Simply adding the word “fruit” makes sugar healthier: The misleading effect of symbolic information on the perceived healthiness of food

Accepted Manuscript Simply adding the word “fruit” makes sugar healthier: The misleading effect of symbolic information on the perceived healthiness of food Bernadette Sütterlin, Michael Siegrist PII:

S0195-6663(15)00328-1

DOI:

10.1016/j.appet.2015.07.011

Reference:

APPET 2633

To appear in:

Appetite

Received Date: 20 August 2014 Revised Date:

2 July 2015

Accepted Date: 9 July 2015

Please cite this article as: Sütterlin B. & Siegrist M., Simply adding the word “fruit” makes sugar healthier: The misleading effect of symbolic information on the perceived healthiness of food, Appetite (2015), doi: 10.1016/j.appet.2015.07.011. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1 2

Simply adding the word “fruit” makes sugar healthier: The misleading effect of

3

symbolic information on the perceived healthiness of food

4 Bernadette Sütterlin and Michael Siegrist

6

ETH Zurich, Institute for Environmental Decisions (IED), Consumer Behavior, Switzerland

RI PT

5

7 Address for correspondence:

9

Michael Siegrist

SC

8

ETH Zurich

11

Institute for Environmental Decisions (IED), Consumer Behavior

12

Universitaetstrasse 22, CHN J76.3

13

CH-8092 Zurich

14

Switzerland

15

E-mail: [email protected]

16

Telephone: +41 (0)44 632 63 21

AC C

EP

TE D

M AN U

10

1

ACCEPTED MANUSCRIPT Abstract

18

People may use simple heuristics to assess the healthiness of food products. For instance, the

19

information that a product contains “fruit sugar” (in German, “fruit sugar” is the colloquial

20

term for fructose) could be interpreted as a cue that the product is relatively healthy, since the

21

term “fruit” symbolizes healthiness. This can have a misleading effect on the perceived

22

healthiness of a product. In Experiment 1, participants (N = 164) were asked to evaluate the

23

healthiness of one of two breakfast cereals based on the information provided in a nutrition

24

table. For one group, the label “fruit sugar” was used; for the other, the label “sugar” was

25

used. Results suggest that the phrase “fruit sugar” listed as an ingredient of the breakfast

26

cereal resulted in a more positive perception of the healthiness of the cereal compared with

27

the ingredient labeled “sugar.” In Experiment 2 (N = 202), the results of Experiment 1 were

28

replicated with a within-subjects design in which participants evaluated the two products

29

simultaneously. Experiment 3 (N = 251) ruled out the alternative explanation that the effect

30

could be due to differing inferences about the product’s ingredients based on the label used,

31

that is, that the product labeled with “fruit sugar” contains fruit. Finally, in Experiment 4 (N =

32

162), the results show that the perceived healthiness associated with the labeling of the

33

ingredient “sugar” (“fruit sugar” vs. “sugar”) mediates the observed effect. Results of the four

34

experiments indicate that symbolic information is an important factor that can influence

35

people’s health perceptions of food. These findings have implications for marketing and

36

public health.

SC

M AN U

TE D

EP

AC C

37

RI PT

17

38

Keywords: Health perception, fructose, fruit sugar, heuristics, health halo effect, symbolic

39

information

2

ACCEPTED MANUSCRIPT 40

Simply adding the word “fruit” makes sugar healthier: The misleading effect of

41

symbolic information on the perceived healthiness of food

42

1. Introduction

43

Most consumers seem to have limited nutrition knowledge (Parmenter & Wardle, 1999). For instance, a Swiss study showed that two thirds of the study participants

45

erroneously believed that brown sugar is much healthier than white sugar (Dickson-

46

Spillmann, Siegrist, & Keller, 2011). If people lack the necessary knowledge to make

47

informed decisions, they have to rely on substitutes for knowledge. In such situations, people

48

may rely on simple heuristics to make decisions (Tversky & Kahneman, 1974). A possible

49

side effect related to the use of heuristics is that wrong conclusions are drawn. In the present

50

research, we examined whether a product labeled as containing “fruit sugar” 1 (in German,

51

“fruit sugar” is the colloquial term for fructose) is perceived to be healthier than a product

52

labeled as containing sugar, since the term “fruit” symbolizes healthiness. Such a result

53

would suggest that consumers rely on symbolic information (e.g., the word “fruit” in “fruit

54

sugar”) to judge non-observable food properties, such as the healthiness of a specific food

55

product. Furthermore, we examined the underlying mechanisms and whether the effect also

56

extends to other product expectations, such as taste.

EP

TE D

M AN U

SC

RI PT

44

A general feature of heuristic judgment is attribute substitution (Kahneman &

58

Frederick, 2005). A judgment is mediated by a heuristic when the attribute of the object a

59

person wants to judge (target attribute) is not readily accessible and a person assesses this

60

target attribute by substituting it with a seemingly semantically or associatively related

61

property that comes more easily to mind (heuristic attribute). Informational attributes with

62

high symbolic significance can serve as such heuristic attributes. Since the substituted

63

heuristic attribute differs from the target attribute, the use of heuristics may result in biased

64

decisions.

AC C

57

3

ACCEPTED MANUSCRIPT 65

The influence of the symbolic significance of information on people’s information processing and evaluation of behaviors, attitudinal objects, and incidents has already been

67

demonstrated in previous studies. Recent research, for example, showed that due to the

68

tendency to focus on information with a strong symbolic meaning, people underestimate the

69

energy consumption of consumers engaging in behaviors with a strong symbolic meaning of

70

energy-friendliness, while they overestimate the energy consumption of consumers who show

71

behaviors with a strong symbolic meaning of energy-unfriendliness (Sütterlin & Siegrist,

72

2014). In line with the assumption that people rely mainly on symbolically meaningful

73

information, another study found that people are more concerned about negative outcomes of

74

hazards, and evaluate them more negatively when they are caused by humans, compared with

75

the identical negative outcomes caused by nature (Siegrist & Sütterlin, 2014). Due to the

76

symbolic nature, the judgments were mainly based on the information about the cause,

77

resulting in different evaluations of the very same outcome. In judgments of the healthiness of

78

food, the word “fruit” has high symbolic significance, since “fruit” is commonly perceived as

79

symbolizing healthiness (Briz et al., 2008). Therefore, if the phrase “fruit sugar” evokes

80

healthy associations because it includes the word “fruit,” this may positively influence the

81

perceived healthiness of a product that contains “fruit sugar,” compared with a product that is

82

described as containing “sugar.”

EP

TE D

M AN U

SC

RI PT

66

What type of symbolic information is relevant when it comes to the perceived

84

healthiness of food? Characteristics such as synthetic and natural seem to be important for

85

evaluating food products (Evans, de Challemaison, & Cox, 2010; Rozin, 2005). Consumers in

86

the Western world have a clear preference for natural products and seem to perceive synthetic

87

products in a negative way. Consequently, symbolic information that indicates the naturalness

88

of a food or reminds people that the food contains artificial ingredients may also influence the

89

perceived healthiness of a product and its sensory evaluation. A symbolic meaning attributed

90

to an aspect of a product or to a term used in the description of the ingredients (e.g., “fruit

AC C

83

4

ACCEPTED MANUSCRIPT sugar”) is based on people’s interpretation of this aspect (e.g., natural ingredient that is

92

relatively healthy). The meaning is symbolic because it transcends the available facts, and

93

stereotypical information associated with the food shapes perception. In line with this

94

reasoning, findings of a recent qualitative study, for example, indicated that children tend to

95

perceive honey to be healthier than white sugar, because they associate it with naturalness

96

(Brierley & Elliott, 2015).

