[Note: Some tables and charts are not available here. E-mail the author for a copy of the paper.]

Authors' Note: Mike Stephenson and Susan Colon are doctoral candidates in the Department of Communication at the University of Kentucky, where Philip Palmgreen and Lewis Donohew are Professors of Communication. Elizabeth Pugzles Lorch and Rick Hoyle are Associate Professors in the Department of Psychology and the University of Kentucky. This research is supported by Grant DA 06892-08 from the National Institute on Drug Abuse.

Sensation seeking, which research indicates is a biologically-based personality variable, is strongly related to both drug use and preferences for highly novel, arousing, and/or unconventional messages and TV programs. This connection is the basis of a targeting strategy in an anti-marijuana public service announcement campaign in a medium-sized market aimed at high sensation seeking adolescents. Data from the first half of the media campaign suggest that the project's anti-marijuana messages are reaching the target audience's marijuana-related beliefs, attitudes, and behaviors. The preliminary results are presented and implications for future campaigns are discussed.

It's like everywhere you walk you see somebody lighting up or smoking or something so they think that everybody else is doing it. - 8^th grade Anglo female

Beginning in the early 1990s, research by Johnston and associates' (1994) Monitoring the Future study revealed that more and more adolescents were beginning to experiment with marijuana. Since then the annual national-level report on substance use has documented a steady increase in adolescent 30-day and lifetime marijuana use. Concerned with this and the increased use of other illicit drugs, Congress recently appropriated 1-billion dollars to be spent on a five-year nationwide media campaign targeting, among other illicit drugs, marijuana use among 9 to 17 year-olds (Palmgreen & Donohew, in press). The Partnership for a Drug Free America already conducts a massive media campaign attempting to reach children and teenagers to educate them about the harmful effects of marijuana and other illicit drugs. Quite clearly, then, media campaigns appear to be one of the more preferred methods used to curb illicit substance use.

One approach that has been used to design drug prevention campaigns is the SENTAR (which stands for sensation seeking targeting) approach. Users of the SENTAR approach target a biologically-based trait called sensation seeking. "It is well established that high sensation seekers (HSS), including the important target group of HSS adolescents, are particularly drawn to the stimulation and/or mood altering effects of a variety of drugs" including marijuana (Palmgreen & Donohew, in press). Most importantly, however, is that HSS have a distinct preference for specific media message characteristics, such as intensity and novelty (Donohew, Palmgreen, & Lorch, 1991). What follows is an overview of the most recent SENTAR project that features an anti-marijuana theme, as well as preliminary results of the ongoing two-city anti-marijuana media campaign. First, however, it is helpful to review the theoretical foundation of this campaign, sensation seeking.

Sensation seeking is a biologically-based trait "based on the idea that persons differ reliably in their preferences for or aversions to stimuli or experiences with high-arousal potential" (Zuckerman, 1988, p. 174). Thus, individuals with a high need for sensation desire "complex experiences and [exhibit] the willingness to take risks for these experiences" (Bardo & Mueller, 1991, p. 196). This biological drive for novel, complex, and intense sensations and experiences is typically satisfied by a willingness to take more social risks (e.g., impulsive behaviors, sexual promiscuity), physical risks (e.g., skydiving, bungee jumping, driving fast), legal risks (e.g., getting arrested and put in jail), and financial risks (e.g., paying fines, impulsive purchases) (Zuckerman, 1979, 1994). In contrast, individuals with a low need for sensation are not only likely to take fewer social, physical, legal, and financial risks, but they also differ in preferences for visual complexity and music choice (Zuckerman, 1988).

Empirically, Zuckerman has determined that these differences in need for sensation can be measured with four subscales: thrill and adventure seeking, experience seeking, disinhibition, and boredom susceptibility. Thrill and adventure seeking (TAS) accounts for the sensation gained through participation in physically risky behaviors, such as parachuting or bungee jumping. Experience seeking (ES) is indicative of the non-conforming lifestyle sought by HSS often characterized by a liking for unusual music or abstract art, as well as making spontaneous decisions or exploring strange places. A third subscale, disinhibition (DI), represents a desire for sensation through socially risky behaviors, such as wild parties, social drinking or drug use, or a variety of sexual partners. Finally, boredom susceptibility (BS) expresses an aversion to routine and conventional activities, conditions, or people. The first three represent the need for novelty and stimulation, while the latter is indicative of "the negative reaction that occurs as a result of a lack of novelty" (Bardo et al., 1995, p. 26).

Numerous studies have determined that the variation in one's level of sensation seeking predicts behavioral differences, especially illicit drug use. HSS are more likely to begin experimenting and using drugs earlier than LSS, as well as use higher levels of a variety of different drugs (Zuckerman, 1979, 1983, 1994; Segal, Huba, & Singer, 1980). Specifically, Donohew (1988, 1990) found that HSS reported more illicit drug use than LSS. Relevant to this investigation of adolescent marijuana use was Donohew's findings that junior high HSS were four times as likely as LSS to use marijuana; in senior high, HSS were three times more likely to use marijuana than LSS. Other studies have drawn a similar connection between sensation seeking and substance abuse (Teichman et al. 1988, 1989). Notably, Andrucci, Archer, Pancoast, and Gordon (1989) determined that sensation seeking is the strongest predictor of drug use and the only variable in their study to predict single versus polydrug use. The logic underlying this link between sensation seeking and drug use implies that "individual differences in the 'need' for novelty exist and that these differences may predict the risk for drug abuse" (Bardo et al., 1995, p. 24). These behavioral differences for illicit drug use between HSS and LSS make sensation seeking an attractive individual difference variable for designing prevention campaigns.

