using communication research to design effective messages for public health: the cases of hpv...

Post on 14-Dec-2015

215 Views

Category:

Documents

1 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Using Communication Research to Design Effective Messages for

Public Health: The cases of HPV vaccine and anti-smoking PSAs

1

Joseph N. Cappella

Annenberg School for Communication

University of Pennsylvania

Effects of Public Information in Cancer

an NCI Center of Excellence in Cancer Communication Research

at the Annenberg School for Communication, University of Pennsylvania

Presented OBSSR, NIH

June 15, 2009

Message: Effects versus Design

• Effects: content and consequents

• Design:– Components– Analysis– Chosen: theoretically & based on real world

content– Target audience: effects– Translation: Re-engineer for public health

• Avoid• Select

2

Message Components

• Content: what the message is about– Persuasive

• Claims• Offer reasons for and reject reasons against• Implicit or explicit

• Format: how the “what” is presented– Equivalence

• Logical• Topical, conceptual, propositional

3

Message Effects via Design

• Choose content & format

• To achieve acceptance – (by target audience)

4

Conventional (theoretical) Wisdom

• Strong arguments (involved targets)

• Attention-getting messages

• Public education overcomes resistance– Psychological – Social

5

For Example

• Content– Strong, weak arguments (Petty & Cacioppo, 1986)– Altruistic versus self-interested

• Format– Framing: effective 70%; ineffective 30%– Sensation value

6

Goals for Today

• Are hi sensation value PSAs effective for smokers?

• Are PSAs using smoking cues effective?

• Are there arguments about smoking that appeal to all smokers?

• How do health messages about the HPV vaccine interact with public discourse about the vaccine?

7

MSV X AS(Strasser, Cappella et al 2009)

• H: Outcomes enhanced for Hi-hi

• Especially for hi SS smokers

• Physio, intention, beliefs re smoking

Message Sensation Value(Morgan et al, 2003)

• Content: Acted out, narrative, unexpected format, twist ending,

• Visual: cuts, edits, faces, unusual colors, intense moments (implicit, explicit)

• Auditory: music, sound saturation, unusual sounds, esp slow or fast voices

• Validity? 9

DESIGN

• 200 adult smokers (lowest M=19/day)

• MSV (2) X Arg Strg (2) X SS (2)

• 4 ads per condition (16)

• Careful selection from set of 600, 100 tested

• Method: real ads, multiple per condition

Key Results

• For high MSV – Corrugator Hi > Lo (p < .02)– Personal efficacy

• (low SS) Lo > Hi (p < .04); • (hi SS = lo SS, hi MSV)

• For high AS ads– Skin conductance strong > weak (p = .05)– Heart rate (strong > weak) (p=.02)

Believable MSV?

• Subtle test

• Recall accuracy

• Recall RT

• Brain response

fMRI, MSV, & Recall(Langleben, Loughead, Hakun, Ruparel, Strasser, Halloway, Cappella, & Lerman,

2009)

• Sample: 18 regular smokers, 18-48, 13 cigs/day, 12 M

• PSAs: 8 pre-selected from set of 99; anti-smoking

• Design: 2 arg strength X 2 MSV, w/i subjects, 2 examples of each

• Procedures: 8 PSAs, 8 control, pseudo-random order, recognition

16

Results: Recognition

Results: Brain response• The hiMSV PSAs were associated with extensive

activation in the occipital (including the fusiform gyrus) cortex and the parahippocampus, while the loMSV PSAs were accompanied by higher orbitofrontal, superior and inferior frontal, temporal and posterior parietal activation.

• Thus, the activations associated with loMSV PSA are suggestive of deeper cognitive processing than the hiMSV PSA. Moreover, absence of differences between the loMSV and hiMSV in the anterior cingulate cortex and higher posterior parietal activation suggests that higher MSV does not translate to higher endogenous attention.

Conclude

• High MSV can distract

• Cognitive resources to irrelevant features

• Consistent with Kang, Cappella, & Fishbein (2006)

19

Kang, Cappella & Fishbein 2006• 60 Anti-marijuana Ads • Adolescents • Two ad features

– Message sensation Value (MSV, Morgan et al., 2003)

– Argument strength• MSV -- attention

– Focus on argument, stronger argument more persuasive

– Focus on stylistic features, distracter, weak argument more persuasive

– MSV as a moderator of argument quality on ad effectiveness

20

Key Finding

21

Why?

• Dominant thoughts disrupted by hi MSV ads

• MSV features distracting from core processing

• Which features?

• Under what conditions?

• What else could distract?

