely fashion, a checklist of potential criteria was pre-prepared, with any new criteria being added as it arose. After the sorting task, the participants were again asked to describe the criteria they had used to sort the snacks. After completing the task, they answered a short questionnaire in which their snacking habits and self-reported height and weight were assessed. Meanwhile, the study manager noted the snacks’ positions on a three-metre strip of cash register paper for later measurement and data entry. When the participants had completed the study, they were provided with financial compensation (CHF 10) and a small snack. 2.5. Statistical analysis All data were analysed using IBM SPSS Statistics, Version 22 (SPSS. Inc., Chicago, IL, USA). Distance data were analysed with a multidimensional scaling analysis (MDS) to represent (dis-) similarities amongst objects in a low-dimensional space (Borg & Groenen, 2005). As a one-dimensional solution was obtained and as the MDS scores were almost perfectly correlated with the mean health perceptions, a further regression analysis on aggregated data was conducted using the mean health perceptions as the dependent variable. Regression coefficients (B), standard errors of the coefficients (SE B), and standardised coefficients (b) were
reported. Variance inflation factors (VIF), which quantify the severity of multi-collinearity in the regression, were also reported. To identify and profile the snack groups, a hierarchical cluster analysis was conducted. Snack clusters were identified based on sorting data by applying Ward’s method. Ward’s method for forming clusters joins objects based on minimising the minimal increment in the within or error sum of squares. Squared Euclidean distances were used as proximity measures in the clustering procedure. The dendrogram similarity scales generated by SPSS ranged from zero (high similarity) to 25 (low similarity). Means (M), standard deviations (SD), and ranks were reported. For non-parametric data, Spearman’s correlation coefficients (RS) were reported and Wilcoxon signed-rank tests were used for comparisons of dependent data.
3. Results Apples were evaluated as the healthiest snack, while gummy bears were seen as the unhealthiest. Table 2 summarises the overall means and ranks of all the snacks. The snacks were not perceived differently between boys and girls, except for the pretzel with butter, which was perceived as healthier by boys (girls: M = 130 ± 58 cm; boys: M = 157 ± 55 cm; t(71) = 2.26, P = .027). The data were therefore not further analysed by gender. An open question was used to investigate the criteria adolescents used in their evaluations. Sugar (or sweetness) was mentioned by almost all of the participants (n = 70, 95.9%). Most also mentioned chocolate (n = 56, 76.7%) and fat (n = 53, 72.6%) as negative criteria, while fruit (n = 52, 74%) and nuts (n = 36, 49.5%) were said to be positive criteria. Portion size or the amount of food was mentioned by about one-third of the participants (n = 24, 32.9%). The criteria reported by more than 10% of the sample as being used to evaluate the healthiness or unhealthiness of snack foods are summarised in Table 1. The actual nutrient contents of the criteria that were mentioned by more than 30% of the participants were used to conduct a hierarchical regression analysis to predict the mean healthiness estimates (see Table 3). Previously, a 50% cut-off was used (Bucher &
Table 1 Frequency of the criteria adolescents mentioned as relevant for sorting 37 snacks (n = 73) in order from least healthy to most healthy, based on how healthy they perceived the food item.
Sugar/sweet Chocolate Fruit (-content) Fat Nut or seed content Amount/portion size Wholemeal Butter Salt Bread Additives (sweeteners, colours, additives, E-numbers) Familiarity Energy (-content)/calories Saturated fat Natural Vitamins Milk Homemade Eating contexts Carbohydrates Satiating Texture
n
%
70 56 54 53 36 24 20 20 20 20 20 19 18 15 15 13 12 11 10 9 8 8
95.9 76.7 74.0 72.6 49.3 32.9 27.4 27.4 27.4 27.4 27.4 26.0 24.7 20.5 20.5 17.8 16.4 15.1 13.7 12.3 11.0 11.0
97
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101 Table 2 Snack characteristics, adolescents’ mean health estimates and nutrient profile (NP) scores. Snack
Apple (140 g) Banana (120 g) Fruit salad (180 g) Whole meal roll with grains (120 g) Trail mix (150 g) Plain fat free yoghurt (180 g) Sesame wheat cracker (42 g) Apple crisps (20 g) Fruit bar (50 g) Yoghurt with muesli big (200 g) Yoghurt with muesli (150 g) Sesame bar (45 g) Fruit and nut bar small (30 g) Strawberry yoghurt (180 g) Crisp bread with cheese filling (37 g) Cereal nut bar (40 g) Croissant (45 g) Pretzel with butter (70 g) Salted crackers (21 g) Salted peanuts (50 g) Apple pie (quiche) (135 g) Danish pastry with apricot (75 g) Chocolate cereal bar (37 g) Strawberry flan (80 g) Blueberry muffin (45 g) Cookie with jelly (80 g) Nut chocolate pastry (25 g) Nut chocolate pastry (75 g) Piece of dark chocolate (12 g) Milk chocolate piece small (5 g) Chocolate bar with milk filling (21 g) Milk chocolate bar (23 g) Potato chips (30 g) Taffy (22 g) Lollypop (12 g) Chocolate peanut bar (50 g) Gummy bears (100 g) a
Nutrient profile scorea
Health estimate (cm) Rank
Mean
SD
NP
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
265 247 241 223 221 215 209 199 196 185 179 179 176 176 172 171 142 141 139 139 130 111 105 105 103 99 91 82 66 59 55 56 57 57 53 51 46
23 44 45 38 39 38 39 44 45 42 43 47 47 44 47 46 51 53 46 61 52 48 47 38 42 37 39 37 38 25 27 23 36 35 35 30 29
5 3 4 0 4 4 8 3 8 1 3 13 2 3 15 1 19 17 23 4 4 14 24 5 13 19 11 11 18 22 28 21 9 14 14 25 14
Classification according to NP score
Healthy Healthy Healthy Healthy Healthy Healthy Less healthy Healthy Less healthy Healthy Healthy Less healthy Healthy Healthy Less healthy Healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy
Nutrient profile scores were calculated according to (Lobstein & Davies, 2009; Rayner et al., 2009).
Table 3 Prediction of health perceptions by fruit, sugar and total fat, nut/seed content (n = 37 snacks). B
SE B
b
VIF
Step 1 (Constant) Sugar content (g/100 g) Fat content (g/100 g)
1.280 2.149
16.60 0.340 0.571
0.485** 0.484**
1.004 1.004
Step 2 (Constant) Sugar content (g/100 g) Fat content (g/100 g) Fruit content (percent) Nut/seed content (percent)
1.347 1.797 0.930 0.905
14.68 0.249 0.523 0.205 0.273
0.511*** 0.405** 0.488*** 0.350**
1.011 1.573 1.311 1.259
Note. VIF stands for Variance Inflation factor. R2 = .44 for Step 1, DR2 = .28 for Step 2 (P < .001), R2 = .72 for Step 2. ** P < .005. *** P < .001.