Message Sensation Value. Donohew, Palmgreen, and Duncan (1980) have suggested that one's need for sensation not only translates into differences in drug use, but also into differences in the desired level of stimulation from exposure to information. This Activation Model of Information Exposure suggests that individuals "enter information exposure situations with the expectation of achieving or maintaining [an] optimal state of activation" (Donohew et al., 1991, p. 214). If individuals are unable to maintain this optimal level of arousal, they will seek other acceptable stimuli that will meet this need. Consequently, when an individual's desired arousal level is exceeded because a message is "too threatening or otherwise exciting," the individual will become uncomfortable and seek out a more preferable level of stimulation (Donohew et al., 1988, p. 206). In contrast, when an individual's desired arousal level is too low because "the message is too boring," the individual similarly turns away to find the appropriate arousal level (p. 206). Thus, "it is only when the message satisfies a desired level of arousal that individuals are likely to stay with it" (p. 207). This model reflects the assumptions made by Zuckerman's (1979) conception of sensation seeking and Berlyne's (1971) arousal potential, such that individuals differ in their optimal level of arousal "or that point at which attention and resultant pleasure is maximized" and "may differ among individuals based on biological conditions" (Everett & Palmgreen, 1995, p. 228). HSS, then, possess a higher optimal level of arousal than do LSS. Extending this to media preferences, Donohew and colleagues (1980) would suggest that HSS prefer to expose themselves to messages that are more stimulating, exciting, and arousing. In contrast, LSS prefer to view messages that are slower, subdued, pleasant, and more simplistic. This contrasting variation in media preferences has been labeled by Palmgreen and colleagues (1991) as message sensation value (MSV), or "the degree to which formal and content audiovisual features of a message elicit sensory, affective, and arousal responses" (p. 219).

A series of focus groups with college age students confirmed the variation in media preferences between HSS and LSS (Donohew et al., 1991; Palmgreen et al., 1991). In general, this formative research determined that HSS have a preference for advertisements containing high sensation value (HSV) message characteristics, while LSS preferred advertisements that included low sensation value (LSV) message characteristics. Specifically, HSS indicated a preference for messages that were presented in a novel or unusual format, had unusual uses of formal features such as more extreme-close ups and more sound effects, included a greater frequency of editing, contained faster and more frequent movement, included more intense music, and incorporated higher levels of suspense and drama. In contrast, LSS preferred messages that had more closure at the end of a story. Interestingly, a series of characteristics were preferred by both HSS and LSS, including humor, a good story line, and believable characters. In sum, then, HSS were more attracted than LSS to HSV-type messages and programming as opposed to LSV messages and programming. Thus, consistent with Zuckerman's (1979) notion that high sensation seekers have higher levels of need for sensation, high sensation seekers "also tend to tolerate or even require stronger messages for attracting and holding their attention" (Donohew et al., 1991, p. 212).

Drug Prevention Message Research. The findings from these focus groups were used to conduct a program of research investigating the effectiveness of anti-drug prevention messages that varied in message sensation value. The following series of studies on sensation seeking, media preferences, and illicit drug use confirmed the effectiveness of the SENTAR approach. First, Donohew and his colleagues (1991) determined that HSV messages were more effective with HSS in inducing participant's (young adults) intentions to call a hotline, whereas LSV messages were more effective with LSS. Extending this research to the area of attention and context of programming, Lorch et al. (1994) found that HSS young adults not only paid more attention (using an eyes-on-screen measure) to HSV programs (vs. LSV programs), but that they were more likely to attend to messages (anti-drug PSAs) that were embedded in such programming. In contrast, LSS preferred LSV messages, although they did show some preference for both HSV and LSV programs. Third, combining the findings from the previous research, Palmgreen et al. (1995) conducted a field study in which air time was purchased over a six-month period during HSV programming to broadcast several HSV PSAs aimed at older teens and young adults. The PSAs showed a hotline number individuals could call to obtain a copy of "The Thrill Seeker's Guide to the Bluegrass," a booklet listing locations for many high sensation seeking activities in the area. Results indicate that 73% of the hotline callers were HSS, that more HSS saw at least one of the PSAs, and that HSS were more likely to report seeing the campaign PSAs than other anti-drug PSAs airing concurrently. Fourth, Everett and Palmgreen (1995) in a study of the influence of anti-cocaine PSAs, found that the interaction between sensation seeking and message sensation value was the most important source of variance in anti-cocaine attitudes, intentions to use cocaine, and message recall. Their results were consistent with previous studies in that HSV PSAs were more effective than LSV PSAs for HSS by enhancing recall, promoting more anti-cocaine attitudes, and reducing intentions to try cocaine. The opposite pattern was displayed by LSS. Thus, this research over the last decade has provided convincing evidence that HSS are more attentive to and more persuaded by prevention messages containing a distinct set of media message characteristics (Palmgreen et al., 1995). More specifically, novel messages with highly sensory, stimulating, and dramatic features elicit and maintain the interest of HSS. "Such messages are more likely to capture the attention of HSS, to hold it, and to encourage future exposure to such messages" (Everett & Palmgreen, 1995, p. 229). With solid empirical evidence validating this method to drug prevention, the SENTAR approach was used to develop a drug prevention media campaign targeting adolescent marijuana use that is presently ongoing in Kentucky and Tennessee. A review of this campaign follows.

The present drug prevention effort is an anti-marijuana media campaign targeting adolescents in the 8^th through the 11^th grades. The media campaign is being conducted in two similar cities -- Lexington, Kentucky and Knoxville, Tennessee. The campaign features five locally developed and professionally produced 30-second HSV PSAs. Each message features one or more of the consequences associated with smoking marijuana, such as lung damage, increased sickness, decreased motivation, lower grades, legal trouble, loss of coordination, psychological and physical dependence, and troubled relationship with friends and family (see Sussman, Stacy, Dent, Simon, & Johnson, 1996, for a review). The goal of the campaign, in addition to assessing the effectiveness of a televised PSA prevention campaign, is to reach at-risk adolescents and produce significant changes in marijuana-related beliefs, attitudes, and behaviors.

This field study employs an innovative and methodologically rigorous controlled time-series design. The first part of the campaign, an intensive 4-month effort from January through April, 1997, was conducted in Lexington, Kentucky while Knoxville, Tennessee served as the control city. A 4-month booster campaign (January through April, 1998) is underway in Lexington while simultaneously running for the first time in Knoxville. Interviews with 100 adolescents in both cities began in May, 1996, 8 months prior to the first campaign, and will continue 8 months after the subsequent campaign in both cities. The last cohort will be interviewed in December, 1998.

In addition to using sophisticated methodology, targeting a specific group of at-risk adolescents, and designing messages with specific characteristics that appeal to the target audience, the SENTAR campaign employs three other important principles of effective campaign development: formative research, message placement, and purchased advertising time (Palmgreen & Donohew, in press; also see Perloff, 1993 and Rogers & Storey, 1987 for an overview of effective campaign design). These three, as applied to this campaign, are briefly discussed.

