Adolescent Reactance and Anti-Smoking
Campaigns: A Theoretical Approach
Joseph Grandpre, Eusebio M. Alvaro, Michael Burgoon,
Claude H. Miller, and John R. Hall
University of Arizona–College of Medicine
Children between the ages of 9 and 15 are a high-risk group for tobacco use. The
Centers for Disease Control estimates that first use of cigarettes among adolescents
has risen 30% over the past decade, and that more than 1.2 million people age <18 be-
came daily smokers in 1996 alone. Moreover, research indicating that awareness and
liking of cigarette advertisements is higher among adolescents than adults under-
scores the need to devote more effort to understanding reactions to tobacco-related
messages. Adding to this problem is the fact that the early gains of some successful
anti-tobacco interventions disappear as adolescents age.
Drawing on the theory of psychological reactance, a number of hypotheses were
tested that addressed the impact of pro- and anti-smoking messages on a variety of
outcomes, including participants’intended behaviors, evaluation of message sources,
and seeking of disconfirming information. All the messages were created and deliv
ered to 4th-, 7th-, and 10th-grade students via personal computers. The pattern of re
sults supports the theoretically derived hypotheses, indicating that grade level and
message type had a significant impact on the processing of tobacco-related mes
sages. Implications and suggestions for future tobacco prevention campaigns are dis
By most measures, anti-smoking public health campaigns have always succeeded
with younger children (Botvin & Eng, 1980). Educational programs are effective
in informing them as to the negative consequences of smoking. Given the plethora
of anti-smoking campaigns over the past two decades, today’s younger children
are more informed, hold more negative attitudes regarding smoking, and have
greater expressed intentions not to smoke than ever before. At present, we have,
overall, an informed adolescent population cognizant of the negative effects of
HEALTH COMMUNICATION, 15(3), 349–366
Copyright © 2003, Lawrence Erlbaum Associates, Inc.
Requests for reprints should be sent to Joseph Grandpre, Wyoming Department of Health, 6101
Yellowstone Road, Suite 259A, Cheyenne, WY 82007. E-mail: firstname.lastname@example.org
smoking. Yet one problem persists, with a few exceptions (e.g., Dent et al., 1995;
Worden, Flynn, Soloman, & Secker-Walker 1996); the early gains of many suc
cessful substance abuse interventions dissipate as adolescents age (Brown, 2001;
Rosenbaum & Hanson, 1998; Oei, 1987). A sizable percentage of adolescents
(20%–30%) initiate smoking behaviors as early as middle school, despite
anti-smoking beliefs, attitudes, knowledge, and intentions (CDC, 1998; Oei, 1987;
Pletcher & Schwarz, 2000).
Numerous causes of adolescent smoking uptake have been explored with peer
pressure continuing to be among the most accepted factors (e.g., Barber, Bolitho,
& Bertrand, 1999; Pearson & Lynn, 2000; Simons-Morton, Haynie, Crump,
Eitel, & Saylor, 2001). However, recent studies have questioned the notion that
“peer pressure” is the most potent predictor of smoking initiation among adoles
cents (Denscombe, 2001; Unger, 2000), and others have pointed instead to ciga
rette advertising as a more credible culprit (Pierce, Choi, Gilpin, Farkas, &
Berry, 1998; Romer & Jamieson, 2001; Stanton, Currie, Oei, & Silva, 1996;
U.S. Public Health Services Centers for Disease Control and Prevention, 1994).
Though one can question the relative import of peer pressure and other
psychosocial variables, it is clear that mass-mediated communication is perva-
sive and persuasive. The practical significance of these findings is more apparent
when juxtaposed with current data that suggest an upward trend in smoking be-
havior among certain portions of the adolescent population (Coogan et al., 1998;
MMRW, 2000; Pierce & Gilpin, 1995). This trend suggests that key moderating
variables remain unidentified.
The problem is relatively straightforward. Despite significant tobacco knowl-
edge and anti-tobacco attitudes and anti-tobacco behavioral intentions among
youth, a significant percentage of youth take up smoking during adolescence. One
commonly proposed solution is earlier intervention. However, such an approach
will do little to further increase already-high tobacco knowledge and anti-tobacco
attitudes and behaviors. It would appear that some change occurs during adoles
cence, and a novel solution would be to develop interventions that are responsive to
It has long been recognized that desire for independence and individuality
along with a concomitant disavowal of authority are characteristic of adoles
cence; moreover, adolescents perceive their individual efforts as being hampered
by external control forces (Botvin & Eng, 1980). Thus, messages restricting ado
lescent freedom conflict with needs for independence and compound percep
tions of external control. Psychological reactance theory (PRT; Brehm, 1966;
Brehm & Brehm, 1981) provides researchers with a framework for understand
ing and preventing adolescents’ negative reactions to anti-smoking messages in
that the theory was specifically formulated to address threats to attitudinal and
behavioral freedoms. Moreover, PRT may also explain adolescent receptivity to
GRANDPRE ET AL.
According to the theory of psychological reactance, persuasive communication
poses a potential threat to freedom. Essentially, reactance is motivated by the indi
vidual’s basic need for self-determination in effecting his or her own environment.
This need for effectanceand autonomy is predicated on the basic assumption that, re
garding certain specifiable areas of behavior, people have a distinct and strong pref
erence to perceive themselves as responsible for their own fate. The theory predicts
that, when an individual’s perceived freedom is threatened by a proscribed attitude
or behavior, the individualwill experiencea motivating pressure toward reestablish
ing the threatened freedom (Heilman & Toffler, 1976). The theory posits (Brehm,
1966; Brehm & Brehm, 1981) that people are aware of and feel they have the ability
to perform or engage in a set of actions or free behaviors at the present moment or
some time in thefuture.Further,these free behaviors can,accordingto their ability to
satisfy certain needs, vary in prominence and importance. Brehm (1966) hypothe
sized that thestrength of psychological reactance is manifested in apositiverelation-
ship between the degree of threat to a behavior and the importance of that behavior.