97

RI PT

91

A closely related phenomenon is the health halo effect, which may lead consumers to make incorrect inferences (Andrews, Burton, & Netemeyer, 2000; Roe, Levy, & Derby,

99

1999). A health halo effect occurs when the perception of an attribute of a product influences

SC

98

the health evaluation of an unrelated product attribute. Research suggests that health claims

101

can cause the perception of inappropriate health benefits (Roe et al., 1999). Perceiving food as

102

healthy may result in underestimating their calories. For instance, the calories in food at fast-

103

food restaurants that are perceived to be healthy were underestimated (Chandon & Wansink,

104

2007). This health halo effect has resulted in higher calorie consumption when food was

105

provided by “healthy” restaurants rather than “unhealthy” restaurants. Health claims also have

106

an impact on product evaluation and the perceived naturalness (Lähteenmäki et al., 2010).

107

Furthermore, health labels have been shown to influence people’s expectations of a product’s

108

taste (Liem, Toraman Aydin, & Zandstra, 2012). However, in studies demonstrating health

109

halo effects, participants did not receive numerical information about the products.

110

Participants had to evaluate food properties (e.g., calories) in a state of high ambiguity.

111

Studies need to demonstrate that symbolic information also influences the interpretation of

112

concrete numerical information, without providing health claims that are explicitly targeted to

113

push consumers’ thoughts in a certain direction.

114

AC C

EP

TE D

M AN U

100

The goal of the present study was to test whether information interpretation and,

115

consequently, perception of a food product differ depending on whether symbolic information

116

is included, even if identical information about the nutrients and calories of the product is 5

ACCEPTED MANUSCRIPT provided. We hypothesized that due to the positive symbolic meaning of the term “fruit,” the

118

ingredient “fruit sugar” would be perceived as healthier than (the unspecific term) “sugar.”

119

Thus, products labeled as containing “fruit sugar” may be perceived as healthier than products

120

labeled as containing sugar. We expected such a result even if the nutrition profiles of the

121

products are otherwise identical.

122

2. Experiment 1

123

RI PT

117

In the first experiment, we used a between-subjects design to examine whether

participants in fact evaluate the identical nutrition information of breakfast cereals differently

125

with regard to healthiness when the labels “sugar” versus “fruit sugar” are used in the

126

nutrition table for the ingredient sugar. We postulated that, due to the symbolic meaning of

127

healthiness associated with the term “fruit,” the ingredient sugar is perceived to be healthier

128

when labeled as “fruit sugar” than when labeled with the umbrella term “sugar,” and that, as a

129

result, cereals presented as containing “fruit sugar” are assessed as healthier than cereals

130

presented as containing “sugar.”

131

2.1. Methods

132

2.1.1. Participants

M AN U

TE D

Experiment 1 involved 164 participants. Participants represented a convenience sample

EP

133

SC

124

of people living in the German-speaking part of Switzerland who agreed to participate in

135

online studies conducted by our group. The sample consisted of 61 (37%) women and 103

136

(63%) men. The mean age was 55 years (SD = 14).

137

2.1.2. Materials and Procedure

138

AC C

134

Participants received an e-mail asking them to fill out a short online questionnaire. In

139

addition to the experiment presented here, participants answered questions not related to the

140

research question of this study.

141

Participants received the following information: “Miss Meier wants to buy breakfast

142

cereals for her children. The following information is displayed on the package.” After this 6

ACCEPTED MANUSCRIPT introductory text, a nutrition table containing the nutrition information presented in Table 1

144

was shown. Participants were presented with either the nutrition table with the label “fruit

145

sugar” for the ingredient sugar (see Table 1, left side) or the one with the unspecific term

146

“sugar” (see Table 1, right side). They were then asked to answer the following question

147

using a slider: “How healthy are these breakfast cereals in your opinion?” The possible

148

responses ranged from “not healthy at all” (0) to “very healthy” (100).

Study participants were randomly assigned to the “fruit sugar” condition or the “sugar”

150

condition.

151

2.2. Results and Discussion

Experiment 1 showed that participants who received the nutrition table with the label

M AN U

152

SC

149

RI PT

143

153

“sugar” assessed the cereal as less healthy compared with participants who received the

154

nutrition table with the label “fruit sugar.” The difference between the two groups was

155

significant, t(162) = 2.16, p = .032. Means and standard deviations are shown in Table 2. In studies examining the health halo effect, the information thought to influence

TE D

156

participants’ perception was in the foreground and stood out from the other information (e.g.,

158

a health claim or a brand); no further information about the food product was provided

159

(Gravel et al., 2012; Schuldt & Hannahan, 2013; Schuldt, Muller, & Schwarz, 2012). Table 1

160

shows that we used a very subtle manipulation, and provided the same concrete nutrition

161

information about the food product. Nevertheless, the symbolic information resulted in a

162

different perception of the product. Furthermore, in contrast to the studies on the health halo

163

effect, no specific product attribute (e.g., ingredient) was especially emphasized, only a

164

different labeling of the same attribute in the nutrition table was used.

165

3. Experiment 2

166

AC C

EP

157

In Experiment 1, we used a between-subjects design; participants evaluated either the

167

product with the nutrition table with the label “sugar” or the one with the label “fruit sugar.”

168

This result could also be explained by the evaluability hypothesis described by Hsee (1998). 7

ACCEPTED MANUSCRIPT For example, he showed that in a separate evaluation condition, in which the participants were

170

presented with the information about only one object of evaluation (in this case, a cup of ice

171

cream), the participants’ willingness to pay for the ice cream was influenced by the irrelevant

172

but easier to evaluate dimension of cup size. That is, they based their decisions on whether the

173

cup was overfilled (5 oz cup with 7 oz of ice cream) or underfilled (10 oz cup with 8 oz of ice

174

cream). In a joint evaluation condition, however, participants’ willingness to pay for the ice

175

cream was influenced by the relevant dimension, that is, the actual amount of ice cream. The

176

study results suggest that people generally use the dimension that is easiest to evaluate as a

177

criterion for making a decision. It is therefore possible that people do not have any idea

178

whether a given amount of sugar or calories is “a lot” when presented with a product. The

179

information “fruit sugar” may be easier to evaluate due to its symbolic meaning, and may be

180

interpreted as a signal that the product does not contain a lot of sugar or calories. Therefore,

181

the numerical information was probably not taken into account, or to a lesser extent. To rule

182

out this alternative explanation, we conducted an experiment with a within-subjects design in

183

which participants could compare and evaluate two breakfast cereals: one with the label

184

“sugar” and one with the label “fruit sugar.” This is a realistic scenario because consumers

185

often compare products. Therefore, examining whether the same results can be observed in a

186

joint evaluation (i.e., within-subjects design) and in separate evaluations (i.e., between-

187

subjects design) is important.

188

3.1. Methods

189

3.1.1. Participants

190

AC C

EP

TE D

M AN U

SC

RI PT

169

Experiment 2 involved 202 participants. Participants represented a convenience sample

191

of people living in the German-speaking part of Switzerland who agreed to participate in

192

online studies conducted by our group. The sample for Experiment 2 consisted of 95 (47%)

193

women and 107 (53%) men. The mean age was 54 years (SD = 15).

194

3.1.2. Materials and Procedure 8

ACCEPTED MANUSCRIPT 195

Participants received an e-mail asking them to fill out a short online questionnaire. In

196

addition to the experiment presented here, participants answered questions not related to the

197

research question of this study.

198

In Experiment 2, a within-subjects design was used; participants evaluated breakfast cereals with a nutrition table with the label “fruit sugar” (“fruit sugar” condition) as well as

200

cereals with a nutrition table with the label “sugar” (“sugar” condition). Participants received

201

the following information: “Miss Meier wants to buy breakfast cereals for her children. The

202

following information is displayed on the package.” After this introductory text, the

203

description of the first product with the respective nutrition information was shown, followed

204

by the second product with the respective nutrition information. The nutrition information

205

depicted in Table 1 was used for the two products (i.e., conditions). The two product

206

descriptions with the respective nutrition tables that differed only in the labeling of the

207

ingredient sugar (“fruit sugar” vs. “sugar”) were presented simultaneously on the same screen,

208

one above the other. This allowed for a direct comparison of the described attributes, and

209

ensured the evaluability of the information about the amount of sugar and calories, which was

210

identical in both nutrition tables.

