This article was downloaded by: [Central Michigan University]On: 16 October 2014, At: 11:51Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Health CommunicationPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/hhth20

Increasing the Effectiveness of Messages PromotingResponsible Undergraduate Drinking: Tailoring toPersonality and Matching to ContextValerie K. York a , Laura A. Brannon b & Megan M. Miller ba Office of Educational Innovation and Evaluation , College of Education, Kansas StateUniversityb Department of Psychology , Kansas State UniversityPublished online: 27 Sep 2011.

To cite this article: Valerie K. York , Laura A. Brannon & Megan M. Miller (2012) Increasing the Effectiveness of MessagesPromoting Responsible Undergraduate Drinking: Tailoring to Personality and Matching to Context, Health Communication,27:3, 302-309, DOI: 10.1080/10410236.2011.585450

To link to this article: http://dx.doi.org/10.1080/10410236.2011.585450

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Health Communication, 27: 302–309, 2012Copyright © Taylor & Francis Group, LLCISSN: 1041-0236 print / 1532-7027 onlineDOI: 10.1080/10410236.2011.585450

Increasing the Effectiveness of Messages PromotingResponsible Undergraduate Drinking: Tailoring to Personality

and Matching to Context

Valerie K. YorkOffice of Educational Innovation and Evaluation, College of Education

Kansas State University

Laura A. Brannon and Megan M. MillerDepartment of PsychologyKansas State University

This study addressed the serious problem of college student binge drinking by identifying fac-tors that improve the effectiveness of messages encouraging responsible drinking presentedthrough a website simulation. We tested schema matching (i.e., whether the message matchesthe person’s self-schema type or not) and two types of context matching (i.e., whether the mes-sage matches the topic or values of the message context) to determine their relative influenceon the effectiveness of the message. We expected that messages matched to any of these fac-tors would be more effective than messages not matched. Schema matching reduced intentionsto drink while staying in/home, but topic matching reduced intentions to drink when goingout, suggesting that different factors are important for messages targeting drinking behaviorin different locations. Significant interactions between topic matching and value matching onmessage evaluation variables indicated that the message should not match the message contexttoo closely. That is, there appears to be a matching threshold: Increasing the number of factorsthe message matches does not increase message effectiveness, possibly because it makes themessage too redundant with the surrounding content.

Between 40 and 45% of undergraduates binge drink1

(Johnston, O’Malley, Bachman, & Schulenberg, 2008;Wechsler et al., 2002). Numerous negative consequencesare associated with binge drinking, such as perform-ing poorly in school, arguing with friends, engaging inunplanned/irresponsible sexual activities, and getting introuble with police (Wechsler et al., 2002). The concernis especially driven by the more serious consequences ofheavy drinking. Annually, nearly 600,000 college studentssuffer unintentional alcohol-related injuries, 1,700 of whichresult in death, including death from motor vehicle accidents;

Correspondence should be addressed to Laura A. Brannon, AssociateProfessor of Psychology, Kansas State University, 492 Bluemont Hall,Manhattan, KS 66506-5302. E-mail: lbrannon@ksu.edu

1Binge drinking is defined as consuming four drinks on one occasionfor females and five drinks on one occasion for males (Wechsler et al.,2002).

nearly 700,000 students are assaulted by a peer who had beendrinking; and 97,000 suffer alcohol-related sexual assault orrape (Hingson, Heeren, Winter, & Wechsler, 2005).

Some colleges require students to complete Internet-based programs (e.g., Walters, Vader, & Harris, 2007),educating students about the effects of heavy alcohol con-sumption, normative college student drinking behavior, andhow to drink safely (Hustad, Barnett, Borsari, & Jackson,2010). Some programs also incorporate audience segmenta-tion (e.g., AlcoholEdu: Wall, 2007) or behavioral feedback(e.g., eChug: Walters et al., 2007).

Although online programs have shown promise of reduc-ing irresponsible college alcohol consumption, we argue thatthese programs may benefit from incorporating individu-alized feedback that is more personally meaningful thanbehavioral or normative feedback. Specifically, we aimedto demonstrate the effectiveness of anti-binge drinking mes-sages tailored to a student’s self-schema (i.e., his or her core

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

MATCHING MESSAGES TO PERSON AND CONTEXT 303

values and beliefs). We also examined how the congruencybetween the message and its surrounding context impactsmessage effectiveness.