Siegrist, 2015), however, a less strict cut-off of 30% was chosen for this analysis to avoid missing a relevant predictor. Fruit, nut, fat and sugar content as well as portion size of the snacks were included as predictors. Chocolate (content) was not included as a predictor because it caused multi-collinearity with other (e.g. fat and sugar). Nutrient variables, content variables, and portion sizes were entered in a stepwise regression model. A stepwise approach was
chosen, to examine the respective relevance of the type of criteria entered. We split sugar and fat from fruit and nut into two steps, because fruits contain sugar and fat and nut are correlated. Sugar and fat content (g/100 g) were entered as the first step (R2 = .44), while fruit and nut/seed content (percent) were added as a second step (DR2 = .28 for step 2 (P < .001), R2 = .72 for step 2). Adding the portion sizes (entered as g/portion or kJ/portion) did not significantly improve the model (data not shown). Fruit content was the strongest positive predictor of healthiness ratings (b = .49, P < .001). Nut/seed content was also a positive predictor (b = .35, P = .002), while sugar (b = .51, P < .001) and total fat content (b = .35, P = .002) were negative predictors. The model with these four predictors explained 71% of the variance, which compares with only 58% of variance explained in the regression of the NP scores on the mean health perceptions. Portion size (entered as g/portion or kJ/portion, data shown) was not a significant predictor and was therefore not included in the final regression model. Rank comparisons of health perceptions between nutritionally similar products that only differed in terms of portion size indicated that bigger portions were evaluated as less healthy for the nut pastry (75 g vs. 25 g, Z = 3.3, P = .001) and the gummy bears (100 g) compared to the taffys (22 g, Z = 2.3, P = .21). The larger portion of yoghurt with muesli (180 g vs. 150 g, Z = 2.6, P = .009) was perceived as healthier than the smaller portion. The mean ranks of all snacks are summarised in Table 2. To identify the underlying structure behind the sorting data, we performed a hierarchical cluster analysis using the squared
98
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
Fig. 1. Dendrogram of the hierarchical cluster analysis using the Ward linkage method for the snack sorting task performed by n = 73 adolescents. Note, that distances are rescaled to a range of 0–25, which is a standard procedure (Mooi, Sarstedt, & SpringerLink (Online service), 2011). Nutrient profile scores of the snacks are listed for comparison. Bold values indicate products that were clustered differently than expected.
Euclidean distances method in order to place greater weight on objects that were further apart for stability verification. Ward’s method, which joins objects based on minimising the minimal increment in the within or error sum of squares, was used to form clusters of snacks (Fig. 1). Visual inspection of the dendrogram indicates that the snacks are grouped into two clusters, with further subdivisions being present. Notably, four less healthy snack foods with high NP scores (sesame bar, crispbread with cheese, fruit bar, and sesame wheat crackers) clustered with relatively more healthy snacks. Fig. 2 plots the mean healthiness estimates against the NP scores, indicating that snacks above the reference line were evaluated by the adolescents as being healthier compared to their NP, while foods below the line were evaluated as less healthy in comparison to their NP. This highlights that the unhealthiness of the salted crackers, croissant and cereal chocolate bar was underestimated, while the unhealthiness was overestimated for the taffy, gummy bears and lolly pop. In general, there is good agreement between the NP scores and the estimates for the yoghurt and chocolate snack foods.
4. Discussion The aim of the current study was to investigate how adolescents evaluate the healthiness of a range of common snack foods. The results indicate that adolescents use the sugar, total fat, fruit, and nut or seed content as the main criteria by which they judge the healthiness of snack foods. The study manager noted that the nutrient labels were only used occasionally and that participants were therefore mostly relying on their prior knowledge and beliefs about the products. The degree to which the labels were used in the evaluations was not quantified in this study. Although, label usage was not the aim of the present study, future work should evaluate the influence of access to nutrition information on adolescents’ perceptions of snack healthiness. A comparison of the adolescents’ assessments with the objective nutrient profile scores, which reflect expert rankings (Lobstein & Davies, 2009), indicates that adolescents do not consider the amount of salt, protein, saturated fat or fibre in the products, although they are all important nutrients in terms of the overall healthiness of snack foods and health in general. In line
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
99
Fig. 2. Plot of mean health perception (cm) against nutrient profile scores of 37 snacks. Thirty-seven adolescents sorted snacks from unhealthy (0 cm) to healthy by. The line indicates the linear regression of the health estimates on the nutrient profile scores (R2 = .58). The dashed line indicates the reference line trough through the unhealthiest possible point (NP = 35, 0 cm) and the healthiest possible point (NP = 15, 300 cm) of both scales.
with previous research (Croll et al., 2001), we found that the adolescents frequently mentioned specific food types, for example ‘fruit’, ‘chocolate’ or ‘bread’, as criteria used in their evaluation, rather than ingredients or nutrients. Indeed, vitamins, carbohydrate and saturated fat were rarely mentioned. Although portion sizes were quite frequently mentioned as a criterion, the total weight or energy of the snacks was not a significant predictor of health perceptions. This might also be related to the fact that more of a ‘‘healthy food” (i.e., yoghurt) was perceived as ‘‘good,” but more of an ‘‘unhealthy food” (i.e., nut pastry) was perceived as bad. The present findings indicate, that the relationship between health perception and amount of food or portion size is not linear. Further research needs to investigate the nature of this relationship in more detail. Rozin et al., reported that adults held a common belief that if something was unhealthy in a large portion then it was also unhealthy in low or trace amounts (Rozin, Ashmore, & Markwith, 1996). For example, they believed that a tiny amount of chocolate contains more calories than a large portion of bread (Rozin et al., 1996). This suggests that certain ingredients, such as fruit or chocolate, bias consumers’ health perception of that food, and that consumers thus pay less attention to portion size when evaluating the healthiness or unhealthiness of snack foods. This could have important practical implications and suggest that smaller portion sizes of snacks could be used to unconsciously reduce intake. Also, it appears that targeted marketing can be used to promote the consumption of healthy snack choices. In general, the adolescents had no problem evaluating and identifying very healthy foods, such as fruit, or less healthy snacks such as chocolate or sweets. Yet, the cluster analysis revealed that the participants experienced difficulties evaluating intermediate prod-
