$54 million and $140.7 million over the next 10 years depending largely on the extent of political support from the governor and the legislature (Independent Evaluation Consortium 2001; Pierce et al. 1998b). An ambitious media campaign began in 1990 that addressed secondhand smoke issues, youth prevention (largely through ads critical of the tobacco industry), and smoking cessation (Independent Evaluation Consortium 2001); only a minority of media expenditures were directed at youth.

Evaluations of the California effort suggested that the campaign had an impact on smoking prevalence in the first 3 years of the effort. Adult smoking prevalence dropped from 22.7 percent to 18 percent from 1989 to 1993, a rate of decline that was about double that of the United States as a whole (Pierce et al. 1998b). According to Friend and Levy (Friend and Levy 2002), similar effect estimates were based on analyses by the Centers for Disease Control and Prevention (CDC 1996). Subsequently, prevalence rates flattened, with no evidence of declines, suggesting that the effects occurred in the first 4 years of the campaign or that the effects were dampened due to reductions in expenditures by California.

Complicating the overall picture, based on self-reported exposure to the campaign, several studies did not find clear evidence for reductions in prevalence associated with campaign exposure (Pierce et al. 1998a; Popham et al. 1994), although another study suggested that the media campaign influenced decisions to quit smoking in California (Popham et al. 1993). It also should be noted that this California campaign was not primarily youth-focused, in contrast to the Massachusetts and Florida efforts (see below).

Massachusetts has been engaged in an ambitious tobacco control program since January 1993, which includes increased taxes that have been used to fund an extensive paid media campaign. Amounts available for these efforts were as high as $43 million in 1995, but have declined since. Strong population-based evidence exists for the effectiveness of this comprehensive effort in reducing adult smoking prevalence (Biener et al. 2000). Media efforts included television, radio, and billboard antismoking advertising, directed primarily at youth. The effectiveness of the media component of this effort was evaluated based on a 4-year longitudinal panel survey of 592 youth (ages 12–15 years at baseline), conducted from 1993 to 1997. Campaign impact was assessed by measuring recall of campaign advertising and comparing progression to established smoking among nonsmoking youth who recalled campaign advertisements to those who did not. A variety of control variables were applied, including parental smoking at home, hours of television watched, and so forth. Results suggested that younger adolescents (ages 12–13 years) reporting exposure to television advertisements at baseline were only about half as likely to progress to regular smoking as were the unexposed (odds ratio [OR] = .49, p < .05). No such differences were found for older adolescents (OR = .94) and no effects of radio or billboard exposure were found for either group (Biener and Siegel 2000).

Limitations of this study include the limited control over possible variables that might be confounded with exposure at baseline. Moreover, measures of campaign message recall or recognition are inherently subject to problems of endogeneity, in which propensity to attend to and recall such messages may be related—positively or negatively—to a possible interest in smoking (Slater 2004). However, the prospective design, lack of consistency of results for older versus younger adolescents, and lack of evidence for effects due to susceptibility differences or differential attrition argue against attributing results to confounding relationships. It should also be noted that the effects of the campaign on adults, as opposed to youth only, are not assessed in