USING TAILORING TO INCREASE MESSAGEEFFECTIVENESS

Compared to using generalized mass communication,health-related attitude and behavior change may be moreeffectively accomplished with the use of audience seg-mentation (Noar, Harrington, & Aldrich, 2009). However,tailoring health information to individual characteristics hasbeen shown in numerous empirical tests to be even moreinfluential than using group-targeting techniques for modi-fying attitude and behavior for a variety of health behaviors(Dijkstra, 2008; Noar et al., 2009; Ten Wolde et al., 2008).Tailoring involves personalization (i.e., using an individual’sname or other identifiable aspects), adaptation (i.e., present-ing information that is relevant to individual characteristics,such as gender), and feedback (i.e., responding to an individ-ual’s psychological or behavioral state) or a combination ofthese three (Dijkstra, 2008). Considering factors such as anindividual’s perceptions of response efficacy and perceptionsof self-efficacy for performing certain health-related behav-iors also has been emphasized (Murray-Johnson & Witte,2003; Parrott, Monahan, Ainsworth, & Steiner, 1998).

Although the mechanisms behind the efficacy of tailor-ing are not well understood, one common explanation isthat individuals are more likely to attend to tailored thanto nontailored information, leading to increased informa-tion processing (Hawkins, Kreuter, Resnicow, Fishbein, &Dijkstra, 2008; Noar et al., 2009). Further, central process-ing of information (i.e., elaborating on information, thinkingabout it critically) is more likely than peripheral processing(i.e., relying on heuristics and visual cues) for personally rel-evant information, increasing the likelihood of stable attitudechange and thus, future behavior change (Dijkstra, 2008;Hawkins et al., 2008; Noar et al., 2009). However, if tailoredinformation is perceived as an indication of source credi-bility, peripheral processing also may increase persuasion(Hawkins et al., 2008).

According to schema correspondence theory, informa-tion can be made self-relevant by tailoring to self-schema,the structured cognitive organization of information aboutthe self that a person uses to interpret and categorize newinformation about the self2 (Brock, Brannon, & Bridgwater,1990). Schema tailoring is supported by several indepen-dent studies testing health- or consumer-related messagestailored to the same four self-schemas used in our study:Responsible, Adventuresome, Relationship-Oriented, andInquisitive. Participants receiving a message matched totheir self-schema have been found to be just as persuaded

2A person’s self-schema is more encompassing of the individual’s rep-resentation of him- or herself than are somewhat related concepts such asthe function the person’s attitudes serve (see Petty, Wheeler, & Bizer, 2000).

as those receiving a behavioral feedback message and morepersuaded than those receiving a message matched to a dif-ferent self-schema or a schema-neutral message (Brannon &Brock, 1994; Brannon & McCabe, 2002; Brock et al., 1990;Pease, Brannon, & Pilling, 2006; Pilling & Brannon, 2007).

Importance of Message–Context Congruencyin Message Effectiveness

A second goal of our study was to examine how the congru-ency between an anti-binge drinking message and the sur-rounding context impacts message effectiveness. Literatureon expectancy disconfirmation theory suggests that peoplethink more deeply about information when their expecta-tions are violated (Maheswaran & Chaiken, 1991), as when amessage is incongruent with its context. However, increasingmessage–context congruency may increase message effec-tiveness because the content of the context may activate(or prime) knowledge structures necessary for processingthe similar content of the message (Sanbonmatsu & Fazio,1991). These activated knowledge structures will be read-ily accessible, and thus will be used to easily process andremember the message. Also, when perceiving informationas relevant, versus irrelevant, to a particular context, peopleexperience more favorable attitudes and behavior change inrelation to the information presented (Lee & Labroo, 2004).

We examined message–context congruency in terms ofboth the topic and the value, the rationale for which wasbased largely on consumer literature identifying the variousaspects of a product that influence product-related decisions(Brannon & Brock, 2006). Specifically, a product can beconsidered in terms of both the function it serves and theattitude or personality associated with the product (Shavitt,1989). For example, a sports car and a minivan serve thesame basic function (transportation), but we associate dif-ferent attitudes or personalities with each (e.g., adventurevs. practicality, respectively). In our study, message topic orbehavior is similar to product function, although the valueassociated with the message topic or behavior is similar toproduct personality.

Using the Internet to Communicate Health Information

The Internet is becoming a popular tool for distributinghealth information because it quickly and easily allows formessage tailoring (Dijkstra, 2008; Noar et al., 2009; Rimal& Adkins, 2003). With the Internet, tailored communicationcan be carried out on a large scale and with less expense thanmass mailings or personal counseling sessions. Given thesebenefits, we used a website simulation to test anti-bingedrinking messages.

STUDY PURPOSES AND HYPOTHESES

We aimed to examine how the effectiveness of an anti-bingedrinking message is influenced by matching the messageto an individual’s self-schema (schema matching), as well

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

304 YORK, BRANNON, AND MILLER

as to the topic and the value associated with the webpagein which the message is presented. We predicted a self-schema-matched message would be more effective than aschema-neutral message. Second, we expected matching tothe topic of the webpage would be more effective than notmatching to the topic, and matching to the value associatedwith the webpage would be more effective than not matchingto the value. We also examined whether any combination ofschema, topic, and value matching adds to message effec-tiveness above and beyond matching to just one of thesefactors. This component of our study was purely exploratory,and we did not have any specific predictions for interac-tion effects. As is typical in tailoring research, we includedparticipant gender as a predictor. We were interested to seewhether one gender is more sensitive than the other to topicmatching than to value matching and vice versa.

