Social Science & Medicine 98 (2013) 204e213

Contents lists available at ScienceDirect

Social Science & Medicine

journal homepage: www.elsevier.com/locate/socscimed

Potential effectiveness of anti-smoking advertisement types in ten low

and middle income countries: Do demographics, smoking
characteristics and cultural differences matter?
Sarah Durkin a, Megan Bayly a, Trish Cotter b, Sandra Mullin b, Melanie Wakefield a, *
a
Cancer Council Victoria, 1 Rathdowne Street, Carlton, VIC 3053, Australia
b
World Lung Foundation, 61 Broadway, Suite 2800, NY 10006, USA

a r t i c l e i n f o a b s t r a c t

Article history: Unlike high income countries, there is limited research to guide selection of anti-tobacco mass media
Available online 3 October 2013 campaigns in low and middle income countries, although some work suggests that messages empha-
sizing serious health harms perform better than other message types. This study aimed to determine
Keywords: whether certain types of anti-smoking advertisements are more likely to be accepted and perceived as
Tobacco effective across smokers in 10 low to middle income countries. 2399 18e34 year old smokers were
Low and middle income countries
recruited in Bangladesh, China, Egypt, India, Indonesia, Mexico, Philippines, Russia, Turkey and Vietnam
Mass media campaigns
to view and rate 10 anti-tobacco ads. Five ads were shown in all countries and five ads were chosen by
Adults
Cultural congruence
country representatives, providing a total of 37 anti-smoking ads across all countries (10 graphic health
effects ads, 6 simulated health effects, 8 emotional stories of health effects, 7 other health effects and 6
non-health effects). Smokers rated ads on a series of 5-point scales containing aggregated measures of
Message Acceptance and Perceived Effectiveness. All ads and materials were translated into the local
language of the testing regions. In multivariate analysis, graphic health effects ads were most likely to be
accepted and perceived as effective, followed by simulated health effects ads, health effects stories, other
health effects ads, and then non-health effects ads. Interaction analyses indicated that graphic health
effects ads were less likely to differ in acceptance or perceived effectiveness across countries, gender, age,
education, parental status and amount smoked, and were less likely to be affected by cultural differences
between characters and contexts in ads and those within each country. Ads that did not emphasize the
health effects of smoking were most prone to inconsistent impact across countries and population
subgroups. Graphic ads about the negative health effects of smoking may be most suitable for wide
population broadcast in low and middle income countries.
Ó 2013 Elsevier Ltd. All rights reserved.

Introduction supported by FCTC Article 12 on public education and training

Over 80% of the world’s adult male smokers now live in low and
middle income countries (Eriksen, MacKay, & Ross, 2012). In order
to reduce global smoking prevalence as rapidly as possible, it is
imperative to fully implement comprehensive evidence-based to-
bacco control strategies, such as those articulated in the WHO
Framework Convention on Tobacco Control (FCTC). Anti-tobacco
mass media campaigns increase quitting and reduce smoking
prevalence in high-income countries (Durkin, Brennan, &
Wakefield, 2012; National Cancer Institute, 2008) and are
* Corresponding author. Centre for Behavioural Research in Cancer, Cancer
Council Victoria, 1 Rathdowne Street, Carlton, VIC 3053, Australia.
E-mail address: melanie.wakefield@cancervic.org.au (M. Wakefield).
(World Health Organization, 2003). Mass media campaigns directly
prompt quitting in adult smokers, as well as reduce the broader
social acceptability of smoking and build public support for the
implementation of other key tobacco control policies (Durkin et al.,
2012; Wakefield, Loken, & Hornik, 2010).
Recent reviews highlight that the effectiveness of anti-tobacco
mass media campaigns depends upon the types of messages used
and the extent to which campaigns can be funded to consistently
reach the majority of the population (Durkin et al., 2012; National
Cancer Institute, 2008). Population-based and advertisement (ad)
rating studies of different message themes in high income coun-
tries indicate that negative health effects messages, most of which
feature graphic imagery (visceral pictures of external and internal
body parts riddled with smoking related disease), testimonial
stories and/or elicit negative emotions, tend to be more effective