22

Smoking Cues

• In anti-smoking PSAs

• 40% roughly

• When outside PSAs create urge

• What about inside PSAs– When args vs smoking are strong – Approach -- avoid

23

Methods

• Participants– Screening criteria– N=96, 54% male, age =33, 14

yrs edu, 59% Caucasian, 17 cig/day, 29 days smoking in the previous 30 days

– N=82 follow-up

Argument Strength (between-subject)

Smoking Cue (within-subject)

No Yes

Strong 3 PSAs 3 PSAs

Weak

3 PSAs 3 PSAs

Yahui KangKang, et al, 2009

Smoking Urge

3.103.153.203.253.303.353.403.453.503.55

pre no-cueads

post no-cue ads

pre smk-cue ads

post smk-cue ads

Smoking Cue

Smoking Urge

Weak Argument

Strong Argument

-0.40

-0.30

-0.20

-0.10

0.00

0.10

0.20

0.30

0.40

no-cue ads smoking-cue ads

Smoking Cue

Effectiveness Index

Weak Argument

Strong Argument

Learning

4.20

4.25

4.30

4.35

4.40

4.45

4.50

4.55

4.60

no-cue ads smoking-cueads

follow-up

Time

Ad Learning from Smoking-cue

Ads

Weak Argument

Strong Argument

Eye Tracking Smoking Cues

28

Design

• 3(I-squared high vs. I-squared low vs. smoking) x 43 (repeats of cues in each category taken from the 16 PSAs available)

• Participants (N=84)– 44 male– 47 African American, 27 white– Age 18 to 65 years, mean=36.9

30F(2,62)=3.59, p<0.05Sanders-Jackson, et al, unpublished

Conclusions

• Variance of looking is less with smoking cues (and hi Information Introduced, I2)

• Cue—Visual Correlation:– Mean Pearson r = .37 (range: -23 to .98)

• For active smoking scenes (more urge), hi arg strength lowers attention to cues (smaller r)

31

Next Steps

• Three cue (none, peripheral, central) X two argument (high, low) factorial, smokers– Urge, physio, eye tracking (Strasser, Lerman &

Cappella, CECCR II)

– fMRI (Loughead, NCI CECCR II)

• Former smokers

32

The Problem with Arguments

• Argument strength important in acceptance in previous work

• Strong-weak is rated by smokers

• No a priori predictions

• Are there structural diffs between effective and ineffective anti-smoking arguments?

33

Anti-smoking Arguments

• Archive of > 1000 ads• 199 selected: English, 30 sec, adult targets, neg

consequences, treatment seeking

• Argument “extraction”• 8 item arg strength (Zhao et al under review)

• 2004– 99 arguments, 300 adult smokers, 12 of 99 random, mall

intercept

• 2008– 100 args, 487 adult smokers, 8 of 100 random, KN sample

34

Young Min Baek

Data• Texts of arguments (10 predictors)

– Automap (Carley) and LIWC (Pennebacker)– Synonym sets

• Individual differences (9 predictors)– Demographic characteristics (5)– # of cigarettes/last seven days– Need for cognition– Perceived vulnerability (2)– SOC, intention to quit combined

35

36

 2004 survey

(n = 99)2008 survey

(n = 100)  

Word-category Mean SD Mean SD test statistic

People (Intimate) .52 .80 .56 .96 t = .39

People (Distant) .55 .95 1.21 1.37 t = 3.98***

Smoking cessation .58 .77 .16 .47 t = -4.61***

SHS (secondhand smoking) .08 .27 .22 .50 t = 2.42*

Causation words 1.21 1.05 .92 1.18 t = -1.84

Death-related words .40 .65 .49 .77 t = .85

Disease/Body .77 1.08 1.00 1.28 t = 1.39

Poison/Chemical .16 .55 .39 .94 t = 2.09*

Tobacco Company .19 .55 .08 .31 t = -1.78

Lifestyle/Cosmetics .60 1.04 .28 .57 t = -2.66**

Argument strength evaluated 3.74 .30 3.24 .27 t = 12.25*Note. * p < .05 ** p < .01 *** p < .001. 99 arguments are evaluated in 2004; and 100 arguments in 2008.