SC

M AN U

TE D

To correct for possible order effects, for half of the participants, the nutrition table for

EP

211

RI PT

199

the product labeled with “fruit sugar” was presented above the nutrition table for the product

213

labeled with “sugar.” For the other half of the participants, the order was reversed. The

214

presentation order to which the participants were assigned was randomly determined. After

215

each product description, the participants were asked to answer the following question using a

216

slider: “How healthy are these breakfast cereals in your opinion?” The possible responses

217

ranged from “not healthy at all” (0) to “very healthy” (100).

218

3.2. Results and Discussion

219 220

AC C

212

All participants evaluated both products. The order of the product descriptions (i.e., nutrition tables) did not affect participants’ evaluations. Therefore, pooled data were 9

ACCEPTED MANUSCRIPT 221

analyzed. As shown in Table 2, participants perceived the cereals with the label “fruit sugar”

222

as healthier than the cereals with the label “sugar.” Results of a paired t-test show that

223

participants evaluated the two products significantly differently, t(201) = 5.28, p < .001.

224

Although we used a joint evaluation design in Experiment 2, allowing the participants to directly compare the description attributes and the (identical) information on the amount of

226

sugar and calories, the results of Experiment 1 were successfully replicated. Therefore, we

227

can rule out evaluability as an alternative explanation for our results. The same effect can be

228

observed in the case of separate (Experiment 1) and joint (Experiment 2) evaluations.

229

4. Experiment 3

SC

RI PT

225

Based on the information given to the participants in Experiments 1 and 2, we cannot

231

rule out that the participants in the “fruit sugar” condition assumed that the cereal contained

232

some fruit and, therefore, perceived them to be healthier. Consequently, the ingredient

233

“fruits,” not the ingredient “fruit sugar,” may have influenced participants’ responses. To rule

234

out this alternative explanation of the findings, in Experiment 3, a package of the cereals was

235

shown to inform participants that the cereals did not contain any fruits (see Figure 1).

TE D

M AN U

230

One goal of Experiment 3 was to replicate the findings of Experiments 1 and 2, this

237

time making it clear to the participants that the cereal did not contain any fruits. To test for the

238

robustness of the misleading effect of the symbolic meaning of the term “fruit sugar” on the

239

perceived healthiness of the products, we used somewhat different numerical information

240

about the nutrients of the cereals. A higher sugar content was indicated. Increasing the amount

241

of sugar could make this information more prevalent, and possibly shift participants’ focus

242

toward the indicated amount of sugar. This could result in greater consideration of the amount

243

of sugar when evaluating a product’s healthiness and, consequently, in better judgments.

244

4.1. Method

245

4.1.1. Participants

AC C

EP

236

10

ACCEPTED MANUSCRIPT 246

Experiment 3 involved 251 participants. Participants represented a convenience sample of people living in the German-speaking part of Switzerland who agreed to participate in

248

online studies conducted by our group. The sample consisted of 104 (41%) women and 150

249

(59%) men. The mean age was 58 years (SD = 13).

250

4.1.2. Materials and Procedure

251

RI PT

247

Participants received an e-mail asking them to fill out a short online questionnaire. In addition to the experiment presented here, participants answered questions not related to this

253

research question.

254

SC

252

In this experiment, after the introductory text, a corn flakes package was shown and, subsequently, the nutrition table was presented. Figure 1 shows the text and information

256

provided to the participants. The design and color of the corn flakes package did not have any

257

similarities to existing packages on the Swiss market. The picture on the package indicated

258

that the cereal did not contain any fruits. Furthermore, to make the information about the

259

amount of sugar more prevalent, in the nutrition tables a higher amount of sugar (double digit

260

number) was indicated. As in the two previous experiments, the nutrition information

261

presented in the “fruit sugar” and the “sugar” conditions was identical, except for the labeling

262

of the ingredient sugar.

TE D

EP

263

M AN U

255

After the product description, participants were asked to answer the following question using a slider: “How healthy are these breakfast cereals in your opinion?” The possible

265

responses ranged from “not healthy at all” (0) to “very healthy” (100).

266

AC C

264

Participants were randomly assigned to the “fruit sugar” (n = 127) or the “sugar”

267

condition (n = 124).

268

4.2. Results and Discussion

269

In line with the previous experiments, in the “fruit sugar” condition, the cereals were

270

perceived to be significantly healthier (M = 39.61, SD = 21.52), compared with the “sugar”

271

condition (M = 32.51, SD = 19.32), t(249) = 2.75, p = .006. 11

ACCEPTED MANUSCRIPT 272

Again, the same nutrition information was evaluated as less healthy when the label “sugar” was used in the nutrition table, compared with the label “fruit sugar.” In this

274

experiment, a picture of the package made it clear to the participants that the cereals did not

275

contain fruits. Therefore, we can rule out the alternative explanation that in the “fruit sugar”

276

condition, the cereals were perceived to be healthier because participants believed that the

277

cereals contained some fruit. Furthermore, pointing to the robustness of the effect, increasing

278

the amount of sugar to make this information more prevalent did not result in a greater

279

consideration of the amount of sugar. Consequently, the misleading effect of the symbolic

280

meaning of “fruit sugar” on the perceived healthiness of a product persisted, and was equally

281

pronounced.

282

5. Experiment 4

SC

M AN U

283

RI PT

273

To this point, we have provided strong evidence for the existence and robustness of the misleading effect of the label “fruit sugar” on the perceived healthiness of products. However,

285

that this misleading effect is in fact attributable to the symbolic meaning of healthiness

286

associated with the term “fruit,” respectively, the label “fruit sugar,” has yet to be

287

demonstrated. Consequently, a primary goal of Experiment 4 was to provide evidence for the

288

assumption that the healthiness evaluation of the cereals was influenced, that is, mediated by

289

the symbolic meaning of the information provided to the participants (i.e., the labeling of the

290

ingredient sugar).

EP

AC C

291

TE D

284

With regard to the underlying mechanisms, the question about possible enforcing

292

factors that could increase people’s susceptibility to the fallacy arises. Previous research

293

showed that health-conscious consumers and consumers with high self-control tend to engage

294

in more elaborate decision-making behavior, and are less subject to judgment biases that

295

could arise from a reliance on heuristic cues, when it comes to explicit beliefs and cognitively

296

shaped evaluations regarding health-related issues (Mai & Hoffmann, 2015). As a result, they

297

profit less from information presentation measures aimed at facilitating the evaluation of 12

ACCEPTED MANUSCRIPT concrete nutrition information, such as the amount of calories (Koenigstorfer, Groeppel-

299

Klein, & Kamm, 2014). Given these findings, it would be of interest to see whether this

300

moderating influence of health-consciousness holds true for people’s susceptibility to the

301

misleading effect of the “fruit sugar” labeling. Attempting to shed more light on the

302

mechanisms underlying the postulated misleading effect of the symbolic meaning of the

303

labels used for ingredients, Experiment 4 was additionally aimed to investigate whether more

304

health-conscious consumers are less prone to this fallacy than less health-conscious

305

consumers, or whether it is equally prevalent.

SC

306

RI PT

298

Summing up, the first goal of Experiment 4 was to provide more insights into the mechanisms underlying the misleading effect of the label “fruit sugar.” For this purpose, it

308

was aimed at more profoundly analyzing the differential perceptions associated with the terms

309

“fruit sugar” versus “sugar,” substantiating our claim that the effect is due to the symbolic

310

meaning assigned to the label “fruit sugar,” and identifying factors that exert a moderating

311

impact on the pronunciation of the fallacy.

TE D

312

M AN U

307

Front-of-package food claims are generally associated with favorable impressions about food products bearing such a claim, and they exert a substantial impact on consumers’

314

perceptions of a product’s healthiness (Roe et al., 1999). The presence of front-of-package

315

food claims can have a misleading effect on consumers’ product evaluations, and result in

316

suboptimal decision-making, because consumers tend to attribute inappropriate nutritional

317

characteristics and health benefits to products (Labiner-Wolfe, Jordan Lin, & Verrill, 2010).