Basically a threat to a freedom increases not only the attractiveness of that freedom
but the attempts to exercise the freedom as well (Fogarty, 1997). As personal free-
doms (as well as the threats to those freedoms) are cognitive in nature, the reaction to
a threatened freedom can be different for each threat. However, various responses to
the arousal of reactance have been conclusively demonstrated.
One way to restore a threatened freedom is to simply engage in the forbidden
behavior or embrace the attitude threatened by the proscription (Brehm, 1966).
This response is of primary import for the purposes of this study or for the design-
ers of any persuasive message campaign. Such a response is termed “restoration”
since it satisfies or restores the targets’ need for self-determination and control. A
second response involves increasing the attractiveness of the threatened freedom
(Brehm & Sensening, 1996; Worchel, 1974) even if the threat results from the in
troduction of a new, more attractive, alternative or reward as a choice option or in
centive. In such an event, the threatened behavior’s increased attraction may stimu
late information seeking to confirm or gauge the level of attraction. Additionally,
reactance may often be followed by aggression or hostility aimed at the threaten
ing agent (Wicklund, 1974; Worchel & Brehm, 1970). In this instance the response
would also be expected to include concomitant message and source derogation as a
potent form of restoration of the threatened freedom.
DEVELOPMENTAL CHANGES INFLUENCING
REACTANCE TO AUTHORITY
Prior to adolescence, children believe that adults are very knowledgeable, and
most children accept what adult authority figures tell them without question
ADOLESCENT REACTANCE 351
(Caissy, 1994). However, with onset of puberty, maturational factors lead children
to desire, experience, and express the need for attitudinal freedom as a way of
adopting an adult identity. Once children enter puberty they take the liberty of
forming their own judgments and begin to observe the world more critically. They
tend to question what their parents or other authorities say, and their receptivity to
messages from adults decreases drastically.
Furthermore, a number of cognitive and psychosocial risk factors emerge dur
ing early adolescence, and these factors may lead to the onset of smoking and other
risky behaviors (Caissy, 1994). Such factors include the following: (a) social com
parison with peer groups as one contributing factor; (b) sensitivity to peer pressure;
(c) a tendency to overestimate the prevalence of smoking among peers; (d) the ten
dency to engage in shallow, oversimplified thinking; (e) the emulation of adult be
havior combined with a lack of understanding of the responsibilities and conse
quences of such behavior; (f) feelings of invincibility; and (g) a rebellion against
authority and adult control (Caissy, 1994; Ruble, 1983; Tinsley, 1992). These fac
tors suggest that adolescents perceive themselves as more capable of realizing and
making choices while perhaps being less aware of the limitations of these choices.
Both processes increase the likelihood that adolescents will experience reactance
to anti-smoking messages, whereas pre-adolescents are less likely to experience
Brehm (1966) has described reactance as situationally generated; however, re-
cently, researchers have examined reactance as a personality trait (e.g., Burgoon,
Alvaro, Grandpre, & Voloudakis, in press; Poorman, 2000). Factors associated
with freedom of choice, conformity, behavioral freedom, and response to advice
underlie this approach to reactance as a trait (Hong & Page, 1989). Studies show-
ing that reactance is highest among younger individuals (e.g., Hong,
Giannakopoulous, Laing, & Williams, 1994) may indicate the presence of the un
derlying factors characterizing this trait during adolescence. This may well lead to
a peaking of reactance in adolescence because of their desire for independence, in
dividuality, and disavowal of authority.
One additional theoretical note is of importance; a significant body of research
strongly indicates that participants who are aware of the intent to persuade on the
part of the influencing agent will be less persuadable (e.g., Walster & Festinger,
1962). Conversely, the less explicit the intent, the more receptive the participant
will be to the influence (Weinstein, Grubb, & Vautier, 1986). Thus, perceived per
suasive intent along with explicitly persuasive messages may result in powerful
threats to freedom (White, 1959). This effect is likely to be robust during adoles
cence in that, prior to this age, children are often confused by the persuasive intent
of advertising. For example, it has been demonstrated that young children are un
able to correctly utilize a variety of attribution principles when making judgments
about causality and motivation (Sedlack & Kurtz, 1981), whereas for older chil
dren the persuasive intent of a message can take several concurrent avenues and
GRANDPRE ET AL.
still be understood (Moore & Lutz, 2000). In adolescence, enhanced needs for in
dependence and the expression of individuality (Botvin & Eng, 1980) may contrib
ute to both the salience of “controlling” cues and the concomitant attribution of
manipulation to messages sources utilizing such cues.
REACTANCE AND ANTI-SMOKING MESSAGES
Reactance theory, then, may explain how anti-smoking messages produce more
reactance in adolescents than in younger children. First, reactance is a function of
the importance of threatened freedoms: As children age there is a concomitant in
crease in the need for self-determination, self-definition, and the assertion of indi
viduality, making freedom in general more important with age. Second, reactance
is a function of the number of freedoms threatened: As children age, a number of
cognitive–developmental factors may be conducive to constructing links between
threats to smoking and a more general perceived threat to all freedoms. Third, for
reactance to occur, individuals must be aware that freedoms exist and they must
have the competence to pursue those freedoms: By adolescence, children have de-
veloped a healthy appreciation for individual liberty and have begun to build confi-
dence in themselves and their own effectance. Moreover, by the time they enter
middle school, children have been provided with a substantial amount of informa-
tion regarding both the pros and cons of smoking behaviors from school curricula,
family, friends, and the media; they likely feel they are competent to make such be-
havior choices. Therefore, faced with explicitly controlling messages, adolescents’
perceptions of external manipulation, coupled with a belief in their own compe
tence, may result in increased reactance toward those messages. These consider
ations lead to the formation of the first hypothesis.
