Combining in-school and community-based media efforts: reducing marijuana and alcohol uptake among younger adolescents
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
HEALTH EDUCATION RESEARCH Vol.21 no.1 2006
Theory & Practice Pages 157–167
Advance Access publication 30 September 2005
Combining in-school and community-based media
efforts: reducing marijuana and alcohol uptake among
younger adolescents
Michael D. Slater1,7, Kathleen J. Kelly2, Ruth W. Edwards3, Pamela J. Thurman3,
Barbara A. Plested3, Thomas J. Keefe4, Frank R. Lawrence5 and
Kimberly L. Henry6
Abstract sults suggest that an appropriately designed in-
school and community-based media effort can
This study tests the impact of an in-school reduce youth substance uptake. Effectiveness
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
mediated communication campaign based on does not depend on the presence of an in-school
social marketing principles, in combination prevention curriculum.
with a participatory, community-based media
effort, on marijuana, alcohol and tobacco
uptake among middle-school students. Eight Introduction
media treatment and eight control communities
throughout the US were randomly assigned to Despite some encouraging downward trends, use
condition. Within both media treatment and of substances including marijuana and alcohol
media control communities, one school received remains widespread among American adolescents.
a research-based prevention curriculum and Early initiation is commonplace, with 14.6% of
one school did not, resulting in a crossed, split- eighth graders reporting marijuana use and 38.7%
plot design. Four waves of longitudinal data reporting alcohol use in the past year (Johnston
were collected over 2 years in each school and et al., 2002). Reducing the rate of early uptake is
were analyzed using generalized linear mixed especially important given evidence that early
models to account for clustering effects. Youth initiation is predictive of a variety of negative
in intervention communities (N 5 4216) showed outcomes (Grant and Dawson, 1997).
fewer users at final post-test for marijuana A premise of the present study, consistent with
[odds ratio (OR) 5 0.50, P 5 0.019], alcohol a social-ecological framework (Berkman and
(OR 5 0.40, P 5 0.009) and cigarettes (OR 5 Kawachi, 2000), is that the norms and expectations
0.49, P 5 0.039), one-tailed. Growth trajectory that influence substance uptake among younger
results were significant for marijuana (P 5 adolescents are formed through a variety of social
0.040), marginal for alcohol (P 5 0.051) and experiences, including experience in school and in
non-significant for cigarettes (P 5 0.114). Re- the larger community. Reinforcement of non-use
norms and expectations, therefore, should ideally
be echoed and reinforced across these social envi-
1
School of Communication, The Ohio State University,
ronments (Flay, 2000). In the present study, we test
Columbus, OH 43210-1339, USA, 2Department of an intervention that includes an in-school media
Marketing, 3Tri-Ethnic Center for Prevention Research and campaign reinforced by participatory, community-
4
Department of Environmental Health, Colorado State based media efforts. This intervention is crossed
University, 5HDD Methodology Center, The Pennsylvania with implementation of a research-based prevention
State University and 6Institute of Behavioural Science,
University of Colorado
curriculum in selected schools.
7
Correspondence to: M. D. Slater; There is evidence that carefully designed,
E-mail: slater.59@osu.edu community-wide anti-drug advertising efforts can
Ó The Author 2005. Published by Oxford University Press. All rights reserved. doi:10.1093/her/cyh056
For permissions, please email: journals.permissions@oxfordjournals.orgM. D. Slater et al.
reduce youth marijuana (Palmgreen et al., 2001a) roaches, their record is mixed (Merzel and D’Afflitti,
and cigarette (Farrelly et al., 2002) use, and that 2003). Participatory community efforts focused
community-wide anti-smoking advertising in on mobilizing media, events and other commun-
conjunction with in-school prevention curricula ication strategies have the potential to reinforce
can reduce smoking uptake (Flynn et al., 1992, in-school communication efforts. We therefore
1994). There are two primary challenges to the expected a main effect for the combined community/
effectiveness of such media-based prevention ef- in-school media treatment on reducing increase in
forts: obtaining sufficient exposure to the messages substance uptake.
to achieve measurable impact (Hornik, 2002), and Similarly, we expected the prevention curricu-
identifying and executing message strategies that lum intervention to reduce substance uptake as well
can achieve such impact given adequate exposure (Tobler and Stratton, 1997). As Flay (Flay, 2000)
(Worden, 1999; Pechmann et al., 2003). notes, however, prevention curriculum effects tend
Ensuring exposure to anti-drug messages is typ- to decay and may require reinforcement (e.g. via
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
ically expensive, requiring paid advertising to en- media) elsewhere in the environment. Therefore,
sure delivery of the message to the desired audience. we also examined interaction effects of the media
However, the school environment provides a unique intervention and the school prevention curriculum.