318

Consumers with high product involvement were found to be less susceptible to the misleading

319

effect of product claims (Lee, Shimizu, Kniffin, & Wansink, 2013). Their higher motivation

320

to deal with this topic in depth leads to a more comprehensive and critical evaluation of the

321

information provided (Petty & Cacioppo, 1986). In light of these findings, a second goal of

322

Experiment 4 was to investigate whether the additional presence of a front-of-package claim

AC C

EP

313

13

ACCEPTED MANUSCRIPT 323

that the cereal contained “fruit sugar” would have a reinforcing effect on the fallacy of the

324

symbolic meaning, especially when it comes to consumers with low health consciousness.

325

When judging food products, criteria other than healthiness are also evaluated. The evaluation of these criteria may be derived based on different product characteristics, such as

327

package design, front-of-package claims and labels, as well as based on information on

328

nutritional components (cf. Orquin & Scholderer, 2015). For example, consumers

329

automatically draw inferences from claims about a food’s healthiness on the tastiness. They

330

tend to associate “unhealthy” foods with being tasty (Raghunathan, Naylor, & Hoyer, 2006).

331

Consequently, a final aim of Experiment 4 was to find out whether the fallacy related to the

332

symbolic meaning of the label “fruit sugar” also extends to the other evaluation criteria of

333

food, or whether it is specifically related to perceived healthiness.

334

5.1. Method

335

5.1.1. Participants

SC

M AN U

As in the previous experiments, the data were collected using a convenience sample of

TE D

336

RI PT

326

residents of the German-speaking part of Switzerland who agreed to take part in online

338

studies of our group on a regular basis. A total of 162 people participated in the study. The

339

sample consisted of 53 women (33%) and 109 (67%) men, with an average age of 59 years

340

(SD = 13).

341

5.1.2. Materials and Procedure

AC C

342

EP

337

An invitation email was sent to the participants, asking them to take part in a short

343

online study. In addition to the experiment presented here, participants also answered

344

questions not related to these research questions.

345

The experiment consisted of three conditions. For the “fruit sugar” and the “sugar”

346

conditions, scenario descriptions and information (including the picture of the package of

347

cereal) identical to Experiment 3 were used. The text and information provided in the third

348

condition, the “fruit sugar & claim” condition, were the same as in the “fruit sugar” condition, 14

ACCEPTED MANUSCRIPT except for the fact that in the “fruit sugar & claim” condition, on the corn flakes package, the

350

food claim “contains 100% fruit sugar” was depicted on the front. Figure 2 shows the picture

351

of the corn flakes package presented to the participants in the “fruit sugar & claim” condition.

352

Participants were randomly assigned to one out of the three experimental conditions.

353

Following the product description, participants were asked to indicate their perceived

354

healthiness of the corn flakes using a slider ranging from “not healthy at all” (0) to “very

355

healthy” (100). The question read as follows: “How healthy are these breakfast cereals in your

356

opinion?” Moreover, they were asked to indicate how they judged the cereals with regard to

357

naturalness, from “not natural at all” (0) to “very natural” (100); how they judged the

358

nutritional value of these cereals, from “very low” (0) to “very high” (100); and how they

359

would judged the taste of these cereals, from “very bad” (0) to “very good” (100).

SC

M AN U

360

RI PT

349

To provide evidence for the differential perceptions associated with the terms “fruit sugar” and “sugar” related to their symbolic meaning of healthiness, previous to the

362

experimental part with the breakfast cereals, participants were asked the following questions:

363

“How healthy is fruit sugar [sugar] in your opinion?” (“not healthy at all” [0] to “very

364

healthy” [100]); “How high do you judge the nutritional value of fruit sugar [sugar]? (“very

365

low” [0] to “very high” [100]); and “How strongly does the consumption of food products

366

containing fruit sugar [sugar] symbolize health consciousness in your opinion? (“very weak

367

symbol” [0] to “very strong symbol” [100]). The participants assigned to the conditions “fruit

368

sugar” or “fruit sugar & claim” answered these questions with regard to “fruit sugar”. In the

369

questions provided to the participants assigned to the condition “sugar,” the term “fruit sugar”

370

was replaced by “sugar.”

AC C

EP

TE D

361

371

Health consciousness, which was assumed to represent a moderator, was measured by

372

three items formulated in the form of statements: “I am interested in nutritional issues.”; “A

373

healthy diet is important to me.”; and “I am not interested in the topic of nutrition.”

374

Participants had to rate, on a 6-point scale ranging from “applies not at all” (1) to “completely 15

ACCEPTED MANUSCRIPT 375

applies” (6), how much the statement applied to them. The reliability of the health

376

consciousness measurement was satisfying, with a Cronbach’s alpha of .74.

377

To control for a possible social desirability bias that might occur, especially for topics that are strongly subject to social norms, such as health and environmental friendliness,

379

socially desirable response patterns were assessed. For this purpose, we utilized the short

380

form of the Marlowe-Crowne Social Desirability Scale M-C 1, proposed by Strahan and

381

Gerbasi (1972). This short form was shown to be most appropriate for online studies, for

382

which, due to time constraints, it is essential to draw on short forms of the scale that are of

383

sufficient psychometric quality (Vésteinsdóttir, Reips, Joinson, & Thorsdottir, 2015). The

384

internal consistency of the short social desirability scale was α = .56, which corresponds to the

385

reliability scores found in other studies (Strahan & Gerbasi, 1972; Vésteinsdóttir et al., 2015).

386

5.2. Results and Discussion

387

5.2.1. The misleading effect of the label “fruit sugar”

M AN U

SC

RI PT

378

The data were analyzed for each evaluation criterion (healthiness, naturalness,

389

nutritional value, taste) of the cereals separately using one-way analysis of variance

390

(ANOVA) with the factor experimental condition (“fruit sugar” vs. “sugar” vs. “fruit sugar &

391

claim”). Analyses revealed a significant effect of the factor experimental condition on the

392

perceived healthiness of the cereals, F(2, 159) = 3.91, p = .022. The cereals were judged

393

significantly healthier when the label “fruit sugar” compared to “sugar” was used. The same

394

was true for the condition with the additional “fruit sugar” claim. The conditions “fruit sugar”

395

and “fruit sugar & claim” did not significantly differ in the perceived healthiness of the

396

cereals. Detailed results are displayed in Table 3. The three experimental conditions did not

397

differ with regard to the other evaluation criteria: naturalness (F[2, 159] = 0.71, p = .493),

398

nutritional value (F[2, 159] = 2.31, p = .103), and taste (F[2, 159] = 0.18, p = .837). No

399

biasing influence of socially desirable response patterns was detected. 2

AC C

EP

TE D

388

16

ACCEPTED MANUSCRIPT 400

To test for a possible moderating effect of health consciousness, participants were grouped into low and high health consciousness using a median split. Subsequently, an

402

ANOVA with the factor experimental condition and the factor health consciousness group

403

(low vs. high health consciousness) was performed. The analysis yielded significant main

404

effects for the factors experimental condition, F(2, 154) = 3.61, p = .029, and health

405

consciousness group, F(1, 154) = 31.59, p < .001. However, there was no interaction effect,

406

F(2, 154) = 1.09, p = .337, indicating that the participants low and high in health

407

consciousness were equally susceptible to the misleading effect of the label “fruit sugar.” The

408

two health consciousness groups only differed in their healthiness judgments in general. The

409

health conscious participants over all of the experimental conditions judged the cereal’s

410

healthiness to be lower. 3

SC

M AN U

411

RI PT

401

According to the findings, the misleading effect of the symbolic meaning of the label “fruit sugar” seems to appear specifically when it comes to judgments related to the product’s

413

healthiness. Additionally, adding an explicit front-of-package claim that the product contains

414

“fruit sugar” did not significantly reinforce the biasing effect. No matter whether the symbolic

415

information is supposedly unobtrusively displayed in the nutrition table, or prominently

416

positioned on the front of the package, once the symbolic information has been read, it exerts

417

an equally misleading effect on the perceived healthiness. Furthermore, health consciousness

418

was not found to have a moderating influence on the misleading effect of the labeling. Health-

419

conscious consumers are as equally susceptible to the misleading effect as consumers who are

420

less interested in health-related issues. Consequently, when confronted with a labeling of

421

ingredients that has a strong symbolic meaning, a more profound evaluation of the depicted

422

nutritional information (as occurs with health conscious consumers) does not guarantee an

423

accurate judgment of the product’s healthiness. Also, in this case, the misleading effect of the

424

symbolic meaning of the used labeling still prevails, and reduces judgment accuracy.