H1: There will be a nonmonotonic relation between grade level and (a)
negative evaluation of persuasive messages, (b) derogation of persua
sive sources, and (c) behavioral intentions to perform behaviors coun
ter to those advocated by the message, peaking in 7th grade.
Reactance theory can also inform researchers as to potential message factors re
sponsible for modifying reactance. Given that reactance is a function of perceived
persuasive intent, messages perceived as explicitly persuasive will result in greater
reactance than more implicit messages. Second, restoring freedom increases per
suasion; and third, informing receivers of their ability to disagree reduces
reactance (Brehm & Brehm, 1981). Therefore, messages minimizing persuasive
intent and stressing freedom of choice may influence reactance regardless of mes
sage content—as specified in the second hypothesis.
ADOLESCENT REACTANCE 353
H2: Persuasive messages emphasizing freedom of choice (implicit) will
result in (a) more positive message evaluation, (b) less derogation of
persuasive sources, and (c) greater behavioral intentions to perform
the behavior advocated by the message.
To utilize reactance theory as an explanatory framework, it is crucial to illus
trate that explicitly threatening persuasive messages, regardless of the pro- or
anti-tobacco position advocated in the message content, will arouse reactance.
To this end, a research program testing a reactance theory explanation should
test both anti- and pro-smoking messages. This is a central concern, as reactance
is predicated on message features associated with threats to freedom, not on
message content. It is also the case that tobacco advertising appears to intention
ally steer clear of these types of threats and possible reactance by presenting an
image of tobacco that is inviting and nonthreatening to adolescents. Conse
quently, to establish the effect of reactance and message type regardless of the
position advocated by the message, new messages were constructed for both po-
sitions (pro- and anti-smoking).
Design and Participants
The design for this study was a 3 (4th, 7th, 10th grades) × 2 (message type im-
plicit–explicit) × 2 (message position pro–anti-smoking) factorial for a total of
12 cells. This study targeted public school children at three grade levels: 4, 7,
and 10, in a metropolitan city located in the southwestern United States. Study
participants included a total of 612 students (N = 205 in 4th, 256 in 7th, and 151
in 10th grade) attending 22 different elementary, middle, and high schools. Be
cause the computer randomly assigned students to one of the four message con
ditions, each condition received almost exactly the same number of students
(anti-explicit = 154, anti-implicit = 156, pro-implicit = 154, and pro-explicit =
153). However, to completely ensure that the computerized techniques utilized
in this study were successful in randomly assigning participants to experimental
conditions, a chi-square analysis was employed to test for systematic differences
between conditions. There was no significant difference in the number of partici
pants in each condition for the three grade levels (see Table 1). Additionally, the
students in this study came from a broad range of socioeconomic and ethnic
backgrounds (approximately 66% White, 16% Hispanic, 4% Native American,
2% African American, 2% Asian, and 10% other). Consent was obtained from
district and school administration, student’s parent–guardians, and assent was
obtained from the students.
GRANDPRE ET AL.
The creation and presentation of each experimental message (pro-implicit, pro-ex
plicit, anti-implicit, and anti-explicit) was accomplished through the use of current
computer technology. These messages utilized interactive multimedia computer
programs to merge photographic stills, music, full-motion video, audio voice-over,
and text to give each message the appearance of extant print and electronic
pro-smoking and anti-smoking information campaigns. The use of computer tech-
nology allowed the researchers to control the delivery of the same independent
variables across message type and position while maintaining high production val-
ues. To this end, each message maintained a consistent presentation of visual ele-
ments, (e.g., video, stills, and text) and audio style (e.g., use of narration and mu-
sic), as well as the overall affective tone. This consistency was necessary given that
extant messages vary according to position advocated; anti-smoking messages
take advantage of all available media whereas pro-smoking messages are restricted
to print media only. The use of computer technology was also beneficial in that it
permitted the researchers to present the experimental messages and collect all rele
vant data in the same medium at the same time.
It is important to note that, as already mentioned, this study employed both im
plicit and explicit pro-smoking messages to test the theoretical hypotheses. This
was done not only to clearly show that the type of message currently used by to
bacco companies (implicit) is effective in persuading adolescents, but also to show
that explicit messages elicit reactance regardless of the position of the message
(pro or anti). Care was taken to ensure that the pro-smoking messages were no
more or less persuasive than the anti-smoking messages but, at the same time, were
no more persuasive than other television commercials that adolescents are exposed
to on a daily basis. As stated earlier, both the pro- and anti-smoking messages (im
plicit or explicit) used the same visual images; only the voice-over elements were
different. The voice-over element for the explicit pro-smoking message stated such
things as “Smoking is cool” or “Smoking makes you look mature” but did not ex
ADOLESCENT REACTANCE 355
Number of Participants in Each Condition by Grade Level
4th 7th 10th Total
Anti-smoking implicit 48 65 39 152
Anti-smoking explicit 54 62 39 155
Pro-smoking implicit 50 67 36 153
Pro-smoking explicit 53 62 37 152
Total 205 256 151 612
(6) = .864, p = .990.
plicitly tell the students to go purchase a pack of cigarettes, unlike many other tele
vision advertising that children are exposed to every day. It was never the intent of
the researchers to advocate cigarette smoking; however, to fully test the study hy
potheses, pro-smoking messages had to be included.