ucts as well as some specific snack foods with high (worse) nutrient profile scores, including the fruit bar, which contained large amounts of sugar, or the cheese-filled crispbread, which was high in saturated fat and sodium. The current study had some limitations that need to be addressed. Currently there is a lack of an adequate nutrient profile scoring system taking into account portion size. It is possible that the results would have differed, if more products that varied in their portion size only had been included. In a recent study, Rizk and Treat (2015) found that female students displayed a strong sensitivity to the portion sizes of unhealthy food, but that their sensitivity declined as the portion size increased. However, future research should further assess systematic variations in product portion sizes in addition to specific education on the importance of mindful portion size choices in children and adolescents. Furthermore, there is also a need for the development of a valid nutrient profiling system, which considers portion size. The participants were recruited from the Zurich area and they may not represent the Swiss adolescent population or adolescents in general. In this study, the adolescents were all of normal weight and so it is possible that overweight subjects may have sorted the snacks differently. In addition, the criteria used when evaluating the healthiness of a snack or food are likely to be dependent on culture. Future research should therefore investigate which criteria are most relevant in other populations and weight status groups. Nevertheless, to the best of our knowledge, this is the first study to assess adolescents’ perceptions of the healthiness of different snacks using a sorting task. We used a standardised approach to evaluate adolescents’ health perceptions of foods. The results of the current study highlight potential gaps in the nutrition knowledge of adolescents and so suggest focus areas for nutrition
100
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
education interventions for this group. In addition, the results suggest that public health campaigns targeting this group should focus on promoting healthy snack choices and appropriate portion sizes, particularly when choosing composite or processed foods. The findings of this study could be combined with knowledge concerning adolescents’ product preferences and buying intentions (Norgaard, Sorensen, & Brunso, 2014) to promote healthy snack choices. Also, future research should investigate the relationship between actual nutrient content and the influence of health claims on consumers’ health perceptions of products. 5. Conclusion Adolescents’ intuition or knowledge of what is healthy proved quite accurate. Consequently, a continued focus on what is healthy and what is not healthy may not be as useful as a focus on healthy portion sizes or on behavioural changes that adolescents (and/or parents) can make. Future studies should focus specifically on how the health perceptions of snack foods relate to both preferences and portion size. Further research on the health and nutrition perceptions among different consumer groups could inform an approach for reformulating snacks and their serving sizes in order to improve their nutrient profile as well as their level of consumer acceptance. Financial support This research was partially supported by the Swiss National Science Foundation (SNSF). The SNSF had no role in the design, analysis or writing of this article. Conflict of interest There are no conflicts of interest for this study. Authorship T.B. designed and directed the study and the analyses, conducted the analyses and wrote the manuscript by incorporating critical inputs from S.D., M.S. and C.C. All authors approved the final manuscript. Acknowledgements We are grateful to the adolescents who participated in this research and to Eveline Frey who assisted in the data acquisition. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodqual.2016.02. 001. References Bearth, A., Cousin, M. E., & Siegrist, M. (2014). The consumer’s perception of artificial food additives: Influences on acceptance, risk and benefit perceptions. Food Quality and Preference, 38, 14–23. Borg, I., & Groenen, P. J. F. (2005). Modern multidimensional scaling. Theory and applications (2nd ed.). New York, USA: Springer. Brug, J. (2008). Determinants of healthy eating: Motivation, abilities and environmental opportunities. Family Practice, 25, I50–I55. Bucher, T., Müller, B., & Siegrist, M. (2015). What is healthy food? Objective nutrient profile scores and subjective lay evaluations in comparison. Appetite, 95, 408–414. Bucher, T., & Siegrist, M. (2015). Children’s and parents’ health perception of different soft drinks. British Journal of Nutrition, 113(3), 526–535.
Croll, J. K., Neumark-Sztainer, D., & Story, M. (2001). Healthy eating: What does it mean to adolescents? Journal of Nutrition Education, 33(4), 193–198. EUFIC. (2006). Child and adolescent nutrition. Retrieved 22.07.14, from. Fitzgerald, A., Heary, C., Nixon, E., & Kelly, C. (2010). Factors influencing the food choices of Irish children and adolescents: A qualitative investigation. Health Promotion International, 25(3), 289–298. Hampl, J. S., Heaton, C. L., & Taylor, C. A. (2003). Snacking patterns influence energy and nutrient intakes but not body mass index. Journal of Human Nutrition & Dietetics, 16(1), 3–11. Hartmann, C., Siegrist, M., & van der Horst, K. (2013). Snack frequency: Associations with healthy and unhealthy food choices. Public Health Nutrition, 16(8), 1487–1496. Hill, J. O., & Peters, J. C. (1998). Environmental contributions to the obesity epidemic. Science, 280(5368), 1371–1374. Johnson, G. H., & Anderson, G. H. (2010). Snacking definitions: Impact on interpretation of the literature and dietary recommendations. Critical Reviews in Food Science and Nutrition, 50(9), 848–871. Lahteenmaki, L. (2013). Claiming health in food products. Food Quality and Preference, 27(2), 196–201. Larson, N. I., Neumark-Sztainer, D. R., Story, M. T., Wall, M. M., Harnack, L. J., & Eisenberg, M. E. (2008). Fast food intake: Longitudinal trends during the transition to young adulthood and correlates of intake. Journal of Adolescent Health, 43(1), 79–86. Lobstein, T., & Davies, S. (2009). Defining and labelling ‘healthy’ and ‘unhealthy’ food. Public Health Nutrition, 12(3), 331–340. Mooi, E., & Sarstedt, M., (2011). A Concise Guide to Market Research the Process, Data, and Methods Using IBM SPSS Statistics. Retrieved from SpringerLink . Nielsen, S. J., & Popkin, B. M. (2003). Patterns and trends in portion size in the US between 1977 and 1996: Is eating out bad for us? FASEB Journal, 17(4). A287– A287. Norgaard, M. K., Sorensen, B. T., & Brunso, K. (2014). A concept test of novel healthy snacks among adolescents: Antecedents of preferences and buying intentions. Food Quality and Preference, 33, 17–26. Piernas, C., & Popkin, B. M. (2010). Snacking increased among U.S. adults between 1977 and 2006. Journal of Nutrition, 140(2), 325–332. Popkin, B. M., Kim, S., Rusev, E. R., Du, S., & Zizza, C. (2006). Measuring