آیا بدبینی یک عامل خطر برای خلق و خوی افسرده در میان افراد مسن ساکن جامعه است؟
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38203||2001||18 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Behaviour Research and Therapy, Volume 39, Issue 3, March 2001, Pages 255–272
Abstract This study examined two senses in which pessimism might be a risk factor for depressive mood among older adults. The first was that a pessimistic explanatory style would predict changes toward depressive mood when combined with stressful life events. The second was that predictive pessimism, or thinking that bad events will happen in the future, would predict changes in depressive symptoms. We found an interaction between explanatory style and life stressors, but it was the optimists who were at higher risk for depressive symptoms after negative life events. We also found support for predictive pessimism, however, as a predictor of depressive symptoms over time.
1. Introduction Explanatory style theory posits that a pessimistic way of explaining the causes of life events is a risk factor for depression (Peterson & Seligman, 1984). Despite two decades of research demonstrating that pessimism, defined as a tendency to explain negative events with internal, stable and global causes (Abramson, Seligman & Teasdale, 1978), is indeed related to the development of depressive symptoms in children and young adults, this prediction has not been tested in community-dwelling older adults. The primary goal of this study was therefore to measure explanatory style in this older population, and to determine whether pessimism is a prospective risk factor for the development of depressive symptoms in older adults. In contrast to some theoretical work (Gatz, Kasl-Godfrey & Karel, 1996) which claims that psychological risk factors for depression might become less important as people get older, we hypothesized that pessimistic explanatory style would indeed predict increases in depressive symptoms in this age group. Whereas the explanatory style approach to pessimism and its opposite, optimism, has not previously been extended into old age, several other approaches to optimism have been used with older samples. These include dispositional optimism (Mroczek, Spiro, Aldwin, Ozer & Bossé, 1993) and personal optimism (Reker, 1997 and Reker and Wong, 1985). We believe that, despite their different measurement instruments, these other approaches to optimism share a focus on expectations for the future, in contrast to the explanatory style approach’s measurement of attributions for past events. Recent research has found explanatory style to be moderately correlated with more future-oriented measures of pessimism (Hjelle, Belongia & Nesser, 1996), but it nonetheless seemed important to evaluate these two constructs separately in our longitudinal analyses. Instead of selecting among possible approaches and related instruments measuring hypothetical future events, we decided to take a more real-life approach to pessimistic predictions by using events actually expected to occur in the near future. The second hypothesis of our study was therefore that pessimistic predictions for actual negative life events in the near future would predict depression. Timing of measurement is critical in prospective longitudinal studies such as this one. Our hypotheses require a period of time during which significant life events may happen for many older adults. The most comprehensive project on life events in community-dwelling older adults used six month testing intervals to maximize recall accuracy, and found that many participants in their large sample experienced several life events over the course of six months (Murrell, Norris & Hutchins, 1984). Therefore, six-month intervals appeared to be appropriate for testing interactions involving life events, and we tested the interaction of explanatory style and life events at six month and one-year follow-up. In contrast, we considered predictive pessimism to be independent of actual life events experienced, suggesting that we should test the hypothesis over a brief period such as one month during which few life events will have occurred. Frese (1992) cautions that the effects of psychological moderator variables may change from short intervals to longer ones. In accordance with this caution, we therefore tested both notions of pessimism over both shorter (one month) and longer (six months and one year) periods to ensure that there were no temporally shifting relationships. Furthermore, we note that our concern in this study was with the prediction of changes in depressive symptoms, treating depression as a continuous variable. We now discuss the two notions of pessimism in more detail, with special attention to how these constructs might be important in life-span perspective and how they might have impact on depressive mood in different time frames. 