In addition, each of the messages—especially the explicit pro-smoking—were
subjected to the scrutiny of each school principle and teacher that participated in
this study along with the superintendent of each participating school district. Ac
tive consent forms were sent home with students only after the superintendents,
principles, and teachers were satisfied that exposing students to a single pro-smok
ing message would not lead to an uptake of tobacco by the student.
In fact, many of the school administrators favored the inclusion of the
pro-smoking message from a media literacy perspective. They viewed the explicit
pro-smoking message as an example of what the tobacco companies are “really”
trying tell adolescents in their more implicit magazine and billboard advertise
ments, and they thought the students might be more critical of future pro-smoking
advertisements after viewing that message.
Additionally, all messages were pilot tested utilizing students enrolled in a sum-
mer activities program at a local university prior to collecting data in the field. This
pilot test was designed not only as a manipulation check for the study messages,
but also to ensure the smooth functioning of the computer program and associated
data collection. In addition, the researchers needed to evaluate whether 4th-grade
students would be able to read and understand the questionnaire and instructions,
as well as estimate the amount of time needed for an individual student to complete
the survey prior to contacting area school districts as to their possible participation
in this study. The results of the pilot testing did indicate a difference in the student’s
evaluations of implicit and explicit messages. In addition, some problems with the
computer program and reading level of the questionnaire were revealed; however,
these problems were quickly addressed and a new version of the program was
loaded onto each of the computers.
Six desktop computers (supplied and transported by the investigation team) were
setup in a secure and quiet place at each school (e.g., school library) and the experi
ment was conducted under supervision of study investigators. Participants were
seated at a computer and instructed by an investigator as to how to start the mes
sage treatment. Each message began with an informed assent statement delivered
in text form, which asked participants to indicate assent to participate by clicking
with the mouse on a hot button, which then initiated the experimental stimuli. The
assent statement was followed by a user-friendly computer-generated question
naire designed to gather preliminary demographic data on the student.
GRANDPRE ET AL.
Once the participants had answered the demographic questions, they were ran
domly assigned to one of the four message conditions by the computer, and the
specific message for that condition began. Each message lasted approximately 2.5
min and contained the exact same visual elements as the other three messages. The
participants were supplied with headphones to listen to the audio portion of the
message, which was the only difference between the messages. After viewing the
experimental message, participants answered a series of questions measuring the
various outcome and mediating variables. Items were presented one screen at a
time by the computer program and participants indicated their responses via the
computer mouse. Once they completed the questionnaire, participants read a tex
tual debriefing about the purpose of the study, and what the investigators hoped to
learn from this research. The entire investigation, from introduction through de
briefing, lasted an average of 20–25 min per student depending on the reading abil
ity of the student. Once the debriefing was complete, the students were thanked for
their participation and returned to their classrooms.
Primary Outcome Measures
This study had three primary dependent variables of interest: source evaluation,
message evaluation, and behavioral intention.
Three different variables were used to investigate the
student’s appraisal of the message. The first variable (video evaluation) evaluated
the video components of the message using two semantic differential scales. Par-
ticipants were asked to assess the video on 5-point scales ranging from 1 (very
good)to5(very bad)and1(valuable)to5(worthless). A Cronbach’s α of .85 was
obtained for this scale.
Two other 5-point Likert-type scales ranging from 5 (strongly agree)to1
(strongly disagree) produced the overall message evaluation variable (Cronbach’s
α = .72). The first item asked the participant if they believed the message was try
ing to control their behavior. The second item asked whether the participant be
lieved the message was trying to manipulate what they thought. These items (con
trol and manipulation) were also analyzed independently of the overall message
evaluation variable. Finally, to examine the student’s perceived decisional free
dom, they were asked if they thought that the message they viewed allowed for de
cisional freedom on their part, using a 5-point Likert-type scale ranging from 1
(strongly agree)to5(strongly disagree). For each variable a higher score denotes a
more negative evaluation.
The student’s evaluation of the source of the message
was assessed using a source credibility scale modified from McCroskey (1966).
Participants were asked to indicate their evaluation of the source on 5-point seman
ADOLESCENT REACTANCE 357
tic differential scales anchored with four bipolar items. These scales included as
sessments of the source as dishonest–honest, untrustworthy–trustworthy, unin
formed–informed, and stupid–smart. Ratings of source evaluation were obtained
by calculating a mean score of these items for each participant. Reliability analysis
obtained a Cronbach’s α of .78 for source evaluation. A higher mean score indi
cates a more negative evaluation of the source by the participant.
Intention to smoke was assessed employing three sepa
rate items using 5-point Likert-type scales ranging from 1 (definitely yes)to5(def
initely no). Adolescents were asked if they (a) intended to smoke a cigarette within
the next year, (b) if they would smoke a cigarette offered by a friend, (c) if they
might try a cigarette soon. These three items were combined into a single overall
measure of intent to smoke (Cronbach’s α = .81). In each case a higher score indi
cates a greater predilection to smoke. The three items making up the overall intent
measure were also analyzed independently.
Clear differences in smoking patterns by gender and
ethnicity have been documented (U.S. Department of Health and Human Services,
1993). These factors may affect perceptions of one’s freedom; therefore, we mea-
sured both ethnicity and gender. We also confirmed what grade the student was in
and what languages they spoke.