the full economic costs of diet, physical activity and obesity-related chronic diseases. Obesity Reviews, 7(3), 271–293. Rangan, A. M., Kwan, J., Flood, V. M., Louie, J. C., & Gill, T. P. (2011). Changes in ‘extra’ food intake among Australian children between 1995 and 2007. Obesity Research & Clinical Practice, 5(1), e1–e78. Rayner, M., Peter Scarborough, & Tim Lobstein. (2009). The UK Ofcom Nutrient Profiling Model. Defining ‘healthy’ and ‘unhealthy’ foods and drinks for TV advertising to children. Retrieved 05.05.2015, 2015, from . Rizk, M. T., & Treat, T. A. (2014). An indirect approach to the measurement of nutrient-specific perceptions of food healthiness. Annals of Behavioral Medicine, 48(1), 17–25. Rizk, M. T., & Treat, T. A. (2015). Perceptions of food healthiness among free-living women. Appetite, 95, 390–398. Rolls, B. J. (2014). What is the role of portion control in weight management? International Journal of Obesity, 38, S1–S8. Rolls, B. J., Morris, E. L., & Roe, L. S. (2002). Portion size of food affects energy intake in normal-weight and overweight men and women. American Journal of Clinical Nutrition, 76(6), 1207–1213. Rozin, P., Ashmore, M., & Markwith, M. (1996). Lay American conceptions of nutrition: Dose insensitivity, categorical thinking, contagion, and the monotonic mind. Health Psychology, 15(6), 438–447. Steenhuis, I. H. M., Leeuwis, F. H., & Vermeer, W. M. (2010). Small, medium, large or supersize: Trends in food portion sizes in The Netherlands. Public Health Nutrition, 13(6), 852–857. Steenhuis, I. H. M., & Vermeer, W. M. (2009). Portion size: Review and framework for interventions. International Journal of Behavioral Nutrition and Physical Activity, 6, 58. Strachan, J., & Pavie-Latour, V. (2008). Food for thought: Shouldn’t we actually target food advertising more towards kids and not less? International Journal of Market Research, 50(1), 13–27. Swiss Federal Statistical Office, (2012). Sustainable development – Monet: health – overweight. Retrieved 05.05.2015, 2015, from . Visschers, V. H. M., & Siegrist, M. (2010). When reduced fat increases preference. How fat reduction in nutrition tables and numeracy skills affect food choices. Appetite, 55(3), 730–733. Wansink, B., & Johnson, K. A. (2015). The clean plate club: About 92% of self-served food is eaten. International Journal of Obesity, 39(2), 371–374. Wansink, B., & van Ittersum, K. (2007). Portion size me: Downsizing our consumption norms. Journal of the American Dietetic Association, 107(7), 1103–1106. Wansink, B., & van Ittersum, K. (2013). Portion size me: Plate-size induced consumption norms and win-win solutions for reducing food intake and waste. Journal of Experimental Psychology-Applied, 19(4), 320–332. Wansink, B., & van Kleef, E. (2014). Dinner rituals that correlate with child and adult BMI. Obesity, 22(5), E91–E95.
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101 WHO, (2014). Global Health Observatory (GHO) data. Mean Body Mass Index (BMI), Situation and trends. Retrieved 22.07.14, from
101
Young, L. R., & Nestle, M. (2003). Expanding portion sizes in the US marketplace: Implications for nutrition counseling. Journal of the American Dietetic Association, 103(2), 231–234. Young, L. R., & Nestle, M. (2007). Portion sizes and obesity: Responses of fast-food companies. Journal of Public Health Policy, 28(2), 238–248. Zizza, C., Siega-Riz, A. M., & Popkin, B. M. (2001). Significant increase in young adults’ snacking between 1977–1978 and 1994–1996 represents a cause for concern! Preventive Medicine, 32(4), 303–310.
Contents lists available at ScienceDirect
Food Quality and Preference journal homepage: www.elsevier.com/locate/foodqual
Adolescents’ perception of the healthiness of snacks Tamara Bucher a,b,⇑, Clare Collins b, Sabine Diem a, Michael Siegrist a a
Consumer Behavior, Institute of Food, Nutrition and Health (IFNH), ETH Zürich, Universitätstrasse 22, 8092 Zurich, Switzerland Nutrition and Dietetics, School of Health Sciences, Faculty of Health and Medicine, Priority Research Centre in Physical Activity and Nutrition, The University of Newcastle, University Drive, Newcastle, Callaghan, NSW 2300, Australia b
a r t i c l e
i n f o
Article history: Received 29 May 2015 Received in revised form 12 January 2016 Accepted 2 February 2016 Available online 3 February 2016 Keywords: Snacks Adolescents Health perception Portion size Hierarchical cluster analysis Multidimensional scaling Snacking Sorting task
a b s t r a c t Changes in snacking habits in developed countries are a growing cause for concern, since foods and beverages commonly consumed as snacks, tend to be both energy dense and nutrient poor. Adolescents are characterised by frequent snack consumption. Therefore, promoting more healthful snack choices to adolescents is important for optimising nutrient intake and lowering the risk of chronic disease. The ability to evaluate the healthiness of snacks is essential to making healthy choices. Previous research has shown that health claims can influence consumers’ perceptions of food products. However, little is yet known about consumers’ perceptions of how nutritious or healthy specific foods or beverages are. This knowledge is important for planning successful interventions and designing healthy snacks that will also appeal to population groups with a higher dietary risk, including adolescents. The aim was to investigate how adolescents evaluate the healthiness of snacks currently available for consumption in school environments. Seventy-five adolescents participated in a sorting task and evaluated the healthiness of 37 representative snacks. The data were analysed using hierarchical multiple regression and cluster analysis. The sugar (b = .51, P < .001), fruit (b = .49, P < .001), total fat (b = .41, P = .002) and nut content (b = .35, P = .002) were significant predictors of snacks’ perceived healthiness. The findings of this study are important for tailoring future interventions to promote healthy eating and setting priorities for nutrition education. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Overweight and obesity are directly responsible for at least 2.8 million deaths worldwide each year (WHO, 2014), while the health care costs stemming from poor dietary patterns account for more than 3.6% of the gross national product in developed countries (Popkin, Kim, Rusev, Du, & Zizza, 2006). In Switzerland, more than 40% of all adults and about 20% of children are overweight (Federal Statistical Office, 2012). Increases in the portion sizes of products (Young & Nestle, 2007) and changes in eating patterns, including more frequent snacking, have been identified as contributing to the obesity epidemic (Hill & Peters, 1998; Young & Nestle, 2002). Several studies show that portions sizes, especially those of energy-dense, nutrient-poor (EDNP) foods and snacks, have become larger in recent decades (Nielsen & Popkin, 2003; Steenhuis, Leeuwis, & Vermeer, 2010; Young & Nestle, 2003). ⇑ Corresponding author at: Priority Research Centre in Physical Activity and Nutrition, The University of Newcastle, University Drive, Newcastle, Callaghan, NSW 2300, Australia. E-mail address: [email protected] (T. Bucher). http://dx.doi.org/10.1016/j.foodqual.2016.02.001 0950-3293/Ó 2016 Elsevier Ltd. All rights reserved.