1.1. Pessimistic explanatory style as a predictor of depression There are two primary reasons that it is critical to extend explanatory style work into the last few decades of the life-span. First, few researchers use theoretical models to test hypotheses concerning depression and affect in late life (see Reker, 1997, for an exception). Second, prevention programs based on explanatory style have prevented depression in at-risk children (Gillham, Reivich, Jaycox & Seligman, 1995). While few examples exist of gerontological research on prevention, it may be an idea whose time has come, especially as this group becomes a much larger part of the American population (Carstensen and Pasupathi, 1993, Gatz, 1995 and Murrell and Meeks, 1991). However, it will only make good sense to do prevention in this way if pessimism predicts depression and poor health in late life. It appears that a pessimistic explanatory style is stable across adulthood (Burns & Seligman, 1989) and relates to worse immune function among older people (Kamen-Siegel, Rodin, Seligman & Dwyer, 1991). However, no prospective evidence exists to link pessimism and depression in older adults. Therefore, we evaluated explanatory style and predictive pessimism as risk factors for depression in old age. Despite cross-sectional evidence that has not found any relationship between depression in older adults and attributions for actual negative life events (Patrick & Moore, 1986) or a general optimistic outlook (Reker, 1997), we expected that a pessimistic explanatory style would be a risk factor in older people when followed for long time periods, as life events happen relatively infrequently. The Attributional Style Questionnaire (ASQ; Peterson et al., 1982) is the traditional method of measuring a person’s explanatory style. It consists of hypothetical events in the affiliation and achievement domains. These domains are differentially important throughout the life-span. In their studies of students responding to midterm exam grades, Metalsky and colleagues used an explanatory style composite score involving only items related to negative achievement events, in order to approximate the actual nature of the life stressor in question (Metalsky et al., 1987 and Metalsky et al., 1993). According to a large-scale epidemiological study, the most frequent negative life events among older adults are ‘self or family member needed to go into hospital’ and ‘good friend died’; almost 40% of their sample experienced a close friend’s death over the course of a year (Murrell et al., 1984). These events are primarily from the affiliation and health domains. In contrast, older people report very few achievement-related major life events. Therefore, it seems appropriate to consider the two domains separately, and to focus in particular on the more relevant affiliation-related items. Peterson et al. (1982) posit that the two domains are highly related in college students and the scales should therefore be combined unless there is a persuasive reason to do otherwise. We believe that the domains are more distinguishable in older adults than in college students, and that the life event data gives a persuasive justification for separating the scales in this research. Thus, we hypothesized that a pessimistic explanatory style regarding interpersonal events would be a risk factor for depressive symptoms when combined with negative life events that are predominantly interpersonal in nature for older adults. 1.2. Predictive pessimism The second sense of pessimism involves whether a person’s expectations that negative life events will happen to them in the near future is a risk factor for later depression. This notion of predictive pessimism complements work which defines optimism based on people’s general future expectancies (Reker and Wong, 1985 and Scheier and Carver, 1985). Reker (1997) recently conducted a study testing the relationship between a future-oriented sense of optimism and depressive symptoms in community-dwelling and institutionalized older adults. Interestingly, optimism only contributed unique variance to depressive symptomatology among institutionalized older people, but not among community-dwelling older adults. However, Bromberger and Matthews (1996) found that middle aged women who endorsed a more negative generalized view of their future became more depressed after going through stressful events. The actual life events that an older person anticipates in her future may be more useful as a measure of predictive pessimism than very generalized expectancies (Reker & Wong, 1985). Unlike Reker and Wong, however, we were curious about pessimism rather than optimism, and we suspected that a more structured assessment of future expectations (actual possible events that could be anticipated or not, as well as a specified time frame for prediction) would be more useful than their less structured measure. We hypothesized that, over a short time period, people who expect that bad things will happen during that period may indeed be more depressed at the end of the period. Our concern was with sheer number of life events expected rather than with accuracy, as we hypothesized that the very willingness to expect bad events soon might be a specific vulnerability for older people. Our conception of ‘predictive pessimism’ differs significantly from Alloy and Ahrens’ (1987) use of the term in reference to the tendency of depressives to make more unfounded negative predictions for the future than nondepressives. This approach presumes that making negative predictions is a symptom of people already experiencing depressive symptoms; however, the tendency to expect bad events is also a risk factor in some theories of depression (e.g. the hopelessness theory, Abramson, Metalsky & Alloy, 1989). We consider the possibility that expectations for negative life events could be a vulnerability for later depressive symptoms, especially in conditions of limited time such as old age. Moreover, we expect that this future-oriented notion of pessimism will operate independently from the explanatory style approach to pessimism, and that the two constructs will therefore be uncorrelated.
