using cross-sectional surveys to plan message strategies · of using cross-sectional surveys to...

9
This article was downloaded by: [FHI 360], [Michael Williams] On: 16 September 2011, At: 07:08 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Social Marketing Quarterly Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/usmq20 Using cross‐‐sectional surveys to plan message strategies Robert Hornik a & Kimberly Duyck Woolf b a Professor of Communication, and holds the Wilbur Schramm Chair in Communication and Health Policy at the Annenberg School for Communication, University of Pennsylvania b Doctoral candidate at the Annenberg School for Communication, University of Pennsylvania Available online: 28 Apr 2010 To cite this article: Robert Hornik & Kimberly Duyck Woolf (1999): Using cross‐‐sectional surveys to plan message strategies, Social Marketing Quarterly, 5:2, 34-41 To link to this article: http://dx.doi.org/10.1080/15245004.1999.9961044 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan, sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Upload: others

Post on 08-Jul-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

This article was downloaded by: [FHI 360], [Michael Williams]On: 16 September 2011, At: 07:08Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Social Marketing QuarterlyPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/usmq20

Using cross‐‐sectional surveys to plan messagestrategiesRobert Hornik a & Kimberly Duyck Woolf ba Professor of Communication, and holds the Wilbur Schramm Chair in Communication andHealth Policy at the Annenberg School for Communication, University of Pennsylvaniab Doctoral candidate at the Annenberg School for Communication, University of Pennsylvania

Available online: 28 Apr 2010

To cite this article: Robert Hornik & Kimberly Duyck Woolf (1999): Using cross‐‐sectional surveys to plan message strategies,Social Marketing Quarterly, 5:2, 34-41

To link to this article: http://dx.doi.org/10.1080/15245004.1999.9961044

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching and private study purposes. Any substantial or systematicreproduction, re-distribution, re-selling, loan, sub-licensing, systematic supply or distribution in any form toanyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses shouldbe independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly inconnection with or arising out of the use of this material.

Page 2: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

34

CO

ü

OÜ«ļMcdi<ülOÍcoi

USING CROSS-SECTIONALSURVEYS TO PLAN

MESSAGE STRATEGIESBy Robert Hornik and Kimberly Duyck Woolf

INTRODUCTIONThis paper identifies methods

for using cross-sectional survey datato guide the design of the messagestrategies used in the communicationcomponent of a social marketing cam-paign. I t divides the process into twostages: deciding on potential messagestrategies to incorporate them into asurvey instrument and then analyzingthe survey results to provide guidancefor message strategy choice.

The analysis requires looking at theevidence to answer three questions:• Are there many people with the

'wrong' views on the message-relevant belief?

• Is there substantial associationbetween the belief and the outcome?and

• Is the belief one which an educa-tional campaign might affect?

The paper presents three illustrativemessage strategies relevant to cigarettesmoking among adolescents, then stepsthrough the analysis of some exampledata and describes some possibleconclusions. The paper concludes withan examination of the strengths andweaknesses of this approach.

There are many ways that empiricalresearch can help guide the developmentof social marketing programs and theircommunication components. They canbe used to define the target audience, tochoose focus behaviors, to select amongpossible channels, to provide feedback asto day-to-day successes or failures of theprogram as well as to provide fundamen-

tal evaluation of long-term success.However, among the most importantroles is assistance in the developmentof message strategies.

There are manyways that empiricalresearch can helpguide the developmentof social, marketingprograms and theircommunicationcomponents.

Cross-sectional surveys are onemethod for obtaining data to guide themessage strategy development process.Other methods include use of focusgroups, in-depth interviews, literaturereviews, expert informants on thequalitative side, or formal laboratoryor field experimentation on thequantitative side.

Mass communication projectsoperate at a distance from their audi-ences. This contrasts with conventionaleducational efforts. Teachers in aclassroom can hear the questions, readconfused expressions, and tailor theirlessons to meet the needs of theirstudents. I f they choose, health workersin a clinic can ask their clients to repeatthe recommendations for home treatmentof a diarrheal episode, to see whetheror not the clients have understood.