Formative Research for the Anti-Marijuana Campaign. Failing to conduct formative research is a major reason that media campaigns tend not to be effective (Backer, 1990; Rogers & Storey, 1987). Campaigns that fail to assess the existing attitudes, beliefs, and values, as well as the media usage and consumption of the target audience may develop a campaign that contains information not relevant to or desired by the target audience. Flay and Burton (1990) suggest a "need for more research at the front end of the campaign design, particularly needs assessment and formative research," implying that "health campaigners need to place far greater emphasis on marketing research if they are ever to be successful" (p. 145). Many successful campaigns attribute their results to careful precampaign research (e.g., McDivitt, Zimicki, & Hornik, 1997; Flora, Maccoby, & Farquhar, 1989) because they were able to target the audience's salient beliefs (Perloff, 1993).

Because this drug prevention campaign specified professionally producing the anti-marijuana PSAs locally, significant time was invested in conducting formative research. Atkin and Friemuth (1989) suggest that formative research occurs in two states: preproduction and production testing. During preproduction, campaign planners spend time gathering and learning "as much as possible about the intended audience before specifying goals and devising strategies" for the campaign (p. 134). Preproduction for this media campaign was accomplished by conducting a series of after-school focus groups with different cohorts of the target audience --- adolescent high sensation seekers. Because of potentially different developmental and cultural differences, eleven different focus groups were conducted and varied by grade (8^th through 11^th ) and ethnicity (Anglo and African-American). Participants were asked to comment on 10 different anti-drug PSAs and offer their opinions on the effectiveness of the PSAs as well as the formal features of each PSA. Participants were also asked about different drugs, their prevalence in the school and community, and how these drugs were perceived by peers who were both users and non-users.

These focus groups offered considerable insight into the characteristics that adolescents perceive to be most effective in drug prevention PSAs as well as valuable information about the current drug situation in the school and community. Among the information found most valuable by the campaign design team was that (1) HSS showed a distinct preference for those messages that told a story, (2) that HSS felt that depicting the consequences of drug use was one of the more effective ways to reach them, (3) and consistent with the national figures, marijuana was considered to be very prevalent locally, but perceived by the adolescents as being little, if any more harmful than cigarettes or alcohol. With this preproduction information, the campaign design team began conceptualizing and designing PSAs. Several were selected to be transformed into "video boards," or what might be equivalent to a rough draft version of a televised PSA. These video boards were used to conduct production testing.

Production testing, according to Atkin and Friemuth (1989), is the process of "gathering target audience reactions to preliminary versions of messages" (p.141). Once again, a series of focus groups was conducted with the target audience members to gauge their reactions to the PSAs developed by the campaign design team. This process of using the target audience to pretest potential campaign messages "is a central part of formative research --- campaign planners need to develop messages that attract the audience's attention, are understandable and will motivate the audience to respond in the desired manner" (Brown & Einsiedel, 1989, p. 161). Consequently, the major goals were to determine if the PSA concepts would be effective, to eliminate those message considered ineffective, and to refine elements of the messages that showed promise with the target audience. Additionally, "formative evaluation at this stage also assists in creating messages that reflect the target audience's vernacular" (Nowak & Siska, 1995, p. 175). One set of focus groups assessed potential PSAs by showing teenagers the video boards. A second set of focus groups tested ideas in script form. Both methods were effective, providing the design team with the information necessary to have five anti-marijuana PSAs professionally produced for use in the campaign.

In this campaign, the formative research process was invaluable in determining what messages were most likely to be successful in reaching the target audience. Campaign planners "have not typically used systematic approaches at the preproduction stage, as mass media campaign efforts often proceed in the absence of a research foundation" (Atkin & Friemuth, 1989, p. 132). While conducting formative research may take a considerable investment of time and energy, valuable information is gained by interacting with the target audience about the most appropriate and effective ways of conveying campaign messages.

Message Placement. No matter how well one integrates formative research into message design, if the messages are not placed where the target audience will see them, then the campaign is likely to be ineffective. With this in mind, the SENTAR approach utilizes the campaign principle of message placement. Palmgreen and Donohew (in press) suggest that "the most elegant message consigned to a media channel or program context to which no one in the target audience attends is like beautiful hemlock falling in the forest --- except that in a prevention campaign it clearly makes no sound of consequence, nor is anyone there to appreciate its beauty" (p. 20). Therefore, to assist in the placement of the HSV PSAs in programming viewed most by HSS, monthly data was collected from the cohort of HSS adolescents regarding their five favorite television programs and their five favorite channels. This data was combined with local television market viewing data to place the PSAs in programming most likely to be viewed by HSS.

Purchased Advertising Time. In order to place messages where HSS would see them, the SENTAR approach purchased advertising time. Erickson and colleagues (1990) suggest that message placement has been troubling for those who rely on free airtime to publicize their campaigns. Competition for free air time continues to be high, and when it is donated, it is typically at an undesirable time when "a relatively small number of spots are shown...thwarting their ability to motivate and persuade an increasingly fragmented and distracted audience" (p. 242). As a result, the efforts behind developing and disseminating a media campaign may be in vain and outcomes may be discouraging. On top of the limited free airtime is the issue of diversification in the number of channels and programs available today. The three networks once had a monopoly, but with cable television, programming is more diverse and there is a clear need to determine when and where the target audience can be reached through the television. Paid advertising may be the most "powerful tool" for media campaigns so that the campaigners have control over when and where the messages are broadcast (p. 243).

In order to successfully accomplish message placement, a professional media buyer negotiated with local stations and the cable service to obtain approximately $60,000 in airtime in which to place the HSV anti-marijuana PSAs. Additionally, each of these broadcasting companies donated at least one free 30-second spot time for each 30-second spot time purchased by the campaign. Telecable committed to donating two free for each one purchased by the campaign. As a result, not only did this strategy increase chances of reaching the target audience, but the combined purchased and donated air time increased the frequency of times the PSAs were broadcasted.

In sum, the SENTAR approach, used in this two city anti-marijuana drug prevention media campaign, employs the sensation seeking trait as the major segmentation variable, designs prevention messages that are high in sensation value to reach high sensation seekers, utilized formative research to conceptualize and design the campaign, and purchased air time to place messages in HSV programming (Donohew, Lorch, & Palmgreen, 1998; Palmgreen & Donohew, in press). Five professionally produced anti-marijuana PSAs ("Michael," "Dealer," "Downer," "Relationships," and "True Lies") are featured in this campaign and address the consequences of marijuana use. For a description of each, see Table 1. Using Knoxville, Tennessee as a control city, the first of a two-part PSA campaigns was conducted in Lexington, Kentucky for the first four months of 1997.