Further, larger portions consistently lead to an increased intake (Rolls, 2014; Rolls, Morris, & Roe, 2002; Steenhuis & Vermeer, 2009). Snacks are often sold in pre-determined portion sizes, which unconsciously suggests a norm for how much should be consumed (Wansink & van Ittersum, 2007, 2013). Portion sizes likely affect intake because individuals consistently consume the vast majority of what they serve themselves (Wansink & Johnson, 2015). Being mindful of consuming healthy portion sizes is therefore essential. Besides larger portion sizes, the increased frequency of snacking in Western societies is a significant cause of concern (Piernas & Popkin, 2010). While eating in the kitchen or dining room at home is associated with a lower BMI in children and adults (Wansink & van Kleef, 2014), adolescents are thought to be particularly prone to making poor choices in terms of nutrition, as they frequently snack and eat outside of the home (Larson et al., 2008; Rangan, Kwan, Flood, Louie, & Gill, 2011; Zizza, Siega-Riz, & Popkin, 2001). Currently, it is unclear whether a higher frequency of snacking promotes weight gain (Hampl, Heaton, & Taylor, 2003; Hartmann, Siegrist, & van der Horst, 2013; Johnson & Anderson, 2010). Adolescents have high energy and nutrition requirements due to the adolescent growth spurt, especially during puberty (EUFIC, 2006),
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
and on average they engage in higher levels of physical activity (EUFIC, 2006). Eating snacks between main meals can help to satisfy these energy and nutrient requirements (EUFIC, 2006). Snacking per se does not have a negative impact on dietary quality. Rather, it can offer an opportunity for making healthy, lower energy density food choices, which result in a wider variety of foods being included in the diet (Whybrow & Kirk, 1997). A recent study conducted by Hartmann et al. found that in Switzerland, snacking was positively associated with intake of sweets and savouries, but also of fruit (Hartmann et al., 2013). Cluster analysis revealed that high-frequency snack consumption occurs in the context of healthy, as well as unhealthy, dietary behaviours and lifestyle patterns (Hartmann et al., 2013). Therefore, public health interventions should promote nutritious snack consumption among groups at particular risk of poor dietary habits. Environmental exposure, motivation, and ability have been suggested as the three key determinants of healthy versus unhealthy food choices (Brug, 2008). Along with motivation, the ability to evaluate the healthiness of snacks when faced with choosing from a range of options is important for adolescents. To date, very little evidence is available concerning why particular foods or beverages are perceived as nutritious or healthy. Lessons about food and nutrition in school could help young people to gain the knowledge necessary to make informed choices about the meals and snacks they regularly consume (EUFIC, 2006). Nutrition education for school-aged children is important because dietary habits, which affect food preferences, energy consumption and nutrient intake, develop in childhood and particularly during adolescence (EUFIC, 2006). In order to provide sustainable nutrition education and to promote healthy snack choices, it is important to understand how adolescents perceive these foods and to determine which criteria shape their perceptions when deciding whether or not a snack is ‘healthy’. Research has shown that health claims can influence consumers’ perceptions of food products (Lahteenmaki, 2013), although little is known about why particular foods are perceived as healthy or unhealthy (Bucher, Müller, & Siegrist, 2015). A study involving adults found a preference for reduced fat products compared to the full fat versions, even though products with zero fat were not preferred to the low fat varieties (Visschers & Siegrist, 2010). Therefore, fat content may also be relevant to health perception. Indeed, recent studies have found that when judging the healthiness of food, women predominantly relied on fat and fibre content (Rizk & Treat, 2014, 2015). Others have found that being ‘natural’ is an important criterion for food choice, with food additives being considered unnatural, unhealthy, or even a health risk (Bearth, Cousin, & Siegrist, 2014). In a study of high school students, moderation, balance, and variety were important criteria for health (Croll, NeumarkSztainer, & Story, 2001), with the term ‘healthy food’ being associated with high consumption of specific food groups such as ‘fruit and vegetables’ or low consumption of energy-dense, nutrientpoor ‘junk’ food (Croll et al., 2001). Adolescents associated healthy eating with fruit, vegetables, carbohydrates and vitamins in a focus group study conducted with children aged 9- to 18-years-old (Fitzgerald, Heary, Nixon, & Kelly, 2010). A recent study found that sugar content ( ), fruit content (+), caffeine content ( ) and sweetener content ( ) were important criteria for both parents and children when evaluating the healthiness of soft drinks (Bucher & Siegrist, 2015). However, it is currently not known whether similar criteria are important for evaluating snack foods. It seems likely that for these foods, other criteria such as energy from fat or protein and portion size may be relevant, although it remains unclear whether consumers actually consider portion size when evaluating the healthiness of products, since the volumes
95
were kept constant in the previous study of soft drinks (Bucher & Siegrist, 2015). To date, there is no universal definition of the ‘healthfulness’ of a food. This is likely due to the numerous factors that need to be considered, including nutrient content, cooking method, and portion size. Lobstein and Davies (2009) attempted to develop a means to define the healthiness of foods using a nutrient profile method. They included key nutrients that have a negative impact on perceived healthiness (i.e. energy, saturated fat, sugars and sodium), as well as others that have a positive impact (i.e. fruit, vegetable, nut, fibre and protein content), and used this information to calculate an overall score (Lobstein & Davies, 2009). This method allows the direct comparison of the healthiness of different foods across categories. However, portion size is not considered within this approach. Therefore, the aim of the current study was to examine the criteria that adolescents aged between 12- and 16-years-old use to judge the healthiness of a variety of snacks available in their school environment and to compare their evaluations with nutrient profile scores.
2. Methods 2.1. Sample characteristics and recruitment Seventy-five adolescents were recruited via flyers distributed at schools in the city of Zurich and surrounding areas. Written informed consent was obtained from all participants. The participants performed the snack-sorting task individually. Two subjects were excluded, one for not following the instructions and another due to incomplete data. Hence, the final analysis included 73 adolescents aged between 12- and 16-years-old (42 girls, mean age 14.2 ± 1.2 years, mean BMI 20.2 ± 2.3 kg/m2). There were no significant differences in BMI by gender. Three participants (4.1%) were in primary school, while the others (n = 70, 95.6%) attended secondary school, which in Switzerland is split into three levels or sections. More specifically, 60 adolescents (82.2%) attended the ‘Sekundarschule’ (medium level), one (1.4%) attended the ‘Realschule’ (lower level), and nine (12.3%) attended the gymnasium (higher level). Some 60 participants (82.2%) were Swiss, nine (12.3%) had another nationality, and four did not answer the relevant question (5.5% missing). All participants lived in the city or the surrounding suburbs of Zurich.