opportunity to inexpensively ensure a relatively high
level of exposure to anti-use communication. In
Methods
addition, focusing communication efforts within a
school may influence youth perceptions of the norms
and expectations within an environment in which
Design, participants and data collection
they spend much of their day. This study utilized a randomized community design
A variety of message strategies have shown some to assess the effects of the community/in-school
success in influencing substance use-related atti- media intervention, with eight media treatment and
tudes and behaviors (Flynn et al., 1994; Palmgreen eight control communities. Communities random-
et al., 2001a; Pechmann et al., 2003). In this study, ized to the media treatment condition received both
we emphasized non-use as an expression of per- community- and in-school media prevention ef-
sonal identity and the consistency of non-use with forts. Data were collected in two middle or junior
youth aspirations (Slater and Kelly, 2002), in the high schools in each of these 16 communities. In
belief that such messages would not be redundant each community (media treatment or media con-
with already existing information regarding sub- trol), one of the two schools also received an
stance risks, more likely to reinforce non-use norms
and less likely to generate reactivity (Ringold,
2002). Another advantage of this strategy is that 8 Media
8 No Media
Communities
the same messages could address a variety of (Community
Communities
(No coalition
substances (i.e. marijuana, alcohol and tobacco), coalition
or media)
media efforts)
whereas risk messages are typically substance
specific. Given the limited resources of most
communities and schools, developing effective
cross-substance prevention strategies is advanta-
Curriculum 8 Curriculum
geous (Griffin et al., 2003). Treatment Comparison
8 Curriculum 8 Curriculum
The school is nested within the larger commu- Treatment Comparison
Schools (In- Schools (In-
Schools Schools (no
nity (Flay, 2000): participatory, community-based school school
(All Stars treatment of
media plus media/ no
approaches may help change youth substance be- All Stars) All Stars)
only) any kind)
haviors (Aguirre-Molina and Gorman, 1996; Perry
et al., 2002), although, like media prevention app- Fig. 1. Study experimental design.
158Combining in-school and community-based media efforts
in-school prevention curriculum and the other did been prohibitively lengthy. Treatment and control
not, creating a crossed design (see Figure 1); as communities were located in each of the four major
noted below, assignment of schools receiving the regions of the US (northeast, southeast, midwest
prevention curriculum was not fully randomized. and west). Because of the complexities of imple-
Four waves of data collection were conducted: the menting and managing this 2-year intervention,
first prior to initiating the in-school curriculum, recruitment and implementation were staggered
the second immediately following the last session, over a 4-year period, following an initial start-up
the third early in the fall of the second school year, year. The first communities began intervention
and the fourth and final wave in late spring of the activities in the fall of 1999 and the last commu-
second year. Data collection in the control schools nities completed the intervention in spring 2003.
was matched as closely as possible to data collec- Communities that entered the project were randomly
tion times in the schools receiving the prevention assigned to condition using a group-matching
curriculum. strategy to minimize the potential for confounded
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
Students (N = 4216) were recruited to the study effects due to random differences between treat-
using active consent procedures, required in this ment and control communities. Data on community
study given the provision of identifying informa- income, region, size, ethnic make-up, junior high
tion. Sixty-six percent of eligible students returned versus middle school configuration and readiness
signed consent forms and participated in at least one to engage as a community in prevention efforts
survey. A total of 68.6% of these participating (Edwards et al., 2000) were gathered from NCES
students provided data at all four measurement and other databases as well as from data collection
occasions; 16.8% provided data on three, 10.9% within each community. All possible assignment
provided data on two and 3.7% provided data on combinations were generated in the latter 2 re-
just one of the measurement occasions. Missing data cruitment years, including those communities pre-
were primarily the result of absence from school on viously assigned in earlier years (in the first year
the day of the survey or missed survey items. In pure random assignment was used for three com-
addition, the data for individual students who had munities). Treatment or control assignment was
indicators of inconsistent responding or exaggera- based on random selection of one of the combina-
tion at a given measurement occasion were removed tions in which no variables were different at P <
from the dataset. At any one measurement occa- 0.15 (from two such combinations in recruiting
sion, this equated to the removal of less than 2% of year 2 and from 10 such combinations in the
the students. The sample was approximately equal crucial recruiting year 3).