What follows is an overview of the interview methods used to gather campaign data. Then, a series of results are presented to gauge the effect of the campaign thus far, including (1) frequency of advertising, (2) recall certainty and (3) post-campaign effects on beliefs, attitudes, intentions, and behaviors.

Subjects

Adolescents in grades 8 through 12 were recruited from Lexington, Kentucky and Knoxville, Tennessee for participation in this study. Of the 1,601 individuals who participated, 802 were from Kentucky and 799 were from Tennessee. About 100 participants were recruited each month over a 12 month period. Of those participating, about 53% were female and 13% were non-white.

Procedure

To secure adolescent participation, survey workers from a university survey research center contacted a parent or guardian by telephone requesting the child to participate in this study. If parents agreed, and if the adolescent was willing to participate, then an interview time was scheduled to take place in the adolescent's home. Because of the sensitive nature of the interview questions, the interview was conducted somewhere in the adolescent's home where nothing could be overheard. Consent forms were obtained from the parents and the adolescent before interviews began. Parents answered a series of demographic and socioeconomic questions before leaving the room for the confidential interview with the adolescent.

Using a laptop computer, the interviewer used a computer program designed specifically for these interview sessions. Questions appeared on the computer screen, and the responses could then be keyed into the computer. Data were saved to a hard disk. The first part of the adolescent interview session was conducted by the interviewer asking the questions and keying the adolescent's responses into the computer. Given the sensitive nature of some of the questions near the end of the questionnaire, responses to the last half of the interview questions were answered directly by the participants. To accomplish this, the laptop computer was turned toward the adolescents so that their responses could be privately entered into the computer. This assisted with the confidentiality of their responses in that only the participant knew how the questions were answered and that it also prevented responses to sensitive questions being answered aloud. Each interview lasted about 45 minutes. Adolescent participants received certificates for a free movie and a pizza for their participation.

Measures

Beliefs. Twelve marijuana-related beliefs about occasional and regular use of marijuana were assessed on a four-point scale with the response options of disagree strongly, disagree somewhat, agree somewhat, and agree strongly. The twelve items were: (1) "Occasional/Regular use of marijuana is harmful to people's health" (2) "Occasional/Regular use of marijuana helps people deal with their problems" (3) "People who occasionally/regularly use marijuana have trouble getting a boyfriend/girlfriend" (4) "People who occasionally/regularly use marijuana smell bad" (5) "Occasional/Regular use of marijuana leads to the use of more harmful drugs" (6) Occasional/Regular use of marijuana hurts people's coordination" (7) "Occasional/Regular use of marijuana has harmful effects on people's memory" (8) Occasional/Regular use of marijuana can cause lung cancer (9) Occasional/Regular use of marijuana can cause emotional problems (10) "People who occasionally/regularly use marijuana become lazy" and (11) "People who occasionally/regularly use marijuana do worse at school, work, or sports." Belief measures for occasional and regular marijuana use demonstrated good internal consistency (=.91 and =.88 respectively).

Attitudes. Seven marijuana-related attitudes about occasional and regular use were assessed on a four-point scale, with the response options of disagree strongly, disagree somewhat, agree somewhat, and agree strongly. The seven items were: (1) "I have bad feelings toward people who smoke marijuana occasionally/regularly" (2) "I don't want to hang around people who smoke marijuana occasionally/regularly" (3) "Occasional/Regular use of marijuana is harmful" (4) "People should not smoke marijuana occasionally/regularly" (5) "Occasional/regular use of marijuana is okay" (6) "Marijuana would be enjoyable if used occasionally/regularly" and (7) "The idea of occasionally/regularly smoking marijuana scares me." Measures for attitudes toward occasional and regular use demonstrated good internal consistency (=.91 and =.85 respectively).

Intentions. Participants were asked their intent to engage in experimental or regular marijuana use in the future. Specifically, experimental intent was assessed with "In the future, do you think you will try marijuana at least once." Intent to use marijuana regularly was assessed with "In the future, do you think you will use marijuana regularly." Responses to both of these items include definitely will not, probably will not, probably will, and definitely will.

Behavior. Participants were asked about 30-day marijuana use and lifetime marijuana use. Specifically, lifetime use was assessed with "Have you ever used marijuana?" By answering yes or no to this question, participants were then asked to "Please type in the number of days between 0 and 30 that you have used marijuana in the last 30 days."

Viewer Certainty. To assess whether the PSAs had been seen by the target audience, adolescents were asked to read a short description of each of the five PSAs and indicate their certainty in seeing or not seeing the ad. Response options included very certain I did not see it, fairly certain I did not see it, fairly certain I saw it, very certain I saw it.

Sensation Seeking. Sensation seeking was measured using the Sensation Seeking Scale for Adolescents (SSS-A), which was developed specifically for this study (e.g., Hoyle, Stephenson, Palmgreen, Lorch, & Donohew, 1997; Hoyle & Stephenson, 1996). The SSS-A is an adapted version of two validated instruments, the Sensation Seeking Scale, Form V (Zuckerman, 1979) and the Sensation Seeking Scale for Adolescents (Huba, Newcomb, & Bentler, 1981). Huba et al.'s (1981) sensation seeking measures were used for their high loadings on four factors that emerged from analyses of the Sensation Seeking Scale (Zuckerman, 1979). Those four factors - experience seeking, thrill and adventure seeking, disinhibition, and boredom susceptibility - were retained for this scale. Each of the four primary dimensions of sensation seeking was represented by five items. Thus, the completed scale contained 20 questions, which attained good internal consistency (=.82). See Table 2 for scale items.

Neither alcohol nor drug use is mentioned in any of the questions. However, the construct validity of the original measures from which this survey was adapted has been demonstrated in studies of adolescent alcohol and drug use (e.g., Huba et al., 1981; Newcomb, Maddahian, & Bentler, 1986; Stacy, Newcomb, & Bentler, 1993), where moderate to strong correlations were reported between sensation seeking and use of alcohol and drugs.