2.2. Selection of snacks Previous studies have shown that healthy children aged sevenyears-old and above are able to complete sorting tasks involving various beverages (Bucher & Siegrist, 2015), and that they can discriminate between healthy and unhealthy foods to some extent (Strachan & Pavie-Latour, 2008). We therefore expected that adolescents aged 12- to 16-years-old would be able to perform a similar task with snack foods. To select snacks regularly consumed by that age group, the snack supply of several schools was analysed. To this end, two cafeterias and two canteens in Zurich were visited and the manager of one school cafeteria was interviewed via telephone about their snack supply. In addition, students, teachers and staff were asked to list their schools’ available snacks or to provide photographs of vending machines and cafeterias. A list of snacks available at seven schools was collected, and a wide range of different healthy and unhealthy snack foods were selected for this study. For some snacks of a similar composition, a range of portion sizes were added to the selection in order to determine the influence of portion size on healthiness evaluations. For example, two
96
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
types of yoghurt with cereal, ‘Emmi Mixit Classic’ (200 g) and ‘Emmi Knusper Mix Classic’ (150 g), were included. 2.3. Calculation of nutrient profile scores As an objective measure of the healthiness of particular foods, we used the Ofcom Nutrient Profiling (NP) Model (Lobstein & Davies, 2009; Rayner, Peter Scarborough, & Tim Lobstein, 2009). This formula scores negative points for total energy, saturated fat, and sugar and sodium content, while it subtracts positive points for fruits, vegetables and nuts, and fibre and protein content (see Table 2 and Supplementary Table 1). A lower score indicates a healthier food product. The Ofcom NP scores were shown to strongly correlate with expert opinion and so the model is currently used to set boundaries for advertising bans on unhealthy food products for children (Rayner et al., 2009). The details that were used to calculate the NP scores for the snacks in this study are reported in Supplementary Table 1. 2.4. Experimental procedure To investigate the criteria that adolescents use to evaluate the healthiness of snacks, we conducted a snack-sorting task using a similar setting to a previous study that investigated perceptions of soft drinks (Bucher & Siegrist, 2015). Participants were invited to the lab individually and then instructed to arrange the 37 snacks in order along a three-metre line. The line ran from the most ‘unhealthy’, labelled on the left end (0 cm), to the most ‘healthy’, labelled on the right end of the line (300 cm), with tied rankings allowed if participants were unable to differentiate between two or more products. The 37 snacks were randomly arranged on a small table opposite the long table that held the three-metre line. Nutrient labels were present on most of the packaged foods and the adolescents were free to use them (or not) as they saw fit (see Supplementary Table 1). The study manager gave instructions from a neutral standing position so as to avoid influencing participants’ decisions either way. It was explained that the sorting task was based on their personal opinions and that there were no right or wrong answers. The participants were further instructed to think out loud and to explain the criteria they were using to organise the snacks, with the study manager taking notes of the participants’ comments. To record all the relevant criteria in a timely fashion, a checklist of potential criteria was pre-prepared, with any new criteria being added as it arose. After the sorting task, the participants were again asked to describe the criteria they had used to sort the snacks. After completing the task, they answered a short questionnaire in which their snacking habits and self-reported height and weight were assessed. Meanwhile, the study manager noted the snacks’ positions on a three-metre strip of cash register paper for later measurement and data entry. When the participants had completed the study, they were provided with financial compensation (CHF 10) and a small snack. 2.5. Statistical analysis All data were analysed using IBM SPSS Statistics, Version 22 (SPSS. Inc., Chicago, IL, USA). Distance data were analysed with a multidimensional scaling analysis (MDS) to represent (dis-) similarities amongst objects in a low-dimensional space (Borg & Groenen, 2005). As a one-dimensional solution was obtained and as the MDS scores were almost perfectly correlated with the mean health perceptions, a further regression analysis on aggregated data was conducted using the mean health perceptions as the dependent variable. Regression coefficients (B), standard errors of the coefficients (SE B), and standardised coefficients (b) were
reported. Variance inflation factors (VIF), which quantify the severity of multi-collinearity in the regression, were also reported. To identify and profile the snack groups, a hierarchical cluster analysis was conducted. Snack clusters were identified based on sorting data by applying Ward’s method. Ward’s method for forming clusters joins objects based on minimising the minimal increment in the within or error sum of squares. Squared Euclidean distances were used as proximity measures in the clustering procedure. The dendrogram similarity scales generated by SPSS ranged from zero (high similarity) to 25 (low similarity). Means (M), standard deviations (SD), and ranks were reported. For non-parametric data, Spearman’s correlation coefficients (RS) were reported and Wilcoxon signed-rank tests were used for comparisons of dependent data.
3. Results Apples were evaluated as the healthiest snack, while gummy bears were seen as the unhealthiest. Table 2 summarises the overall means and ranks of all the snacks. The snacks were not perceived differently between boys and girls, except for the pretzel with butter, which was perceived as healthier by boys (girls: M = 130 ± 58 cm; boys: M = 157 ± 55 cm; t(71) = 2.26, P = .027). The data were therefore not further analysed by gender. An open question was used to investigate the criteria adolescents used in their evaluations. Sugar (or sweetness) was mentioned by almost all of the participants (n = 70, 95.9%). Most also mentioned chocolate (n = 56, 76.7%) and fat (n = 53, 72.6%) as negative criteria, while fruit (n = 52, 74%) and nuts (n = 36, 49.5%) were said to be positive criteria. Portion size or the amount of food was mentioned by about one-third of the participants (n = 24, 32.9%). The criteria reported by more than 10% of the sample as being used to evaluate the healthiness or unhealthiness of snack foods are summarised in Table 1. The actual nutrient contents of the criteria that were mentioned by more than 30% of the participants were used to conduct a hierarchical regression analysis to predict the mean healthiness estimates (see Table 3). Previously, a 50% cut-off was used (Bucher &
Table 1 Frequency of the criteria adolescents mentioned as relevant for sorting 37 snacks (n = 73) in order from least healthy to most healthy, based on how healthy they perceived the food item.