We measured message effectiveness in terms of drinkingintentions, which have been shown to be positively related tolater drinking behavior (Shim & Maggs, 2005), and attitudetoward the message, which has been shown to be a good pre-dictor of behavioral intentions and actual behavior (Ajzen& Fishbein, 1980; Eagly & Chaiken, 1998). Greater mes-sage effectiveness was defined by intentions for drinking lessand more positive attitudes toward the message. We mea-sured intentions for both going out and staying in drinkingbecause differences between these two contexts might resultin differences in drinking behavior. A report of focus groupsand in-depth interviews with college students suggests thatwhen the reason for drinking is socially driven, students goout to drink, with drinking being the primary activity of theevening (Treise, Wolberg, & Otnes, 1999). Perhaps moreimportant, students report that when they go out to social-ize but do not intend to drink, because they want to feellike a part of the group, they may succumb to binge drink-ing because everyone around them is drinking (Treise et al.,1999). In contrast, when staying in drinking, perhaps sur-rounded by only a few roommates or friends, compared tobeing in a crowded bar or club, students may feel more com-fortable acting in accord with their personal attitudes andvalues and choose to abstain from drinking or limit their con-sumption (Treise et al., 1999). Additionally, because certainnegative consequences of heavy drinking (Wechsler et al.,2002) may be more likely to occur in one of these con-texts versus the other (e.g., drunk driving may be more likelywhen going out, and disturbing or harming roommates maybe more likely when staying in), it is important to separatelyexamine the influence of responsible drinking messages inthese two contexts.

METHOD

Participants

One hundred eighty-four undergraduate students (52%female) participated in our study in partial fulfillment of a

General Psychology course requirement. Students enrolledin all sections of the course had the opportunity to partic-ipate by signing up on a paper sheet posted with sign-upsheets for other studies in the psychology department lobby.Participants’ average age was 19.4 years (SD = 1.8). Justover half of the participants (51%) were binge drinkers.

Design and Procedure

In a 2 × 2 × 2 × 2 between-subjects design, we examinedhow the influence of an anti-binge-drinking message variesby four factors: schema matching (schema-matched orschema-neutral), topic matching (matched or not matchedto the topic of the webpage context), value matching(matched or not to the value associated with the webpagecontext), and participant gender (male or female). Therewere eight total manipulation combinations: The messagecould be presented on one of four webpage contexts, andthe message could be either schema-matched or schema-neutral. Students’ drinking intentions both for drinking inand drinking out and their attitude toward an anti-bingedrinking message served as dependent variables. Participantswere randomly assigned to a computer that determinedtheir assignment to an experimental condition. They weretold they would be viewing a newly developed websiteabout activities and services in their college town andcompleting surveys about their personality and behaviors(including alcohol consumption) and their opinions of thewebsite.

Schema selection survey. Participants were told thatthe researchers were interested to see if different typesof people had different opinions of the website, and theywere asked to provide information about their person-alities (self-schema) and behaviors. We measured self-schema with the same basic methodology used in previousresearch (Brannon & Brock, 1994; Brannon & McCabe,2002; Brock et al., 1990; Pease et al., 2006; Pilling &Brannon, 2007). Participants were presented with four self-schema types (Responsible, Adventuresome, Relationship-Oriented, and Inquisitive), each represented by a string offour identifying adjectives (responsible-dependable-helpful-sensible, adventuresome-skillful-competitive-spontaneous,warm-communicative-compassionate-feeling, and versatile-wise-conceptual-curious); a descriptive summary statementelaborating on the adjectives; and a graphic of people whoseattire and activities were reflective of the schema.3

Activities and interests survey. Next, partici-pants completed the Activities and Interests Survey,which included a measure of baseline drinking behavior.Participants were presented with two open-ended questions

3The construct validity of this measure was demonstrated in a studyin which over 8,000 subjects easily selected the schema type that mostoverlapped with their own self concept (Brock et al., 1990).

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

MATCHING MESSAGES TO PERSON AND CONTEXT 305

asking how many drinks they typically consume when goingout drinking and when staying in to drink. Distracter itemsasked about engagement in other activities.