Data Analysis Strategy

• Arg, Person, Arg X Person AS

• MLM, specifically HCM

• Retention of interaction terms only if replicated

37

Key Findings

38

Variance(in rated AS)

• Mostly individual (49%)

• Significant argument (9%)

• Little interactions (< 1%)

• Implications:

Argument Effects

• Positive

– People (Close)

– SHS

– Death

• Negative– People (distant)– Cosmetic

(lifestyle)

42

Person X Argument

• SHS– For females (+)– For younger (+)

• Disease/body– For educated (+)

• Chemical/Poison– For educated (+)

43

Person Effects

• Stage of Change (+++)

• Perceived vulnerability to disease (++)

44

Conclusions

• Argument strength highly individualized but – Early SOC– Perceived invulnerability to smoking harm

• Effective arguments focus on – intimates, death, and second hand smoke

• Some targeting: – Young & women: SHS– Educated: disease, poison, chemicals 45

Other Studies

• Emery Collaboration: into the field• Death PSAs: PSAs with death themes

– more fear, perceived risk & effectiveness (adults)

• Narrative & efficacy (print news): – transportation intention to quit

• Fear X efficacy intention to quit– both nec’y for early SOC– efficacy only for later SOC

46

HPV

• Resistance is social, political (also)

• Rolling cross section re HPV– One year, rep sample, monthly– Knowledge and news coverage

• Two experiments– HPV, STIs, promiscuity, rep sample– HPV, pos-neg frame, rep sample

Conclusions re HPV

• Public and public health agendas diverged

• Message framing mattered

• Ideology mattered

• Ideology affected message interpretation

The Public Debate

• Kelly et al

• Media coverage (+) knowledge

HPV Vaccine

• The FDA approved June 8, 2006

• Vaccine prevents infection against 4 strains of HPV

• Recommendation: vaccination to girls and women ages 9-26

• The vaccine effective only if prior to infection with HPV

HPV Vaccine Debate

• Being vaccinated allows for increased and riskier sexual behavior

• Anti-cancer vs. Anti-STI vaccine

• Voluntary vs. Mandatory vaccination

• Liberals in Congress and elsewhere have warned that the Bush administration and religious groups should not interfere with Gardasil's approval or required use. In response, many conservative groups have made statements supporting the vaccine.

• "Despite rumors to the contrary, our organization doesn't oppose the vaccine and we have taken no position regarding mandatory laws," said Wendy Wright, president of Concerned Women of America, a conservative group based in Washington.

• Some groups support the vaccine but oppose mandatory vaccinations because cervical cancer is caused by a sexually transmitted virus.

• "We can prevent it by the best public health method, and that's not having sex before marriage," said Linda Klepacki of Focus on the Family, a Christian advocacy organization based in Colorado Springs.

U.S. Approves Use of Vaccine for Cervical Cancer By GARDINER HARRIS

• Groups wary of drug industry motives find themselves on the same side of the anti-vaccination debate with unexpected political allies: religious and cultural conservatives who oppose mandatory use of the vaccine because they say it would encourage sexual activity by young girls.

• And in Illinois, a bill introduced by a legislator who had the virus the vaccine is intended to prevent prompted a conservative group’s blog to speculate that she had been promiscuous.

• “I’m offended by their ignorance, but if I have to take a hit to educate people, I’m willing to do it,” said the bill’s sponsor, Debbie Halvorson, the Democratic majority leader in the Illinois Senate.

| February 17, 2007 Furor on Rush to Require Cervical Cancer Vaccine By STEPHANIE SAUL and ANDREW POLLACK

Fox News

• PROPOSALS FOR MANDATES HAVE POPPED UP FROM CALIFORNIA TO CONNECTICUT SINCE THE FIRST PIECE OF LEGISLATION WAS INTRODUCED IN SEPTEMBER IN MICHIGAN. MICHIGAN'S BILL WAS NARROWLY DEFEATED LAST MONTH. LAWMAKERS SAID THE REQUIREMENT WOULD INTRUDE ON FAMILIES' PRIVACY, EVEN THOUGH, AS IN MOST STATES' PROPOSALS, PARENTS COULD OPT OUT

• EVEN WITH SUCH OPT-OUT PROVISIONS, MANDATES TAKE AWAY PARENTS' RIGHTS TO MAKE MEDICAL DECISIONS FOR THEIR CHILDREN, SAID LINDA KLEPACKI OF THE COLORADO-BASED EVANGELICAL ORGANIZATION FOCUS ON THE FAMILY. THE GROUP CONTENDS THE VACCINE SHOULD BE AVAILABLE FOR PARENTS WHO WANT IT, BUT NOT FORCED ON THOSE WHO DON'T.

Colbert Report

• Vision America

• If you have sex, God will give you cancer.

• http://www.comedycentral.com/motherload/index.jhtml?ml_video=82235

Religious Groups Respond• Nor we can not overlook the moral dimension,"

Scarborough cautioned, ungrammatically, in a press release. "The governor's action seems to signify that God's moral law regarding sex outside of marriage can be transgressed without consequence."

• Family Research Council, which sponsored last fall's Values Voter Summit, claimed in its e-letter today that Texas had "erupted in protest," and couched its opposition to the vaccine as a parents-rights issue.