425

5.2.2. The mediating role of the perceived healthiness associated with symbolic information

AC C

EP

TE D

412

17

ACCEPTED MANUSCRIPT 426

In line with the assumption that the term “fruit sugar” symbolizes healthiness due to the use of the word “fruit,” which finally results in an increase in the ingredient’s perceived

428

healthiness and a more favorable judgment of other health-related aspects, significant

429

differences in the perceptions associated with the terms “fruit sugar” and “sugar” emerged

430

(see Table 4). Compared to the general term “sugar,” “fruit sugar” was perceived to be

431

healthier, to have a higher nutritional value, and the consumption of products containing “fruit

432

sugar” was considered to be a stronger symbol for healthiness.

RI PT

427

We postulated that the perceived healthiness of cereal was driven by the symbolic

434

meaning of the provided information, that is, the symbolic meaning of healthiness associated

435

with the labeling of the ingredient sugar (“fruit sugar” vs. “sugar”). Based on this assumption,

436

we expected that the impact of the different labeling (i.e., the provision of information with

437

different symbolic meanings) on the perceived healthiness of the cereal would be mediated by

438

the perceived healthiness associated with the labeling, as a result of its symbolic meaning.

439

This simple mediation effect was tested using the method proposed by Preacher and Hayes

440

(2008). The cereals in the experimental conditions “fruit sugar” and “fruit sugar & claim”

441

were described using the label “fruit sugar.” Since these two conditions did not significantly

442

differ in the perceived healthiness of the cereals, the data from these conditions were pooled

443

for the mediation analysis. The variable “labeling of the ingredient sugar” was dummy coded

444

(0 = sugar and 1 = fruit sugar). The results of the mediation analysis are presented in Figure 2.

445

Results show that the labeling of the ingredient sugar (i.e., the symbolic information) had a

446

significant impact on the healthiness associated with the ingredient, which, in turn,

447

significantly influenced the perceived healthiness of the cereal. Controlling for the healthiness

448

associated with the ingredient sugar due to its labeling resulted in a non-significant impact of

449

the labeling on the perceived healthiness of the cereal. Therefore, a full mediation could be

450

observed. A bias-corrected 95% bootstrap confidence interval for the indirect effect based on

451

1,000 bootstrap samples was clearly above zero (3.90 to 12.47).

AC C

EP

TE D

M AN U

SC

433

18

ACCEPTED MANUSCRIPT 452

Results suggest that the misleading effect of the labeling on the product’s perceived healthiness can be explained by the difference in the perceived healthiness of the ingredient

454

sugar, when labeled “fruit sugar” compared to “sugar” (i.e., umbrella term), which is due to

455

the associated symbolic meaning of healthiness. The mediation analysis revealed that the

456

labeling of the ingredient sugar led to a different perception of the healthiness of the

457

ingredient sugar, and that the participants evaluated the healthiness of the breakfast cereals

458

based on the perceived healthiness associated with the labeling of the ingredient sugar.

459

6. General Discussion

SC

RI PT

453

The aim of the present research was to provide evidence that people rely on symbolic

461

information when evaluating food properties. More precisely, it aimed to examine whether

462

presenting consumers with the information that a product contains “fruit sugar” affects

463

product evaluation due to the symbolic nature of the term “fruit.” Results of the present study

464

indicate that the phrase “fruit sugar” listed as an ingredient of breakfast cereals results in a

465

more positive perception of the cereals’ healthiness. The mediation analysis conducted in

466

Experiment 4 revealed that the labeling of the ingredient sugar as “fruit sugar” leads to a

467

higher perceived healthiness of the ingredient, compared with the labeling as “sugar,” and that

468

the healthiness associated with the ingredient as a result of the labeling mediates the perceived

469

healthiness of the food product. The findings impressively show that the healthiness of the

470

very same product, with the very same nutrition information, is perceived differently, simply

471

by changing the labeling of the ingredient sugar. The labeling of the ingredients by making

472

use of symbolic information may, consequently, exert a misleading effect on a consumer’s

473

assessment of the product’s healthiness. The findings suggest that the effect is quite robust. A

474

more profound and comprehensive evaluation of the provided information (as occurs with

475

people with pronounced health consciousness) does not protect against the misleading effect

476

of symbolic information, and does not add to judgment accuracy. This indicates that relying

477

and drawing on the symbolic meaning of information is, to a certain extent, an automatic and

AC C

EP

TE D

M AN U

460

19

ACCEPTED MANUSCRIPT 478

implicit process that cannot easily be correct by increasing people’s health consciousness.

479

Furthermore, evidence was provided that the perception of a subtle cue is sufficient to trigger

480

reliance on symbolic information, when evaluating the healthiness of food products. Why can we claim that the decisions in our experiments were biased? In general, it can

482

be stated that the term “sugar” represents an umbrella term that subsumes the different types

483

of sugar, such as fructose (“fruit sugar”). Bearing this in mind, no difference between the

484

ingredient sugar labeled as “fruit sugar” and labeled with the umbrella term “sugar” should be

485

observed. However, consumers could automatically associate the term “sugar” with sucrose,

486

commonly named “table sugar,” which is a combination of fructose and glucose. Therefore,

487

the question arises whether associating “fruit sugar” with higher healthiness than sucrose is in

488

fact inaccurate, and results in biased judgments. It has been shown that high fructose

489

consumption can induce, among other things, insulin resistance (Rizkalla, 2010), and that it is

490

a driver for pathological cardiac growth and dysfunction (Mirtschink et al., 2015). However,

491

in most of these studies, unrealistically high doses of fructose were used. In a comprehensive

492

review of the literature, no direct evidence was found for more negative effects of high

493

fructose compared with sucrose (Tappy & Le, 2010). Based on the scientific evidence, it can

494

be concluded that it does not matter whether the breakfast cereals contain added fructose or

495

sucrose (Rizkalla, 2010; Tappy & Le, 2010). There is no scientific justification for assessing

496

the cereals containing “fruit sugar” as healthier than the cereals containing sugar. On the

497

contrary, a recent study provided evidence that the brain and behavioral response to fructose,

498

relative to glucose, may promote feeding behavior. It was shown that the ingestion of fructose

499

relative to glucose resulted in a greater brain reactivity to food cues and, in line with this, led

500

to greater hunger and desire for food, as well as greater willingness to give up long-term

501

monetary rewards to obtain immediate high-caloric food rewards (Luo, Monterosso,

502

Sarpelleh, & Page, 2015). It is safe to conclude, therefore, that we have observed biased

AC C

EP

TE D

M AN U

SC

RI PT

481

20

ACCEPTED MANUSCRIPT 503

responses in our experiments, because participants were influenced by the symbolic

504

significance of information that is not relevant to the evaluation task.

505

In the present experimental design, we made sure that the participants had identical information about the two products, except for the manipulated variable (i.e., the labeling of

507

the ingredient sugar). The goal of our experiment was to test whether people perceive “fruit

508

sugar” as healthier than sugar and not whether people assume that the nutrition profile (e.g.,

509

the number of calories or amount of fat) of a product containing “fruit sugar” is different from

510

that of a product containing sugar. Therefore, we used the identical realistic nutrition

511

information for the products and presented it in the form of a nutrition table. In nutrition

512

tables commonly used in Switzerland and other European countries, the unspecific term

513

“sugar,” not “fruit sugar,” is used. However, claims on food packages or in advertisements

514

may emphasize that the product is sweetened with “fruit sugar”. Had we only used the full

515

product packaging and the “fruit sugar” claim, the participants could have concluded that the

516

breakfast cereals labeled as containing “fruit sugar” had a different nutritional profile than the

517

breakfast cereals labeled as containing “sugar.”