The following analyses reflect post-campaign effects assessed over a period of 8 months (May through December, 1997; the campaign ran in Lexington, Kentucky from January to April, 1997 while Knoxville, Tennessee served as a control group). Data on the extent to which the campaign was successful in targeting HSS are drawn from three sources: (1) frequency of advertising as determined from information gathered by the professional media buyer, (2) viewer certainty of exposure to the PSAs, and (3) post-campaign effects on marijuana-related beliefs, attitudes, intentions to use marijuana, and marijuana use or non-use.

Frequency of Advertising

Between January 1 and April 30, 1997, the eight HSV campaign PSAs were broadcast a total of 1,998 times in Lexington. From that total, the campaign paid for 753 commercial airtime slots, while local broadcasting companies ran the PSAs 1,245 times free of charge. Consequently, the campaign PSAs ran about 500 times a month, or 114 times a week for the 17 ½ weeks of the Lexington media campaign.

Viewer Certainty

To assess recall certainty by HSS and LSS in Lexington (since the first campaign did not run in Knoxville), an Analysis of Variance was used with sensation seeking (low vs. high, based on median split) as the independent variable and recall certainty as the dependent variable. A significant main effect for sensation seeking on recall certainty was found for "Michael," F(1,709) = 4.39, p<.05, "Downer," F(1,709) = 12.98, p<.01, for "Relationships," F(1,709) = 19.90, p<.01, and for "True Lies," F(1,709) = 8.67, p<.01, with HSS indicating greater recall certainty ("Michael": 3.08; "Downer": M= 3.03; "Relationships": M=3.11; "True Lies": M=3.27) than LSS ("Michael": 2.89; "Downer": M= 2.67; "Relationships": M=2.73; "True Lies": M=2.97). No significant effect emerged for "Dealer"; production delays relegated it to run only during the last two weeks of the campaign.

It is also worthy to assess recall certainty between the campaign city and the control city. Thus, an Analysis of Variance was used with city (Lexington, Knoxville) as the independent variable and recall certainty by all viewers (both HSS and LSS) as the dependent variable. A significant main effect for city on recall certainty was found for "Michael," F(1,1452) = 618.13, p<.01, "Dealer," F(1,1452) = 13.78, p<.01, "Downer," F(1,1452) = 659.90, p<.01, for "Relationships," F(1,1452) = 587.73, p<.01, and for "True Lies," F(1,1452) = 1030.5, p<.01, with Lexington viewers indicating greater recall certainty ("Michael": 2.99; "Dealer": M=1.76; "Downer": M= 2.87; "Relationships": M=2.93; "True Lies": M=3.13) than Knoxville viewers ("Michael": 1.66; "Dealer": M=1.60; "Downer": M= 1.45; "Relationships": M=1.62; "True Lies": M=1.47).

Campaign Effects

To assess post-campaign effects on eight dependent variables (beliefs about occasional/regular use of marijuana; attitudes toward occasional/regular use of marijuana; intent to experiment with marijuana; intent to use marijuana regularly; 30-day marijuana use; and lifetime marijuana use), Factorial Analysis of Variance was used with city (Lexington, Knoxville), sensation seeking (low, high), and time period (May through December) as the independent variables. All significant main effects and interactions are reported. Particular attention is given to any significant interactions between interview city and time period, where a significant interaction would reveal differences between the two cities over time that can be attributed to the campaign. Because media campaign effects are believed to emerge over the period of approximately one year (Palmgreen & Donohew, in press), effects that are marginally significant or approaching significant are reported for later discussion.

Beliefs About Occasional Use of Marijuana. A significant main effect was detected for interview city on beliefs about occasional marijuana use, F(1,1560) = 25.75, p<.05, where Lexington adolescents (M=2.19) responded with stronger pro-marijuana beliefs about occasional marijuana use than did Knoxville adolescents (M=2.01). A significant main effect was also detected for sensation seeking on beliefs about occasional marijuana use, F(1,1560) = 267.39, p<.05, where HSS (M=2.33) responded with stronger pro-marijuana beliefs than did LSS (M=1.84). No main effect for time period was detected. An interaction between interview city and sensation seeking was significant F(1,1560) = 3.96, p<.05, where differences in level of sensation seeking are more extreme in Knoxville (LSS M=1.76; HSS M=2.35) than in Lexington (LSS M=2.00; HSS M=2.45). Additionally, the interaction between interview city and time period approached significance, F(7,1560) = 1.59, p=.13. Multiple comparison tests (p<.05) suggest that pro-marijuana beliefs about occasional marijuana use increased between the first and eighth post-campaign month (where adolescent beliefs in Knoxville increased significantly more [from M = 1.99 to M = 2.24] than adolescent beliefs in Lexington [from M = 2.17 to M = 2.32]), the second and eighth post-campaign month (where adolescent beliefs in Knoxville increased significantly more [from M = 2.02 to M = 2.24] than adolescent beliefs in Lexington [from M = 2.25 to M = 2.32]), the fourth and eighth post-campaign month (where adolescent beliefs in Knoxville increased significantly more [from M = 2.07 to M = 2.24] than adolescent beliefs in Lexington [from M = 2.21 to M = 2.32]), and the fifth and eighth post-campaign month (where adolescent beliefs in Knoxville increased significantly more [from M = 2.06 to M = 2.24] than adolescent beliefs in Lexington [from M = 2.20 to M = 2.32]). No other interactions were significant.

Beliefs About Regular Use of Marijuana. A significant main effect was detected for interview city on beliefs about regular marijuana use, F(1,1560) = 4.76, p<.05, where Lexington adolescents (M=1.67) responded with significantly stronger pro-marijuana beliefs about regular marijuana use than did Knoxville adolescents (M=1.59). A significant main effect was also detected for sensation seeking on beliefs about regular marijuana use, F(1,1560) = 148.08, p<.05, where HSS (M=1.78) held significantly stronger pro-marijuana beliefs about regular marijuana use than did LSS (M=1.48). No main effect for time period was detected. An interaction between interview city and sensation seeking was also significant, F(7,1560) = 1.24, p<.05, where differences in level of sensation seeking are more extreme in Knoxville (LSS M=1.76; HSS M=2.35) than in Lexington (LSS M=2.00; HSS M=2.45). No other interactions were significant.