by gender (52% female/48% male). The majority of The two schools within each community were,
the sample was white (83.3%); 10.4% of respond- when possible, randomly assigned to either the
ents were African-American, 2.9% were Hispanic curriculum or no curriculum condition. However,
and 3.4% were of some other ethnic background. problems of scheduling and staffing in seven of the
As we were concerned with assessing students in 16 communities precluded assignment of a school
their first year of middle or junior high school, we to the curriculum treatment condition (usually be-
recruited sixth graders from the former and seventh cause key school staff members were new hires
graders from the latter (mean age at baseline = 12.2 and administrators were unwilling to burden them
years). As noted below, we balanced school/grade with a new curriculum). In such cases, school ad-
type between treatment and control conditions. ministrators provided written documentation to assure
Communities were recruited using the National us that assignment was not based on perceived
Center for Educational Statistics (NCES) database, need. As assignment to school condition, unlike
excluding the two largest census groupings, as the the media treatment, was not fully random, infer-
time required to gain approval for inclusion of a ences about prevention curriculum effects are qual-
prevention curriculum in larger districts would have ified accordingly. Because school curriculum was
159M. D. Slater et al.
of secondary concern (i.e. we wanted to examine tion, as discussed below, and adaptations were made
possible interactions of curriculum and media as needed.
intervention effects), this limitation does not One set of materials was developed for distri-
affect our analyses of primary interest. bution in the first year, with another set for the
second year in order to keep the campaign fresh.
Intervention design In the last year, schools in the media treatment
Social marketing principles were used to guide the communities were also offered the opportunity to
development of the media campaign to ensure a localize a poster with the campaign slogan for their
focus on influencing behavior change (Goldberg school by featuring a diverse group of students
et al., 1997; Kotler and Zaltman, 1971). Specifi- from that school. Typically, a school counselor
cally, primary and secondary research was used to or administrative support person was respons-
better understand adolescents’ attitudes, values and ible for distributing these media/social marketing
behaviors regarding substance use, and this knowl- materials.
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
edge guided the message strategy for the in-school The community-based, participatory communica-
media campaign, ‘Be Under Your Own Influence’. tion effort had several components. First, half-day
Additionally, through focus groups and personal community readiness workshops were conducted,
interviews we learned what types of promotional involving people active in prevention efforts in the
items (the products) and media channels (the community as identified using a snowball recruit-
places) were valued and attended to by the adoles- ment approach (Thurman et al., 2003; Slater et al.,
cent audience. These materials accordingly included 2005). In these workshops, trained project staff
print materials such as a series of posters as well as reviewed results of the community readiness assess-
promotional items such as book covers, tray liners, ment that had already been conducted to facilitate
T-shirts, water bottles, rulers and lanyards. assignment of communities to treatment condition
The central premise of our message strategy was and worked with community prevention leaders to
that the primary task of adolescence is attaining identify prevention strategies appropriate to their
greater independence and autonomy. A principal community’s level of readiness. This was followed
benefit of substance use therefore is likely to be the by a half-day session focused on community media
accompanying feeling of rebellious noncompliance by providing training in the use of campaign media
and independence. A principal cost is the risk to materials (including brochures, press releases, ideas
aspirations associated with greater maturity and for special events, posters and radio public service
autonomy. Our campaign was intended, therefore, announcements) (Hansen, 1996). Community pre-
to emphasize the inconsistency of drug (primarily vention leaders developed their own strategies for
marijuana) and alcohol use—and to a lesser extent this media effort and used whatever materials they
tobacco use—with one’s aspirations (Oman et al., either chose or developed on their own. Our in-
2004). In addition, the campaign sought to reframe tention was that community efforts would reinforce
substance use as an activity that impaired rather in-school efforts for youth by underscoring an anti-
than enhanced personal autonomy (Williams et al., drug community norm. A part-time project staff
1999). person developed new materials and provided on-
We thereby sought to decrease the cost and going support as needed for these community
increase the benefit of the non-use choice by ado- media efforts.
lescents. We also often used images such as rock- The in-school intervention was a research-based
climbing and four-wheeling that were appealing to cross-substance prevention curriculum, All StarsTM,
risk-oriented, sensation-seeking youth (Palmgreen which emphasizes non-use norms, commitment not
et al., 2001b). The campaign was monitored via to use and school bonding (Hansen et al., 1996;
qualitative and quantitative process evaluations Harrington et al., 2003). The curriculum involved
several times throughout campaign implementa- 13 sessions in the first year and seven booster
160Combining in-school and community-based media efforts
sessions in the second year; teachers were trained procedure incorporated in SAS version 9.0 (SAS
by experienced All StarsTM staff. Institute, Cary, NC).