SMQ / VOL. V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 3: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

SOCIAL MARKETING QUARTERLY 3 5 .

APPLICATIONS

Communication projects may reach much largeraudiences and often are able to assure higher fidelity ofcore messages than classroom teachers or clinic staff.However, they do not have such easy channels forfeedback from their audiences.

Communication projects may reach muchlarger audiences and often are ableto assure higher fidelity of core messagesthan classroom teachers or clinic staff.However, they do not have such easychannels for feedback from their audiences.As a result, they have a special need forresearch and evaluation, for use indesigning their messages.

We assume that program planners havechosen a behavior and target audience andare ready to define message strategy. Bymessage strategy, we refer to the essentialbelief(s) that a message will be designed toimpart. I t is not the same as the messageitself, which will be the product of a creativeprocess that will turn the strategy into aspecific realization - whether that is atelevision ad, a printed booklet, or a radiosoap opera episode. The development of thestrategy is the intervening step betweenchoosing a focus behavior and messagecreation. I t wil l often be the meat of thecreative brief which producers will dependupon to guide message production.

THE PROCEDUREWe find i t useful to divide the process

of using cross-sectional surveys to assiststrategy development into two distinctstages: the design of the survey, andthe analysis of the survey to make recom-mendations of preferred strategies. Thedesign of the survey begins with thedevelopment of hypotheses about factorsthat might explain the behavior. Plannerscan rely on available theory, on the bestadvice of informants, on discussions withgroups representing the target audience -

or on their own judgment - and based onthese considerations, suggest a range ofpossible causes that they think mightinfluence audience behavior.

For example, in this paper we use desireto stop smoking among 12th grade currentsmokers as the target outcome (we knowthat the desire to stop smoking is highlypredictive of the behavior of frequentlytrying to quit smoking). Some mightsuggest that quitting cigarette smokingamong adolescents depends on their percep-tion of the negative health consequences ofsmoking. Others may argue that rigarettesmoking is a reflection of social expectationsfrom peers - individual use will vary withperceived peer smoking. Still others mightsuggest that overall attitude toward smok-ing, knowledge about ways to quit smoking,or perception of whether or not smokers areperceived as 'cool' is crucial. The process ofgenerating this list is really a creativeprocess, as is hypothesis generation, usually.There is no definitive way to know when thelist is finished, when all the importantpotential explanations have been offered.However, there are some useful guides forgenerating possible candidate explanations.

There is a great deal of codified experi-ence concerning likely determinants ofhealth behavior (for useful reviews seeFishbein, Middlestadt, & Hitchcock, 1991;Hornik, 1991; or Maibach & Parrott, 1995).The categories these authors suggestwill stimulate useful speculation aboutpossible explanations for many types ofbehaviors. One helpful empirical approachis what Ajzen and Fishbein (1980) call'elicitation research.' This process involvesquestioning small samples of respondents.

SMQ / VOL. V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 4: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

3 6 SOCIAL MARKETING QUARTERLY

APPLICATIONS

They are asked to describe what good andbad consequences are likely to accrue fromengaging in a particular behavior and tolist which of various influencers aroundthem would expect them to perform or notperform the behavior. The most consistentlymentioned consequences and influencersrepresent implicit hypotheses about whatmight influence the behavior, and would beadded to the questionnaire.

However, the development of the listof hypotheses for a communication strategyis subject to a constraint that is not presentin the usual listing of possible explanationsfor a behavior. These explanations musthave clear implications for the developmentof a message strategy. Almost always, thatconstraint will mean that each explanationwill touch on a respondent's beliefs orperceptions, since beliefs are the essenceof what messages address (whether thosebeliefs are about the benefits or costs of abehavior, the expectations of others for therespondent's behavior, or the respondent'sbeliefs about him or herself vis-a-vis thebehavior). As a contrasting example,even i f parental education is a predictorof willingness to quit smoking, i t has noobvious implication for a message strategy.

