Evidence of a dose-response relationship between "truth" antismoking ads and youth smoking prevalence.