0277-9536/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.socscimed.2013.09.022
 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213
than other message themes (Durkin et al., 2012; U.S. Department of
Health and Human Services, 2012). There is a small but growing
body of evidence of the effectiveness of anti-tobacco mass media
campaigns in low and middle income countries (Alday et al., 2010;
Murukutla et al., 2011; Perl et al., 2011; Thrasher et al., 2011;
Wakefield et al., 2013), although only one study has directly
compared the effectiveness of different message types among adult
smokers in such countries (Wakefield et al., 2013). That study found
two ads containing graphic imagery, and a third showing a disgust-
provoking simulation of the serious health effects of smoking, were
rated consistently more positively on message acceptance and
perceived ad effectiveness than an emotional story about the health
effects of smoking and a simulation of the effects of emphysema,
across 10 low and middle income countries (Wakefield et al., 2013).
A limitation of the study was that all ads featured the negative
health effects of smoking and there were only one or two repre-
sentations of the different ad types. The current study aims to more
fully explore the effects of different types of anti-tobacco ads, using
a total of 37 ads in the same 10 countries.
As smoking prevalence differs across different demographic
groups e for example by age, gender or socio-economic status
(Eriksen et al., 2012) e it is often assumed that particular campaigns
should be specifically designed for the highest prevalence sub-
groups. However, developing and producing many different ads for
different subgroups requires substantial resources and erodes the
amount of funding available to widely broadcast ads and reach the
majority of smokers (Durkin et al., 2012; U.S. Department of Health
and Human Services, 2012). Campaigns specifically targeted to
subgroups risk stigmatizing the targeted group by implicitly
communicating to the public that this group is in particular need of
the message (Hornik & Ramirez, 2006). Creating targeted messages
for many different groups may also inadvertently reduce the po-
tential for campaigns to support broader societal norm and policy
change, due to public perceptions that the messages are of relevance
only to specific subgroups (Hornik & Ramirez, 2006). Also, targeted
campaigns may not perform better for a particular demographic
subgroup than ‘common denominator’ ads that are relevant to a
broad range of smokers (Durkin et al., 2012; Hornik & Ramirez, 2006;
National Cancer Institute, 2008; U.S. Department of Health and
Human Services, 2012). Therefore, it has been suggested that if
‘common denominator’ messages are found to be equally effective
across different demographic groups, a segmented strategy is un-
necessary and not worth the potential risks as well as extra costs
(Hornik & Ramirez, 2006).
Previous research in high income countries has found few dif-
ferences between the response of different demographic groups to
anti-smoking mass media campaigns among adolescents (National
Cancer Institute, 2008; U.S. Department of Health and Human
Services, 2012) and adults (Bala, Strzeszynski, & Cahill, 2008;
Durkin et al., 2012). A review of campaign effects in adolescents
(National Cancer Institute, 2008) concluded that ad characteristics
are more important than demographic characteristics in deter-
mining ad effectiveness, and that ads that perform well do so
among many population groups. Consistent with this, in adults,
negative health effects ads that have been found to be most effec-
tive across a broad range of smokers have also been found to be
equally effective across different age groups and genders (Durkin
et al., 2012). There is also emerging evidence these health effects
ads are at least equally, if not more effective among lower income
and lower educated smokers (Durkin, Biener, & Wakefield, 2009,
Durkin, Wakefield, & Spittal, 2011; Niederdeppe, Farrelly,
Nonnemaker, Davis, & Wagner, 2011). Decisions about segmenta-
tion of campaign messages are particularly important for the
development of campaigns in low and middle income countries,
where funding is limited. Therefore this study aimed to examine
205
whether certain types of negative health effects and non-health
effects anti-smoking ads are more likely to be accepted and
perceived as effective across a broad cross-section of smokers from
low and middle income countries, or whether there are no ‘com-
mon denominator’ messages and only specific types of ads are
effective for specific demographic subgroups.
It has been suggested that an efficient way of using funds avail-
able for anti-smoking campaigns in low and middle income coun-
tries is to adapt and/or recycle messages already used successfully in
other jurisdictions to reduce costs associated with new campaign
development (Cotter et al., 2010). However, it is possible that the
success of some of these ads may be reduced within different
countries due certain ads being culturally different (Wakefield et al.,
2013). Communication theories indicate personal relevance or self-
referencing where viewers relate events in the message back to their
own life and behaviour (Burnkrant & Unnava, 1989, 1995) are
important in determining message persuasiveness (Dunlop,
Wakefield, & Kashima, 2008, 2010; Epstein, 2003; Green, 2008;
Petty, Cacioppo, Strathman, & Priester, 2005). Messages may have
lower personal relevance and elicit less self-referencing among
viewers if they include people and actors from cultures obviously
different from their own. If smokers see people like themselves in
the ad or if the story is more recognizable, messages may be more
likely to be accepted and perceived as effective. Our study also aimed
to examine whether cultural differences between the characters and
contexts in the ad and the country in which it was tested affected the
perceived impact of different types of anti-tobacco ads, and whether
this differed by demographic and smoker subgroups.
Interest in quitting, quitting history and addiction level have
also been found in some studies to influence responses to anti-
tobacco mass media campaign messages. Those who have greater
motivation to quit smoking and have made previous attempts to
quit are more likely to perceive anti-smoking ads to be relevant and
effective (Davis, Nonnemaker, Farrelly, & Niederdeppe, 2011;
Durkin & Wakefield, 2008; Niederdeppe et al., 2011; Veer, Tutty,
& Willemse, 2008), while those who smoke more or are more
addicted tend to be less likely to report positive responses to anti-
smoking messages (Davis et al., 2011; Strasser et al., 2009). Other
studies have found more behaviourally-based responses to health
effects anti-tobacco campaigns (i.e. quit attempts) to be more
equally generated by sustained media campaign exposure irre-
spective of baseline interest in quitting (Borland & Balmford, 2003;
Wakefield, Spittal, Yong, Durkin, & Borland, 2011). The current
study also aimed to examine if smokers in low and middle income
countries who are more motivated to quit or who have previously
tried to quit respond more positively, and heavier smokers less
positively, to anti-smoking messages, and whether this differs
depending on the type of message.
Method
Design
Smokers in ten low and middle-income countries individually
rated, and then discussed in a group, a set of ten anti-smoking
television ads. Five ads were shown and rated in all countries,
and the remaining five ads differed in each country, providing a
total of 37 different anti-smoking ads across the 10 countries
(online Appendix 1).
Participants
2399 daily cigarette smokers (including female bidi smokers in
India; a bidi is small cigarette, locally produced in India and usually
from cut tobacco rolled in leaf), aged 18e34 years who could read
206 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213

and write in their own language, and who did not work in health
promotion, market research, advertising, or the tobacco industry,
were recruited from the ten countries. In each country, approxi-
mately 240 adult smokers (24 groups of 8e12 smokers) were
recruited using convenience sampling by a local market research
agency through face-to-face or telephone interviewing. Fieldwork
was completed between August 2008 and September 2010. Groups
were segmented by age, gender, location, and order of ad presen-
tation. The sample included more male groups to reflect higher
male smoking prevalence in these countries, and in Bangladesh,
only males were recruited, as the scarcity of female smokers meant
it was impractical to recruit sufficient numbers. Table 1 shows
participant characteristics by country. There were higher pro-
portions of university-educated participants recruited in
Bangladesh, Egypt, India and Mexico.
Anti-smoking advertisements
Five ads were selected from the WLF mass media clearinghouse
of anti-tobacco ads to test in each country. These five anti-smoking
ads were chosen to represent the health effects message theme
previously found to perform well in encouraging smoking cessation
using a range of ad execution styles (National Cancer Institute,
2008), and also had good potential to be efficiently adapted for
use in other countries and had performed well in evaluations. The
five ads that were consistent across all countries were 30-s in
duration with original English text and speech, while the languages
of the 32 different country-chosen ads varied. The ad scripts were
translated by country-based research agencies into the local lan-
guage (or languages in India) and then translated into English and
reviewed. This process was repeated until an accurate and appro-
priate translation was achieved. The original content of ad scripts
was closely preserved, allowing for some variation to maximize
cultural acceptability and comprehension. Ads were then dubbed
into the local language, and end-frames removed or replaced with a
local equivalent. All 37 ads are described in the Appendix 1 and can
be viewed at www.worldlungfoundation.org/mmr.
Over 80% of the 37 ads were about the serious health harms of
smoking. Based on previous ad categorizations of negative health
effects ads (Durkin et al., 2009, 2011; Farrelly, Davis, Nonnemaker,
Kamyab, & Jackson, 2011; Wakefield et al., 2013) one researcher
(MB) categorized each ad into one of four types of health effects ad
or a non-health effects ad category, whether the ad contained a
second hand smoke (SHS) message or not, and also coded the
extent to which each ad contained culturally different characters
and/or contexts compared to the country in which it was tested. In
each country there were 1e2 country-specific experts who oversaw
data collection. As these experts differed from country to country,
the research coordinator provided an outline of what constituted a
local, neutral, culturally similar appearance and culturally different
appearance ad: Neutral ads (n ¼ 6 ads) did not feature images of
people, and included mostly simulations or demonstrations; Local
Culture ads (n ¼ 12 ads) were those tested in the same country in
which they were made; Culturally Similar ads (n ¼ 7 ads), where
those in which the main character of the ad and the context was
culturally or ethnically similar to people in the country in which it
was tested; Culturally Different ads (n ¼ 19 ads) included characters
or contexts that were clearly different to the cultures of the coun-
tries in which they were tested. The research coordinator suggested
a potential category for each ad in each country and the in-country
experts were then asked to closely examine each ad and decide
whether this categorization was accurate or needed to be altered.
There were 5 occasions when the in-country experts differed from
the original suggestion and in each case the in-country expert’s
suggestion was used. Table 2 provides the number of smoker re-
sponses and number of ads in each ad category.
Ad type
Ads categorized as Graphic Health Effects ads described or
depicted the negative health effects of smoking or SHS in a visceral
(disgust-provoking) way, typically using imagery of diseased tissue
and organs. Ads categorized as Simulation Health Effects ads
described the negative health effects of smoking or SHS using
simulated imagery (e.g. Bubblewrap) or ‘augmented’ scenes, where

Table 1
Demographic characteristics of participants.

