An unexpected finding was the relatively poor long-term performance of the NS+CM condition. Participants in this condition initially did as well as those in NS, but at the end of two years were reporting the same outcomes as those in CaseM. A clue to this finding may be found in the between-treatment results for the mediating variables, particularly self-efficacy. At posttreatment all of the treatments showed similar increases in self-efficacy. By 15 and 27 months, however, the NS condition had resulted in greater self-efficacy than the other conditions. Contingency management may have been effective as long as it was in place, but the external reinforcements it provided may not have generated increases in self-efficacy that were sustained after they were discontinued. This may be the result of attributions by participants that their abstinence behavior was externally maintained by the reinforcements received. For self-efficacy to increase, people have to attribute their accomplishments to their own efforts (Bandura, 1986).