However, many hypotheses do carrysuch implications about message strategies.I f the perceived risk of health consequencesis key, then messages about the extent ofrisk might be expected to influence behav-ior. I f perceptions of peer behavior arecentral and i f others are, in fact, not usingcigarettes, messages emphasizing what peersare doing would be appropriate. Thus, thefirst task is to lay out the range of possibleinfluences on a behavior and their impliedmessage strategies.

This list should drive the design of thesurvey, since the largest number of ques-tions will be meant to indicate the extentto which the respondent believes or doesn'tbelieve in the truth of a statement. For thesake of this article, we will assume that thetarget audience is adolescents, and thetarget behavior is quitting smoking. Toillustrate the process of choosing amongmessage strategies, we will assume that

there are only three candidate strategies,one of which emphasizes the risk of a healthconsequence from smoking, the second ofwhich emphasizes a perception that smokingis the norm among friends and the third,which emphasizes general disapproval ofsmoking as a behavior. We use the datafrom the 1996 Monitoring the Future surveyof 12th grade students (Bachman, Johnston,& O'Malley, 1998), selecting those studentswho describe themselves as having smokedin the last 30 days. Our analysis focuses onthe four questions which follow, with theresponse categories in parentheses after thequestions. The first three questions embodyone of the three hypotheses, while thefourth question, asking about 'wanting tostop,' will be used as the outcome variable.

1) How much do you think people riskharming themselves (physically or inother ways) i f they smoke one or morepacks of cigarettes per day? [no risk,slight risk, moderate risk, great risk].

2) How many of your friends would youestimate smoke cigarettes? [none ofthem, a few of them, half of them, mostof them, all of them].

3) Do you disapprove of people (who are18 or older) smoking one or more packsof cigarettes a day? [don't disapprove,disapprove, strongly disapprove].

4) Do you want to stop smoking now?[Yes, No]

The survey was done with 2,466 12th

grade students in the Spring of 1996 by theUniversity of Michigan and the data weredownloaded from the website of the Insti-tute for Survey Research (Bachman et al.1998). Our choice of questions to analyzewas constrained by what was available on acommon form in that data set. To simplifypresentation, all data were recoded intodichotomous variables and presented ascross-tabulations. Only the 602 currentlysmoking respondents, who answered thequestion about wanting to stop smoking,were included in the tables. How wouldthe results of these analyses be turned intospecific message strategy advice?

SMQ / VOL. V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 5: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

SOCIAL MARKETING QUARTERLY 3 7

APPLICATIONS

TABLE 1 - Wanting to Stop Smoking and Belief that Cigarettes are Harmful

Do you wantto stopsmoking?

No

Yes

%(N)

Perceived risk that people will harm themselvesif they smoke one or more packs per day.

None or slight risk

80%

20%

15% (85)

Moderate orgreat risk

58%

42%

85% (483)

%(N)

61%(349)

39%(219)

568

A message strategy is promising insofaras i t satisfies three criteria:

1) There are a substantial number of peoplewho are not in the desired position onthe message strategy-relevant variable.

2) There is a substantial relation betweenthe message strategy-relevant variableand the outcome variable. Differentvalues on the message strategy-relevantvariable should predict who will and whowill not engage in the desired behavior.

3) I t will be feasible to move the targetaudience on the message strategy-relevant variable. I t is a beliefthat might be learned from aneducational campaign.

For each of the three candidate strate-gies, we can address each of those ques-tions. A dichotomized version of eachvariable is used in Table 1, Table 2 orTable 3, which present results for therisk, normative and general disapprovalstrategies, respectively.

EVALUATION OF STRATEGIESWe begin with the risk strategy.

The first criterion is whether there are asubstantial number of people who hold the'wrong' belief or whether the target popula-

tion already holds the 'correct' belief. InTable 1, the answer is fairly clear. Eighty-five percent of the respondents alreadyrecognize the risk of cigarette use. Thispotential strategy doesn't look promisingon those grounds, since there are only15% of the population who hold the'wrong' position.