SC

M AN U

TE D

518

RI PT

506

In the present study, participants evaluated breakfast cereals. 4 This product category was chosen because in Switzerland, breakfast cereal manufacturers often emphasize that their

520

product does not contain additional sugar but is sweetened by honey and maple syrup or that

521

the product contains only “fruit sugar”. In various countries, soft drinks are sweetened with

522

high-fructose corn syrup. Sweetened soft drinks are most likely not perceived as healthy

523

products, which is why we used the more neutral food category of breakfast cereals. Future

524

studies could examine whether similar effects can also be observed for more hedonic products

525

such as soft drinks.

526

AC C

EP

519

The present study is not the first to suggest that symbolic information influences the

527

perception of food (Chandon & Wansink, 2007; Gravel et al., 2012; Lähteenmäki et al., 2010;

528

Schuldt & Hannahan, 2013; Schuldt et al., 2012; Schuldt & Schwarz, 2010; Wansink & 21

ACCEPTED MANUSCRIPT Chandon, 2006). However, our research differs from the existing studies in a number of

530

aspects. First, our manipulation was more subtle than the interventions in most past studies.

531

We did not make an explicit claim regarding the food products directly targeted at pushing

532

consumers’ product evaluation in a certain direction (Chandon & Wansink, 2007; Gravel et

533

al., 2012; Lähteenmäki et al., 2010; Schuldt & Hannahan, 2013; Schuldt et al., 2012; Schuldt

534

& Schwarz, 2010). In our experiments, we provided identical detailed information in the

535

nutrition tables with the same ingredients. Only a different labeling of the ingredient sugar

536

was used, which was effectuated by merely changing one word in the nutrition table.

537

Nevertheless, our manipulation had an impact. This further shows the powerful impact of

538

symbolic information on the perception of food. Second, in some past studies, the information

539

provided was ambiguous to participants (Gravel et al., 2012). It is very difficult to estimate

540

the calories of food without receiving concrete information about its ingredients. In the

541

present study, participants were fully informed about the nutrition profile of the food. Third,

542

past studies did not examine possible mechanisms of the observed effect. In Experiment 4, we

543

tested the perceived healthiness associated with ingredients’ labeling as a possible mechanism

544

underlying the effect of presenting symbolic information on the perception of food. Our

545

results point to the misleading effect of symbolic information, showing that simply by using

546

the symbolically positive label “fruit sugar,” consumers are misled to perceive the ingredient

547

sugar as healthier and, in the end, this results in the perception of the cereals as healthier.

548

Fourth, to the best of our knowledge, despite all the studies that have shown the possible

549

negative effects of high fructose intake (Luo et al., 2015; Tappy & Le, 2010), this is the first

550

study to suggest that the term “fruit sugar” is still perceived in a more positive way compared

551

with the unspecific term “sugar.”

552

AC C

EP

TE D

M AN U

SC

RI PT

529

Several limitations of the present study must be addressed. In all three experiments,

553

convenience samples were used. It is, therefore, unclear to what extent the present findings

554

can be generalized. Furthermore, we focused on the effect of symbolic information on 22

ACCEPTED MANUSCRIPT information processing and the resulting perceived healthiness of products, and the perception

556

of other evaluation criteria (e.g., taste). Several response types can be used to evaluate the

557

impact of symbolic information. In the present study, we examined the effect of the label

558

“fruit sugar” on the perceived healthiness. Another type of response that could be assessed is

559

people’s willingness to buy the product utilizing real food packages. We did not choose this

560

second approach, because some participants may not buy cereal, and the perceived healthiness

561

may be only one factor that influences people’s willingness to buy food products. Given these

562

additional confounding variables, much larger samples would be needed to show the expected

563

effects. Therefore, the focus of the present study was on the perceived healthiness, and not on

564

the purchase decision. In our experiments, we provided evidence for the misleading effect of

565

the labeling of nutrition information. Of course, paying attention to the information presented

566

in the nutrition table does not guarantee for a healthier product choice. However, it is obvious

567

that consumers, first of all, need to correctly interpret and process the nutrition information,

568

before they can use it to make more healthy decisions. Therefore, insights into the processing

569

of nutrition facts and potentially biasing factors are crucial for producers and policy makers,

570

when it comes to the provision of health-related information.

SC

M AN U

TE D

Future research should examine how people judge the healthiness of food in more

EP

571

RI PT

555

detail, and whether there exists other symbolic information that may exert an (misleading)

573

influence on a consumer’s perception of a product’s healthiness. Furthermore, exploring

574

whether the label “fruit sugar” also changes other product expectations (e.g., product quality,

575

environmental friendliness), or the perception of other nutritional values provided on the

576

package (e.g., fat) would be worthwhile. In the present research, we focused on “fruit sugar”;

577

however, examining how other sugars (e.g., honey) are perceived compared to “fruit sugar”

578

using an experimental approach would also be interesting. As mentioned above, the presented

579

experiments investigated the biasing effect of symbolic information on the perceived

580

healthiness of food products, that is, on the processing of nutrition information. Future

AC C

572

23

ACCEPTED MANUSCRIPT 581

research might wish to expand the focus to the analysis of the impact of symbolic information

582

on the final food choice.

583

In sum, results of the present study suggest that consumers perceive a product labeled as containing “fruit sugar” as healthier than one labeled with the unspecific term “sugar.” We

585

can rule out that the label “fruit sugar” made participants believe that the product has fewer

586

calories, less fat, or less sugar compared to the other product, because we provided identical

587

nutrition information for the products. Moreover, in the joint evaluation condition (i.e.,

588

within-subjects design in Experiment 2), it was obvious to participants that the nutrition

589

profiles of the two products were identical. Therefore, our results are in line with our

590

hypothesis that symbolic information is an important factor in people’s evaluation of how

591

healthy a food product is, and may lead to biased judgments. People perceive fruits as healthy

592

food (Bucher, van der Horst, & Siegrist, 2013); therefore, since the word “fruit” symbolizes

593

healthiness, the term “fruit sugar” is associated with higher healthiness. As a result, the

594

ingredient “fruit sugar” and, consequently, the food product are perceived to be relatively

595

healthy.

SC

M AN U

TE D

596

RI PT

584

The implications of these findings for marketing and public health are clear. From a marketing point of view, the emphasis on natural ingredients and ingredients with a positive

598

symbolic meaning may result in a better health perception of food. Not only direct health

599

claims (Lähteenmäki, 2013), but also ingredients may automatically evoke positive

600

associations that have an important impact on a consumer’s information processing and

601

decision-making. From a public health perspective, the challenge is how to educate the public

602

so that consumers are not influenced by misleading symbolic product information, when they

603

evaluate the healthiness of food products. Our findings suggest that consumers are highly

604

susceptible to the symbolic information that food marketers may specifically use in an attempt

605

of greenwashing or nutriwashing to promote their products. Such information may bias

606

consumers’ information processing and product evaluation, and finally, result in suboptimal

AC C

EP

597

24

ACCEPTED MANUSCRIPT purchase decisions. This is especially concerning, given that subtle symbolic cues are already

608

sufficient to induce biased product evaluation, and given that even high involvement

609

consumers, who generally engage in more profound information processing, are susceptible to

610

this fallacy.

AC C

EP

TE D

M AN U

SC

RI PT

607

25

ACCEPTED MANUSCRIPT 611

References

612

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

614

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

615

10.1080/00913367.2000.10673615

616

RI PT

613

Brierley, M., & Elliott, C. (2015). Nutritional components and children's interpretations of packaged food. International Journal of Health Promotion and Education, 1-14. doi:

618

10.1080/14635240.2015.1010654

619

SC

617

Briz, T., Sijtsema, S. J., Jasiulewicz, A., Kyriakidi, A., Dolors Guàrdia, M., van den Berg, I., & van der Lans, I. A. (2008). Barriers to fruit consumption: Driving forces behind

621

consumer behaviour. Scripta Horticulturae, 8, 7-18.