All dependent variables were tested using standard analysis of variance (ANOVA)
procedures with alpha set at .05 in all tests.
Analysis of video evaluation resulted in a significant main effect for grade in the
student’s evaluation of the video, F(2, 609) = 4.144, p = .016, η
= .013. Sev
enth-grade students were most positive about the video itself (M = 2.70), whereas
10th- (M = 2.93) and 4th-graders (M = 3.00) were more negative in their evaluation
of the video. A curvilinear analysis showed a significant quadratic trend, F(912) =
5.32, p = .005, for video evaluation across grade. In addition, there was a signifi
cant main effect for message type, F(1, 613) = 21.079, p < .001, η
= .033, for
video evaluation. Overall, the videos were rated more positively in the implicit
conditions (M = 2.66) than in the explicit (M = 3.06). Parenthetically, students had
a significantly more negative evaluation of the pro-smoking messages than the
anti-smoking video, regardless of message type. Finally, a significant main effect
for grade, F(2, 606) = 4.250, p = .015, η
= .014, was found in the area of self-deci
GRANDPRE ET AL.
sion. Regardless of which message they viewed, 10th-grade students (M = 2.85)
felt significantly less free to make their own choices than 7th- (M = 2.52) or
4th-grade (M = 2.51) students. This difference in decisional freedom resulted in a
significant quadratic trend across grade, F(2, 912) = 7.02, p = .001.
A significant main effect for grade, F(2, 606) = 5.901, p = .003, η
= .020, was
found for source evaluation. Regardless of message type or position, 4th-grade stu
dents had the most negative view of the source (M = 2.92), followed by 7th-graders
(M = 2.64) with 10th-grade students (M = 2.60) having the least negative impres
sion of the source. A curvilinear analysis for source evaluation across grade indi
cated a significant quadratic trend, F(2, 912) = 6.17, p = .002.
There was also a significant main effect for message type, F(1, 613) = 19.874, p
< .001, η
= .03, found for source evaluation. Overall, implicit messages resulted in
more positive source evaluation (M = 2.57) than explicit messages (M = 2.88). In-
cidentally, the students participating in this study had a more negative evaluation of
the source of pro-smoking messages than the source of the anti-smoking messages,
regardless of message type.
There was a significant main effect for overall intent to smoke found between the
three grade levels, F(2, 599) = 58.81, p < .001, η
= .164. Overall, 4th-graders indi-
cated the least intent to smoke (M = 1.22), with 7th-graders in the middle (M =
1.57) and 10th-grade students indicating the most intent to smoke (M = 2.39). A
significant quadratic trend was found for overall intent by grade, F(2, 897) = 83.17,
p < .001.
The three behavioral intent measures were then tested separately to examine
whether this pattern (4th grade lowest, 7th next, and 10th grade highest) held for
each measure of intent. A significant main effect, F(2, 599) = 54.54, p < .001, η
.154, was found for grade and student’s intent to smoke in the next year. When
asked if they intend to smoke a cigarette in the next year 4th-grade students were
again the least likely to smoke (M = 1.22), with 7th-graders in the middle (M =
1.43) and 10th-grade students most likely to smoke in the next year (M = 2.35). A
significant quadratic trend was found for intent to smoke in the next year across
grade, F(2, 897) = 76.33, p < .001.
However, the pattern for grade level changed when the students were asked if
they might try a cigarette soon, F(2, 523) = 5.553, p = .004, η
= .021, or if they
would smoke a cigarette offered by a friend, F(2, 523) = 13.650, p < .001, η
.050. In both cases, 7th-graders were more likely to smoke (M
= 1.50, M
ADOLESCENT REACTANCE 359
= 1.51) than either 10th- (M
= 1.41, M
= 1.47) or 4th-graders
= 1.26, M
= 1.17). Significant quadratic trends for grade were
found for both might try, F(2, 788) = 8.93, p < .001, and friend offers, F(2, 788) =
16.65, p < .001. There were no significant main effects in message type for overall
intent, intention to smoke in the next year, might try a cigarette soon, or if a friend
offered a cigarette.
An omnibus ANOVA analysis on overall message evaluation revealed a significant
interaction effect, F(2, 606) = 4.531, p = .030, η
= .012, for grade by message type
(implicit vs. explicit). In the two implicit conditions, 7th- (M = 2.95) and
10th-graders (M = 2.96) students had a significantly more positive overall evalua
tion of the message than 4th-graders (M = 3.33). On the other hand, in the explicit
conditions 7th-graders were significantly more negative toward the message (M =
3.32) than 4th- (M = 3.14) or 10th-graders (M = 3.17). A curvilinear analysis of
overall message evaluation revealed only a significant linear trend for grade, F(2,
913) = 4.99, p = .026.
Additionally, a significant interaction was found for grade by message type in ex-
amining whethera student mighttry a cigarettesoon, F(2, 520)= 3.607, p= .028, η
.014. Analysis of the means reveledno significant differencesbetween implicit mes-
sage and explicit messages for 4th- (M
= 1.34, M
= 1.18) and 7th-graders
= 1.55, M
= 1.45). However, there was a significant difference be-
tween implicit and explicit messages in the 10th grade, such that these students re-
ported beinglesslikely to try acigarette after viewingan implicit message (M =1.23)
than after viewing an explicit message (M = 1.56; see Figure 1).
GRANDPRE ET AL.
FIGURE 1 Might try a cigarette soon by grade and message type.