Bangladesh China Egypt India Indonesia Mexico Philippines Russia Turkey Vietnam

N 192 240 240 270 240 226 280 231 240 240

Ad order
Order A 50.0 50.0 50.0 50.0 50.0 50.9 50.0 51.1 50.0 50.0
Order B 50.0 50.0 50.0 50.0 50.0 49.1 50.0 48.9 50.0 50.0
Gender
Male 100.0 66.7 66.7 70.4 66.7 65.9 66.1 66.7 66.7 66.7
Female 0 33.3 33.3 29.6 33.3 34.1 33.9 33.3 33.3 33.3
Age
18e24 years 50.0 50.0 50.0 49.6 50.0 50.0 50.0 50.2 50.0 48.3
25e34 years 50.0 50.0 50.0 50.4 50.0 50.0 50.0 49.8 50.0 51.7
Parental status
Parent 17.7 19.2 18.8 28.2 19.2 41.6 40.0 23.8 25.8 17.1
Not a parent 82.3 80.8 81.3 71.9 80.8 58.4 60.0 76.2 74.2 82.9
Completed university degree
Yes 45.8 5.4 44.2 43.3 22.9 38.9 12.1 7.4 11.7 17.9
No 54.2 94.6 55.8 56.7 77.1 61.1 87.9 92.6 88.3 82.1
Cigarettes smoked daily
1e15 78.7 67.5 52.5 93.3 82.5 94.7 82.5 64.9 52.1 84.6
16 or more 21.4 32.5 47.5 6.7 17.5 5.3 17.5 35.1 47.9 15.4
Thinking about quitting in next 12 months
Yes 62.0 39.2 63.8 34.8 42.9 40.3 68.6 50.7 62.1 47.1
No 38.0 60.8 36.3 65.2 57.1 59.7 31.4 49.4 37.9 52.9
Previously tried to quit
Yes 66.2 52.9 72.5 43.3 72.1 58.0 62.9 72.3 63.8 63.3
No 33.9 47.1 27.5 56.7 27.9 42.0 37.1 27.7 36.3 36.7
Location
Smaller urban city 50.0 50.0 50.0 49.6 50.0 47.8 50.0 50.2 50.0 50.0
Large urban city 50.0 50.0 50.0 50.4 50.0 52.2 50.0 49.8 50.0 50.0
 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213
Table 2
Number of smoker responses (number of different ads) for Ad Type and Ad-Cultural Differences.a
Health effects e Graphic Health effects e Simulation
Similar 692 (3) 240 (1)
Local culture 720 (3) 280 (1)
Neutral 2399 (1) 4744 (3)
Culturally different 4337 (6) 760 (3)
Total 8148 (10) 6024 (6)
a
Ad-Cultural Difference categories are not mutually exclusive, given particular ads were culturally similar in some countries but not others, therefore the number of
different ads within each category may not add to the total number of different ads for a given ad type.
computerized imagery is used to enhance an image (e.g. When you
smoke). Emotional Health Effects Story ads described or depicted the
consequences of smoking or SHS using emotional narratives, and
included real personal testimonials (e.g. Zita), or emotional dra-
matizations (e.g. Worse). Other Health Effects ads described the
negative health effects of smoking or SHS in a generic way (e.g.
Poison), and did not meet the definitions of the Health effects
Graphic, Emotional story and Simulation categories. Non-Health
Effects ads included humorous, anti-industry, financial benefits
messages, and non-emotional stories that related to either the so-
cial unacceptability of personal smoking or SHS.
Cultural appearance difference
To allow examination of the interaction between ad type and
difference in the cultural appearance of characters, the ‘Similar’ and
‘Local’ categories were first collapsed due to empty or small cell sizes
(there were no ads that fit into both ‘Similar’ and ‘Other’; ‘Neutral’
and ‘Health effects story’; or ‘Neutral’ and ‘Other’ ad categorizations).
Preliminary examinations of the effects of ‘Similar’ or ‘Local’ and
‘Neutral’ ad types indicated they did not differ significantly from
each other on Message Acceptance or Perceived Effectiveness
(‘Similar or Local’ compared to ‘Neutral’ on Message Acceptance
OR ¼ 1.09, p ¼ 0.273; on Perceived Effectiveness OR ¼ 1.00, p ¼ 1.00).
So, for the purposes of analyses, those ads that were ‘Not Culturally
Different’ (local, similar, or neutral) were combined and compared to
those that were ‘Culturally Different’ (Table 2).
Measures
Smokers’ gender, age (18e24 years; 25e34 years), parental status
(no; yes), level of educational attainment (below university educa-
tion; completed university education), daily cigarette/bidi con-
sumption (1e15; 16 or more per day), intention to quit in the next 12
months (no; yes), ever tried to quit (no; yes), and residential location
(small urban city; large urban city) were collected at the beginning of
the rating session. The advertisement ratings were repeated for each
ad, and incorporated items used in other ad rating studies (Davis
et al., 2011; Dillard, Shen, & Vail, 2007; Durkin & Wakefield, 2008;
Wakefield, Durkin, Murphy, & Cotter, 2007; Wakefield et al., 2003).
These questions measured the extent to which participants thought
each ad was ‘easy to understand’, ‘believable’, ‘relevant’, and ‘effec-
tive’, whether the ad taught them ‘something new’, and made them
‘stop and think’, feel ‘concerned about their smoking’, and ‘moti-
vated to try to quit smoking’. Each item was measured on a 5-point
scale where 1 represented ‘strongly disagree’, 2 ‘slightly disagree’, 3
‘neither agree nor disagree’, 4 ‘slightly agree’, and 5 ‘strongly agree’.
We undertook an iterative process of translation and back-
translation of the ad rating items, ensuring they were meaning-
fully translated into the local language/s, while accurately repre-
senting the original questionnaire to facilitate cross-country
consistency. Rating items were collapsed into 2 scales based on
principal components analyses conducted across all countries using
the 5 country-consistent ads, as reported previously (Wakefield
207
Health effects e Stories Health effects e Other Non-health effects
457 (2) 240 (1) 0 (0)
240 (1) 670 (3) 942 (4)
0 (0) 422 (2) 0 (0)
6132 (7) 226 (1) 480 (2)
6829 (8) 1558 (7) 1422 (6)
et al., 2013). The first scale examined Message Acceptance (MA)
(Cronbach’s a ¼ 0.80), and comprised Understand and Believable
(scores from these two items were averaged together). The second
scale examined Perceived Effectiveness (PE) (Cronbach’s a ¼ 0.94)
and was comprised of: Taught me something new; Stop and think;
Relevant; Concerned about my smoking; Makes me more likely to
try to quit; and Effective (averaged together). Scores on the MA and
PE scales were negatively skewed and so were binary coded allo-
cating a score greater than 3.5 (versus 3.5 and below) to represent an
average positive response on each composite scale allowing us to use
logistic regression analyses which does not require normally
distributed variables. As response tendency may have differed be-
tween countries, a set of sensitivity analyses were conducted. For
these sensitivity analyses, each country’s median score was used to
classify whether a smoker gave a positive response on each item
making up the MA and PE scales. For these sensitivity analyses a
positive response on MA was coded if smokers responded at or above
their country median on Understand and Believe, while a positive
response on PE was coded if smokers responded at or above their
country median on at least 4 of the 6 items that comprise PE. This
classification method resulted in fewer smokers reporting positive
responses overall, however the pattern of results reported here was
consistent with those found using these alternative measures of MA
and PE. Therefore we chose to report the findings using a standard
cut off for positive response that was used across all countries and
that is consistent with the methods used in the previous paper
reporting findings from this study (Wakefield et al., 2013).
Procedure
Full details of the procedure are available in a previous paper
(Wakefield et al., 2013). Briefly, after completing demographic
items and rating a practice ad, smokers were shown the first anti-
smoking advertisement twice, after which they completed the ad
rating questions. This process was repeated for the remaining nine
ads. Participants were instructed not to talk to each other when
watching and rating the ads. The five ads that were shown in all
countries were alternated with the five country-choice ads. To
counterbalance potential effects of ad viewing order, half the
groups viewed the ads in the reverse order.
Data analysis
Multivariate logistic regression analyses examining the effect of ad
type, demographic subgroup, smoker subgroup and cultural differ-
ence were performed separately for the MA and PE outcome mea-
sures. Robust standard errors to control for individuals each having
rated multiple ads were used (Kirkwood & Stern, 2001). All models
included the following covariates: SHS message, ad order, gender, age,
parental status, education level, average daily cigarette consumption,
thinking about quitting smoking in the next year, having made a
previous quit attempt, country and residential location. A series of
two-way interaction models were used to assess if responses to each
208 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213