623 624

Bucher, T., van der Horst, K., & Siegrist, M. (2013). Fruit for dessert. How people compose healthier meals. Appetite, 60(0), 74-80. doi: 10.1016/j.appet.2012.10.003 Chandon, P., & Wansink, B. (2007). The biasing health halos of fast-food restaurant health

TE D

622

M AN U

620

625

claims: Lower calorie estimates and higher side-dish consumption intentions. Journal

626

of Consumer Research, 34(3), 301-314. doi: 10.1086/519499 Dickson-Spillmann, M., Siegrist, M., & Keller, C. (2011). Development and validation of a

EP

627

short, consumer-oriented nutrition knowledge questionnaire. Appetite, 56(3), 617-620.

629

doi: http://dx.doi.org/10.1016/j.appet.2011.01.034

630 631 632 633

AC C

628

Evans, G., de Challemaison, B., & Cox, D. N. (2010). Consumers’ ratings of the natural and unnatural qualities of foods. Appetite, 54(3), 557-563. doi: http://dx.doi.org/10.1016/j.appet.2010.02.014 Gravel, K., Doucet, É., Peter Herman, C., Pomerleau, S., Bourlaud, A.-S., & Provencher, V.

634

(2012). “Healthy,” “diet,” or “hedonic”. How nutrition claims affect food-related

635

perceptions and intake? Appetite, 59(3), 877-884. doi:

636

http://dx.doi.org/10.1016/j.appet.2012.08.028 26

ACCEPTED MANUSCRIPT 637 638 639

Hsee, C. K. (1998). Less is better: When low-value options are valued more highly than highvalue options. Journal of Behavioral Decision Making, 11(2), 107-121. Kahneman, D., & Frederick, S. (2005). A model of heuristic judgment. In K. J. Holyoak & R. G. Morrison (Eds.), The Cambridge handbook of thinking and reasoning (pp. 267-

641

293). Cambridge, UK: Cambridge University Press.

642

RI PT

640

Koenigstorfer, J., Groeppel-Klein, A., & Kamm, F. (2014). Healthful food decision making in response to traffic light color-coded nutrition labeling. Journal of Public Policy &

644

Marketing, 33(1), 65-77. doi: http://journals.ama.org/doi/abs/10.1509/jppm.12.091

645

SC

643

Labiner-Wolfe, J., Jordan Lin, C.-T., & Verrill, L. (2010). Effect of low-carbohydrate claims on consumer perceptions about food products' healthfulness and helpfulness for

647

weight management. Journal of Nutrition Education and Behavior, 42(5), 315-320.

649 650

Lähteenmäki, L. (2013). Claiming health in food products. Food Quality and Preference, 27(2), 196-201. doi: http://dx.doi.org/10.1016/j.foodqual.2012.03.006 Lähteenmäki, L., Lampila, P., Grunert, K., Boztug, Y., Ueland, Ø., Åström, A., &

TE D

648

M AN U

646

Martinsdóttir, E. (2010). Impact of health-related claims on the perception of other

652

product attributes. Food Policy, 35(3), 230-239. doi:

653

http://dx.doi.org/10.1016/j.foodpol.2009.12.007

655 656 657

Lee, W.-c. J., Shimizu, M., Kniffin, K. M., & Wansink, B. (2013). You taste what you see: Do organic labels bias taste perceptions? Food Quality and Preference, 29(1), 33-39.

AC C

654

EP

651

doi: http://dx.doi.org/10.1016/j.foodqual.2013.01.010

Liem, D. G., Toraman Aydin, N., & Zandstra, E. H. (2012). Effects of health labels on

658

expected and actual taste perception of soup. Food Quality and Preference, 25(2),

659

192-197. doi: http://dx.doi.org/10.1016/j.foodqual.2012.02.015

660 661

Luo, S., Monterosso, J. R., Sarpelleh, K., & Page, K. A. (2015). Differential effects of fructose versus glucose on brain and appetitive responses to food cues and decisions

27

ACCEPTED MANUSCRIPT 662

for food rewards. Proceedings of the National Academy of Sciences, 112(20), 6509-

663

6514. doi: 10.1073/pnas.1503358112

664

Mai, R., & Hoffmann, S. (2015). How to combat the unhealthy = tasty intuition: The influencing role of health consciousness. Journal of Public Policy & Marketing, 34(1),

666

63-83. doi: http://dx.doi.org/10.1509/jppm.14.006

667

RI PT

665

Mirtschink, P., Krishnan, J., Grimm, F., Sarre, A., Horl, M., Kayikci, M., . . . Krek, W.

(2015). HIF-driven SF3B1 induces KHK-C to enforce fructolysis and heart disease.

669

Nature, advance online publication. doi: 10.1038/nature14508

670

SC

668

Orquin, J. L., & Scholderer, J. (2015). Consumer judgments of explicit and implied health claims on foods: Misguided but not misled. Food Policy, 51(0), 144-157. doi:

672

http://dx.doi.org/10.1016/j.foodpol.2015.01.001

674 675

Parmenter, K., & Wardle, J. (1999). Development of a general nutrition knowledge questionnaire for adults. European Journal of Clinical Nutrition, 53(4), 298-308. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L.

TE D

673

M AN U

671

676

Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 19, pp. 123-205).

677

New York: Academic Press.

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing

EP

678

and comparing indirect effects in multiple mediator models. Behavior Research

680

Methods, 40(3), 879-891. doi: 10.3758/BRM.40.3.879

681 682

AC C

679

Raghunathan, R., Naylor, R. W., & Hoyer, W. D. (2006). The unhealthy = tasty intuition and its effects on taste inferences, enjoyment, and choice of food products. Journal of

683

Marketing, 70(4), 170-184. doi:

684

http://journals.ama.org/doi/abs/10.1509/jmkg.70.4.170

685 686

Rizkalla, S. W. (2010). Health implications of fructose consumption: A review of recent data. Nutrition and Metabolism, 7, 82. doi: 10.1186/1743-7075-7-82

28

ACCEPTED MANUSCRIPT 687

Roe, B., Levy, A. S., & Derby, B. M. (1999). The impact of health claims on consumer search

688

and product evaluation outcomes: Results from FDA experimental data. Journal of

689

Public Policy & Marketing, 18(1), 89-105. doi: 10.2307/30000511

690

Rozin, P. (2005). The meaning of “natural”: Process more important than content.

692

Psychological Science, 16(8), 652-658. doi: 10.1111/j.1467-9280.2005.01589.x

RI PT

691

Schuldt, J. P., & Hannahan, M. (2013). When good deeds leave a bad taste. Negative inferences from ethical food claims. Appetite, 62(0), 76-83. doi:

694

http://dx.doi.org/10.1016/j.appet.2012.11.004

695

SC

693

Schuldt, J. P., Muller, D., & Schwarz, N. (2012). The “fair trade” effect: Health halos from social ethics claims. Social Psychological and Personality Science, 3(5), 581-589. doi:

697

10.1177/1948550611431643

698

M AN U

696

Schuldt, J. P., & Schwarz, N. (2010). The "organic" path to obesity? Organic claims influence calorie judgments and exercise recommendations. Judgment and Decision Making,

700

5(3), 144-150.