Despite a profusion of state and national anti-smoking campaigns, adolescent cig
arette smoking continues to be a major health concern in the United States. Be
tween 1989 and 1996 the incidence of first use of cigarettes among adolescents
rose by 30%, and it is estimated that over 1.2 million people aged <18 became
daily smokers in 1996 alone (CDC, 1998). Additionally, according to the 1999 Na
tional Youth Tobacco Survey, 12.8% of middle school students and 34.8% of high
school students currently use some form of tobacco (CDC, 2000), with approxi
mately 3,000 youths becoming established smokers each day (Gilpin, Choi, Berry,
& Pierce, 1999). Clearly messages about the harmful effects of cigarette smoking
are out there, but it seems that young people are not heeding these warnings. Con
sequently, one must ask the question, are there any unidentified factors that may af
fect adolescent reactions to tobacco prevention messages. This study set out to
help answer this question by focusing on the types of messages employed in anti-
and pro-smoking media campaigns. Reactance theory provided a template to ex-
amine whether the type of language normatively utilized in anti-smoking cam-
paigns simply antagonizes young people. By determining what elements make for
an ineffective anti-smoking message, it is hoped that we can produce more effica-
cious tobacco prevention messages and campaigns in the future.
Hypothesis 1 was developed with the understanding that
reactance is driven by desires for self-determination and independence. Given the
combination of the onset of adolescence, the need for self-determinism (both be
ginning at approximately 12 years of age), and a lack of freedom-restoration skills,
it was assumed that reactance would be greatest in the 7th grade.
Participants in the 10th grade perceived the messages as impinging on their
freedoms significantly more than students in the 7th and 4th grades do.
Tenth-grade students felt that the messages they watched did not allow for deci
sional freedom regarding whether they should or should not smoke cigarettes.
When evaluating the quality of the video 7th-graders were significantly more fa
vorable than 4th- or 10th-graders, meaning the aesthetic qualities of the video ap
pealed most to the 7th-grade students. Feeling that the messages impinged on their
freedom to smoke, 10th-graders may have evaluated the video quality more nega
tively. Participants in the 4th grade may not have been able to correctly interpret
the video evaluation questions. Counter to expectations, 4th-grade students be
lieved that the people who made the video were more dishonest, untrustworthy,
uniformed, and stupid than did 7th- or 10th-graders. Lastly, when looking at a
composite measure of smoking intentions, 10th-graders were more likely to
smoke. However, analyses of the specific items in this composite measure revealed
ADOLESCENT REACTANCE 361
that, whereas 10th-graders had the highest intent to smoke next year, participants
in the 7th grade were more likely to try a cigarette soon and smoke a cigarette of
fered by a friend. The smoking intent results for 10th-grade students are mitigated
by the fact that there were more current smokers in 10th grade who, of course, had
a greater intention of smoking in the near future. The fact that students in the 7th
grade were more likely to try tobacco in the future and would smoke a cigarette of
fered by a friend suggests that, though these students are not currently smoking,
they are very susceptible to tobacco uptake in the future.
Taken together, the aforementioned pattern of results suggests that a
nonmonatonic relation exists between reactance and grade level. However, the na
ture of this relation is more complicated than anticipated by the stated hypothesis.
Due to significant variability in developmental factors, such as the onset of pu
berty, grade level may not have been a reliable indicator of participant’s need for
independence and self-determination. Nonetheless, it does appear that reactance
was more pronounced at the higher grade levels. This pattern of results is more
clearly interpretable when discussing the obtained interactions between grade
level and message type (see later).
The second and more central hypothesis in this study gener-
ally proposed that implicit persuasive messages would result in less reactance than
explicit persuasive messages. This hypothesis was derived from the theoretical
proposition that messages that minimize freedom of choice and have an obvious
persuasive goal will result in greater reactance in message recipients. This hypoth-
esis received overall support, especially regarding message and source evaluation.
Strong support is provided by student’s responses to video evaluation items. As
expected, explicit messages were rated more negatively than implicit messages.
Support for the hypothesis was further bolstered by analyses concerning student’s
perceptions of the message source. As expected, implicit messages resulted in
more positive evaluations of message source than explicit messages.
Although not formally hypothesized, a number of interactions
consistent with the theoretical rationale were obtained. These results present a
clearer picture of how reactance functioned in determining responses to the experi
mental messages. Though there is mixed evidence that reactance is greatest at the
beginning of adolescence, the obtained interactions suggest that specific message
factors (e.g., controlling language) cue reactive responses. Moreover, these re
sponses are most prevalent at the beginning of adolescence. Specifically, 7th-grad
ers in the explicit condition evaluated the message more negatively than 4th- and
10th-graders. However, in the implicit condition both 7th- and 10th-graders evalu
ated the message more positively than did 4th-graders.
In assessing behavioral intentions, (might try a cigarette soon) grade level of the
student also played a significant role. Messages functioned as expected for
362 GRANDPRE ET AL.
10th-graders, with implicit messages resulting in less intent to try a cigarette soon
than explicit messages. However, message type had no significant impact on either
4th- or 7th-grader’s behavioral intentions.
Conclusion and Implications
Overall, as expected from the framework of reactance theory, grade level (age)
played a significant role in determining responses to persuasive messages. Given
increased tobacco-related knowledge, increased awareness of behavioral options
and an increased ability to engage in tobacco-related behaviors, it was expected
that reactance to persuasive messages would increase with age, peaking in children
around the age of 12 to 13. This is predicated on a newly acquired awareness of the
importance of individuality, self-expression, and self-determination.