We go on to consider the secondcriterion: Does status on the messagestrategy variable actually predict behavior(or in this case intention to undertakea behavior)? Here, the data are moreoptimistic; while 20% of those who don'tperceive much risk want to stop smoking,42% of those who do perceive the risk wantto stop. A typical way of summarizing thisrelationship is the relative odds ratio. Thisis calculated by comparing the ratios in thetwo columns: (.42/.58)/(.20/.80) which inthis case is equal to 2.90. Thus the oddsof wanting to quit smoking are 2.9 timesgreater i f one perceives the risk of smokingthan i f one doesn't.

The final criterion is whether or notthe predictor belief is 'movable.' In thejudgment of the planner, will i t be possibleto convince people of the risk of harm fromsmoking? In the end, this is a judgmentcall. On the one hand, the fact that manypeople already hold this belief makes i tpromising on this criterion. I f 85% of

SMQ / VOL. V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 6: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

3 8 SOCIAL MARKETING QUARTERLY

APPLICATIONS

the population already believe that the riskis substantial, shouldn't i t be possible toconvince the others? A contrary view wouldbe that i f there was enough public informa-tion in the environment to convince 85% ofthe population, the 15% who were notconvinced must be quite resistant tomessages. On those grounds, a new set ofmessages might not be expected to behelpful. We return to deciding whether ornot this strategy is promising after weexamine the remaining strategies.

This analysis of the first messagestrategy contrasts with the sense one canmake of the data in Table 2, which examinesthe association between normative percep-tions and willingness to quit smoking.

Against the first criterion, whether ornot there is anyone left to convince, thispredictor variable fares much better than thefirst. Sixty-three percent of this sample ofsmokers believe that most of their friendsare smoking. Against the second criterion,the presence of a substantial associationbetween the belief and the outcome (inten-tions), this strategy doesn't look very good:39% of those who have few friends smokingintend to quit themselves; virtually thesame proportion, 37%, of those who havemany friends smoking intend to quit.Again, this can be summarized as relativeodds, the ratio of the numbers in each

column (.39/.61)/(.37/.63), equal to 1.09,essentially showing no relationship.

This lack of a relationship suggests thata social expectations strategy would not beso promising. However, we examine thefinal criterion, whether or not people aremovable on the targeted belief. As before,judgment comes to the fore. Even i f therewas a relationship, is i t reasonable to expectto convince people that their friends do notsmoke when their direct perceptions are thatthey do? I f the respondents are accuratelyreporting their close friends' behavior, i tseems doubtful that any communicationstrategy will convince them otherwise.

There are two substantive commentsabout this result worth noting. On the onehand, despite the lack of relationship withdesire to quit smoking here, in data notpresented here, friends' smoking behavioris highly related to respondent's currentsmoking behavior. This suggests that peerinfluences are worth attention i f the issue isinitiation of smoking. I t may also be thatfriends' smoking behavior is not the same aspeer expectations that the respondent stopsmoking. Friends may be saying that therespondent ought to quit, even i f they aresmoking themselves. I f we had a measureof perceived expectation of others that onestop smoking, there might have beena relationship.

Table 2 - Wanting to Stop Smoking and Friends' Smoking

Do you wantto stopsmoking?

No

Yes

% (N)

Belief that friends smoke

Most of them,all of them.

63%

37%

63% (334)

None of them,a few of them,some of them

61%

39%

37% (194)

%(N)

62%(327)

38%(201)

528

SMQ / VOL. V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 7: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

SOCIAL MARKETING QUARTERLY 3 9

APPLICATIONS

Table 3 - Wanting to Stop Smoking and Disapproval of Smoking

Do youwant tostop smoking?

No

Yes

% (N)

Disapproval of people smoking oneor more packs of cigarettes a day

Don't Disapprove

67%

33%

73% (284)

Disapprove, StronglyDisapprove

54%

46%

27% (103)

%(N)

64%(246)

36%(141)

387*

*Note: Disapproval questions asked only of a subset of the sample.