Attitudes Toward Occasional Use of Marijuana. A significant main effect was detected on attitudes toward occasional marijuana use for interview city, F(1,1560) = 20.52, p<.05, where Lexington adolescents (M=2.29) responded with significantly stronger pro-marijuana attitudes toward occasional marijuana use than did Knoxville adolescents (M=2.13). Additionally, a significant main effect was detected for sensation seeking, F(1,1560) = 389.16, p<.05, where HSS (M=2.57) reported significantly stronger pro-marijuana attitudes toward occasional marijuana use than did LSS (M=1.99). No main effect for time period was detected. No two-way interactions were significant, although the interaction between interview city and time period was approaching significance, F(7,1560) = 1.71, p=.10. Multiple comparison tests (p<.05) reveal significant differences between the first and eighth post-campaign month (where pro-marijuana attitudes increased at a significantly greater rate for Knoxville [from M = 2.08 to M = 2.33] than for Lexington [from M = 2.20 to M = 2.32]), and between the third and eight post-campaign month (where pro-marijuana attitudes increased at a significantly greater rate for Knoxville [from M = 2.08 to M = 2.33] while declining slightly in Lexington [from M = 2.35 to M = 2.32]). No other interactions were statistically significant.

Attitudes About Regular Use of Marijuana. A significant main effect was detected for sensation seeking, F(1,1560) = 208.10, p<.05, where HSS (M=1.99) indicated significantly stronger pro-marijuana attitudes toward regular marijuana use than did LSS (M=1.50). No main effects for interview city or time period were detected, and there were no significant interactions.

Intention to Try Marijuana at Least Once. A significant main effect was detected for interview city, F(1,1475) = 19.47, p<.05, where Lexington adolescents (M=2.26) were significantly more likely than Knoxville adolescents (M=2.03) to try marijuana at least once. Additionally, a main effect was detected for sensation seeking, F(1,1475) = 298.26, p<.05, where HSS (M=2.56) were significantly more likely LSS (M=1.66) to try using marijuana at least once. No main effect for time period was detected. No interactions were significant.

Intention to Use Marijuana Regularly. A significant main effect was detected for interview city on intentions to use marijuana regularly, F(1,1475) = 13.00, p<.05, where Lexington adolescents (M=1.37) had slightly stronger intentions to use marijuana regularly than Knoxville adolescents (M=1.25). Additionally, a significant main effect was detected for sensation seeking on intention to use marijuana regularly, F(1,1475) = 107.83, p<.05, where HSS adolescents (M=1.48) were significantly more likely to use marijuana regularly than LSS (M=1.12). The main effect for time period was approaching significance, F(7,1475) = 107.93, p=.08. No interactions were statistically significant.

30-Day Marijuana Use. A significant main effect was detected for sensation seeking on 30-day marijuana use, F(1,1530) = 118.97, p<.05, where HSS (M=.28) were significantly more likely than LSS (M=.07) to have used marijuana in the past 30 days. No main effects for interview city or time period were detected, and there were no significant interactions.

Lifetime Marijuana Use. A significant main effect was detected for interview city on lifetime marijuana use, F(1,1530) = 3.99, p<.05, where Lexington adolescents (M=.36) were significantly more likely than Knoxville adolescents (M=.30) to have used marijuana in their lifetime. Additionally, a significant main effect was detected for sensation seeking on lifetime marijuana use, F(1,1530) = 150.11, p<.05, where HSS adolescents (M=.45) were significantly more likely than LSS (M=.18) to have used marijuana in their lifetime. A main effect for time period on lifetime marijuana use was approaching significance, F(7,1530) = 1.98, p=.06. Additionally, a significant interaction between interview city and time period on lifetime marijuana use was detected, F(7,1530) = 2.04, p<.05. Multiple comparison tests (p<.05) reveal significant differences between the first and eighth post-campaign month, where the number of adolescents reporting lifetime marijuana use increased at a significantly greater rate over the post-campaign time period for Knoxville (from M = .23 to M = .43) than for Lexington (from M = .30 to M = .43). Additionally, significant differences were detected between the second and eighth post-campaign month, where the number of adolescents reporting lifetime marijuana use increased at a greater rate for Knoxville (from M = .21 to M = .43) than in Lexington (from M = .36 to M = .43). No other interactions were significant.

The data on several fronts from this post-campaign analysis suggest that the four-month anti-marijuana media campaign targeting HSS adolescents in Lexington, Kentucky had an initial effect on marijuana-related beliefs, attitudes, intentions, and behaviors. While the campaign appears to have already slowed pro-marijuana beliefs and attitudes about occasional marijuana use in Lexington more than in the control city, other outcome variables are indicative of trends toward statistical significance. This is consistent with the long-held notion that media effects are not immediate, but rather emerge over time (e.g., Wartella & Middlestadt, 1991). Given the arduous task of persuading teenagers about the consequences of using a drug that has attained general acceptance among many of their peers, these preliminary results are promising.

First, results indicate that the campaign was successful in reaching the target group of high sensation seeking adolescents. With the exception of one of the five PSAs, which was delayed in production, HSS in Lexington appear to be more than "fairly certain" that they have seen the campaign PSAs. In some cases, LSS also give some indication that they have seen the PSAs as well. These findings can be attributed in part to message placement and purchased advertising time, two campaign design principles endorsed by the SENTAR approach (e.g., Palmgreen & Donohew, in press). Consistent with earlier SENTAR research, purchasing airtime and strategically placing the PSAs in high sensation value programming strongly increased the chances that the target group would be exposed to the campaign messages. To a certain degree, these results can also be attributed to the cooperation of the local television stations and cable company for committing at least one free PSA airing for each purchased time slot, although PSAs are less likely to air in HSV programming. Nevertheless, the combined frequency of paid and donated airtime strongly contributed to the high levels of campaign PSA awareness.

Beyond exposure and recall certainty, these results also indicate that the anti-marijuana media campaign is penetrating certain outcome variables in the campaign city. For example, lifetime marijuana use by Knoxville adolescents has increased at a faster rate over the eight month post-campaign period than it has in Lexington where the campaign aired for four months. Likewise, statistical trends suggest that adolescents in Knoxville may be developing stronger pro-marijuana beliefs about occasional marijuana use and stronger pro-marijuana attitudes toward occasional marijuana use at a faster rate than those adolescents in Lexington. While Lexington adolescents began as generally more pro-marijuana than Knoxville adolescents, these trends suggest that the two cities may soon be equal or trade positions. This is especially encouraging since the five campaign PSAs feature specific consequences associated with marijuana use and intended to affect marijuana-related beliefs and attitudes. Thus, this drug prevention effort has positively affected HSS adolescents in Lexington, providing an encouraging scenario for the ongoing follow-up booster campaign.

