Research BriefThe Effect of Food Label Cues on Perceptions of Qualityand Purchase Intentions among High-InvolvementConsumers with Varying Levels of Nutrition KnowledgeAmber Walters, MS; Marilee Long, PhD

DepartmenCollins, COAddress foCommunicPhone: (97�2012 SOdoi:10.1016

350

ABSTRACT

Objective: To determine whether differences in nutrition knowledge affected how women (a high-involvement group) interpreted intrinsic cues (ingredient list) and extrinsic cues (‘‘all natural’’ label) onfood labels.Methods: A 2 (intrinsic cue) � 2 (extrinsic cue) � 2 (nutrition knowledge expert vs novice) within-subject factorial design was used. Participants were 106 female college students (61 experts, 45 novices).Dependent variables were perception of product quality and purchase intention.Results: As predicted by the elaboration likelihood model, experts used central route processing toscrutinize intrinsic cues and make judgments about food products. Novices used peripheral route process-ing to make simple inferences about the extrinsic cues in labels.Conclusions and Implications: Consumers’ levels of nutrition knowledge influenced their ability toprocess food labels. The United States Food and Drug Administration should regulate the ‘‘all natural’’food label, because this claim is likely to mislead most consumers.KeyWords: food labeling, nutrition, persuasion, mental processes (J Nutr Educ Behav. 2012;44:350-354.)

INTRODUCTION

Concerns about the veracity ofnutrition claims in food labeling andconsumers' ability to evaluate theseclaims have become more wide-spread.1-3 As the number and typesof nutrition claims increase, researchinto how consumers process theseclaims becomes more important. Thecurrent study investigates howcontradictory nutrition claims onfood labels influence perceptions offood product quality and purchaseintentions among consumers withdifferent levels of nutritionknowledge.

Research indicates that ability toprocess information is influenced byknowledge and the extent to whichthat knowledge can be retrieved.4

Experts have more knowledge thannovices and are better able to deploythat knowledge to understand mes-sages.5 For example, research suggests

t of Journalism and Technical Comm

r correspondence: Marilee Long, PhDation, C216 Clark Bldg, Colorado0) 491-6463; Fax: (970) 491-2908; E-CIETY FOR NUTRITION EDUC/j.jneb.2011.08.008

consumers with more nutritionknowledge are more likely to uselabel information related to fat,calories, and ingredients.6,7 Otherstudies have found that greaternutrition knowledge makes claims of‘‘healthfulness’’ more salient inassessments of food products.8

The elaboration likelihood modelof persuasion (ELM) provides a frame-work for understanding the influenceof knowledge on food label process-ing.9 It posits that processing occursthrough either a central or a peripheralroute. Under the central route, theperson draws on experience andknowledge to scrutinize the informa-tion and generate favorable and/orunfavorable responses.10 Centralroute processing occurs when an indi-vidual has high involvement with thetopic (ie, the topic is personally rele-vant) and has the ability to evaluatethe information. Conversely, if a per-son has low involvement and/or lacks

unication, Colorado State University, Fort

, Department of Journalism and TechnicalState University, Fort Collins, CO 80523;mail: malong@lamar.colostate.eduATION AND BEHAVIOR

Journal of Nutrition Education and Beh

experience and knowledge to evaluatethe information, peripheral routeprocessing occurs. Under this route,the person's responses are based onsimple inferences from peripheralcues, which are salient aspects of themessage that are not central to themessage argument (eg, the picture ofan Olympic athlete on the front ofa cereal box could cause consumersto infer that the cereal will helpthem be better athletes).

Two types of product label infor-mation that will be processed differ-entially by experts and novices areintrinsic and extrinsic cues.11 Intrin-sic cues are product-related attributes,such as ingredients, that cannot bemanipulated without altering thephysical properties of the product.12

Conversely, extrinsic cues refer toproduct-related information apartfrom the physical properties, such ashealth and nutrition label claims.According to the ELM, when con-sumers are highly involved and havethe ability to process information,their product evaluations will bebased on intrinsic cues, which requiremore knowledge to interpret. Con-versely, when consumers have lowinvolvement and/or their ability toprocess information is limited, theirproduct evaluations will be based on

avior � Volume 44, Number 4, 2012

Journal of Nutrition Education and Behavior � Volume 44, Number 4, 2012 Walters and Long 351

simple inferences from extrinsic cues,which act as peripheral cues.