A final example considers the issue ofgeneralized disapproval of smoking. Table 3presents the data. Against the first crite-rion; having some members of the audiencewho hold the 'wrong' view, this predictordoes well, with 73% who 'don't disapprove,'perhaps not surprising for a sample ofsmokers. Against the second criterion,association with desire to stop smoking,the variable does only moderately well,with 33% of those who don't disapprove and46% of those who do disapprove wanting toquit. The relative odds would be (.46/.54)/(.33/.67)= 1.73, not as strong as in the firsttable, but stil l of some size. For the lastcriterion, 'movableness,' many messageplanners would be hopeful. While mostsmokers don't disapprove, the fact thatone quarter of them do disapprovesuggests that i t is something that evensmokers might accept.

So against the three criteria, the threemessage strategies fare differently. The firststrategy is probably not promising: While i thas the strongest relationship with behavior,there are few people left unconvinced of thisbelief to promise much reward. This issupported by an estimate of how muchchange in the population would occur i f acampaign focusing on this message strategywas completely successful and everyone

believed the statement. In Table 1, 39% ofall of the respondents want to stop smoking.Imagine that every one of the people whodid not believe in the risk of smoking becameconvinced, and looked just like the peoplewho are believers in Table 1. Then, we wouldexpect that 42% of them would intend toquit. The maximum change we could expectfrom a completely successful implementationof this strategy would be a change from 39%to 42% wanting to quit, a minimal change.

The second strategy is quite promisingon the first criterion, lots of people tochange, but i t fails on the second, sincethere is no relationship. Also, if, in reality,friends are smoking, i t would seem difficultto imagine messages which would producechange in that belief.

The third strategy is perhaps the mostpromising, given the data. There are alarge number of people to move, there isa moderate relationship between the beliefand the intended behavior, and i t maybe possible to develop a set of messagesthat wil l increase disapproval of smoking.Complete success would mean there wouldbe 46% rather than 36% of smokers wantingto quit now, a worthwhile change.

Clearly, these analyses are only illustra-tive, and they are constrained by thesecondary data available to us, thus the

SMQ / VOL. V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 8: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

4 0 SOCIAL MARKETING QUARTERLY

APPLICATIONS

choices of hypotheses we could test.Certainly, other potential explanations forbehavior, and thus other message strategies,might have been tested had we started theprocess from scratch. Nonetheless, they givesome notion of what the process of analysisand interpretation might look like whileusing widely available data.

If 85% of the populationalready believe that therisk is substantial,shouldn't it be possibleto convince the others?A contrary view would bethat if there was enoughpublic information in theenvironment: to convince85% of the population, the15% who were notconvinced must be quiteresistant to messages.

IMPLICATIONS FORSOCIAL MARKETING

This analysis illustrates the process amessage planner might go through if he orshe wanted to exploit survey data to informmessage strategy choice. The strengths ofthis approach are substantial. By laying outhypotheses explicitly and subjecting them tosystematic empirical tests on a representa-tive sample of the population, one willproduce evidence that can help eliminatesome potential approaches and providesupport for others. Surveys provide evidencethat is somewhat independent of theobserver, reducing dependence on a priorisuppositions about the audience that mayhave been based on anecdotal evidence orinsights derived from conversations withunrepresentative informants.

There are some limitations to thisapproach, of course. The implication of this

analysis is that i f one changes a predictorbelief, one can expect change in thepredicted outcome. But, these are cross-sectional data, and one ought not confuseobserved association with claims of causa-tion. There remains a risk that some otherconfounding factor accounts for an observedrelation or that there is reverse causation -with intentions to stop smoking causingbeliefs, rather than vice versa. The presenceof an association cannot guarantee that achange in the predictor belief will produce achange in the outcome.

