dangers and by information about those dangers. Where the substance or behavior is addictive or habit-forming, this may generate a long-term increase in the number of users. In addition, lower perceived harmfulness may reduce incentives to quit or desist. Relapse may also be encouraged by the perception of less dangerous means of continuing the desired behavior. Lower perceived harmfulness may also increase intensity of use.
Changing the riskiness of an act is known to alter the behavior of the population. Generally the change is in the form of compensation, i.e., higher risk will reduce the prevalence of the behavior while lower risk will increase that prevalence. Engineers tend to overestimate the benefits from safety devices, since they ignore that behavioral adaptation. On auto safety, Evans (1991) has noted: “If the safety change affects vehicle performance, it is likely to be used to increase mobility. Thus, improved braking or handling characteristics are likely to lead to increased speeds, closer following, and faster cornering. Safety may also increase, but by less than if there had been no behavioral response.” In its most extreme form, this kind of risk compensation has been labeled risk homeostasis—a term that implies implicit or explicit efforts to maintain a constant level of risk (Wilde, 1982).
These changes are logical possibilities. How substantial they are is an empirical matter. So is the extent of generalizability across product domains and populations. The introduction of a safer automobile may, for example, have negligible effect on the driving behavior of older drivers but sharply increase speeding by younger drivers, while condom availability may increase sexual behavior of older adolescents more than it affects that of younger adolescents. The psychological mechanisms generating alcohol and cigarette dependence may be different enough that harm reduction interventions in general are more effective in one field than in the other.
This lack of generalizability raises a question as to the relevance of examples from other fields to tobacco interventions. But tobacco harm reduction involves a large variety of potential interventions. They differ in some important dimensions, just as do automobile seat belts and needle exchange programs. The examples can help identify the dimensions that influence the outcomes of harm reduction interventions.
The harm reduction framework can be applied not only in a number of different policy areas but to a variety of forms of interventions:
Lowering the inherent harmfulness of a broad class of products