However, as previously discussed, whereas reactance is evident in 7th-grade
students, the predicted peak in reactance may not occur until later in adolescence
(10th grade). Nonetheless, the pattern of support garnered for the study’s hypothe-
ses illustrates that messages targeting young children may not elicit the same posi-
tive responses from adolescents. Whereas younger message recipients may be ac-
customed to, or more tolerant of, behavioral restrictions, adolescents are less
receptive to messages targeting behavioral changes. Results of this study indicate
that such messages, although containing structural elements that appeal to adoles-
cents, may nevertheless be met with the rejection of message content.
These findings highlight a number of practical implications for the develop-
ment of future tobacco prevention campaigns targeting adolescents. First, preven-
tative interventions at earlier ages—a commonly prescribed recommendation—
may do little in curbing adolescents’ initiation of tobacco use. Young students
(4th-graders) appear to be quite knowledgeable of the harmful effect of tobacco
and have positive attitudes toward remaining smoke free. Yet, by adolescence, the
same messages that appear to be accepted in 4th grade are derogated and rejected.
Second, further research is required in maximizing the fit between audience
subgroups and those message features that are most effective in persuading those
groups. Messages and campaigns that are effective with 8-year-old audiences may
have little positive impact on teens. Worse yet, these messages may even boomer
ang and have negative effects on adolescents’ health behaviors. The other major
prediction of this study was that implicit messages would result in significantly
less reactance than explicit messages. Though support was mixed, the overall pat
tern was consistent with predictions. Especially significant is the fact that students
were more positive toward the message source and the video itself in the implicit
conditions. Given the important role played by the source of a persuasive message,
these findings deserve further consideration. Adolescents simply do not like hav
ing their choices limited and their options clearly delineated.
ADOLESCENT REACTANCE 363
These results suggest that it may be fruitful to allow adolescents more free
dom to make their own choices regarding healthy behaviors. Such an approach
would allow for the utilization of the significant storehouse of pre-existing
knowledge and positive attitudes toward being tobacco free. Existing campaigns
often make clever use of current technology in producing messages that cer
tainly gain the attention of adolescents. However, we must be careful not to
alienate the at-risk audience members by the content of our messages. In a
worst-case scenario, we would successfully capture the interest of an at-risk teen
only to have that teen turn their full attention to restoring a sense of lost freedom
by taking up contra-advocated behaviors. Instead, it may be possible to make in
triguing and attention-getting messages that stimulate thought about what it
means to be healthy, attractive, accepted, and independent. By allowing adoles
cents the freedom to actively consider diverse health choices, resultant decisions
may be seen as being wholly self-initiated.
Though the observed effect sizes in this study were small, it should be noted
that the differences between the explicit and implicit anti-smoking messages in
this study were very subtle. Add to this subtlety the fact that Arizona is conduct-
ing its own highly saturated statewide tobacco prevention campaign—with over
90% of teens having seen at least one commercial (Burgoon, Hendricks, Alvaro,
et al., 2000)—it is noteworthy that any significant effects were obtained. In real-
ity, given the plethora of pro- and anti-tobacco advertising seen by a child nearly
everyday, one short (2.5 min) anti-smoking message should not be expected to
produce any discernable differences between students who view implicit or ex-
plicit messages; however, this is exactly what happened in this study. This leads
the investigators of this study to believe that, though the effect sizes are not
large, the differences between implicit and explicit messages are real and in need
of further study.
Seemingly, the greatest challenge for designers of future health-related cam
paigns is to create implicit messages that result in desired outcomes without stipu
lating the parameters of possible options in the messages themselves. In the end,
internalized decisions about smoking and other health behaviors will be those that
are freely chosen.
This research is supported in part by Arizona Disease Control and Research Com
mission (ADCRC), Grant 9904, awarded to Michael Burgoon.
All opinions expressed are those of the authors, not necessarily those of the
GRANDPRE ET AL.
Barber, J. G., Bolitho, F., & Bertrand, L. D. (1999). The predictors of adolescent smoking. Journal of
Social Service Research, 26, 51–66.
Botvin, G. J., & Eng, A. (1980). A comprehensive school-based smoking prevention program. Journal
of School Health, 50, 209–213.
Brehm, J. W. (1966). A theory of psychological reactance. New York: Academic.
Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory of freedom and control. San
Diego, CA: Academic.
Brehm, J. W., & Sensenig, J. (1966). Social influence as a function of attempted and implied usurption
of choice. Journal of Personality and Social Psychology, 4, 703–707.
Brown, J. H. (2001). Youth, drugs and resilience education. Journal of Drug Education, 31, 83–122.
Burgoon, M., Alvaro, E., Grandpre, J., & Voloudakis, M. (2002). Revisiting the theory of psychological
reactance: Communicating threats to attitudinal freedom. In J. Dillard & M. Pfau (Eds.), The
persuasion handbook: Theory and practice.
Burgoon, M., Hendriks, A., Alvaro, E. M., Miller, C., Grandpre, J. R., Denning, V., et al. (2000). One
year later: Evaluation of Arizona’s tobacco prevention and cessation media campaign. Paper deliv
ered at the annual meeting of the American Public Health Association, Boston.
Caissy, G. A. (1994). Early adolescence. New York: Plenum.
Centers for Disease Control and Prevention (CDC). (1998). Incidence of initiation of cigarette smok-
ing—United States, 1965–1996. Morbidity & Mortality Weekly Report (MMWR), 47, 837–840.
Center for Disease Control and Prevention (CDC). (2000). Tobacco use among middle and high school
students—United States, 1999. Morbidity and Mortality Weekly Report, 49, 49–52.
Coogan, P. F., Adams, M., Geller, A. C., Brooks, D., Miller, D. R., Lew, R. A., et al. (1998). Factors as-
sociated with smoking among children and adolescents in Connecticut. American Journal of Preven-
tative Medicine, 15(1), 17–24.