Sugar/sweet Chocolate Fruit (-content) Fat Nut or seed content Amount/portion size Wholemeal Butter Salt Bread Additives (sweeteners, colours, additives, E-numbers) Familiarity Energy (-content)/calories Saturated fat Natural Vitamins Milk Homemade Eating contexts Carbohydrates Satiating Texture
n
%
70 56 54 53 36 24 20 20 20 20 20 19 18 15 15 13 12 11 10 9 8 8
95.9 76.7 74.0 72.6 49.3 32.9 27.4 27.4 27.4 27.4 27.4 26.0 24.7 20.5 20.5 17.8 16.4 15.1 13.7 12.3 11.0 11.0
97
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101 Table 2 Snack characteristics, adolescents’ mean health estimates and nutrient profile (NP) scores. Snack
Apple (140 g) Banana (120 g) Fruit salad (180 g) Whole meal roll with grains (120 g) Trail mix (150 g) Plain fat free yoghurt (180 g) Sesame wheat cracker (42 g) Apple crisps (20 g) Fruit bar (50 g) Yoghurt with muesli big (200 g) Yoghurt with muesli (150 g) Sesame bar (45 g) Fruit and nut bar small (30 g) Strawberry yoghurt (180 g) Crisp bread with cheese filling (37 g) Cereal nut bar (40 g) Croissant (45 g) Pretzel with butter (70 g) Salted crackers (21 g) Salted peanuts (50 g) Apple pie (quiche) (135 g) Danish pastry with apricot (75 g) Chocolate cereal bar (37 g) Strawberry flan (80 g) Blueberry muffin (45 g) Cookie with jelly (80 g) Nut chocolate pastry (25 g) Nut chocolate pastry (75 g) Piece of dark chocolate (12 g) Milk chocolate piece small (5 g) Chocolate bar with milk filling (21 g) Milk chocolate bar (23 g) Potato chips (30 g) Taffy (22 g) Lollypop (12 g) Chocolate peanut bar (50 g) Gummy bears (100 g) a
Nutrient profile scorea
Health estimate (cm) Rank
Mean
SD
NP
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
265 247 241 223 221 215 209 199 196 185 179 179 176 176 172 171 142 141 139 139 130 111 105 105 103 99 91 82 66 59 55 56 57 57 53 51 46
23 44 45 38 39 38 39 44 45 42 43 47 47 44 47 46 51 53 46 61 52 48 47 38 42 37 39 37 38 25 27 23 36 35 35 30 29
5 3 4 0 4 4 8 3 8 1 3 13 2 3 15 1 19 17 23 4 4 14 24 5 13 19 11 11 18 22 28 21 9 14 14 25 14
Classification according to NP score
Healthy Healthy Healthy Healthy Healthy Healthy Less healthy Healthy Less healthy Healthy Healthy Less healthy Healthy Healthy Less healthy Healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy Less healthy
Nutrient profile scores were calculated according to (Lobstein & Davies, 2009; Rayner et al., 2009).
Table 3 Prediction of health perceptions by fruit, sugar and total fat, nut/seed content (n = 37 snacks). B
SE B
b
VIF
Step 1 (Constant) Sugar content (g/100 g) Fat content (g/100 g)
1.280 2.149
16.60 0.340 0.571
0.485** 0.484**
1.004 1.004
Step 2 (Constant) Sugar content (g/100 g) Fat content (g/100 g) Fruit content (percent) Nut/seed content (percent)
1.347 1.797 0.930 0.905
14.68 0.249 0.523 0.205 0.273
0.511*** 0.405** 0.488*** 0.350**
1.011 1.573 1.311 1.259
Note. VIF stands for Variance Inflation factor. R2 = .44 for Step 1, DR2 = .28 for Step 2 (P < .001), R2 = .72 for Step 2. ** P < .005. *** P < .001.
Siegrist, 2015), however, a less strict cut-off of 30% was chosen for this analysis to avoid missing a relevant predictor. Fruit, nut, fat and sugar content as well as portion size of the snacks were included as predictors. Chocolate (content) was not included as a predictor because it caused multi-collinearity with other (e.g. fat and sugar). Nutrient variables, content variables, and portion sizes were entered in a stepwise regression model. A stepwise approach was
chosen, to examine the respective relevance of the type of criteria entered. We split sugar and fat from fruit and nut into two steps, because fruits contain sugar and fat and nut are correlated. Sugar and fat content (g/100 g) were entered as the first step (R2 = .44), while fruit and nut/seed content (percent) were added as a second step (DR2 = .28 for step 2 (P < .001), R2 = .72 for step 2). Adding the portion sizes (entered as g/portion or kJ/portion) did not significantly improve the model (data not shown). Fruit content was the strongest positive predictor of healthiness ratings (b = .49, P < .001). Nut/seed content was also a positive predictor (b = .35, P = .002), while sugar (b = .51, P < .001) and total fat content (b = .35, P = .002) were negative predictors. The model with these four predictors explained 71% of the variance, which compares with only 58% of variance explained in the regression of the NP scores on the mean health perceptions. Portion size (entered as g/portion or kJ/portion, data shown) was not a significant predictor and was therefore not included in the final regression model. Rank comparisons of health perceptions between nutritionally similar products that only differed in terms of portion size indicated that bigger portions were evaluated as less healthy for the nut pastry (75 g vs. 25 g, Z = 3.3, P = .001) and the gummy bears (100 g) compared to the taffys (22 g, Z = 2.3, P = .21). The larger portion of yoghurt with muesli (180 g vs. 150 g, Z = 2.6, P = .009) was perceived as healthier than the smaller portion. The mean ranks of all snacks are summarised in Table 2. To identify the underlying structure behind the sorting data, we performed a hierarchical cluster analysis using the squared
98
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
Fig. 1. Dendrogram of the hierarchical cluster analysis using the Ward linkage method for the snack sorting task performed by n = 73 adolescents. Note, that distances are rescaled to a range of 0–25, which is a standard procedure (Mooi, Sarstedt, & SpringerLink (Online service), 2011). Nutrient profile scores of the snacks are listed for comparison. Bold values indicate products that were clustered differently than expected.
Euclidean distances method in order to place greater weight on objects that were further apart for stability verification. Ward’s method, which joins objects based on minimising the minimal increment in the within or error sum of squares, was used to form clusters of snacks (Fig. 1). Visual inspection of the dendrogram indicates that the snacks are grouped into two clusters, with further subdivisions being present. Notably, four less healthy snack foods with high NP scores (sesame bar, crispbread with cheese, fruit bar, and sesame wheat crackers) clustered with relatively more healthy snacks. Fig. 2 plots the mean healthiness estimates against the NP scores, indicating that snacks above the reference line were evaluated by the adolescents as being healthier compared to their NP, while foods below the line were evaluated as less healthy in comparison to their NP. This highlights that the unhealthiness of the salted crackers, croissant and cereal chocolate bar was underestimated, while the unhealthiness was overestimated for the taffy, gummy bears and lolly pop. In general, there is good agreement between the NP scores and the estimates for the yoghurt and chocolate snack foods.