Research suggests that experts notonly use intrinsic cues more thannovices, but they are also more likelyto pay attention to mismatches be-tween intrinsic and extrinsic cues.For example, when expert and novicephotographers evaluated advertise-ments for cameras, expert photogra-phers noticed mismatches betweenthe type of camera advertised (anextrinsic cue) and the features listedfor the camera (intrinsic cues),whereas novice photographers didnot.13

The current study investigated thesituation in which consumers' in-volvement in nutrition was high buttheir ability to process product infor-mation varied. Women were used inthis study because they can be consid-ered a high-involvement group fornutrition information. Researchshows that women, when comparedto men, consider eating healthfullymore important7 and are more likelyto use food labels.14-17 For this study,the following hypotheses were posed:

Hypothesis 1: When the extrinsic andintrinsic cues for food products donot match, experts will have lowerperceived quality ratings for thoseproducts than will novices.

Hypothesis 2: When the extrinsic andintrinsic cues for food products donot match, experts will have lowerpurchase intentions for those prod-ucts than will novices.

Hypothesis 3: Experts' ratings of per-ceived product quality will behigher when the extrinsic and in-trinsic cuesmatch than when thesecues do not match.

Hypothesis 4: Experts' purchase inten-tions will be higher for productswhen the extrinsic and intrinsiccues match than when these cuesdo not match.

Another question involves theeffect of intrinsic cues alone. That is,in the absence of an extrinsic cue, doexperts examine intrinsic cues?The following research questionsaddressed this situation:

ResearchQuestion 1: In the absence ofan extrinsic cue and the presence ofan intrinsic cue, will experts andnovices differ in their perceptionof perceived quality of a product?

ResearchQuestion 2: In the absence ofan extrinsic cue and the presence ofan intrinsic cue, will experts andnovices differ in their purchase in-tention toward a product?

The debate over ‘‘all natural’’ label-ing on food products that containhigh fructose corn syrup (HFCS) pro-vides an opportunity to test the effectsof extrinsic and intrinsic cues on con-sumers' purchase intentions and per-ceptions of quality. ‘‘All natural’’labeling is popular; a recent study offood labels found it among the top 5nutrition marketing approaches.2

High-fructose corn syrup is a commoningredient in food products in theUnited States,18 and nutritionists andother consumer groups, such as theCenter for Science in the Public Inter-est, considerHFCS to be an artificial in-gredient.19 Research on how ‘‘allnatural’’ labeling influences consumerperceptions and purchase intentionsis limited, especially in situations inwhich label claims conflict with theproduct's ingredients (eg, the packag-ing contains an ‘‘all natural’’ label, yetthe ingredients include HFCS). Forthis study, ‘‘all natural’’ was used as anextrinsic cue and ‘‘high-fructose cornsyrup’’ was used as an intrinsic cue.

METHODS

A 2 (presence or absence of extrinsiccue) � 2 (presence or absence of in-trinsic cue) � 2 (expert vs novice)within-subject design was used. The4 products chosen for the study (gra-nola bars, lemonade, vanilla yogurt,and salad dressing) are representativeof food products with ‘‘all natural’’ la-beling. Because familiarity with brandnames can influence product evalua-tions, a brand name was created forthis study (Sun Market).

Participants were given simulatedlabels that were not attached to foodproducts. The extrinsic cue was ma-nipulated by placing the phrase ‘‘allnatural’’ on the product label. The ex-trinsic cue was present when thephrase was on the label, and it wasabsent when the phrase was removed.

The intrinsic cue was manipulatedby the placement of ‘‘high-fructosecorn syrup’’ in the ingredients list.The intrinsic cue was present when‘‘high-fructose corn syrup’’ appearedas the third ingredient in the list,

and it was absent when ‘‘high-fructosecorn syrup’’ was replaced by ‘‘honey’’(Figure).

Nutrition knowledge expertise wascreated by assigning participants whohad completed an upper-division hu-man nutrition course to the expertcondition. Participants who had notcompleted this course were assignedto the novice condition.

For this study, 2 dependent vari-ables were measured: perceived prod-uct quality and purchase intention.Ratings of 3 aspects of quality (nutri-tion, healthfulness, and naturalness)were each measured using five 7-point Likert statements (eg, ‘‘Thisproduct is nutritious’’). Purchase in-tention was measured using 4 state-ments (eg, ‘‘I would buy thisproduct,’’ ‘‘I would like to try thisproduct’’) that were evaluated usinga 7-point Likert scale. Scales for eachdependent variable were constructedby summing the scores from the ap-propriate items. Cronbach a for the 2scales was 0.88 or higher. Addition-ally, the dependent variables werenormally distributed according toKolmogorov-Smirnov tests.