Matching to context topic and value. Next, partici-pants were taken to the homepage of the website simulationand were instructed to view the website in a consistentfashion. On the left side of the screen were links to four web-pages (Bars, Health Organizations, Spiritual Organizations,and Points of Interest). The webpages represented the fourpossible combinations of topic and value matching and non-matching: topic matching/value nonmatching (Bars), topicmatching/value matching (Health Organizations), topicnonmatching/value matching (Spiritual Organizations), andtopic nonmatching/value nonmatching (Points of Interest).Participants were randomly assigned to topic and valuecongruence conditions and were presented with an anti-binge drinking message on the webpage associated withtheir assigned topic and value conditions. For example, ifassigned to the topic matching and value nonmatching con-ditions, a participant saw the message on the Bars pagebecause the topic of the message (i.e., drinking alcohol) iscongruent with the topic of the page, but the value associatedwith the message (i.e., drink responsibly) is not congruentwith the value associated with the page (i.e., drink for fun).The link to the page on which the message appeared waspresented last in the list of webpage links. When partici-pants opened the final webpage, the responsible drinkingmessage appeared on the left-hand side of the screen, andthe webpage content appeared in the center of the screen.

Matching to schema. Four distinct messages werecreated by the researchers using rhetorical styles to repre-sent each of the four schema types.4 Although all messagescontained the same basic information, they contained differ-ent persuasive arguments for why binge drinking was notconsistent with the values of the schema types. Half of theparticipants were randomly assigned to the schema-matchedmessage condition and were presented with an anti-bingedrinking message tailored to the schema type they selectedearlier in the study. For example, if a participant identifiedwith the “Adventuresome” self-schema and was assigned tothe schema-matched condition, he or she received the fol-lowing message: “Alcohol is actually a depressant. Whentoo much is consumed, alcohol dulls the senses and slowsa person down, thereby making it impossible for them tobe the life of the party. Binge drinking actually limits theamount of fun and excitement you can experience in anight. Live the Exciting Life, Don’t Binge Drink!” Theother half of the participants were randomly assigned to aschema-neutral message condition and received the follow-ing message: “Drinking to excess leads to many problems in

4All messages used in the study have been pretested and used in otherpublished work (Pilling & Brannon, 2007).

a person’s life. Different types of people experience differ-ent types of problems, but most people who drink to excessexperience a variety of difficulties. These can interfere withthe person’s quality of life. Drinking in moderation is thebest choice. Don’t Binge Drink!”

Drinking intentions. Participants were told that wewanted to know whether the website affected their inter-ests, and were asked to indicate their intentions for engag-ing in the same behaviors they were asked about earlier.Participants reported the number of drinks they intended tohave when going out and when staying in drinking in thenext week.

Evaluations of the message. Participants also weretold that the researchers wanted to display health-relatedmessages on the website and were interested in the stu-dents’ opinions of the message they had viewed. Participantsresponded to three items assessing their general atti-tudes toward the message using a 7-point Likert-typescale: “I (1 = dislike, 7 = like) the message;” “I react(1 = unfavorably, 7 = favorably) to the message;” and “Ifeel (1 = negatively, 7 = positively) toward the message.”Two items assessed the extent to which participants feltthe message was appropriate in the website context: “Howappropriate is the message in the context of the website youviewed?” (1 = very inappropriate, 7 = very appropriate);“How relevant was the message to the content of the websiteyou viewed?” (1 = very irrelevant, 7 = very relevant).

Social desirability. Because binge drinking is not asocially desirable behavior, participants might misreporttheir drinking behavior, intentions for future drinking behav-ior, and their attitudes toward an anti-binge drinking mes-sage. Therefore, we asked participants to complete a socialdesirability measure in which they responded true or falseto 10 items (Strahan & Gerbasi, 1972). Scores on this mea-sure range from 0 to 10, with higher scores indicating moresocially desirable responses. Social desirability was includedas a covariate in our analyses.

Manipulation checks. Using a 7-point Likert-typescale (1 = not at all similar, 7 = completely similar), par-ticipants indicated how similar they thought the topic andvalue of the responsible drinking message were to the topicand value of the webpage on which the message appeared.Finally, participants were thanked and debriefed.

RESULTS

Manipulation Checks

Participants perceived topic matching webpages (Bars andHealth Organizations, M = 5.08, SD = 1.31) to match themessage topic better than the topic nonmatching webpages(Spiritual Organizations and Points of Interest, M = 3.12,

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

306 YORK, BRANNON, AND MILLER

SD = 1.70), F(1, 182) = 76.21, p < .001, and valuematching webpages (Health Organizations and SpiritualOrganizations, M = 4.79, SD = 1.45) to match the messagevalue better than the value nonmatching webpages (Bars andPoints of Interest, M = 3.77, SD = 1.82), F(1, 182) = 17.72,p < .001.

Drinking Intentions

We conducted two 2 (schema-matching: schema-matched/

schema-neutral)×2 (topic matching: matched/nonmatched)× 2 (value matching: matched/nonmatched) × 2 (gender:males/females) between-subjects analyses of covariance(ANCOVAs) on the number of alcoholic beverages partic-ipants intended to drink when going out drinking and whenstaying in drinking during the next week. The two covari-ates in the analyses were the participant’s baseline numberof alcoholic beverages typically consumed on a drinkingoccasion (either when going out drinking or when stay-ing in drinking, respectively) and the participant’s socialdesirability score.