4. Discussion The aim of the current study was to investigate how adolescents evaluate the healthiness of a range of common snack foods. The results indicate that adolescents use the sugar, total fat, fruit, and nut or seed content as the main criteria by which they judge the healthiness of snack foods. The study manager noted that the nutrient labels were only used occasionally and that participants were therefore mostly relying on their prior knowledge and beliefs about the products. The degree to which the labels were used in the evaluations was not quantified in this study. Although, label usage was not the aim of the present study, future work should evaluate the influence of access to nutrition information on adolescents’ perceptions of snack healthiness. A comparison of the adolescents’ assessments with the objective nutrient profile scores, which reflect expert rankings (Lobstein & Davies, 2009), indicates that adolescents do not consider the amount of salt, protein, saturated fat or fibre in the products, although they are all important nutrients in terms of the overall healthiness of snack foods and health in general. In line
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
99
Fig. 2. Plot of mean health perception (cm) against nutrient profile scores of 37 snacks. Thirty-seven adolescents sorted snacks from unhealthy (0 cm) to healthy by. The line indicates the linear regression of the health estimates on the nutrient profile scores (R2 = .58). The dashed line indicates the reference line trough through the unhealthiest possible point (NP = 35, 0 cm) and the healthiest possible point (NP = 15, 300 cm) of both scales.
with previous research (Croll et al., 2001), we found that the adolescents frequently mentioned specific food types, for example ‘fruit’, ‘chocolate’ or ‘bread’, as criteria used in their evaluation, rather than ingredients or nutrients. Indeed, vitamins, carbohydrate and saturated fat were rarely mentioned. Although portion sizes were quite frequently mentioned as a criterion, the total weight or energy of the snacks was not a significant predictor of health perceptions. This might also be related to the fact that more of a ‘‘healthy food” (i.e., yoghurt) was perceived as ‘‘good,” but more of an ‘‘unhealthy food” (i.e., nut pastry) was perceived as bad. The present findings indicate, that the relationship between health perception and amount of food or portion size is not linear. Further research needs to investigate the nature of this relationship in more detail. Rozin et al., reported that adults held a common belief that if something was unhealthy in a large portion then it was also unhealthy in low or trace amounts (Rozin, Ashmore, & Markwith, 1996). For example, they believed that a tiny amount of chocolate contains more calories than a large portion of bread (Rozin et al., 1996). This suggests that certain ingredients, such as fruit or chocolate, bias consumers’ health perception of that food, and that consumers thus pay less attention to portion size when evaluating the healthiness or unhealthiness of snack foods. This could have important practical implications and suggest that smaller portion sizes of snacks could be used to unconsciously reduce intake. Also, it appears that targeted marketing can be used to promote the consumption of healthy snack choices. In general, the adolescents had no problem evaluating and identifying very healthy foods, such as fruit, or less healthy snacks such as chocolate or sweets. Yet, the cluster analysis revealed that the participants experienced difficulties evaluating intermediate prod-
ucts as well as some specific snack foods with high (worse) nutrient profile scores, including the fruit bar, which contained large amounts of sugar, or the cheese-filled crispbread, which was high in saturated fat and sodium. The current study had some limitations that need to be addressed. Currently there is a lack of an adequate nutrient profile scoring system taking into account portion size. It is possible that the results would have differed, if more products that varied in their portion size only had been included. In a recent study, Rizk and Treat (2015) found that female students displayed a strong sensitivity to the portion sizes of unhealthy food, but that their sensitivity declined as the portion size increased. However, future research should further assess systematic variations in product portion sizes in addition to specific education on the importance of mindful portion size choices in children and adolescents. Furthermore, there is also a need for the development of a valid nutrient profiling system, which considers portion size. The participants were recruited from the Zurich area and they may not represent the Swiss adolescent population or adolescents in general. In this study, the adolescents were all of normal weight and so it is possible that overweight subjects may have sorted the snacks differently. In addition, the criteria used when evaluating the healthiness of a snack or food are likely to be dependent on culture. Future research should therefore investigate which criteria are most relevant in other populations and weight status groups. Nevertheless, to the best of our knowledge, this is the first study to assess adolescents’ perceptions of the healthiness of different snacks using a sorting task. We used a standardised approach to evaluate adolescents’ health perceptions of foods. The results of the current study highlight potential gaps in the nutrition knowledge of adolescents and so suggest focus areas for nutrition
100
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101
education interventions for this group. In addition, the results suggest that public health campaigns targeting this group should focus on promoting healthy snack choices and appropriate portion sizes, particularly when choosing composite or processed foods. The findings of this study could be combined with knowledge concerning adolescents’ product preferences and buying intentions (Norgaard, Sorensen, & Brunso, 2014) to promote healthy snack choices. Also, future research should investigate the relationship between actual nutrient content and the influence of health claims on consumers’ health perceptions of products. 5. Conclusion Adolescents’ intuition or knowledge of what is healthy proved quite accurate. Consequently, a continued focus on what is healthy and what is not healthy may not be as useful as a focus on healthy portion sizes or on behavioural changes that adolescents (and/or parents) can make. Future studies should focus specifically on how the health perceptions of snack foods relate to both preferences and portion size. Further research on the health and nutrition perceptions among different consumer groups could inform an approach for reformulating snacks and their serving sizes in order to improve their nutrient profile as well as their level of consumer acceptance. Financial support This research was partially supported by the Swiss National Science Foundation (SNSF). The SNSF had no role in the design, analysis or writing of this article. Conflict of interest There are no conflicts of interest for this study. Authorship T.B. designed and directed the study and the analyses, conducted the analyses and wrote the manuscript by incorporating critical inputs from S.D., M.S. and C.C. All authors approved the final manuscript. Acknowledgements We are grateful to the adolescents who participated in this research and to Eveline Frey who assisted in the data acquisition. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodqual.2016.02. 001. References Bearth, A., Cousin, M. E., & Siegrist, M. (2014). The consumer’s perception of artificial food additives: Influences on acceptance, risk and benefit perceptions. Food Quality and Preference, 38, 14–23. Borg, I., & Groenen, P. J. F. (2005). Modern multidimensional scaling. Theory and applications (2nd ed.). New York, USA: Springer. Brug, J. (2008). Determinants of healthy eating: Motivation, abilities and environmental opportunities. Family Practice, 25, I50–I55. Bucher, T., Müller, B., & Siegrist, M. (2015). What is healthy food? Objective nutrient profile scores and subjective lay evaluations in comparison. Appetite, 95, 408–414. Bucher, T., & Siegrist, M. (2015). Children’s and parents’ health perception of different soft drinks. British Journal of Nutrition, 113(3), 526–535.
Croll, J. K., Neumark-Sztainer, D., & Story, M. (2001). Healthy eating: What does it mean to adolescents? Journal of Nutrition Education, 33(4), 193–198. EUFIC. (2006). Child and adolescent nutrition. Retrieved 22.07.14, from
T. Bucher et al. / Food Quality and Preference 50 (2016) 94–101 WHO, (2014). Global Health Observatory (GHO) data. Mean Body Mass Index (BMI), Situation and trends. Retrieved 22.07.14, from
101
Young, L. R., & Nestle, M. (2003). Expanding portion sizes in the US marketplace: Implications for nutrition counseling. Journal of the American Dietetic Association, 103(2), 231–234. Young, L. R., & Nestle, M. (2007). Portion sizes and obesity: Responses of fast-food companies. Journal of Public Health Policy, 28(2), 238–248. Zizza, C., Siega-Riz, A. M., & Popkin, B. M. (2001). Significant increase in young adults’ snacking between 1977–1978 and 1994–1996 represents a cause for concern! Preventive Medicine, 32(4), 303–310.