ad type were consistent across each country, between each of the
demographic attributes and smoking characteristics and by Ad-
cultural difference. For the two outcomes, a two-way interaction
term was included by multiplying the 5-category ad type variable
separately by the 10-country variable, then by each of the categorical
demographic variables, then by each of the categorical smoking
characteristic variables, and finally by the cultural difference variable
(e.g., age  ad type in one model, gender  ad type in another model,
and so on). A global test of the significance of each interaction was
conducted after each model to examine whether the interaction
contributed significant additional explanation of variance above and
beyond the model without the inclusion of the interaction. From these
covariate-adjusted interaction models, average predicted probabili-
ties of positive ad ratings on the two outcomes were calculated for
each ad, and it is these covariate adjusted probabilities that are re-
ported throughout the Results section.
Sensitivity analyses were also conducted including only the 32
ads that were not rated in every country. As the substantive find-
ings of these analyses did not differ from the analyses including all
37 ads, we provide results for the full set of ads.
Results
Ad ratings
Message Acceptance
Table 3 indicates a significant effect of ad type for MA
(c2 ¼ 300.91, p < 0.001) and Fig. 1 plots the adjusted proportions of
positive response for each ad type overall and for each country. Post
hoc comparisons showed Graphic Health Effects ads were signifi-
cantly more likely to be rated highly on MA than all other ad types
(covariate adjusted proportion 91%; all p values <0.001). Simulated
Health Effects (85%), Emotional Health Effects Stories (85%) and
Other Health Effects (84%) ads were all equally likely to be accepted,
but were each more likely than the Non-Health Effects ads (75%; all
p values <0.001) to be accepted. Table 3 shows that males and those
not thinking about quitting in the next year were significantly less
likely to accept anti-tobacco messages overall. Ads that were local,
culturally similar or neutral were more likely to be rated positively
on MA (90%) compared to culturally different ads (83%, p < 0.001).
Perceived Effectiveness
As shown in Table 3, there was a significant effect of ad type
found for PE (c2 ¼ 525.11, p < 0.001) and Fig. 2 plots the adjusted
proportions of positive response for each ad type overall and for
each country. Post hoc comparisons showed Graphic Health Effects
ads were most likely to be rated highly on PE, significantly more so
than all other ad types (covariate adjusted proportion 90%; all p
values <0.001). Simulated Health effects ads (84%) were next most
likely to be highly rated (significantly higher than each of the three
other ad types; all p values <0.001), followed by Emotional Health
Effects Stories (80%; higher than Other Health Effects and Non-
Health Effects ads, all p values <0.001), Other Health effects ads
(75%; higher than Non-Health Effects ads only, p < 0.001), and then
ads in the Non-Health Effects category (69%; lower than all other ad
types, all p values <0.001). Table 3 shows that males, those with

Table 3
Multivariate logistic regression analyses of positive Message Acceptance and Perceived Effectiveness ratings by ad type, demographic and smoker characteristics, and Ad-
cultural difference.

Message Acceptance Perceived Effectiveness

Odds ratio (95% CI) p Odds ratio (95% CI) p

Ad Type
Health effects e Graphic 3.65 (3.13, 4.27) 0.000 4.36 (3.77, 5.03) 0.000
Health effects e Simulation 1.95 (1.68, 2.27) 0.000 2.58 (2.24, 2.97) 0.000
Health effects e Emotional Story 1.97 (1.71, 2.26) 0.000 1.92 (1.69, 2.18) 0.000
Health effects e Other 1.88 (1.55, 2.28) 0.000 1.35 (1.16, 1.58) 0.000
Non-Health Effects (ref) 1.00 1.00
Demographic subgroups
Education
No university degree (ref) 1.00 1.00
Completed university degree 0.87 (0.75, 1.00) 0.058 0.66 (0.56, 0.77) 0.000
Gender
Female (ref) 1.00 1.00
Male 0.72 (0.62, 0.83) 0.000 0.82 (0.70, 0.96) 0.011
Age
25e34 years (ref) 1.00 1.00
18e24 years 1.15 (0.99, 1.33) 0.067 1.03 (0.88, 1.20) 0.709
Location
Smaller urban city (ref) 1.00 1.00
Large urban city 1.12 (0.99, 1.27) 0.068 1.27 (1.11, 1.44) 0.000
Parental status
Not a parent (ref) 1.00 1.00
Parent 1.00 (0.85, 1.19) 0.957 1.26 (1.04, 1.51) 0.017
Smoker subgroups
Thinking about quitting
No (ref) 1.00 1.00
Yes 1.26 (1.10, 1.43) 0.001 1.74 (1.52, 1.98) 0.000
Cigarettes smoked daily
1e15 (ref) 1.00 1.00
16 or more 1.03 (0.89, 1.20) 0.716 0.84 (0.72, 0.98) 0.030
Previously tried to quit
No (ref) 1.00 1.00
Yes 1.10 (0.96, 1.26) 0.173 1.26 (1.10, 1.45) 0.001
Ad-Cultural Difference
Culturally Different (ref) 1.00 1.00
Not Culturally Different 1.85 (1.67, 2.05) 0.000 1.46 (1.34, 1.60) 0.000