• "[T]he issue at hand is not whether to make the drug available to young women — few would argue otherwise — but whether or not it should be a requirement of school attendance for schoolgirls and who should literally call the shots," wrote FRC president Tony Perkins. "Parents should not have to 'opt-out'; rather they should be able to 'opt-in' their daughters for the vaccination."

Methods

• Annenberg National Health Communication Survey, a monthly online survey with a nationally-representative sample of adults

• Participants selected through RDD, one participant per household

• Survey ran in June 2006

• Survey completion rate= 74%

Methods

• Survey consisted of three parts:– Experimental manipulation of information– Attitudes toward HPV vaccine policies

• Participants were randomized to read one of three versions (framing manipulations) describing the vaccine

• Sample: N=635; 51% female; 75% non-Hispanic white, mean age=47.6 yrs

Framing Manipulations

HPV causes cervical cancer

HPV is a sexually transmitted infection

The vaccine may/may not lead to increased sexual promiscuity

Message 1 Message 2 Message 3

Intentions to Vaccinate(1= very unlikely, 5= very likely)

PAY, p< .20NO COST, p< .005

Policy, Framing Version, and Ideology

Ideology Mean (std)

Cervical Ext. Conservative

Conservative

Moderate

Liberal

Ext. Liberal

2.71 (1.496)

3.34 (1.295)

3.77 (1.150)

3.74 (0.976)

4.00 (1.732)Vaccines for the Uninsured

STI Ext. Conservative

Conservative

Moderate

Liberal

Ext. Liberal

3.57 (1.618)

3.66 (1.200)

3.65 (1.131)

3.67 (1.226)

4.00 (1.001)

Sexual Ext. Conservative

Conservative

Moderate

Liberal

Ext. Liberal

2.08 (1.379)

2.78 (1.488)

3.60 (1.218)

4.14 (1.229)

4.43 (0.787)

Ideology: p < .001

Intention to use: Ideology X Frame

1

1.5

2

2.5

3

3.5

4

4.5

5

Ext Con Con Mod Lib Ext Lib

Ideoology

Intention

Cervical

STI

Sexual

Attribute Framing

• HPV vaccine

• Media and Public Health Frames varied

• Affect judged effectiveness?

• Affect policy judgments?

Cabral Bigman

Framing

Media coverage -- positive attribute frame:

“A vaccine licensed by the Food and Drug Administration in June protects against two strains of HPV that cause nearly 70 percent of cervical cancer cases. The same vaccine, Gardasil, produced by Merck &. Co., also prevents two other strains linked to 90 percent of genital wart cases.”

(Howard Price, J. 2007, Feb 28, “Nearly 25% of females ages 14-59 have HPV, study finds” The Washington Times)

Framing

Media coverage -- negative attribute frame:

“Millions of women have annual Pap smears to test for cervical cancer, and tens of thousands undergo further expensive testing and procedures after receiving false positive tests. Such testing will continue in part because the vaccine’s preventive effects are years away but also because Gardasil does not protect against viral strains that cause up to 30 percent of cervical cancers.”

(Harris, G., 2006, June 30, Panel Unanimously Recommends Cervical Cancer Vaccine for Girls 11 and Up, The New York Times)

Framing

Media coverage -- mixed attribute frame

“Called Gardasil, the vaccine has been proved to prevent two types of human papilloma virus infection, the strains responsible for 70 percent of cervical cancers. It is also effective against about 90 percent of the HPV strains that cause genital warts. Even then, Gardasil cannot protect against 30 percent of HPV strains that cause cervical cancer, so regular Pap smears are still recommended for cancer screening.”

(Griffith, D. 2006, Aug. 31, Some parents not sold on cervical cancer shot. Sacramento Bee)

How Information Presented

2007 Data

F(4, 323)=5.91, <.001Pos > Cont > Neg, p< .05

Framing Effects for Mandate and Non-mandate Related Items.

Note. Black bars reflect t-test differences of p≤.05.

2007 & 2008

78

Framing Effects for Mandate Opinion Support by Ideology and Year

Note. Framing effects reflect the mean difference between positive and negative frame conditions with prior mandate opinion as a covariate. 

• Amy Leader

• Cabral Bigman

• Caryn Lerman

• Hyun Suk Kim

• Jean Brechman

• Chul-joo Lee

Collaborators

• Amy Leader• Cabral Bigman• Young Min Baek• Caryn Lerman• Hyun Suk Kim• Jean Brechman• Chul-joo Lee

• Andrew Strasser• Alyssa Bindman• Heather Forquer• Mario Giorno• Yahui Kang• Marty Fishbein• Robert Hornik

CECCR PI

Lessons

• Conventional wisdom

• Resistance: social & psychological

• Design: art & science

• Science: long way to go

• But …

top related