There are weaknesses in this approach,surely. They constrain one's confidencein the recommendations that derive fromthe analysis. However, the issue here isnot whether this analysis will provide anincontrovertible foundation for messageplanning. The criterion for assessing theworth of this approach is whether i t im-proves judgment, making the planner lesslikely to go off in a problematic direction.I t should be valuable, in this way, at bothstages. The process of generating candidatestrategies will force planners to be explicitabout what they might consider. Theprocess of analyzing the results againstestablished criteria will lead to the rejectionof some approaches and support for thecandidacy of others.

Unlike some qualitative approaches,for example, focus groups or in-depthinterviews, survey research may not be souseful in generating the first list of ideasfor possible message strategies. Indeed, i twill often make sense to delay survey workuntil exploratory research is complete andcan be used as the basis for the developmentof candidate message strategies. However,the survey approach has the advantage ofbeing systematic and offers the ability toexamine candidate strategies on a represen-tative sample of the target population.

In sum, cross-sectional survey researchcan both inform and be informed by thejudgment of the planners. We think that i tpromises to improve the quality of messagestrategy planning.

SMQ / VOL V / NO. 2 / JUNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11

Page 9: Using Cross-Sectional Surveys to Plan Message Strategies · of using cross-sectional surveys to assist strategy development into two distinct stages: the design of the survey, and

SOCIAL MARKETING QUARTERLY 4 1

APPLICATIONS

REFERENCESAJZEN, I . & FISHBEIN, M. (1980). Understanding attitudes and predicting social behavior.Englewood Cliffs, NJ: Prentice Hall, Inc.

BACHMAN, J. G., JOHNSTON, L. D. & O'MALLEY, P. M. (1998). Monitoring the Future: Acontinuing study of American youth (12th grade survey) 1996 [Computer fi le]. Conducted byUniversity of Michigan, Survey Research Center. ICPSR ed. Ann Arbor, MI: Inter-UniversityConsortium for Political and Social Research [Producer and Distributor].

FISHBEIN, M., MIDDLESTADT, S. E. & HITCHCOCK, P. 3. (1991). Using information to changesexually transmitted disease-related behaviors: An analysis based on the theory of reasonedaction. In J.N. Wasserheit, S.O. Aral & K.K Holmes (Eds.), Research issues in human behaviorand sexually transmitted diseases in the AIDS era (pp. 243-257). Washington, DC: AmericanSociety for Microbiology.

HORNIK, R. (1991). Alternative models of behavior change. In J.N. Wasserheit, S.O. Aral &K.K Holmes (Eds.), Research issues in human behavior and sexually transmitted diseases inthe AIDS era (pp. 201-218). Washington, DC: American Society for Microbiology.

MAIBACH, E. & PARROTT, R. L. (Eds.). (1995). Designing health messages: Approaches fromcommunication theory and public health practice. Thousand Oaks, CA: Sage.

ABOUT THE AUTHORSRobert Hornik, Ph.D., is a Professor of Communication, and holds the Wilbur Schramm Chairin Communication and Health Policy at the Annenberg School for Communication, Universityof Pennsylvania. He has led efforts to design and evaluate large-scale public health communi-cation and education programs. Some major projects for which Hornik has been principalinvestigator include USAID-sponsored evaluations of national AIDS education programs in fourdeveloping countries, of communication for child survival programs in 10 developing countries,and CDC-sponsored research on determinants of immunization status in Philadelphia. He iscurrently co-principal investigator and scientific director for the evaluation of the NationalYouth Anti-drug Media Campaign, as well as principal investigator on two evaluations ofdomestic violence prevention projects, one in Philadelphia, and one nationwide.

Kimberly Duyck Woolf, M.A., provided research assistance with this paper. She is adoctoral candidate at the Annenberg School for Communication, University of Pennsylvania.Her research interests include developmental differences in comprehension of media messagesand the effects such messages have on children's attitudes and behavior.

Note: The authors are grateful for the helpful comments from Martin Fishbein.

SMQ / VOL. V / NO. 2 / 3UNE 1999

Dow

nloa

ded

by [

FHI

360]

, [M

icha

el W

illia

ms]

at 0

7:08

16

Sept

embe

r 20

11