Note: All analyses adjusted for participant country, inclusion of second-hand smoke theme, ad presentation order, and for each individual having rated multiple ads.
 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213 209

Fig. 1. Adjusted proportion of positive responses on Message Acceptance by country.

higher education, without children, from smaller urban cities,
heavier smokers, those who had not previously tried to quit and
who were not thinking about quitting in the next year each were
less likely to perceive anti-smoking ads overall to be effective. Ads
that were local, culturally similar or neutral were more likely to be
rated positively on PE (86%) compared to culturally different ads
(81%, p < 0.001).
Ad Type by Country Interactions
There was a significant interaction between ad type and country
on Message Acceptance (c2 ¼ 249.94, p < 0.001). Similar to the
overall pattern of effects, post hoc tests indicated that Graphic
Health Effects ads were rated higher or equal to other ads on MA in
each country. Simulated and Other Health Effects ads were gener-
ally rated next highest and similarly to each other across countries,
except in China and Russia where Other Health Effects ads were
rated higher. Health Effects stories and Non-Health Effects ads
varied widely in MA from country to country. Health Effects Stories
rated equal to highest in Mexico and the Philippines, but lowest in
Russia, Vietnam, Bangladesh, China, India and Turkey. Non-Health
Effects ads rated equal to the highest in Bangladesh, China and
India, but by far the lowest in Egypt and Indonesia.
There was also a significant interaction between ad type and
country on Perceived Effectiveness (c2 ¼ 225.84, p < 0.001).
Graphic Health Effects ads were rated higher or equal to other ad
types on PE across all countries. Simulated Health Effects ads were
typically rated higher than or equal to the remaining ad styles in
each country except in Mexico, where they were rated slightly
lower than Health Effects Stories. Health Effects Stories, Other
Health Effects ads and Non-Health Effects ads varied widely in
ratings from country-to-country. Health Effects Stories were rated
equal to highest in the Philippines and Mexico, but lowest in
Bangladesh, China and Turkey. Other Health Effects ads were rated
equal to highest in China but lowest in Russia, Mexico, and Viet-
nam. Non-Health effects ads were rated equal to highest in China,
but were rated lowest or equal lowest in Egypt, India and Indonesia.
Ad Type by Demographic Interactions
Table 4 shows significant interactions between ad type and
gender, and parental status were found on message acceptance (all
p values <0.05), while a trend towards a significant interaction was
found for age (p < 0.10). Fig. 3 provides the differences in covariate-
adjusted proportion of positive ratings on ad acceptance between
each of the demographic subgroups. For MA, females rated Graphic
Health effects ads on average only 1% point higher than males,
Emotional Health Effects Stories and Non-Health Effects ads only 3%
points higher, but Simulated Health Effects ads 6% points higher,
and Other Health effects ads 9% points higher. Parents and non-
parents did not substantially differ in their average ratings of
Graphic, Simulated, Emotional Health Effects Stories and Non-

Fig. 2. Adjusted proportion of positive responses on Perceived Effectiveness by country.

210 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213

Table 4
Ad type by Demographic, Smoking and Ad-Cultural Difference Interaction effects from multivariate logistic regression analyses of positive Message Acceptance and Perceived
Effectiveness ratings.

Message Acceptance Perceived Effectiveness

Interaction c2 p Interaction c2 p

Ad Type by Country Interactions 249.94 0.000 225.84 0.000

Ad Type and Demographic Interactions
Education*Ad type 1.36 0.716 8.47 0.037
Gender*Ad type 12.15 0.016 8.89 0.064
Age*Ad type 8.69 0.069 9.74 0.045
Location*Ad type 6.83 0.145 2.27 0.687
Parental status*Ad type 13.48 0.009 16.17 0.003
Ad Type by Smoking Subgroup Interactions
Thinking about quitting*Ad type 3.15 0.533 12.89 0.012
Cigarettes smoked daily*Ad type 15.17 0.004 7.24 0.124
Previously tried to quit*Ad type 8.30 0.081 11.22 0.024
Ad-Cultural Difference by Ad type Interaction 130.87 0.000 67.72 0.000

Note: All analyses adjusted for participant country, inclusion of second-hand smoke theme, ad presentation order, and other demographic and smoking subgroups, Ad-cultural
difference, and for each individual having rated multiple ads.

Health Effects ads (0e3% point differences), however parents rated
Other Health Effects ads 6% points higher than those without
children. There were only very small differences between younger
and older smokers on each of the Health Effects ad types (range
0%e2% points), while Non-Health Effects ads were rated 6% points
higher on average by younger smokers than older smokers.
Table 4 shows significant interactions between ad type and
education level, age, and parental status on PE (all p values <0.05),
and a trend towards an interaction for gender (p < 0.10). Fig. 4
provides the differences in proportion of positive ratings on
perceived ad effectiveness between each of the demographic
subgroups (covariate adjusted proportions). For PE, there was little
difference between lower and higher educated smokers’ ratings of
Graphic, Simulated, and Other Health Effects ads (2e3% points
difference), but lower educated smokers rated Emotional Health
Effects Stories and Non-Health Effects ads higher than educated
smokers (7% and 9% points higher, respectively). Younger smokers
rated Graphic Health Effects ads similarly to older smokers (1%
point higher), while older smokers rated all other ad types slightly
higher (2e3% points). Parents and those without children rated
Graphic Health Effects ads similarly (1% point difference), but par-
ents rated Simulated (5% points), Emotional Health Effects Stories
(8% points), Other Health Effects (9% points), and Non-Health Ef-
fects ads (8% points) higher than those without children. There
were no differences between females and males on their average PE
rating of Graphic Health Effects ads, while females rated all other ad
types somewhat higher than males (3e6% points).
Ad Type by Smoker Characteristic Interactions
A significant interaction between ad type and cigarette con-
sumption level was found on MA (p < 0.01), while a trend toward
a significant interaction was found for previous quit attempts
(p < 0.10) (Table 3). Lighter smokers did not differ substantially
from heavier smokers on each of the Health Effects ad types
(range 1%e3% point differences), however lighter smokers rated
Non-Health Effects ads 14% points higher on average than heavier
smokers. There were no differences in MA response to the
Graphic, Simulation and Health Effects Story ads between smokers
who had made a previous quit attempt and those who had not
(Fig. 1), but smokers who made a previous attempt rated Other
Health effects ads slightly higher (2% points), and Non-Health
Effects ads slightly lower (3% points) than those who had not
made an attempt.