Finally, while less important, it is also worth noting that these results support the well-researched idea that HSS are much more likely to engage in thrill-seeking behaviors, such as using drugs. In this study, HSS hold significantly stronger pro-marijuana beliefs and attitudes for both occasional and regular use. Further, HSS are more inclined to both experiment with marijuana as well as use marijuana regularly. Finally, the data suggest that HSS are much more likely to experiment with marijuana, as supported by 30-day and lifetime use of marijuana. Consequently, HSS remain a viable target for this and future drug prevention efforts.

This preliminary analysis is not without its limitations. First, the ideal statistical assessment of this data is a time series analysis, which will be used for campaign evaluation when the larger study involving 32 months of data collection is complete. However, time series analysis is inappropriate for a preliminary assessment such as this one with a restricted number of available data points. Thus, with the existing data, it is difficult to examine the relationship of pre-campaign and during campaign data to the post-campaign effects. A related issue, then, is the use of Analysis of Variance to assess the effects of the campaign thus far. Using monthly mean scores makes ANOVA less sensitive to detecting campaign effects. For example, a significant interaction was not detected between interview city and time period for 30-day marijuana use. However, if one examines the time series trends in Figure 1, it is quite clear that Knoxville adolescent marijuana use is increasing at a faster rate than Lexington adolescent use. While Figure 1 depicts the two-month trend for a longer period of time and analyses conducted for this report included only eight post-campaign months, it is difficult not to believe that the campaign is having an influence not detected by the analysis here. Time series analysis will be utilized once both campaigns have been conducted and the entire 32 months of data has been collected.

These preliminary findings suggest that this anti-marijuana media campaign is reaching the target audience of high sensation seeking adolescents and significantly affecting certain specific outcomes targeted by the PSAs in this campaign. Other targeted outcomes show trends toward statistical significance. These findings are promising and lend support to the idea that theoretically-based, well-designed media campaigns can be effective, and that prevention strategies based on the link between sensation seeking and drug abuse continue to generate promising results.

Andrucci, G. L., Archer, R. P., Pancoast, D. L., & Gordon, R. A. (1989). The relationship of MMPI and sensation seeking scales to adolescent drug use. Journal of Personality Assessment, 53, 253-266.

Atkin, C. K., & Friemuth, V. (1989). Formative evaluation research in campaign design. In R. E. Rice & C. K. Atkin (Eds.), Public Communication Campaigns. Newbury Park: Sage.

Backer, T. E. (1990). Comparative synthesis of mass media health behavior campaigns. Knowledge: Creation, Diffusion, Utilization, 11, 315-329.

Bardo, M., T., & Mueller, C. W. (1991). Sensation seeking and drug abuse prevention from a biological perspective. In L. Donohew, H. E. Sypher, & W. J. Bukoski (Eds.), Persuasive Communication and Drug Abuse Prevention (pp. 195-207). Hillsdale, NJ: Lawrence Erlbaum.

Bardo, M. T., Donohew, R. L., & Harrington, N. G. (1996). Psychobiology of novelty seeking and drug seeking behavior. Behavioural Brain Research, 77, 23-43.

Berlyne, D. E. (1971). Aesthetics and psychophysiology. New York: Appleton-Century-Crofts.

Brown, J. D., & Einsiedel, E. F. (1990). Public health campaigns: Mass media strategies. In E. B. Ray & R. L. Donohew (Eds.), Communication and Health: Systems and Applications. Hillsdale, NJ: Lawrence Erlbaum.

Donohew, L. (1988). Effects of drug abuse message styles: Final report. (Grant No. DA03462). Rockville, MD: National Institute on Drug Abuse.

Donohew, L. (1990). Public health campaigns: Individual message strategies and a model. In E. B. Ray & L. Donohew (Eds.), Communication and Health: Systems and Applications. Hillsdale, NJ: Lawrence Erlbaum.

Donohew, L., Lorch, E., & Palmgreen, P. (1991). Sensation seeking and targeting of televised anti-drug PSAs. In L. Donohew, H. E. Sypher, & W. J. Bukoski (Eds.), Persuasive Communication and Drug Abuse Prevention (pp. 209-226). Hillsdale, NJ: Lawrence Erlbaum.

Donohew, L., Palmgreen, P., & Duncan, J. (1980). An activation model of information exposure. Communication Monographs, 47, 295-303.

Donohew, L., Palmgreen, P., & Lorch, E. P. (1994). Attention, need for sensation, and health communication campaigns. American Behavioral Scientists, 38(2), 310-322.

Donohew, L., Palmgreen, P., & Lorch, E. P. (1998). Applications of a theoretic model of information exposure to health interventions. Human Communication Research, 24(3), 454-468.

Donohew, L., Helm, D., Lawrence, P., & Shatzer, M. (1988). Sensation seeking and targeting of televised anti-drug PSAs. In L. Donohew, H. E. Sypher, & W. Bukoski (Eds.), Persuasive Communication and Drug Abuse Prevention (pp. 209-226). Clifton, NJ: Humana Press.

Everett, M. W., & Palmgreen, P. (1995). Influences of sensation seeking, message sensation value, and program context on effectiveness of anticocaine public service announcements. Health Communication, 7(3), 225-248.

Erickson, A. C., McKenna, J. W., & Romano, R. M. (1990). Past lessons and new uses of the mass media in reducing tobacco consumption. Public Health Reports, 105, 239-244.

Flay, B. R., & Burton, D. (1990). Effective mass communication strategies for health campaigns. In C. Atkin & L. Wallack (Eds.), Mass communication and public health: Complexities and conflicts (pp. 129-146). Newbury Park, CA: Sage.

Flora, J. A., Maccoby, N., & Farquhar, J. W. (1989). Communication campaigns to prevent cardiovascular disease: The Stanford community studies. In R. E. Rice & C. K. Atkin (Eds.) Public communication campaigns (pp. 233-252). Newbury Park: Sage.

Hoyle, R. H., & Stephenson, M. T. (1996). Brief measure of sensation seeking for adolescents. Paper presented to the Midwestern Psychological Association, Chicago, Illinois.

Hoyle, R. H., Stephenson, M. T., Palmgreen, P., Lorch, E. P., & Donohew, R. L. (1997). Brief measure of sensation seeking for research on adolescents. Manuscript under review.