Going out drinking. The ANCOVA for drinking inten-tions when going out revealed a significant main effect fortopic matching, F(1, 160) = 4.85, p < .05, partial η2 = .029.Participants who saw the message on a topic matchingwebpage (Bars or Health Organizations, M = 3.59, SE =0.12) intended to drink fewer drinks when going out thandid those who viewed a message on a topic nonmatch-ing webpage (Spiritual Organizations or Points of Interest,M = 3.97, SE = 0.12). We also found a significant interac-tion between schema matching and gender, F(1, 160) = 4.27,p < .05, partial η2 = .026. Females who received a schema-matched message (M = 3.87, SE = 0.17) intended to drinkabout the same as females who received a schema-neutralmessage (M = 3.79, SE = 0.17), but males who received aschema-matched message (M = 3.41, SE = 0.18) intendedto drink less when going out compared to males whoreceived a schema-neutral message (M = 4.04, SE = 0.18).The ANCOVA did not show significant main effects forschema matching [F(1, 160) = 2.47, ns], value matching[F(1, 160) = 0.22, ns], or gender [F(1, 160) = 0.34, ns], nordid it reveal any other significant interaction effects.

Staying in drinking. The ANCOVA for drinking inten-tions when staying in revealed a significant schema match-ing effect, F(1, 160) = 4.54, p < .05, partial η2 = .028.Participants who received a self-schema-matched message(M = 1.75, SE = 0.16) intended to drink less than did partic-ipants who received a schema-neutral message (M = 2.22,SE = 0.16). No main effects were revealed for topic match-ing [F(1, 160) = 0.23, ns], value matching [F(1, 160) = 0.14,ns], or gender [F(1, 160) = 0.92, ns], nor were there anysignificant interaction effects.

Evaluations of the Message

We conducted two 2 (schema matching: schema matched/

schema neutral) × 2 (topic matching: matched/nonmatched) × 2 (value matching: matched/nonmatched) × 2(gender: males/females) between-subjects ANCOVAson the message evaluation variables: attitude toward themessage and appropriateness of the message in the webpagecontext. In both ANCOVAs, baseline alcohol consumptionboth for going out drinking and for staying in drinking andsocial desirability were entered as covariates.

Attitude toward the message. The ANCOVA ofattitude toward the message revealed a significant maineffect of gender, F(1, 165) = 5.74, p < .05, partial η2 = .034.Females (M = 5.03, SE = 0.12) rated the message morepositively than did males (M = 4.60, SE = 0.12). We alsofound a significant topic matching × value matching inter-action, F(1, 165) = 5.08, p < .05, partial η2 = .030. Themessage was rated more positively if it was presented in atopic matching/value nonmatching context (Bars, M = 4.97,SE = 0.17) or a value matching/topic nonmatching context(Spiritual Organizations, M = 5.04, SE = 0.16) than ifpresented in a context that was both topic and value match-ing (Health Organizations, M = 4.57, SE = 0.16) or bothtopic and value nonmatching (Points of Interest, M = 4.69,SE = 0.17). We found no other significant interactioneffects or significant main effects of schema matching [F(1,165) = 3.16, ns], topic matching [F(1, 165) = 0.30, ns], orvalue matching [F(1, 165) = 0.02, ns].

Appropriateness of the message in the webpagecontext. The ANCOVA of appropriateness of the messagein the webpage context revealed a significant main effectfor topic matching, F(1, 165) = 18.59, p < .001, partialη2 = .101. Participants who viewed the message in a topicmatching webpage (Bars or Health Organization, M = 4.73,SE = 0.17) rated the message as more appropriate in thecontext than did those who viewed the message in a topicnonmatching webpage (Spiritual Organizations or Points ofInterest, M = 3.71, SE = 0.17). However, this main effectwas qualified by a significant topic matching by valuematching interaction, F(1, 165) = 5.90, p < .05, partialη2 = .034. Participants rated the message as more appropri-ately placed in a topic matching/value nonmatching context(Bars, M = 4.98, SE = 0.25) than in a topic matching/valuematching context (Health Organizations, M = 4.48,SE = 0.23). Similarly, they rated the message as moreappropriately placed in a topic nonmatching/value matchingcontext (Spiritual Organizations, M = 4.03, SE = 0.23) thanin a topic nonmatching/value nonmatching context (Pointsof Interest, M = 3.39, SE = 0.23). We found no othersignificant interaction effects or any significant main effectsof schema matching [F(1, 165) = 0.01, ns], value matching[F(1, 165) = 0.09, ns], or gender [F(1, 165) = 0.00, ns].