To testwhetherwomenwere indeeda high-involvement group, nutritioninvolvement was measured usingitems taken from the Personal Involve-ment Inventory.20 Participants wereasked to rate 8 adjectives related to nu-trition (eg,matters tome,beneficial, es-sential) using a 7-point unipolar scale.

To mask the specific aims of thisstudy and reduce the potential fordemand characteristics, participantsassessed their involvement in 3 addi-tional topics (alcohol use, safe sex,andglobalwarming).Toavoidprimacyand recency effects, nutrition was thesecond topic that participants rated.

People often differ in their prefer-ences for food products, and thesepreferences may affect people's prod-uct quality perceptions and purchaseintentions. To control for these prefer-ences, product liking was measured.Participants were asked to rate ona 7-point scale the extent to whichthey liked each product.

Consumer activist groups and foodmanufacturers have attempted to raisepublicawarenessofHFCS. For example,theCornRefinersAssociation launchedan information campaign in 2008 con-sisting of television commercials, full-page print advertisements in national

Figure. Sample food label with intrinsic cue of ‘‘high fructose corn syrup’’ and withoutextrinsic cue of ‘‘all natural.’’

352 Walters and Long Journal of Nutrition Education and Behavior � Volume 44, Number 4, 2012

newspapers and magazines, and infor-mational Web sites. To control forawareness of HFCS, participants wereasked to indicate whether they hadheard of HFCS and whether they hadseen any commercials about HFCS.

Before this study was conducted, itwas approved by the InstitutionalReview Board at Colorado State Uni-versity.

Participants

The sample consisted of 106 femalestudents (61 experts, 45 novices) atColorado State University who wereenrolled in either an upper-divisionjournalism class or an upper-divisionmedical nutrition therapy course(which had an upper-division coursein human nutrition as a prerequisite).Students were recruited in class, andparticipation was voluntary. Meanage for participants was 22.

Procedure

A 4� 4 Greco-Latin Square design wasused to create a packet containing 4

labels for each study participant anda self-administered questionnaire.Testing occurred in classrooms. Partic-ipants first answered a series of demo-graphic questions and nutritioninvolvement questions. Then, afterreading each label, participants evalu-ated the particular product using a se-ries of items that measured perceivedquality, purchase intention, and prod-uct liking. When all 4 product evalua-tions were completed, respondentsanswered questions to assess theirawareness of HFCS.

RESULTS

Before the hypotheses were tested, the3 control variables (awareness ofHFCS, product liking, and nutritioninvolvement) were tested to rulethem out as alternative explanationsfor study results. Awareness of HFCSwas not a likely alternative explana-tion for the results of this study. Be-cause 97% of participants indicatedthey had heard about HFCS prior tothe study, any differences in the de-pendent variables could not reason-ably be attributed to differences in

awareness of HFCS. Additionally,a nonsignificant independent-samples t test showed that participantswho reported that they had seen HFCScommercials (68 participants) did notevaluate the 4 products differentlythan participants who had not seenHFCS commercials (37 participants).

Differences in the liking of the 4products used in the study were alsoruled out as an alternative explana-tion. Product liking was only mod-estly correlated (0.22 or less) withparticipants' purchase intentions andperceptions of the quality of products.

To examine differences betweenexperts' and novices' involvementwith nutrition, a scale was created us-ing 7 of the 8 original items; 1 itemwas deleted because of low reliability(final Cronbach a ¼ .73). For thisscale, low scores indicate high in-volvement. Both experts' and novices'mean nutrition involvement scoreswere high (experts ¼ 11.7, SD ¼7.09; novices ¼ 15.62, SD ¼ 7.02).However, because they differed signif-icantly (t (104) ¼ 2.82, P < .001), cor-relations were run to determinewhether involvement predicted dif-ferences in the dependent variables.Results were not statistically signifi-cant. Further, the maximum effectsize was 0.01. Therefore, althoughlevel of nutrition involvement dif-fered between experts and novices, itwas not included in the analysis.

Hypothesis Testing

Our first hypothesis, which predictedthat experts would have a lower per-ceived quality for products when theextrinsic cue and the intrinsic cuewere mismatched than would nov-ices, was supported. Experts' perceivedquality (mean [M]¼ 14.72) was signif-icantly different from that for novices(M ¼ 18.54, t[102] ¼ �3.16, P ¼ .001).