The imputation model was as rich as our analytic
Measures model (Schafer, 1999), reducing the chance that the
For the lifetime incidence of alcohol intoxication imputation would bias the results. In addition, the
score, students responded to three questions: ‘‘Have imputation model included auxiliary items that
you ever gotten drunk?’’, ‘‘How old were you the were not used in the analyses, but were useful in
first time you got drunk?’’, ‘‘How often in the last predicting missing values (i.e. attitudes, normative
month have you gotten drunk?’’. For the lifetime beliefs, demographic variables, etc.) The parameter
smoking score, students responded to three ques- estimates reported below reflect combined esti-
tions: ‘‘Have you ever smoked cigarettes?’’, ‘‘Do mates from analyses done on 10 imputed data
you smoke cigarettes?’’, ‘‘In using cigarettes are sets. The 10 data sets yielded a relative efficiency
you a ... (non-user, very light user, light user, estimate of 95% for our parameter estimates
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
moderate user, heavy user or very heavy user)?’’. (Rubin, 1987).
For the lifetime marijuana score, students responded
to five questions: ‘‘Have you ever tried mari- Statistical analysis
juana?’’, ‘‘How often in the past month have you The unit of randomization in our design (the
used marijuana?’’, ‘‘How old were you the first time community) was used to compute degrees of
you used marijuana?’’, ‘‘Have you ever used mar- freedom for the test statistics (Murray et al., 1998,
ijuana when alone?’’, ‘‘In using marijuana are you 2004). This approach permits relatively unqualified
a ... (non-user, very light user, light user, moderate assertions of support for causal claims, in that the
user, heavy user or very heavy user)?’’. Items were degrees of freedom reflect a true community-
from the American Drug and Alcohol SurveyTM, randomized experimental design.
used by permission of the Rocky Mountain Behav- The model used was a four-level (measurement
ioral Science Institute. occasion within individual within school within
An affirmative response to any of the items community) random-intercept model. Random-
resulted in a score of ‘‘1’’ for that particular lifetime slope models were initially attempted, but failed
use score, while students who indicated in all the to converge because the variance among slopes
items that they had never tried the substance re- approached zero. Slopes were therefore treated as
ceived a score of ‘‘0’’. As one would expect, given fixed (global tests of model fit indicated the fixed
that students experimenting with use might endorse slope model fit the data better than a random slope
past trial but not past month use, reliability across model). Two-stage analyses were not used to permit
these items was quite good, but not perfect. The estimation of random slopes because such analyses
a values for the lifetime marijuana use measure ignore variability in subordinate cluster sizes
varied from 0.88 to 0.92 across the four waves; (school size was quite variable in this study, so
a varied from 0.83 to 0.88 for lifetime intoxication, this limitation would have distorted results in a non-
and from 0.88 to 0.90 for lifetime cigarette use. trivial way) and because of problems associated
The timing of the data collection varied among with estimation of standard errors in two-stage
participants. Therefore, our model was constructed analyses (Verbeke and Molenberghs, 2000). To
to allow for intervals of different lengths between address non-linearity issues, over-dispersion (i.e.
measurements (Brown and Prescott, 1999; Wallace conditional variance larger than implied by the
and Green, 2002). model) and the clustering effects, we used gener-
alized linear mixed models (McCullagh and Nelder,
Missing data 1989; Rotnitzky and Jewell, 1990; Hastie and
For our study, missing data were treated using Pregibon, 1993; Lee and Nelder, 1996; McCulloch
multiple imputation (MI). We employed the MI and Searle, 2001).
161M. D. Slater et al.
The intra-class correlations (ICCs) indicated that One test of our hypothesis regarding effects at
relatively little variation in the outcome variables the conclusion of our intervention is a test of the
(0.01% for marijuana, 0.02% for alcohol and 1.3% treatment’s main effect on Wave 4 intercepts; the
for cigarettes) was explained at the school level. model intercept was placed at the last measurement
Clustering at the community level accounted for occasion (Raudenbush and Bryk, 2002). This test
5.7% of the variance for marijuana, 8.5% for has the advantages of superior statistical power and
alcohol and 19.5% for cigarettes; all analyses in- easily interpreted odds ratios (ORs). The treatment 3
corporated the random intercepts for individual, time interaction provides useful additional infor-
school and community. mation regarding the effect of treatment on the
The fixed-effects portion of the model treated linear rate of change. It also provides a more rig-
substance use as a function of media treatment, orous, although less statistically powerful, test of
curriculum treatment, time and treatment interact- hypotheses, because it is not subject to baseline
ing with time. Other interactions (e.g. media treat- differences in outcome measures. (However, it
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
ment 3 school curriculum treatment and the should be noted that these baseline differences did
various higher-order interactions) were also as- not approach statistical significance).