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

MATCHING MESSAGES TO PERSON AND CONTEXT 307

DISCUSSION

Drinking Intentions

We found schema matching positively influenced drinkingintentions in the context of staying in to drink. This findingsupports our prediction that different types of appeals maybe necessary for influencing drinking behavior in differentsituations. Specifically, with fewer social pressures guidingtheir behavior, college students may feel more comfortableacting in accordance with their personal attitudes and val-ues when staying in to drink than when going out drinking(Treise et al., 1999). This may be why we found self-schemamessages, which emphasize how irresponsible drinking isinconsistent with one’s personal values, to influence drinkingintentions when staying in but not when going out. Thus, inthe context of going out drinking, tailored appeals aimed atreducing alcohol consumption may not be powerful enoughto overcome the influence of social pressures.

Social factors may be especially influential on femaledrinking behavior in the context of going out to drink,as indicated by a significant interaction between schemamatching and gender. Males intended to drink less whengoing out if they received a schema-matched messageversus a schema-neutral message, but in both of thesemessage conditions, female drinking intention approachedthe consumption amount for binge drinking (schema-matched message, M = 3.87, SE = 0.17; schema neutralmessage, M = 3.79, SE = 0.17). Because relationships aremore strongly integrated into the identity and self-esteemof females than males, females may be more stronglyinfluenced than males to consume alcohol in social situa-tions for reasons such as to avoid being degraded for notdrinking, to feel more comfortable interacting with others,and to appear more attractive to the opposite sex (Gleason,1994). Therefore, interventions may need to incorporategender-based segmentation to encourage college students todrink responsibly when going out drinking. A self-schematailored approach may be effective at influencing males, butan approach that homes in on social influences or the impor-tance of relationships may be necessary for influencingfemale drinking behavior in this context.

The significant influence of topic matching on drinkingintentions when going out but not when staying in may beexplained by priming effects (Herr, 1989; Sanbonmatsu &Fazio, 1991). Because a webpage about bars is more con-gruent with the concept of going out drinking than with theconcept of staying in drinking, participants who saw an anti-binge drinking message on the Bars (topic matching) web-page may have been primed to think more deeply about theirown drinking behavior in terms of going out, thus result-ing in a stronger influence on their intentions for drinkingwhen going out. Similarly, the Health Organizations (topicmatching) webpage included information about alcohol-related organizations (e.g., Mothers Against Drunk Driving)

promoting the use of public transportation and designateddrivers to prevent drinking-related risky behavior such asdriving under the influence. Because the alcohol-relatedhealth organizations focused on risks more commonly asso-ciated with drinking when going out versus staying in,seeing an anti-binge drinking message on this page may haveprimed thoughts about changing drinking behavior in thecontext of going out more so than in the context of staying in.

The lack of value matching effects may be explainedby a moderating effect of “level of involvement” withthe target behavior, such that an effect of value matchingmay be more likely to occur when message recipients aremotivated to think deeply about the value being conveyed(De Pelsmacker, Geuens, & Anckaert, 2002). It is quitepossible that undergraduate students have low motivation tothink deeply about responsible drinking, especially duringparticipation in a research study. Similarly, when thinkingabout drinking alcohol, the value of drinking responsiblymay be salient to some students, but others may associate adifferent value (e.g., having fun) with alcohol consumption.Still, value matching may be important when examining abehavior for which the associated value is generally con-sistent and salient to students. For example, exerting one’sconstitutional right and responsibility to vote might be morestrongly influenced if a persuasive message encouragingstudents to vote is placed on a webpage that imbues valuessimilar to those associated with the message (e.g., a localgovernment webpage).

Evaluations of the Message

Together, expectancy confirmation theory (Herr, 1989)and expectancy disconfirmation theory (Maheswaran &Chaiken, 1991) may explain why participants seemed toprefer only some message–context congruency. Participantstended to more positively evaluate the message if it wasonly partially matched to its surrounding context (i.e., eithertopic or value-matched) than if it was completely matched(i.e., topic and value-matched) or if it was completelynonmatched (i.e., topic and value nonmatched). Expectancyconfirmation theory would suggest that students preferred atleast some message–context congruency because they maybe more likely to expect a responsible drinking messageto appear on a page related to drinking behavior (topicmatching) or a page about drinking responsibly (valuematching). On the other hand, the expectancy disconfirma-tion literature suggests that when a message is inconsistentwith a message recipient’s expectations, the recipient spendstime thinking about how to resolve this conflict, resulting inan increase in message processing (Maheswaran & Chaiken,1991). This may explain why participants tended to morepositively evaluate the message when it was only partiallycongruent with its surrounding context. Placing a messagein a context with which it is at least somewhat congruent

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

308 YORK, BRANNON, AND MILLER

may be beneficial because students may react negativelytoward the message if it has no congruency with the context.At the same time, however, making the message–contextrelationship somewhat incongruent may help grab themessage recipient’s attention and lead to greater cognitiveprocessing of the message.

