The WAGER Vol. 9(11) – Attribution, Addiction, and Gambling: Series Conclusion

Series Special Editor: Sarah E. Nelson

Attributions play a role in all aspects of addiction. Attributions for wins and losses can influence the development of gambling problems (see WAGER 9(6)); attributions about peers’ drinking behavior can affect a person’s own drinking behavior (see WAGER 9(7)); being labeled as a heavy smoker can alter people’s attributions about their smoking (see WAGER 9(8)); people’s attributions about their own addictive behavior differ in predictable ways from attributions they make about others’ addictive behaviors (see WAGER 9(9)); and the attributions people make about their addictions can predict their own chances for recovery or likelihood of relapse (see WAGER 9(10)). This week’s WAGER reviews an article that attempts to model the way attributions change across the course of addiction (Davies, 1996).
Davies’ theory of attribution change rests on the idea that the explanations people make for their behaviors are functional: people make different attributions for the same  event in different contexts (i.e., depending on the setting and the goals of the interaction) (Davies, 1996; Schlenker & Weigold, 1992; Tedeschi & Reiss, 1981). For example, in a group of heroin users, Davies found that subjects made different attributions for their own heroin use to an interviewer who substance using habits were unknown than they did to a fellow heroin user (Davies & Baker, 1987). When it comes to explaining our own behavior, the attributions we make often reflect an ego-defensive bias: that is, attributions about the self serve to protect self esteem, meet self-presentational goals, and preserve self-concept (Schlenker, Weigold, & Hallam, 1990). Davies’ model reflects how attributions might serve this self-protective function in relation to an addiction.

In several different studies, Davies and his colleagues conducted interviews in Scotland with drug and alcohol users both in and out of treatment. Based loosely on these interviews, Davies outlined five attributional stages through which a person might progress as an addiction develops. Each stage is marked by a different attribution style for substance-using behavior. In Davies’ model, attributions can vary in terms of purposiveness (i.e., how intentional the behavior is portrayed), hedonism (i.e., how positively the behavior is described), contradictoriness (i.e., whether attributions contradict across the course of the interview), and the inclusion of addicted self-ascription (i.e., whether attributions make use of the concept of addiction as an explanation for behavior). (1)

During the first stage, presumably before the substance using behaviors become a problem, subjects’ attributions for their drug or alcohol use are high on purposiveness and hedonism – people enjoy using the substance and consider it under their control. During stage two, as problems begin to surface, subjects’ discourse becomes contradictory, varying from context to context between the positive and negative aspects of drug use, and the controlled and uncontrolled aspects of their using behavior. These attributions reflect the ambivalence that emerges during the development of addiction (e.g., Shaffer, 1997). During stage three, people refer to themselves as addicted, explain their substance use as out of their control, and view it as negative. At stage four, people begin to reject the usefulness of the concept of addiction in explaining their behavior and their discourse again becomes mixed and contradictory. Finally, people are able to proceed to a fifth stage that is either positive or negative. In either version, their attributions are relatively stable (i.e., the attributions don’t contradict from one context to another) and do not refer to substance using behavior in terms of addiction. In the positive version, people might have given up drugs or alcohol, but return to a view of their past behavior as controllable and a description of their use that highlights both the positive and negative aspects of that behavior. In the negative version, although the concept of addiction has been dropped, people continue to use substances and see themselves as “down and out” – their behavior is uncontrollable and their substance use is negative. Although these stages tend to relate to the progression of an addiction, people can move back and forth between stages. The one exception to this, according to Davies, is an irreversible transition from stage two to stage 3, which often occurs when subjects enter treatment and may persist long after; consider the Alcoholics Anonymous practice of participants introducing themselves, “Hi, my name is X and I’m an alcoholic.” .


To test this model of attributions, interviews with drug and alcohol users were transcribed and coders rated the attributions given in each interview in terms of the dimensions outlined in the model. The investigators assigned each respondent to one of the six stages based upon those ratings. Consensus between four judges rating the same twenty subjects was good: average agreement between the judges was 71% (.’s ranged from .49 to .75). In all cases, the judges never disagreed by more than one stage.
Although Davies demonstrated the reliability of his model (i.e., ability of coders to identify the attribution patterns associated with each stage) and stated that these stages related to the stages of an addiction, he did not provide information about hwo the attributional stages correspond to the actual temporal progression of addiction in his interviewees (e.g., whether the majority of subjects classified as stage 3 were in treatment at the time of the interview). Given his claim that movement between at least two of the stages is irreversible (a claim that contradicts established research on addiction stages — see Prochaska, Norcross, & DiClemente, 1994; Shaffer, 1997), this research is needed to verify the model. Also, although he developed his model based on years of observations and interviews of substance users, in this paper he only tests it on twenty interviewees. Given the theoretical basis of the model (i.e., that attributions vary according to context), it is important to test this model and its stages in different samples of substance users and different settings.

Davies’ model of attribution change needs to be validated, but is important for the questions it raises. How do these attributional stages relate to the stages of change described in more well-studied models (e.g., precontemplation, contemplation, preparation or determination, action, maintenance, and termination; Prochaska, Norcross, & DiClemente, 1994)? If these attributional patterns do reliably correspond to different stages of an addiction, it is important to determine

whether these attributions predict change (e.g., provide explanations that refer to being addicted as precursors of treatment-seeking behavior) or reflect change (e.g., provide explanations that refer to being addicted as an attempt to understand and explain past behavior within the treatment context). Both possibilities (i.e., predictive and reflective) and the research reviewed in this WAGER series stress the importance of people’s subjective understanding and interpretation of behavior in guiding future behavior. This attribution-behavior cycle is a crucial, often neglected piece of the study of addictions.