Fig. 3. Adjusted proportion of positive responses on Message Acceptance by demographic and smoker subgroups, and Ad-cultural difference.
 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213
Fig. 4. Adjusted proportion of positive responses on Perceived Effectiveness by demographic and smoker subgroups, and Ad-cultural difference.
Table 3 shows significant interactions were found between ad
type and quitting intentions and previous quit attempts on PE (all p
values <0.05). Those who intended to quit within the next year
rated all of the different types of ads higher in PE than those who
did not, however this difference was least pronounced for the Non-
Health Effects ads (4% points higher, versus a range 6e9% points
higher for the different ads about Health Effects) (Fig. 2). Similarly,
those who made previous quit attempts rated Graphic, Simulated,
Emotional Health Effects Stories and Other Health Effects ads
higher (2e4% points), but Non-Health Effects ads only 1% point
higher than those who had not made a previous quit attempt.
Ad Type by Ad-Cultural Difference Interactions
Significant interactions were found between Ad-cultural dif-
ference and ad type for both MA and PE (all p values <0.001). There
were only very small differences between culturally different and
non-culturally different Graphic and Simulated Health Effects ads
on MA (1%e2% point differences), whereas culturally different
Health Effects Stories were rated lower (9% points), and culturally
different Other Health Effects and Non-Health Effects ads were
rated substantially lower than non-culturally different versions of
these ads (22% and 29% points lower, respectively). For PE, cultur-
ally different Other Health Effects and Non-Health Effects ads were
also rated substantially lower than non-culturally different versions
of these ads (11% and 16% points, respectively). Smaller differences
were found between culturally different and non-culturally
different versions of Simulated and Emotional Health Effects Stor-
ies (5% and 7% points, respectively), while there was no meaningful
difference between the culturally different and non-culturally
different versions of Graphic Health Effects ads (1% point differ-
ence) on PE.
Discussion
Overall, graphic health effects ads were most likely to be
accepted and perceived as effective, followed by simulated health
effects ads, health effects stories and other health effects ads. Non-
health effects ads were the least likely to be accepted and perceived
as effective overall. Consistent with our previous examination of a
sub-set of these ads (Wakefield et al., 2013) graphic health effects
211
ads were not only rated most highly overall, but were also least
likely to show differential ratings across countries. These findings
add to the growing body of research indicating ads that evoke
strong emotional reactions, such as fear and disgust, are perceived
as more effective and more likely to be associated with reduced
smoking behaviours than ads that do not evoke strong emotions
(Durkin et al., 2012; National Cancer Institute, 2008; U.S.
Department of Health and Human Services, 2012). Often govern-
ments are inclined to broadcast positive public health messages
however these findings indicate that strongly emotional graphic
health effects ads may be the best value for money for driving down
national smoking rates. Future research could explore whether
these types of ads are the best option for other pressing public
health issues in developing countries, such as HIV prevention.
Extending previous findings, we also found that the graphic
health effect ads were least likely to vary across individual
subgroups and were least subject to Ad-cultural difference effects.
These findings indicate these messages can influence different
demographic subgroups equally and so are ideal ‘common de-
nominator’ anti-tobacco messages, likely to be appropriate for
broad use in many low and middle income countries. The univer-
sality of the graphic images of internal organs in these ads may
make them highly relevant to many populations and elicit common
emotional responses among viewers. If organizations aim to reduce
smoking in particular subgroups, funds could be used to ensure
these graphic ads are aired more frequently on channels which are
popular with those in these subgroups, given high reach and fre-
quency of exposure are essential for campaign impact (Durkin et al.,
2012; Hornik & Ramirez, 2006).
Simulated health effects ads were rated highly on message
acceptance and perceived effectiveness across most countries,
while emotional stories varied widely in terms of acceptance and
perceived effectiveness across different countries consistent with
our previous analyses (Wakefield et al., 2013). This previous ex-
amination indicated that where the story-based ad rated poorly, it
was due to problems with understanding and acceptance. The
findings reported here include evaluation data from another 7
emotional story based ads, and support the previous finding; in
those countries where these emotional story-based ads rated
poorly on message acceptance they also rated poorly on perceived
effectiveness. ‘Other health effects’ ads rated well on message
212 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213
acceptance across the six countries in which they were tested, but
had more variable perceived effectiveness ratings. For example,
these types of ads rated highly on message acceptance but rela-
tively low on perceived effectiveness in Russia, Vietnam and Egypt.
This pattern of effects is not surprising given the lower level of
emotion evoked by these ads, with previous research indicating
lower levels of emotion are associated with lower ad perceived
effectiveness ratings (Durkin et al., 2012).
Simulated health effects, emotional stories of the health effects,
and ‘other’ health effects messages rated similarly across most
demographic subgroups on message acceptance. These types of
messages showed slightly more variability on perceived effective-
ness across demographic groups, with parents, older smokers and
females perceiving these types of messages as more effective. In
addition, those with lower education perceived emotional stories of
the health effects of smoking more effective than those with higher
education. This is consistent with previous research that has indi-
cated personal testimonials and emotional narratives are more
effective among those with lower education (Durkin et al., 2009,
2011; Niederdeppe et al., 2011).
The simulated and emotional story-based ads were also less
likely to be accepted or perceived as effective if they featured
culturally different people or contexts. However, as these types of
messages have been found to be among the most influential types
of ads in high-income countries (Durkin et al., 2012), it is highly
likely that local or culturally adapted versions of these types of
messages may also be very effective in low and middle income
countries.
Consistent with several studies from high income countries
(Durkin & Wakefield, 2008; Niederdeppe et al., 2011), health effects
ads were also most likely to be perceived as effective by those
motivated to quit in the next year and those who had made a
previous quit attempt. Like other subgroup differences in response
to health effects ads, although statistically significant, the degree of