Limitations, Implications, and Future Research

Our participants were required to view an anti-binge drink-ing message in a webpage chosen by the researchers. In reallife, however, students choose the websites they visit, andthey might react more favorably to a health-promoting mes-sage presented on a website they choose to visit versus on awebsite they do not self-select. Future research could involvea field study in which participants are presented with themessage in a webpage they select themselves.

While the significant effects on drinking intentions sup-ported our predictions, the benefits of message personaliza-tion and context matching would be further supported byreplicating our findings in research using actual behavior asthe primary dependent variable. However, given that this wasa preliminary study, it was appropriate to assess messageeffectiveness with drinking intentions and attitude towardthe message, which are good predictors of actual behavior(Eagly & Chaiken, 1998; Shim & Maggs, 2005).

Future message–context congruency research couldexamine the influence of an intermediate level of congru-ency, rather than simply matched or not. For example, in ourstudy, the Health Organizations page had a relatively highlevel of congruency with the anti-binge drinking message.To reduce the degree of congruency between the persuasivemessage and the webpage context, the anti-binge drinkingmessage could be presented on a webpage describing healthorganizations not specifically related to discouraging drink-ing, like the university recreational center or counselingcenter. In this way, the message could match the topic ofhealth, but not match to the exact health-related behavior.This discrepancy between topics may be enough to violateexpectancy and increase message processing (Maheswaran& Chaiken, 1991).

Conclusions

Interventions discouraging undergraduate binge drinkingcan benefit from incorporating tailored messages and frompresenting the information in a context with which it hassome congruency. It is important to consider both thedrinking situation (going out or staying in) and how dif-ferent groups of students (e.g., males versus females) areinfluenced by different appeals across situations. Our find-ings further support message tailoring and demonstratethat situation/context and individual differences are amongthe numerous variables to consider in designing tailoredmessages and responsible drinking programs for college

students. Finally, our findings have important implicationsfor responsible drinking appeals to which students mayhave brief, one-time exposures. Given the additional dangersassociated with drinking while out (e.g., increased likeli-hood of drunk driving) compared to drinking while in, itis promising that including an anti-binge drinking messageon a webpage or in a newspaper advertising bar specialsmay be enough to influence students to reduce their drinkingintentions for that evening.

REFERENCES

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predictingsocial behavior. Englewood Cliffs, NJ: Prentice Hall.

Brannon, L. A., & Brock, T. C. (1994). Test of schema correspondencetheory of persuasion: Effects of matching an appeal to actual, ideal, andproduct “selves.” In E. M. Clark, T. C. Brock, & D. W. Stewart (Eds.),Attention, attitude and affect in response to advertising (pp. 169–188).Hillsdale, NJ: Lawrence Erlbaum Associates.

Brannon, L. A., & Brock, T. C. (2006). Measuring the protypicality ofproduct categories and exemplars: Implications of schema correspon-dence theory. In L. R. Kahle & C. H. Kim (Eds.), Creating images andthe psychology of marketing communication (pp. 31–46). Hillsdale, NJ:Lawrence Erlbaum Associates.

Brannon, L. A., & McCabe, A. E. (2002). Schema-derived persuasion andperception of AIDS risk. Health Marketing Quarterly, 20, 31–48.

Brock, T. C., Brannon, L. A., & Bridgwater, C. (1990). Message effective-ness can be increased by matching appeals to recipient’s self-schemas:Laboratory demonstrations and a national field experiment. In S. J.Agres, J. A. Edell, & T. M. Dubitsky (Eds.), Emotions in advertising:Theoretical and practical explorations (pp. 238–315). Westport, CT:Quorum Books.

De Pelsmacker, P., Geuens, M., & Anckaert, P. (2002). Media context andadvertising effectiveness: The role of context appreciation and context/adsimilarity. Journal of Advertising, 31, 49–61.

Dijkstra, A. (2008). The psychology of tailoring-ingredients in computer-tailored persuasion. Social and Personality Compass, 2, 765–784.

Eagly, A. H., & Chaiken, S. (1998). Attitude structure and function. In D.T. Gilbert, S. T. Fiske & G. Lindzey (Eds.), The handbook of socialpsychology, Vols. 1 and 2 (4th ed., pp. 269–322). New York, NY:McGraw-Hill.

Gleason, N. A. (1994). College women and alcohol: A relational perspec-tive. Journal of American College Health, 42, 279–289.

Hawkins, R. P., Kreuter, M., Resnicow, K., Fishbein, M., & Dijkstra, A.(2008). Understanding tailoring in communicating about health. HealthEducation Research, 23, 454–466.

Herr, P. M. (1989). Priming price: Prior knowledge and context effects.Journal of Consumer Research, 16, 67–75.

Hingson, R., Heeren, T., Winter, M., & Wechsler, H. (2005). Magnitudeof alcohol-related mortality and morbidity among U.S. college studentsages 18–24: Changes from 1998–2001. Annual Review of Public Health,26, 259–279.

