Data from three samples of adults (Ns = 571, 472, and 989) and a sample of adolescents (N = 1,710) supported the possibility that the prevalence of major depression has been increasing in recent birth cohorts, a phenomenon labeled the age-cohort effect (ACE). A significant ACE for relapse was also found in 1 of the adult samples. In addition, early onset age in the adults (prior to age 25) tended to be associated with relapse. Adults in recent birth cohorts were also found to show an elevated prevalence of other disorders. We examined the power of 4 variables (current mood state, social desirability response bias, labeling, and time interval between the episode and the diagnostic interview) to produce these results without an actual increase in the rate of mental disorder. With 1 exception (labeling), the variables were significantly associated with reports of past episodes of disorder and with birth cohort. Controlling for their influence, however, did not reduce the ACE.

The degree to which psychosocial variables associated with depression were also associated with age was examined in 3 samples of community residents 50 years of age or older (N = 4,617). Most of the expected concomitants of depression were found. With only a few exceptions, age was not correlated with depression-related psychosocial variables. Rather, age was most strongly associated with levels in neuropsychological and psychophysiological functioning. In addition, the magnitudes of the correlations in women compared with men and in young-old age groups compared with old-age groups were examined. A number of significant differences emerged, and their implications for theories of depression are noted.

Observers assessing the probability of an interpretation for a behavioral event may (a) assess the probability that certain inferences can be drawn from the event (inference set) or (b) assess the probability that some explanation can cause the event (explanation set). We suggested that inference set subjects would be more likely than explanation set subjects to discount less plausible interpretations in favor of more plausible interpretations. In three studies observers either estimated the probability that some inferences can be drawn from an event or estimated the probability that some explanation can be the cause for the event. As predicted, the inference set produced a higher level of discounting. Studies 1 and 3 also showed that future-oriented observers made attributions similar to those made in the inference set. However, this effect was open to alternative interpretations in Study 1 and failed to reach statistical significance in Study 3. There was also an indication that inference set subjects were more likely to make correspondent attributions. Additional tests of the effects of time orientation and the possible relation between the inference-explanation distinction and actor-observer differences were discussed.The common attribution task requires observers to provide an interpretation for some behavioral event. A great deal is known about the factors that influence the outcomes of this task. In contrast, the meta-theory of the attribution process has attracted less attention. What do observers look for when they search for an interpretation? What do observers estimate when they assess the probability of an interpretation? What is an interpretation? These questions represent an important yet relatively unknown aspect of attribution theory.Recently, Zuckerman, Eghrari, and Lambrecht (1986) suggested that an interpretation for an event can be seen either as an inference that might be drawn from the event or as an explanation leading to the event. They also suggested that whether observers search for an inference or for an explanation determines a set, which then affects the nature of the attribution process. In an inference set, observers estimate whether some knowledge can be derived from the event. This estimate approximates the probability that certain factors are present if the behavior is present (i.e., the probability that the inference is a necessary condition). In an explanation set, observers estimate the likelihood that a cause or reason could produce the event. This estimate approximates the probability that certain factors can produce the event (i.e., the probability that the explanation is a sufficient condition). Note that the probabilities defining the two attribution sets are only approximated. We think that the logic of necessary and sufficient causes represents an abstract ideal of what observers do. It is unlikely, however, that this logic serves as an explicit criterion for formulating or testing inferences and explanations.

No abstract

No abstract