Dent, C. W., Sussman, S., Stacy, A. W., Craig, S.,Burton, D., & Flay, B. R.(1995). Two-year behavior out-
comesof project towardnotobaccouse. JournalofConsulting& Clinical Psychology, 63,676–677.
Denscombe, M. (2001). Peer group pressure, young people and smoking: New developments and pol-
icy implications. Drugs-Education Prevention & Policy, 8, 7–32.
Fogarty, J. S. (1997). Reactance theory and patient noncompliance. Social Science and Medicine, 45,
Gilpin, E. A., Choi, W. S., Berry, C., & Pierce, J. P. (1999). How many adolescents start smoking each
day in the United States? Journal of Adolescent Health, 2, 248–255.
Heilman, M. E., & Toffler, B. L. (1976). Reacting to reactance: An interpersonal interpretation of the
need for freedom. Journal of Experimental Psychology, 12, 519–529.
Hong, S. M., Giannakopoulous, E., Laing, D., & Williams, N. A. (1994). Psychological reactance: Ef
fects of age and gender. Journal of Social Psychology, 134, 223–228.
Hong, S. M., & Page, S. (1989). A psychological reactance scale: Development, factor structure and re
liability. Psychological Reports, 64, 1323–1326.
McCroskey, J. (1966). Scales for the measurement of ethos. Speech Monographs, 33, 65–72.
Moore, E. S., & Lutz, R. J. (2000). Children, advertising, and product experiences: A multimethod in
quiry. Journal of Consumer Research, 27, 31–48.
Oei, T. P. (1987). Smoking prevention program for children: A review. Journal of Drug Education, 17,
Pearson, M., & Lynn, M. (2000). Smoke rings: Social network analysis of friendship groups, smoking
and drug-taking. Drugs-Education Prevention & Policy 7, 21–37.
Pierce, J. P., & Gilpin, E. A. (1995). A historical analysis of tobacco marketing and the uptake of smok
ing by youth in the United States: 1890–1977. Health Psychology, 14, 500–508.
Pierce, J. P., Choi, W. S., Gilpin, E. A., Farkas, A. J., & Berry, C. C. (1998). Tobacco industry promotion
of cigarettes and adolescent smoking. Journal of the American Medical Association, 279, 893–899.
Pletcher, J. R., & Schwarz, D. F. (2000). Current concepts in adolescent smoking. Current Opinion in
Pediatrics, 12, 444–449.
Poorman, M. O. (2000). The impact of reactance on the treatment progress of incarcerated male adoles
cent offenders. Dissertation Abstracts International, 60, 3576.
Romer, D., & Jamieson, P. (2001). Advertising, smoker imagery, and the diffusion of smoking behav
ior. In P. Slovic (Ed.), Smoking: Risk, perception & policy. Thousand Oaks, CA: Sage.
Rosenbaum, D. P., & Hanson, G. S. (1998). Assessing the effects of school-based drug education: A
six-year multilevel analysis of project D.A.R.E. Journal of Research in Crime and Delinquency, 35,
Ruble, D. N. (1983). The development of social-comparison processes and their role in achieve
ment-related self-socialization. In E. T. Higgins, D. Ruble, & W. W. Hartup (Eds.), Social cognition
and social development: A sociocultural perspective. Cambridge, NY: Cambridge University Press.
Sedlack, A. J., & Kurtz, S. T. (1981). A review of children’s use of causal inference principles. Child
Development, 52, 759–784.
Simons-Morton, B., Haynie, D. L., Crump, A. D., Eitel, P., & Saylor, K. E. (2001). Health Education &
Behavior, 28, 95–107.
Stanton, W. R., Curri G. D., Oei, T., & Silva, P. A. (1996). A developmental approach to influences on
adolescent’s smoking and quitting. Journal of Applied Developmental Psychology, 17, 307–319.
Tinsley, B. J. (1992). Multiple influences on the acquisition and socialization of children’s health atti-
tudes and behavior. An integrative review. Child Development, 63, 1043–1069.
U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health,
National Institute on Drug Abuse. (1993). National survey results on drug use from Monitoring the
future survey 1975–1992, Vol. 1. Publication No. 93–3598.
U.S. Public Health Services Centers for Disease Control and Prevention, Office on Smoking and
Health. (1994). Preventing tobacco use among young people: A report of the Surgeon General
(DHHS Publication). Washington, DC: U.S. Government Printing Office.
Walster, E., & Festinger, L. (1962). The effectiveness of “overheard” persuasive communications.
Journal of Personality and Social Psychology, 71, 285–290.
Weinstein, N. D., Grubb, P. D., & Vautier, J. S. (1986). Increasing automobile seat belt use: An interven
tion emphasizing risk susceptibility. Journal of Applied Psychology, 71, 285–290.
White, R. W. (1959). Motivation reconsidered: The concept of competence. Psychological Review, 66,
Wicklund, R. A. (1974). Freedom and reactance. New York: Wiley.
Worchel, S. (1974). The effect of three types of arbitrary thwarting on the instigation to aggression.
Journal of Personality, 42, 300–318.
Worchel, S., & Brehm, J. W. (1970). Effects of threats to attitudinal freedom as a function of agreement
with the communicator. Journal of Personality and Social Psychology, 14, 18–22.
Worden, J. K., Flynn, B. S., Solomon, L. J., & Secker-Walker, R. H. (1996). Using mass media to pre
vent cigarette smoking among adolescent girls. Health Education Quarterly, 24, 453–468.
366 GRANDPRE ET AL.