Smoking and Tobacco Control Monograph No. 7 
nicotine are highly correlated across the full range of tar and nicotine yields.) 
People who smoke low- and ultralow-yield cigarettes may be more health 
conscious, have better diets, and be interested in smoking less. A random 
sample of persons does not select ultralow-yield cigarettes. 
Despite their advantages, experimental brand-switching studies have 
important limitations. Outside of laboratories, smokers select their own 
brands. There is a free market for most purchases of cigarettes. An 
unsatisfying brand is likely to be rejected for a satisfying brand. Persons 
trying an ultralow-yield cigarette may feel that they are puffing on air, so 
they decide not to smoke these cigarettes and probably will not buy more 
than one pack. Some compensatory smoking techniques (e.g., vent blocking 
[Kozlowski et al., 1980 and 1989]) may take time to be learned by trial and 
error. Short-term studies (i.e., less than 1 week of exposure on lower yield 
brands) do not provide an adequate indication of the nature of compensatory 
smoking in self-selected smokers. All reviewed studies involved brand 
manipulations (change of "treatment" or brand in experimental study) 
of more than 7 days. 
Studies of brand switching also have biased samples. Who does and 
does not volunteer for these studies? One of the five studies reviewed 
(Guyatt et al., 1989) showed a dramatic number of dropouts following 
informed consent. Of the people who went to at least one session in this 
study, 81 percent dropped out. Another study on brand switching (Benowitz 
et al., 1986a) required that participants be hospitalized for 14 days. Some 
smokers, knowing that they were going to get ultralow-yield cigarettes, 
either might not have wanted to smoke them or spend 14 days in the 
hospital. One must wonder who would be available to participate in a 
14-day study requiring confinement to a hospital room. Most studies of 
brand switching also have small samples (mean = 22 subjects). As for 
demographic differences, there is no way to represent the complexities of 
age, sex, race, and education adequately in a sample of 22 participants. 
According to the boundary model of dmg regulation, plasma nicotine 
levels are not precisely regulated (Kozlowski and Herman, 1984); there are 
aversive upper and lower limits or boundaries on intake for dependent 
smokers. At the upper limit, when people are smoking a great deal, it is 
difficult for them to smoke more due to overdose or toxic effects of nicotine. 
When they are smoking a little, it is hard for them to smoke less than the 
lower limit because of insufficient nicotine intake. However, within these 
broad limits or boundaries, psychosocial factors primarily (i.e., the presence 
of others smoking) determine nicotine ingestion, and dose manipulations 
tend to have a smaller effect on smoking behavior (Kozlowski and Herman, 
1984; Kozlowski, 1989) and how smokers feel (Benowitz et al., 1986b). 
RESEARCH fable 1 shows the five studies reviewed and gives a summary of their 
results. The following studies were not included in the review because they 
were either too short term or used cigarette holders, which could interfere 
with natural smoking behavior: Benowitz and colleagues (1986a), Kolonen 
162 