sessed. Only those interactions that were significant We report one-tailed tests of significance because
for at least one substance outcome were included in our hypotheses are directional and because we
the final model (summarized in Table I). conducted preliminary examination of the data
Table I. Fixed effects for four-level random effects models using Murray et al. d.f. recommendations
Log-odds SE t d.f. P OR
Marijuana
intercept ÿ1.528 0.214 ÿ7.14 14Combining in-school and community-based media efforts
prior to project completion to ensure that our inter- treatment communities at the last measurement
vention was not causing iatrogenic effects. Had occasion, i.e. substance use uptake for youth in
we observed such effects, the experiment would treatment communities was half or less than that
have been terminated. Therefore, use of two-tailed of control communities by Wave 4. The Wave 4
tests for these data is superfluous as effects results for the media treatment were clearly signif-
opposite the hypothesized direction would not icant for marijuana [t (14) = 2.30, P = 0.019] and
have had the opportunity to be assessed in this alcohol [t(14) = 2.72, P = 0.009, one-tailed]. Even
model. though the Wave 4 effects for cigarettes were
statistically significant, they were less robust than
were effects on the other substances [t(14) = 1.90,
Results P = 0.039]. The percent of youth using each sub-
stance by study condition at Time 4 is shown in
Process evaluation and exposure Figure 1.
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
manipulation check The treatment 3 time interaction, as noted above,
Qualitative results using in-depth interviews with provides a more rigorous though less statistically
key community coalition and school district partic- powerful and less easily interpreted test of treatment
ipants indicated that the community and in-school effect. For the media treatment 3 time interaction
media interventions were successfully implemen- on marijuana, effects were significant [t(14) = 1.89,
ted, although with some variation in intensity, in all P = 0.040]. The media treatment 3 time interaction
treatment communities. The mix of media and was marginally significant for alcohol [t(14) = 1.75,
communication approaches used varied based on P = 0.051] and was not statistically significant for
community interest, capabilities, resources and cigarettes [t(14) = 1.26, P = 0.114]. Figure 2
needs, consistent with a participatory model. Quan- illustrates the media treatment effect on rate of
titative assessment of treatment versus control ex- change in marijuana, alcohol and cigarette use.
posure differences was possible only with respect to Effects of the curriculum were statistically sig-
a sampling of advertisement-type messages that nificant at Wave 4 for each of the three substances
were used primarily in the in-school media in-
tervention. Three such messages were reproduced
in the evaluation instrument along with a foil (a
fake) message intended to reduce false recognition,
response set and research demand bias problems
associated with recognition measurement (Slater
and Kelly, 2002). After adjusting for recognition of
the foil, students exposed to the media campaign
were more likely to report recognition of selected
campaign messages at all post-test waves (Time 2,
OR=4.70, P < 0.0001; Time 3, OR=6.80, P <
0.0001; Time 4, OR=10.13, P < 0.0001).
Intervention effects
Results using community as the unit of random-
ization (Verbeke and Molenberghs, 2000) in a four-
level random effects model were supportive of
hypothesized community and in-school media ef-
fects (see Table I). The OR for using marijuana was Fig. 2. Percent of youth using each substance by study
0.50, for alcohol 0.40 and for cigarettes 0.49 for condition at final Wave 4 post-test.
163M. D. Slater et al.
as illustrated by Figure 2, culminating in the
relatively large Wave 4 treatment effects. Media
intervention effects on reducing cigarette initiation
were more problematic, as the analysis of traject-
ory effects was not significant. Effects on cigarette
uptake were less robust, perhaps because only some
messages included tobacco use, while all mentioned
drugs (or marijuana specifically) and alcohol.