difference was of small practical consequence, with these ads
scoring highest even for those least interested in quitting or not
having tried to quit previously.
The content of the messages that were not focused on the health
effects of smoking ranged from humorous depictions of smokers, to
anti-industry, financial and non-emotional messages about the
social unacceptability of personal smoking or second hand smoke.
Consistent with previous research, these types of ads were the least
likely to be accepted and perceived as effective compared to ads
about the serious harms of smoking overall (Durkin et al., 2012;
National Cancer Institute, 2008; U.S. Department of Health and
Human Services, 2012). These ads also showed the greatest de-
gree of difference across the five countries in which they were
tested and between different demographic subgroups, were the
least likely to be rated highly by those motivated to quit and were
most likely to be affected by Ad-cultural differences. However, as
there were only a few examples of these non-health effects ads and
they were all included in the one category, further research
comparing each type of these ads to the various health effects
messages is warranted to further explore their utility in low and
middle income countries.
Similar to other ad-rating studies, we used self-report ratings of
message acceptance and perceived effectiveness and so the effects
here are limited to perceptions rather than actual behaviour
change. A series of recent studies that link anti-smoking ad ratings
of perceived effectiveness with subsequent attitude change
(Dillard, Shen, et al., 2007; Dillard, Weber, et al., 2007), quitting
intention change (Bigsby, Cappella, & Seitz, 2012; Brennan, Durkin,
Wakefield, & Kashima, 2013; Davis et al., 2011; Dillard, Shen, et al.,
2007) and reduced smoking behaviour (Brennan et al., 2013) in-
dicates that these ratings are important predictors of the actual
effectiveness of ads. Convenience sampling was used in each
country and so the smokers included here may not be represen-
tative of all smokers within each country, similar to other adver-
tisement rating studies (Durkin et al., 2012). Our aim was to
examine the comparative effectiveness of a range of different ads
across many different low and middle income countries. To permit
valid comparisons between ads we set standard protocol, educa-
tion, gender and other demographic quotas across countries. This
allowed for valid comparisons across ad types ensuring that the
ratings were not unduly influenced by differences in underlying
demographic profiles. In some countries, however, women smokers
were rare and so there were fewer women than planned in some
groups, and no women included in Bangladesh. In India we relaxed
the recruitment criteria to include bidi smokers in order to get a
sufficient number of female smokers. Our in-country fieldwork
agency indicated that most of the female smokers in India were bidi
smokers, however we did not specifically record who was and was
not a bidi smoker, so bidi smokers were treated the same as other
smokers in our analyses. It is possible that bidi smokers’ responses
to the ads differed from non-bidi smokers and future research
should examine this issue.
In addition, due to cross country differences in higher education
rates, there were some countries with higher proportions that had
attained university degrees. Given previous work indicating per-
sonal testimonials and emotional narratives may be more effective
among those with lower education (Durkin et al., 2012), the overall
effectiveness of ads using emotional stories may be underestimated
in the current study. However, this would have been minimized by
including education level as a covariate in each analyses comparing
the effects of different ad types.
To maximize the potential for valid comparisons of ad types
across countries standard procedures were also used in every
country to ensure consistency in the translation of the study pro-
tocol, ad scripts, and questionnaires. Pilot study of two groups in
each country prior to the main data collection were also conducted
to ensure participants could follow the procedure, and to check for
any ad script or questionnaire translation issues.
We were able to replicate the pattern of these findings using an
alternative method for categorizing positive message acceptance
and perceived effectiveness ratings and when including only the 32
ads that had not been examined in our previous analysis (Wakefield
et al., 2013). These sensitivity analyses indicate that these results
are not an artefact of any particular coding method or ad selection
bias and so provide greater confidence in the findings.
Overall, the findings from this study indicate that ads that do not
emphasize the health effects of smoking may be prone to incon-
sistent impact across countries and population subgroups. As these
types of ads may not be relied upon to impact the whole smoker
population they may not be the most efficient use of the funding
required for wide population broadcast. Simulations and emotional
stories about the health effects of smoking, especially those that
feature actors or real people, may need more careful adaptation
than graphic style health effects ads. However, these types of ads
may prove very effective once they have been appropriately
modified, as they have been found to be in high-income countries.
Strong graphic ads about the negative health effects of smoking
may be most suitable for rapid adaptation and wide population
broadcast and may most efficiently contribute to the reduction in
smoking in low and middle income countries.
Role of the funding source
This study was funded by the World Lung Foundation (WLF) as
part of the Bloomberg Initiative. WLF contributed to the study
design, facilitated data collection in each country, and contributed
 S. Durkin et al. / Social Science & Medicine 98 (2013) 204e213
to manuscript preparation and the decision to submit this paper for
publication.
Acknowledgements
We acknowledge the assistance of the International Anti-
Tobacco Advertisement Rating Study Team for the coordination of
data collection in each country: Bangladesh: Sandra Mullin, Tahir
Turk; China: Yvette Chang, Winnie Chen; Egypt: Mohamed
Elghamrawy, Stephen Hamill; India: Shefali Gupta, Sandra Mullin,
Tahir Turk; Indonesia: Yvette Chang; Mexico: Jorge Alday, Claudia
Cedillo; Philippines: Jorge Alday; Russia: Irina Morozova, Rebecca
Perl; Turkey: Stephen Hamill, Mego Lien; Vietnam: Tom Carroll,
Mego Lien.
Appendix 1. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.socscimed.2013.09.022.
References
Alday, J., Murukutla, N., Cedillo, C., Johns, P., Monteiro, A., & Wakefield, M. (2010).
Smoke-free Sao Paulo: a campaign evaluation and the case for sustained mass
media investment. Salud Publica De Mexico, 52(Suppl. 2), S216eS225.
Bala, M., Strzeszynski, L., & Cahill, K. (2008). Mass media interventions for smoking
cessation in adults. Cochrane Database of Systematic Reviews, 23(1), CD004704.
Bigsby, E., Cappella, J. N., & Seitz, H. H. (2012). Efficiently and effectively evaluating
public service announcements: additional evidence for the utility of perceived
effectiveness. Communication Monographs, 80(1), 1e23.
Borland, R., & Balmford, J. (2003). Understanding how mass media campaigns
impact on smokers. Tobacco Control, 12(Suppl. 2), ii45e52.