701 702

TE D

699

Siegrist, M., & Sütterlin, B. (2014). Human and nature-caused hazards: The affect heuristic causes biased decisions. Risk Analysis, 34(8), 1482-1494. doi: 10.1111/risa.12179 Strahan, R., & Gerbasi, K. C. (1972). Short, homogeneous versions of the Marlowe-Crowne

704

Social Desirability Scale. Journal of Clinical Psychology, 28(2), 191-193. doi:

705

10.1002/1097-4679(197204)28:2<191::AID-JCLP2270280220>3.0.CO;2-G

707 708 709 710 711 712

AC C

706

EP

703

Sütterlin, B., & Siegrist, M. (2014). The reliance on symbolically significant behavioral attributes when judging energy consumption behaviors. Journal of Environmental Psychology, 40(0), 259-272. doi: http://dx.doi.org/10.1016/j.jenvp.2014.07.005 Tappy, L., & Le, K. A. (2010). Metabolic effects of fructose and the worldwide increase in obesity. Physiological Reviews, 90(1), 23-46. doi: 10.1152/physrev.00019.2009 Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131. doi: 10.1126/science.185.4157.1124 29

ACCEPTED MANUSCRIPT 713

Vésteinsdóttir, V., Reips, U.-D., Joinson, A., & Thorsdottir, F. (2015). Psychometric

714

properties of measurements obtained with the Marlowe–Crowne Social Desirability

715

Scale in an Icelandic probability based Internet sample. Computers in Human

716

Behavior, 49(0), 608-614. doi: http://dx.doi.org/10.1016/j.chb.2015.03.044

718

Wansink, B., & Chandon, P. (2006). Can “low-fat” nutrition labels lead to obesity? Journal of

RI PT

717

Marketing Research, 43(4), 605-617. doi: 10.1509/jmkr.43.4.605

AC C

EP

TE D

M AN U

SC

719

30

ACCEPTED MANUSCRIPT 720

Footnotes

721

1

722

is used. In German, however, the word “Fruchtzucker,” which translated into English means

723

“fruit sugar,” is commonly used. This word has much closer associations with fruits and

724

naturalness compared with the word “fructose.” Therefore, we use the term “fruit sugar,” the

725

literal translation of the German word “Fruchtzucker,” in this paper.

726

2

727

possible social desirability effects, revealed no significant results for the covariate social

728

desirability, F(1, 154) = 0.38, p = .538. Consequently, the findings on the perceived

729

healthiness were not subject to a social desirability bias. The same was true for the perceived

730

naturalness (p = .957), estimated nutritional value (p = .856), and estimated taste (p = .221).

731

3

Regression analyses resulted in the same conclusion.

732

4

The German word “Frühstücksflocken” was used. This term refers to breakfast cereals and

733

muesli.

RI PT

EP

TE D

M AN U

SC

An analysis of variance, with participants’ social desirability score as a covariate to test for

AC C

734

We are aware that in the English language, the word “fructose,” not the phrase “fruit sugar,”

31

ACCEPTED MANUSCRIPT 735

Table 1

736

Nutrition information provided in the “fruit sugar” (nutrition table on the left) and “sugar”

737

(nutrition table on the right) conditions in Experiments 1 and 2.

738 Nutrition value for a portion (30 g)

Nutrition value for a portion (30 g)

Energy

Energy

514.5 kJ

514.5 kJ

3.6 g

Carbohydrates

Protein

21.6 g

- Fruit sugar - Starch Fat

Carbohydrates

7.8 g

- Sugar

13.8 g

- Starch

2.1 g

3.6 g

21.6 g 7.8 g

13.8 g 2.1 g

AC C

EP

TE D

M AN U

739

Fat

SC

Protein

121.8 kcal

RI PT

121.8 kcal

32

ACCEPTED MANUSCRIPT 740

Table 2

741

Perceived healthiness of breakfast cereals using the label “fruit sugar” vs. “sugar” in the

742

nutrition table. Sugar

M (SD) [N]

M (SD) [N]

48.71 (23.82) [73]

40.16 (26.22) [91]

52.01 (20.37) [202]

45.40 (21.21) [202]

(between-subjects design) Experiment 2 (within-subjects design)

RI PT

Experiment 1

Fruit sugar

Note. The ratings of perceived healthiness could range from 0 (not healthy at all) to 100 (very

744

healthy).

AC C

EP

TE D

M AN U

SC

743

33

ACCEPTED MANUSCRIPT 745

Table 3

746

Perceived healthiness, perceived naturalness, estimated nutritional value, and estimated taste

747

of breakfast cereals in the conditions “fruit sugar” vs. “sugar” vs. “fruit sugar & claim” of

748

Experiment 4. Sugar

Fruit sugar & claim

M (SD) [N]

M (SD) [N]

M (SD) [N]

Perceived healthiness

39.29a (21.50) [52]

29.25b (20.07) [55]

38.58a (21.07) [55]

Perceived naturalness

32.60a (18.03) [52]

30.82a (23.47) [55]

35.60a (21.75) [55]

Estimated nutritional value 48.15ab (21.57) [52] 41.78b (27.29) [55]

51.93a (25.68) [55]

50.52a (21.76) [52]

50.64a (27.96) [55]

52.93a (20.64) [55]

SC

Estimated taste

RI PT

Fruit sugar

Note. Different letters indicate significant differences between particular experimental

750

conditions, p < 0.05, using planned comparisons. The ratings of perceived healthiness could

751

range from 0 (not healthy at all) to 100 (very healthy), ratings of perceived naturalness from 0

752

(not natural at all) to 100 (very natural), ratings of estimated nutritional value from 0 (very

753

low) to 100 (very high), and ratings of estimated taste from 0 (very bad) to 100 (very good).

AC C

EP

TE D

M AN U

749

34

ACCEPTED MANUSCRIPT 754

Table 4

755

Health-related perceptions associated with the labels “fruit sugar” vs. “sugar” in Experiment

756

4. Fruit sugar

Sugar

t-test result

M (SD) [N]

M (SD) [N]

53.63 (19.24) [107] 32.55 (16.68) [55]

t(160) = 6.90, p < .001

Estimated nutritional value

56.35 (21.88) [107] 49.47 (28.78) [55]

t(160) = 1.70, p = .046 a

Symbolic meaning of healthiness 47.33 (23.40) [107] 40.04 (29.44) [55]

t(160) = 1.72, p = .044 a

RI PT

Perceived healthiness

associated with consumption

Note. a One-tailed p-value is indicated. Participants in the “fruit sugar” and the “fruit sugar &

758

claim” conditions answered the questions related to “fruit sugar,” while participants assigned

759

to the “sugar” condition answered the questions related to “sugar.” The ratings of perceived

760

healthiness could range from 0 (not healthy at all) to 100 (very healthy), ratings of estimated

761

nutritional value from 0 (very low) to 100 (very high), and ratings of the symbolic meaning of

762

healthiness associated with consumption from 0 (very weak symbol) to 100 (very strong

763

symbol).

AC C

EP

TE D

M AN U

SC

757

35

ACCEPTED MANUSCRIPT 764

Figure 1

765

Information provided in the “fruit sugar” condition in Experiment 3. In the “sugar” condition,

766

the same information was provided, but the label “fruit sugar” was replaced by the label

767

“sugar.”

768

770

Mrs. Meier wants to buy breakfast cereals for her children.

M AN U

SC

771

RI PT

769

772 773

Mrs. Meier finds the following information on the package:

775

TE D

774

Nutrition value for a portion (30 g) Energy

111.3 kcal

Protein

1.4 g

26.1 g

- Fruit sugar

11.1 g

- Starch

15.0 g

Fat 776

AC C

Carbohydrates

EP

466.0 kJ

0.2 g

36

ACCEPTED MANUSCRIPT 777

Figure 2

778

Picture of the corn flakes package with the claim “contains 100% fruit sugar” (in German:

779

“enthält 100% Fruchtzucker”) on the front, as presented to the participants in the experimental

780

condition “fruit sugar & claim” of Experiment 4.

M AN U

SC

RI PT

781

AC C

EP

TE D

782

37

ACCEPTED MANUSCRIPT Figure 3

784

Results of the mediation analysis for the impact of the labeling of the ingredient sugar (“fruit

785

sugar” vs. “sugar”) on the perceived healthiness of cereals are shown. Direct effects are

786

presented in (a), and the model with the mediator variable “healthiness associated with

787

labeling” is shown in (b). Nonstandardized coefficients (and standard errors in parentheses)

788

are presented. Non-significant paths are depicted as dotted lines. * p < .05, ** p < .01, *** p <

789

.001.

RI PT

783

AC C

791

EP

TE D

M AN U

SC

790

38

ACCEPTED MANUSCRIPT Research Highlights • Symbolic information influences people’s health perceptions of foods • Using the label “fruit sugar” instead of “sugar” increases perceived healthiness of foods

RI PT

• Perceived healthiness associated with ingredients’ labeling affects health perception of foods

AC C

EP

TE D

M AN U

SC

• Health consciousness does not reduce the biasing effect of symbolic information