A particularly encouraging dimension of this
intervention is that it appeared to influence several
substance outcomes, in particular marijuana and
alcohol use. The focus on autonomy and aspirations
(‘Be Under Your Own Influence’) was equally
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
applicable to both substances. Such multi-substance
approaches are particularly advantageous given the
limited resources and time available in most school
Fig. 3. Community treatment effect on rate of change in
marijuana, alcohol and cigarette use (averaged growth curves settings (Griffin et al., 2003).
as a function of elapsed time baseline data collection). The apparent effectiveness of this media inter-
vention may be attributed to several factors. One is
(P < 0.005; Table I). These effects must be inter- the message strategy, emphasizing the ways that
non-use meets immediate adolescent needs re-
preted with caution given the imperfect randomiza-
garding autonomy, and both personal and social
tion of the curriculum treatment described above;
success. Another possibility is that the media in-
moreover, post hoc analyses suggested significant
tervention had the advantage of ubiquity. Among
baseline differences favoring treatment schools.
young teens, beliefs and attitudes are highly dy-
Curriculum 3 media treatment interactions were
namic and changes in beliefs and attitudes tend to
not statistically significant, indicating that the effect
decay (Resnicow and Botvin, 1993), even when the
of curriculum was additive rather than synergistic.
The curriculum 3 time interactions also were not interventions that trigger such changes are rela-
tively intensive. If so, a media intervention in the
statistically significant.
school which remains continuously visible may
serve to keep desirable attitudes salient, accessible
Discussion and more likely to influence behavior (Fazio et al.,
1989), even though the intervention at any one time
Results provide support for the effectiveness of in- cannot be considered an intensive one.
school media efforts combined with participatory Inferences regarding the school curriculum can-
communication efforts at the community level by not be confidently drawn given problems with
final post-test—the odds of uptake at Time 4 were random assignment of the curriculum. However,
approximately twice as high for previously non- this is not a significant concern with respect to our
using members of the control group compared to primary objective of testing the media intervention.
their counterparts in the community-based and in- The pattern of results in Figure 1 suggests treatment
school media treatment group. strategies are not contingent nor synergistic (i.e.
These effects were confirmed by the treatment 3 effectiveness of neither the media nor the curricu-
time interaction for marijuana and were marginally lum treatment depended on the presence of the
supported (at P = 0.051) for alcohol, testing differ- other treatment), although as one would expect
ences between slope trajectories. Initially smaller the strongest effects appeared to be in the com-
treatment/control differences grew larger over time, bined condition. Study results are also qualified by
164Combining in-school and community-based media efforts
possible selection bias associated with use of active We also found, based on coding a minimum of
consent, as some research suggests that at-risk six pre- and post-test interviews with community
youth disproportionately do not provide parental key informants, that there were impacts of the
consent (Unger et al., 2004). Also, as noted earlier, media intervention on community knowledge of
the data structure was most appropriately analyzed the substance use issue and marginally on com-
with random intercept rather than random slope munity climate (Slater et al., 2005). We are con-
models, which could not be estimated due to the ducting further research to identify paths of influence
minimal variability of these slopes. While the small for the in-school and community campaigns, and
ICC for slope variability (less than 1% of variance) to determine the relative contributions of the in-
suggests this should have little impact on results, school and the community campaign components.
it is a limitation on inference, as random slope In the meantime, these results suggest that
analyses provide the most complete analyses of appropriately designed in-school and community-
group randomized trials (Murray et al., 2004). based media efforts can significantly reduce youth
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
During the period of this research the Office of substance uptake, and that such efforts can be used
National Drug Control Policy was engaged in an independently of, or in addition to, classroom
active media campaign on national television prevention curricula.
focused primarily on marijuana use prevention, and
the latter years of this effort also overlapped with
national advertising by the Legacy Foundation and
the truthTM campaign to discourage youth smoking.
Acknowledgements
The effects of these national campaigns on both
treatment and control communities should make for This research was supported by grant DA12360
a more conservative test of hypothesized intervention from the National Institute of Drug Abuse to the
impacts, but should not create any systematic bias. first author. The authors thank the research staff of
Another limitation of the present study is that the the Tri-Ethnic Center, Colorado State University,
relative contributions of the in-school media/social and project manager Linda Stapel for their efforts
marketing effort and the community-based commu- in executing this research project, William Hansen
nication efforts are impossible to disentangle within and Tanglewood Research for providing and
the media treatment condition. Clearly, the combi- implementing the in-school curriculum, and the
nation has considerable potential. However, costs administrators, teachers, students, and community
for taking the in-school intervention to wide-scale prevention leaders in study communities who made
dissemination would probably be much lower than this research possible.
for the community media effort.