Hustad, J. T. P., Barnett, N. P., Borsari, B., & Jackson, K. M. (2010). Web-based alcohol prevention for incoming college students: A randomizedcontrol. Addictive Behaviors, 35, 183–189.

Johnston, L. D., O’Malley, P. M., Bachman, J. G., & Schulenberg, J.E. (2008). Monitoring the Future national survey results on drug use,1975–2007. Volume II: College students and adults ages 19–45 (NIHPublication No. 08-6418B). Bethesda, MD: NIDA.

Lee, A. Y., & Labroo, A. A. (2004). The effect of conceptual and percep-tual fluency on brand evaluation. Journal of Marketing Research, 41,151–165.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014

MATCHING MESSAGES TO PERSON AND CONTEXT 309

Maheswaran, D., & Chaiken, S. (1991). Promoting systematic processingin low-motivation settings: Effect of incongruent information on pro-cessing and judgment. Journal of Personality and Social Psychology, 61,13–25.

Murray-Johnson, L., & Witte, K. (2003). Looking toward the future:Health message design strategies. In T. L. Thompson, A. M. Dorsey,Miller, K. I., & R. Parrott (Eds.), Handbook of health communication(pp. 473–495). Mahwah, NJ: Lawrence Erlbaum Associates.

Noar, S. M., Harrington, N. G., & Aldrich, R. S. (2009). The role of mes-sage tailoring in the development of persuasive health communicationmessages. Communication Yearbook, 33, 73–133.

Parrott, R., Monahan, J., Ainsworth, S., & Steiner, C. (1998).Communicating to farmers about skin cancer: The behavior adaptationmodel. Human Communication Research, 24, 386–409.

Pease, M. E., Brannon, L. A., & Pilling, V. K. (2006). Increasing selec-tive exposure to health messages by targeting person versus behaviorschemas. Health Communication, 19, 231–240.

Petty, R. E., Wheeler, S. C., & Bizer, G. Y. (2000). Attitude functions andpersuasion: An elaboration likelihood approach to matched versus mis-matched messages. In G. R. Maio & J. M. Olson (Eds.), Why we evaluate:Functions of attitudes (pp. 133–162). Mahwah, NJ: Lawrence ErlbaumAssociates.

Pilling, V. K., & Brannon, L. A. (2007). Assessing college students’ atti-tudes toward responsible drinking messages to identify promising bingedrinking intervention strategies. Health Communication, 22, 265–276.

Rimal, R. N., & Adkins, A. D. (2003). Using computers to narrowcasthealth messages: The role of audience segmentation, targeting, and tailor-ing in health promotion. In T. L. Thompson, A. M. Dorsey, Miller, K. I.,& R. Parrott (Eds.), Handbook of health communication (pp. 497–513).Mahwah, NJ: Lawrence Erlbaum Associates.

Sanbonmatsu, D. M., & Fazio, R. H. (1991). Construct accessibility:Determinants, consequences, and implications for the media. In J. Bryant& D. Zillmann (Eds.), Responding to the screen: Reception and reactionprocesses (pp. 45–62). Hillsdale, NJ: Lawrence Erlbaum Associates.

Shavitt, S. (1989). Operationalizing functional theories of attitude. In A.R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.), Attitude struc-ture and function (pp. 311–337). Hillsdale, NJ: Lawrence ErlbaumAssociates.

Shim, S., & Maggs, J. (2005). A cognitive and behavioral hierarchi-cal decision-making model of college students’ alcohol consumption.Psychology and Marketing, 22, 649–668.

Strahan, R., & Gerbasi, K. C. (1972). Short, homogeneous versions ofthe Marlowe–Crowne Social Desirability Scale. Journal of ClinicalPsychology, 28, 191–193.

Ten Wolde, G. B., Dijkstra, A., van Empelen, P., van den Hout, W.,Neven, A. K., & Zitman, F. (2008). Long-term effectiveness of computer-generated tailored patient education on benzodiazepines: A randomizedcontrolled trial. Addiction, 103, 662–670.

Treise, D., Wolburg, J. M., & Otnes, C. O. (1999). Understanding the “socialgifts” of drinking rituals: An alternative framework for PSA developers.Journal of Advertising, 28, 17–31.

Wall, A. F. (2007). Evaluating a health education web site: The case ofAlcoholEdu. NASPA Journal, 44, 692–714.

Walters, S. T., Vader, A. M., & Harris, T. R. (2007). A controlled trialof web-based feedback for heavy drinking college students. PreventionScience, 8, 83–88.

Wechsler, H., Lee, J. E., Kuo, M., Seibring, M., Nelson, T. F., & Lee, H.(2002). Trends in college binge drinking during a period of increasedprevention efforts. Journal of American College Health, Special Issue:Drinking on the college campus, 50, 203–217.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

11:

51 1

6 O

ctob

er 2

014