Our second hypothesis was alsosupported. Experts had lower pur-chase intentions for products whenthe extrinsic cue and the intrinsiccue were mismatched (M ¼ 10.25)than did novices (M ¼ 13.67, t[104]¼ �3.12, P ¼ .001).

Our third hypothesis was also sup-ported. The results of a paired-samplet test showed that experts' perceivedquality was higher for productswhen the extrinsic cue and intrinsic

Journal of Nutrition Education and Behavior � Volume 44, Number 4, 2012 Walters and Long 353

cue matched (M ¼ 23.68) than whenthe extrinsic and intrinsic cues didnot match (M ¼ 14.83, t[58] ¼ 8.04,P < .001). Likewise, results for ourfourth hypothesis showed thatmismatched cues affected experts'purchase intentions. Experts' pur-chase intentions were higher for prod-ucts when cues matched (M ¼ 17.31)than when they did not match (M ¼10.24, t[60] ¼ 7.64, P < .001).

Our first research question askedwhether experts and novices differedin their perceptions of the quality ofaproduct in the absenceof an extrinsiccue and the presence of an intrinsiccue. Novices perceived that productswith the intrinsic cue of HFCS were ofhigher quality (M¼ 16.6) than did ex-perts (M ¼ 14.2, F[1,102] ¼ 5.50, P ¼.02). A similar pattern was found forour second research question, whichaskedwhether experts andnovices dif-fered in their purchase intentions inthe absence of an extrinsic cue andthe presence of an intrinsic cue. Nov-ices weremore likely to purchase theseproducts (M¼ 12.5) than were experts(M ¼ 10.4, F[1,104] ¼ 7.74, P < .001).

DISCUSSION

This study found that experts andnovices differed in their ability to pro-cess food label information. Experts'perceptions of product quality andpurchase intentions were higher to-ward products when the label claimof ‘‘all natural’’ was consistent withthe ingredients. Conversely, ‘‘all natu-ral’’ labeling, regardless of the ingredi-ents listed on the label, positivelyinfluenced novices' product evalua-tions. Overall, these findings are con-sistent with the theoreticalmechanisms of the ELM,9,10 previousresearch on the role of expertise inevaluating products with conflictingcues,13 and previous research on therole of nutrition knowledge in con-sumers' likelihood of using food la-bels.6-8,16

The results also indicate that ex-perts and novices differ in their evalu-ation of products based on intrinsiccues alone. In the absence of the ex-trinsic cue of ‘‘all natural,’’ noviceshad higher perceived quality and pur-chase intention ratings for productswith the intrinsic cue of HFCS thandid experts.

Limitations

This research was limited to women,a high-involvement group, to exam-ine how differences in nutritionknowledge influence this group's abil-ity to process intrinsic information.The study also was limited in that itinvolved young, educated women at1 university. Therefore, generalize-ability of the results is reduced. Thisstudy could be extended by replicat-ing it in other age groups and amongparticipants with a variety of educa-tional backgrounds and levels ofinvolvement in nutrition.

Another limitation was the productlabel manipulation. Labels used for thestudy were not as elaborate as thosetypically used on food products. Forexample, the labels were graphicallysimple, did not have photographs,were printed in black and white, didnot contain nutrition facts, and werenot attached to the food products. Al-though the use of these more simplis-tic labels hurts the ecological validityof the study, the fact that novice con-sumers still appeared to rely solely onthe peripheral cue of ‘‘all natural’’when making their product assess-ments is quite telling. Regardless, repli-cation of the study should beconducted using labels more typicallyfound on food products, which wouldprovide a more realistic situation.

The results of this study are also lim-ited in that only 1 type of extrinsic cueand 1 type of intrinsic cue were used. Itis possible that the results obtainedwere a function of this specific cue ma-nipulation. Furthermore, other attri-butes could be investigated, such as theextent to which extrinsic and intrinsiccues influencebeliefsabout theenviron-mental friendliness of food products.

Finally, this study assumed partici-pants enrolled in an upper-divisionhuman nutrition course were experts.Use of a food label knowledge survey,or other nutrition knowledge test,would have provided further evidenceof expertise.

IMPLICATIONS FORRESEARCH ANDPRACTICE

Although novice consumers wereaware of HFCS, their perceptions ofproduct quality and purchase inten-

tions were influenced by the ‘‘all natu-ral’’ extrinsic cue, regardless of theingredients listed. Given that con-sumers who lack nutrition knowledgeappear to be more likely to use extrin-sic label information, the UnitedStates Food and Drug Administrationneeds to recognize that label claimssuch as ‘‘all natural’’ may mislead themajority of food consumers and thusshould be regulated.