From a research perspective, too, it is important
to establish the relative contributions of each pro- References
gram component (Flay, 2000). We do have evi-
dence that amount of exposure to the media Aguirre-Molina, M. and Gorman, D.M. (1996) Community-
materials within the treatment schools is predictive based approaches for the prevention of alcohol, tobacco and
other drug use. Public Health, 17, 337–358.
of treatment outcomes (Slater and Kelly, 2002). In Ajzen, I. and Fishbein, M. (1980) Understanding Attitudes
particular, we found that students from a subsample and Predicting Social Behavior. Prentice Hall, Englewood
of schools early in the study who recognized having Cliffs, NJ.
Berkman, L. and Kawachi, I. (2000) Social Epidemiology.
seen more of the campaign’s messages had greater Oxford University Press, New York.
aspirations inconsistent with marijuana use and that Brown, H. and Prescott, R. (1999) Applied Mixed Models in
that the effects of aspirations on marijuana use were Medicine. Wiley, Chichester.
Edwards, R.W., Thurman, P.J., Plested, B.A., et al. (2000)
mediated by intentions, consistent with the Theory Community readiness: research to practice. Journal of
of Reasoned Action (Ajzen and Fishbein, 1980). Community Psychology, 28, 291–307.
165M. D. Slater et al.
Farrelly, M.C., Healton, C.G., Davis, K.C., et al. (2002) Murray, D.M., Varnell, S.P. and Blitstein, J.L. (2004) Design
Getting to the truth: evaluating national tobacco counter- and analysis of group-randomized trials: a review of recent
marketing campaigns. American Journal of Public Health, methodological developments. American Journal of Public
92, 901–907. Health, 94, 423–433.
Fazio, R.H., Powell, M.C. and Williams, C.J. (1989) The role of Oman, R.F., Vesely, S., Aspy, C.B., et al. (2004) The potential
attitude accessibility in the attitude-to-behavior process. protective effect of youth assets on adolescent alcohol and
Journal of Consumer Research, 16, 280–288. drug use. American Journal of Public Health, 94, 1425–1430.
Flay, B.R. (2000) Approaches to substance use prevention Palmgreen, P., Donohew, L., Lorch, E.P., et al. (2001a)
utilizing school curriculum plus social environment change. Television campaigns and adolescent marijuana use: tests of
Addictive Behavior, 25, 861–885. sensation seeking targeting. American Journal of Public
Flynn, B.S., Worden, J.K., Seeker-Walker, R.H., et al. (1992) Health, 91, 292–295.
Prevention of cigarette smoking through mass media inter- Palmgreen, P., Donohew, L., Lorch, E.P., et al. (2001b)
ventions and school programs. American Journal of Public Television campaigns and sensation seeking targeting of
Health, 82, 827–835. adolescent marijuana use: a controlled time-series approach.
Flynn, B.S., Worden, J.K., Secker-Walker, R.H., et al. (1994) In Hornik, R.C. (ed.), Public Health Communication:
Mass media and school interventions for cigarette smoking Evidence for Behavior Change. Erlbaum, Mahwah, NJ,
prevention. American Journal of Public Health, 84, pp. 35–56.
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
1148–1150. Pechmann, C., Zhao, G., Goldberg, M., et al. (2003) What to
Goldberg, M.E., Fishbein, M. and Middlestadt, S.E. (1997) convey in antismoking advertisements for adolescents: the use
Social Marketing: Theoretical and Practical Perspectives. of protection motivation theory to identify effective message
Erlbaum, Mahwah, NJ, pp. 291–312. themes. Journal of Marketing, 67(2), 1–18.
Grant, B. and Dawson, D. (1997) Age of onset of alcohol use Perry, C.L., Williams, C.L., Komro, K.A., et al. (2002) Project
and its association with DSM-TV alcohol abuse and depend- Northland: long-term outcomes of community action to
ence: results from the National Longitudinal Alcohol Epidemi- reduce adolescent alcohol use. Health Education Research,
ologic Survey. Journal of Substance Abuse, 9, 103–110. 17, 117–132.
Griffin, K.W., Botvin, G.J., Nichols, T.R., et al. (2003) Raudenbush, S.W. and Bryk, A.S. (2002) Hierarchical Linear
Effectiveness of a universal drug abuse prevention ap- Models: Applications and Data Analysis Methods, 2nd edn.
proach for youth at high risk for substance use initiation. Sage, Thousand Oaks, CA.