Huba, G. J., Newcomb, M. D., & Bentler, P. M. (1981). Comparisons of canonical correlation and interbattery factor analysis on sensation seeking and drug use domains. Applied Psychological Measurement, 5, 291-306.

Johnston, L. D., Bachman, J. G., & O'Malley, P.M. (1994). Monitoring the future: A continuing study of the lifestyles and values of youth. Ann Arbor: Consortium for Political and Social Research.

Lorch, E. P., Palmgreen, P., Donohew, L., Helm, D., Baer, S. A., & D'Silva, M. U. (1994). Program context, sensation seeking, and attention to televised anti-drug public service announcements. Human Communication Research, 20(3), 390-412.

McDivitt, J. A., Zimicki, S., & Hornik, R. C. (1997). Explaining the impact of a communication campaign to change vaccination knowledge and coverage in the Phillipines. Health Communication, 9, 95-118.

Newcomb, M. D., Maddahian, E., & Bentler, P. M. (1986). Risk factors for drug use among adolescents: Concurrent and longitudinal analyses. American Journal of Public Health, 76, 525-531.

Nowak, G. J., & Siska, M. J. (1995). Using research to inform campaign development and message design: Examples from the American Responds to AIDS campaign. In E. Maibach & R. L. Parrott (Eds.), Designing Health Messages. Newbury Park: Sage.

Palmgreen, P., & Donohew, L. (in press). Effective mass media strategies for drug abuse prevention campaigns. In W. J. Bukoski & Z. Sloboda (Eds.), Handbook of Drug Abuse Prevention: Theory, Science, and Practice. Plenum Publications.

Palmgreen, P., Donohew, L., Lorch, E. P., Rogus, M., Helm, D., & Grant, N. (1991). Sensation seeking, message sensation value, and drug use as mediators of PSA effectiveness. Health Communication, 3, 217-227.

Palmgreen, P., Lorch, E. P., Donohew, L, Harrington, N. G., D'Silva, M., & Helm, D. (1995). Reaching at-risk populations in a mass media drug abuse prevention campaign: Sensation seeking as a targeting variable. Drugs and Society, 8(3-4), 29-45.

Perloff, R. M. (1993). The dynamics of persuasion. NJ: Erlbaum.

Rogers, E. M., & Storey, J. D. (1987). Communication campaigns. In C. R. Berger & S. H. Chaffee (Eds.) Handbook of Communication Science (pp. 817-846). Beverly Hills: Sage.

Stacy, A. W., Newcomb, M. D., & Bentler, P. M. (1993). Cognitive motivations and sensation seeking as long-term predictors of drinking problems. Journal of Social and Clinical Psychology, 12, 1-24.

Sussman, S., Stacy, A. W., Dent, C. W., Simon, T. R., & Johnson, C.. (1996). Marijuana use: Current issues and new research directions. Journal of Drug Issues, 26(4), 695-720.

Teichman, M., Barnea, Z., & Rahav, G. (1989). Sensation seeking, state and trait anxiety and depressive mood in adolescent substance users. International Journal of Addictions, 24, 87-99.

Wartella, E., & Middlestadt, S. (1991). The evolution of models of mass communication and persuasion. Health Communication, 3, 205-215.

Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal. Hillsdale, NJ: Lawrence Erlbaum.

Zuckerman, M. (1988). Behavior and biology: Research on sensation seeking and reactions to the media. In L. Donohew, H. E. Sypher, & E. T. Higgins (Eds.), Communication, social cognition, and affect (pp. 173-194). Hillsdale, NJ: Lawrence Erlbaum.

Zuckerman, M. (1994). Behavioral expressions and biosocial bases of sensation seeking. New York: Cambridge University Press.

Table 1

Anti-Marijuana PSA Descriptions

"Michael" - Based on a true story and features an African-American male who liked to smoke marijuana because it made him do crazy things in front of his friends. One day, while smoking pot, Michael and his friends play a game of Russian Roulette --- Michael lost, paralyzing him. He says "I only smoked pot for a few months, but now I'm on drugs for life."

"Dealer" - An in-your-face drug dealer in a school bathroom asks if "you want to get depressed or anxious, lose your girl, flunk out of school, get a little clumsy, smell bad, do something stupid? Then have some weed." Video depicting consequences matches the drug dealer's message.

"Downer" - Three friends are hanging out smoking pot when one of the three falls off the fence. As he uncontrollably rolls down the steep hill, vivid text is superimposed suggesting that marijuana can make you lose your grip on reality, friends, family, school, sports, jobs, and it can really "bust your lungs."

"Relationships" - A teenage girl sits in her boyfriend's room while he puts in a CD. She tries to talk to him about how since he has starting smoking marijuana, all he does is lie around. She also tells him that he is always sick and always forgetting things, and that she had to get a ride home from practice last night with another guy. The boyfriend ignores her and eventually turns up his music, when she gets up and leaves. The tagline says "Marijuana: It's messed up a lot of relationships."

"True Lies" - Opening with a man preaching the evils of smoking marijuana in classic "Reefer Madness" style, one of four girls grabs the remote and turns the TV off. While rolling a joint, she asks "Are they still trying to feed us that junk about weed?" Her three friends tell her that "she needs to listen girl." They teill her that she "could get hung up on it," that it can mess up her lungs, and that she can quit caring about things, like school and her friends. The girl rolling the joint gets up and leaves saying, "Well I can see I'm in the wrong place."

Table 2

Experience Seeking

1. I would like to explore strange places.

5. I like modern or abstract paintings.

9. I like to try new foods.

13. People should dress the way they want.

17. I would like to take off on a trip with no pre-planned routes or timetables.

Thrill and Adventure Seeking

2. I would like to try rock climbing.

6. I like to do frightening things.

10. I would like to try parachute jumping.

14. I would like to ski down a steep mountain.

18. I would like to try bungee jumping.

Disinhibition

3. I like wild parties.

7. I like to be around real party-ers.

11. I would like to live in the fast lane.

15. I like watching sexy scenes in movies.

19. I would love to have new and exciting experiences, even if they are illegal.

Boredom Susceptibility

4. I hate watching a movie for the second time.

8. I get bored seeing the same kids all the time.

12. I get bored with people who always say the same thing.

16. I get restless when I spend too much time at home.

20. I prefer friends who are excitingly unpredictable.