In the absence of regulation, in-creasing novice consumers' nutritionknowledge through education pro-grams can help them discriminateamong products based on intrinsiccue information, such as ingredientsand nutrition facts.

ACKNOWLEDGMENTS

This manuscript is based on the firstauthor's thesis.

REFERENCES

1. Taylor CL, Wilkening VL. How thenutrition food label was developed,part 1: the Nutrition Facts panel. J AmDiet Assoc. 2008;108:437-442.

2. Colby SE, Johnson L, Scheett A,Hoverson B. Nutrition marketing onfood labels. J Nutr Educ Behav. 2010;42:92-98.

3. Labiner-Wolfe J, Lin C-TJ, Verrill L.Effect of low-carbohydrate claims onconsumer perceptions about food prod-ucts’ healthfulness and helpfulness forweight management. J Nutr Educ Behav.2010;42:315-320.

4. Celsi RL, Olson JC. The role ofinvolvement in attention and compre-hension. J ConsumRes. 1988;15:210-224.

5. Gregan-Paxton J. The role of abstractand specific knowledge in the forma-tion of product judgments: an analogi-cal learning perspective. J ConsumPsychol. 2001;11:141-158.

6. Drichoutis AC, Lazaridis P,Nayga RM. Nutrition knowledge andconsumer use of nutrition food labels.European Review of Agricultural Econom-ics. 2005;32:39-118.

7. Jasti S,Kovacs S.Use of trans fat informa-tion on food labels and its determinants ina multiethnic college student population.J Nutr Educ Behav. 2010;42:307-314.

8. Crites SL, Aikman SN. Impact of nutri-tion knowledge on food evaluations.Eur J Clin Nutr. 2005;59:1191-1200.

354 Walters and Long Journal of Nutrition Education and Behavior � Volume 44, Number 4, 2012

9. Petty R, Cacioppo JT. Elaborationlikelihood model of persuasion. In:Berkowitz L, ed.Advances in Experimen-tal Social Psychology. New York, NY:Academic Press; 1986:123-205.

10. Petty RE, Brinol P, Priester JR. Massmedia attitude change: implications ofthe elaboration likelihood model ofpersuasion. In: Bryant J, Oliver MB,eds. Media Effects: Advances in Theoryand Research. 3rd ed. New York, NY:Routledge; 2010:125-164.

11. Olson JC, Jacoby J. Cue utilization inthe quality of perception process. Pro-ceedings of the Third Annual Conferenceof the Association for Consumer Research.Iowa City, IA: Association for Con-sumer Research; 1972:167-179.

12. Richardson PS, Dick AS, Jain AK.Extrinsic and intrinsic cue effects on

perceptions of store brand quality.Journal of Marketing. 1994;58:28-36.

13. Sujan M. Consumer knowledge: effectson evaluation strategies mediatingconsumer judgements. J Consum Res.1985;12:31-46.

14. Nayga RM. Determinants of con-sumers’ use of nutritional informationon food packages. Journal of Agricul-tural and Applied Economics. 1996;28:303-312.

15. Nayga RM. Impact of sociodemo-graphic factors on percieved impor-tance of nutrition in food shopping.J Consum Aff. 1997;31:1-9.

16. DrichoutisAC, Lazaridis P,NaygaRM,Kapsokefalou M, Chryssochoidis G. Atheoretical and empirical investigationof nutritional label use. Eur J HealthEcon. 2008;9:293-304.

17. Mangleburg TF, Grewal D, Bristol T.Socialization, gender, and adolescents’self-reports of their generalized use ofproduct labels. J Consum Aff. 1997;31:255-279.

18. Bocarsly ME, Powell ES, Avena NM,Hoebel BG. High-fructose cornsyrup causes characteristics of obesity inrats: increased body weight, body fat,and triglyceride levels. Pharmacol Bio-chem Behav. 2010;97:101-106.

19. Center for Science in the Public Inter-est. CSPI to Sue Cadbury Schweppesover ‘‘All Natural’’ 7UP. http://www.cspinet.org/new/200605111.html. Pub-lishedMay 11, 2006. AccessedDecember8, 2011.

20. Zaichkowsky JL. Measuring theinvolvement construct. J Consum Res.1985;12:341-352.