Preventative Medicine, 36, 1–7. Resnicow, K. and Botvin, G. (1993) School-based substance use
Hansen, W.B. (1996) Pilot test results comparing the All Stars prevention programs—why do effects decay? Preventative
program with seventh grade D.A.R.E.: program integrity and Medicine, 22, 484–490.
mediating variable analysis. Substance Use and Misuse, 31, Ringold, D.J. (2002) Boomerang effects in response to public
1359–1377. health inventions: Some unintended consequences in the
Harrington, N.G., Lane, D.R., Donohew, L., et al. (2003) alcoholic beverage market. Journal of Consumer Policy, 25,
Persuasive strategies for effective anti-drug messages. 27–63.
Communication Monographs, 70(1), 16–38. Rotnitzky, A. and Jewell, N.P. (1990) Hypothesis testing of
Hastie, T.J. and Pregibon, D. (1993) Generalized linear models. regression parameters in semiparametric generalized linear
In Chambers, J.M. and Hastie, T.J. (eds), Statistical Models. models for cluster correlated data. Biometrika, 77, 485–497.
Chapman & Hall, London, pp. 195–248. Rubin, D.B. (1987) Multiple Imputation for Nonresponse in
Hornik, R.C. (2002) Exposure theory and evidence about all the Surveys. Wiley, New York.
ways it matters. Social Marketing Quarterly, 8(3), 30–37. Schafer, J.L. (1999) Multiple imputation: a primer. Statistical
Johnston, L.D., O’Malley, P.M. and Bachman, J.G. (2002) The Methods in Medical Research, 8, 3–15.
Monitoring the Future National Survey Results on Adolescent Slater, M.D. and Kelly, K.J. (2002) Testing alternative explan-
Drug Use: Overview of Key Findings, 2001. National Institute ations for exposure effects in media campaigns: the case of
of Drug Abuse, Bethesda, MD. a community-based, in-school media drug prevention project.
Kotler, P. and Zaltman, G. (1971) Social marketing: an approach Communication Research, 29, 367–389.
to planned social change. Journal of Marketing, 35, 3–12. Slater, M.D., Edwards, R.W., Plested, B.A., et al. (2005) Using
Lee, Y. and Nelder, J.A. (1996) Hierarchical generalized linear community readiness key informant assessments in a random-
models. Journal of the Royal Statistical Society B, 58, ized group trial: impact of a participatory community media
619–678. intervention. Journal of Community Health, 30, 39–53.
McCullagh, P. and Nelder, J.A. (1989) Generalized Linear Thurman, P.J., Edwards, R.W., Plested, B.A., et al. (2003)
Models, 2nd edn. Chapman & Hall, Boca Raton, LA. Honoring the differences: using community readiness to create
McCulloch, C.E. and Searle, S.R. (2001) Generalized, Linear culturally valid community interventions. In Bernal, G.,
and Mixed Models. Wiley, New York. Trimble, J., Burlew, K. and Leong, F. (eds), Handbook of
Merzel, C. and D’Afflitti, J. (2003) Reconsidering community- Racial and Ethnic Minority Psychology. Sage, Thousand
based health promotion: Promise, performance and potential. Oaks, CA, pp. 591–607.
American Journal of Public Health, 93, 557–574. Tobler, N.S. and Stratton, H.H. (1997) Effectiveness of school-
Murray, D.M., Hannan, P.J., Wolfinger, R.D., et al. (1998) based drug prevention programs: a meta-analysis of the
Analysis of data from group-randomized trials with repeat research. Journal of Primary Prevention, 18, 71–128.
observations on the same groups. Statistics in Medicine, 19, Unger, J.G., Gallaher, P., Palmer, P.H., et al. (2004) No news is
1581–1600. bad news: characteristics of adolescents who provide neither
166Combining in-school and community-based media efforts
parental consent nor refusal for participation in school-based Williams, G.C., Cox, E.M., Kouides, R., et al. (1999) Present-
survey research. Evaluation Reviews, 28, 52–63. ing the facts about smoking to adolescents—effects of an
Verbeke, G. and Molenberghs, G. (2000) Linear Mixed Models autonomy-supportive style. Archives of Pediatrics, 153,
for Longitudinal Data. Springer, New York. 959–964.
Wallace, D. and Green, S.B. (2002) Analysis of repeated Worden, J.K. (1999) Research in using mass media to
measures designs with linear mixed models. In Hershberger, prevent smoking. Nicotine and Tobacco Research, 1,
S.L. (ed.), Modeling Intraindividual Variability with Repeated S117–S121.
Measures Data: Methods and Applications. Erlbaum,
Mahwah, NJ, pp. 103–134. Received on May 23, 2005; accepted on August 15, 2005
Downloaded from http://her.oxfordjournals.org/ by guest on May 9, 2015
167You can also read