Brennan, E., Durkin, S. J., Wakefield, M. A., & Kashima, Y. (2013). Assessing the
effectiveness of antismoking television advertisements: do audience ratings of
perceived effectiveness predict changes in quitting intentions and smoking
behaviours? Tobacco Control. Online first 20 April 2013, http://dx.doi.org/10.
1136/tobaccocontrol-2012-050949.
Burnkrant, R. E., & Unnava, H. R. (1989). Self-referencing: a strategy for increasing
processing of message content. Personality and Social Psychology Bulletin, 15(4),
628e638.
Burnkrant, R. E., & Unnava, H. R. (1995). Effects of self-referencing on persuasion.
Journal of Consumer Research, 22(1), 17e26.
Cotter, T., Perez, D., Dunlop, S., Hung, W. T., Dessaix, A., & Bishop, J. F. (2010). The
case for recycling and adapting anti-tobacco mass media campaigns. Tobacco
Control, 19(6), 514e517.
Davis, K. C., Nonnemaker, J. M., Farrelly, M. C., & Niederdeppe, J. (2011). Exploring
differences in smokers’ perceptions of the effectiveness of cessation media
messages. Tobacco Control, 20(1), 26e33.
Dillard, J. P., Shen, L., & Vail, R. G. (2007). Does perceived message effectiveness
cause persuasion or vice versa? 17 consistent answers. Human Communication
Research, 33(4), 467e488.
Dillard, J. P., Weber, K. M., & Vail, R. G. (2007). The relationship between the
perceived and actual effectiveness of persuasive messages: a meta-analysis
with implications for formative campaign research. Journal of Communication,
57(4), 613e631.
Dunlop, S. M., Wakefield, M., & Kashima, Y. (2008). Can you feel it? Negative
emotion, risk, and narrative in health communication. Media Psychology, 11(1),
52e75.
Dunlop, S. M., Wakefield, M., & Kashima, Y. (2010). Pathways to persuasion:
cognitive and experiential responses to health-promoting mass media mes-
sages. Communication Research, 37(1), 133e164.
Durkin, S. J., Biener, L., & Wakefield, M. A. (2009). Effects of different types of
antismoking ads on reducing disparities in smoking cessation among socio-
economic subgroups. American Journal of Public Health, 99(12), 2217e2223.
Durkin, S. J., Brennan, E., & Wakefield, M. (2012). Mass media campaigns to promote
smoking cessation among adults: an integrative review. Tobacco Control, 21(2),
127e138.
213
Durkin, S. J., & Wakefield, M. (2008). Interrupting a narrative transportation expe-
rience: program placement effects on responses to anti-smoking advertising.
Journal of Health Communication, 13(7), 667e680.
Durkin, S. J., Wakefield, M., & Spittal, M. (2011). Which types of televised anti-
tobacco campaigns prompt more quitline calls from disadvantaged groups?
Health Education Research, 26(6), 998e1009.
Epstein, S. (2003). Cognitive-experiential self-theory of personality. In T. Millon,
M. J. Lerner, & I. B. Weiner (Eds.), Handbook of psychology (pp. 15e184). Hobo-
ken, NJ: John Wiley & Sons.
Eriksen, M., MacKay, J., & Ross, H. (2012). The tobacco atlas. Atlanta, Georgia, USA:
American Cancer Society. Available from http://tobaccoatlas.org/.
Farrelly, M. C., Davis, K. C., Nonnemaker, J. M., Kamyab, K., & Jackson, C. (2011).
Promoting calls to a quitline: quantifying the influence of message theme,
strong negative emotions and graphic images in television advertisements.
Tobacco Control, 20(4), 279e284.
Green, M. C. (2008). Transportation theory. In W. Donsbach (Ed.), The international
encyclopedia of communication. Blackwell Publishing. Retrieved from http://
www.communicationencyclopedia.com/subsriber/tocnode?
id¼g9781405131995_yr2011_chunk_g97814513199525_ss58_1.
Hornik, R. C., & Ramirez, A. S. (2006). Racial/Ethnic disparities and segmentation in
communication campaigns. American Behavioral Scientist, 49(6), 868e884.
Kirkwood, B., & Stern, J. (2001). Essential medical statistics. Oxford: Blackwell
Publishing.
Murukutla, N., Turk, T., Prasad, C. V. S., Saradhi, R., Kaur, J., Gupta, S., et al. (2011).
Results of a national mass media campaign in India to warn against the dangers
of smokeless tobacco consumption. Tobacco Control, 21(1), 12e17.
National Cancer Institute. (2008). The role of the media in promoting and reducing
tobacco use. Tobacco Control Monograph No. 19. Bethesda, MD: US Department
of Health and Human Services, National Institutes of Health, National Cancer
Institute
Niederdeppe, J., Farrelly, M. C., Nonnemaker, J., Davis, K. C., & Wagner, L. (2011).
Socioeconomic variation in recall and perceived effectiveness of campaign ad-
vertisements to promote smoking cessation. Social Science & Medicine, 72(5),
773e780.
Perl, R., Stebenkova, L., Morozova, I., Murukutla, N., Kochetova, V., Kotov, A., et al.
(2011). Mass media campaigns within reach: effective efforts with limited re-
sources in Russia’s capital city. Tobacco Control, 20(6), 439e441.
Petty, R. E., Cacioppo, J. T., Strathman, A. J., & Priester, J. R. (2005). To think or not to
think: exploring two routes to persuasion. In T. C. Brock, & M. C. Green (Eds.),
Persuasion: Psychological insights and perspectives. Thousand Oaks, CA: Sage.
Strasser, A. A., Cappella, J. N., Jepson, C., Fishbein, M., Tang, K. Z., Han, E., et al.
(2009). Experimental evaluation of antitobacco PSAs: effects of message con-
tent and format on physiological and behavioral outcomes. Nicotine & Tobacco
Research, 11(3), 293e302.
Thrasher, J. F., Huang, L., Perez-Hernandez, R., Niederdeppe, J., Arillo-Santillan, E., &
Alday, J. (2011). Evaluation of a social marketing campaign to support Mexico
City’s comprehensive smoke-free law. American Journal of Public Health, 101(2),
328e335.
U.S. Department of Health and Human Services. (2012). Preventing tobacco use
among youth and young adults: A report of the surgeon general. Atlanta, Georgia:
U.S. Department of Health and Human Services, National Center for Chronic
Disease Prevention and Health Promotion, Office on Smoking and Health.
Available from www.surgeongeneral.gov/library/reports/preventing-youth-
tobacco-use/full-report.pdf.
Veer, E., Tutty, M., & Willemse, J. (2008). It’s time to quit: using advertising to
encourage smoking cessation. Journal of Strategic Marketing, 16(4), 315e325.
Wakefield, M., Bayly, M., Durkin, S., Cotter, T., Mullin, S., & Warne, C. (2013).
Smokers’ responses to television advertisements about the serious harms of
tobacco use: pre-testing results from 10 low- to middle-income countries. To-
bacco Control, 22(1), 24e31.
Wakefield, M., Durkin, S., Murphy, M., & Cotter, T. (2007). Pre-testing anti-smoking
commercials. Process for the conduct of market research. Alexandria, NSW: Can-
cer Institute NSW and the Cancer Council Victoria.
Wakefield, M., Durrant, R., Terry-McElrath, Y., Ruel, E., Balch, G. I., Anderson, S., et al.
(2003). Appraisal of anti-smoking advertising by youth at risk for regular
smoking: a comparative study in the United States, Australia, and Britain. To-
bacco Control, 12(Suppl. II), ii82eii86.
Wakefield, M., Loken, B., & Hornik, R. C. (2010). Use of mass media campaigns to
change health behaviour. The Lancet, 376(9748), 1261e1271.
Wakefield, M., Spittal, M. J., Yong, H. H., Durkin, S. J., & Borland, R. (2011). Effects of
mass media campaign exposure intensity and durability on quit attempts in a
population-based cohort study. Health Education Research, 26(6), 988e997.
World Health Organization. (2003). WHO framework convention on tobacco control.
Geneva, Switzerland: World Health Organization. Available from http://
whqlibdoc.who.int/publications/2003/9241591013.pdf.