نتیجه گیری انگلیسی
Results We tested two hypotheses concerning different ways in which pessimism might predict depressive symptoms over time among community-dwelling older adults at three follow-up time points: one month, six months and one year after the initial interview. We will present results separately for the pessimistic explanatory style and predictive pessimism hypotheses. The primary findings were that predictive pessimism predicted changes in depressive symptoms over a month, but that a pessimistic explanatory style did not predict higher levels of depressive symptoms. Indeed, we will present some evidence that extreme optimists may be more at risk for depressive symptoms than pessimists when faced with negative life events. First, though, basic descriptive statistics are presented in Table 1. As predicted, the two notions of pessimism evaluated in this study were uncorrelated: the number of negative events predicted for the next month was not correlated with either the composite score for explanatory style for affiliation events (r=−0.08, n.s.) or for achievement events (r=−0.04, n.s.), suggesting that they are indeed two distinct constructs and that separate analyses for each are justified. Table 1. Descriptive statistics Variable Mean SD N BDI — Baseline 7.14 5.29 71 BDI — 1 month 4.57 3.77 67 BDI — 6 months 6.27 5.42 51 BDI — 1 year 4.69 5.29 48 Affiliation ES 6.86 3.71 71 Achievement ES 5.10 3.73 71 Negative predictions 0.45 0.97 71 Life events over 1 month 1.25 1.32 67 Life events over 6 months 1.77 2.06 51 Life events over 1 yeara 3.04 2.71 45 a Sum of life events for the two 6-month time intervals, so only participants contacted at both 6-month and 1-year follow-up are included. Table options 3.1. Hypothesis 1: pessimistic explanatory style Explanatory style showed some simultaneous relationship to depressive symptoms at baseline interviews. Composite ASQ-OAV score was marginally correlated with BDI score at baseline (r=−0.23, P<0.06), with more optimistic participants having fewer depressive symptoms. While the composite affiliation score was uncorrelated with BDI score (r=−0.12, n.s.), those participants who were more pessimistic on the achievement-related items tended to report more depressive symptoms (r=−0.26, P<0.05). One-month follow-up did not reveal any effects of explanatory style on depressive symptoms. Composite ASQ-OAV score was uncorrelated with BDI score at one-month follow-up (r=−0.18, n.s.), as was composite score on the affiliation items (r=−0.07, n.s.); composite score on the affiliation items was marginally correlated with one-month follow-up BDI (r=−0.24, P<0.06). However, it remained possible that explanatory style might predict levels of depression at one-month follow-up above and beyond baseline depression, either as a main effect or in interaction with life events over the month (diathesis-stress interaction). Therefore, we conducted a separate Analysis of Partial Variance (APV) on one-month follow-up BDI score using each of the three composite ASQ-OAV scores. None of the main effects of composite score, achievement items or affiliation items predicted one-month follow-up BDI score above and beyond baseline BDI score, nor did any of the interactions of these scores with stressful life events experienced over the month. A different picture emerged at six month follow-up, with a diathesis-stress effect emerging such that the interaction of explanatory style for affiliation events and negative life events became a significant predictor of changes in depressive symptoms. There was no significant correlation between BDI score at this follow-up and composite ASQ-OAV score (r=−0.20, n.s.). Affiliation composite score and BDI score remained uncorrelated (r=−0.08, n.s.), while participants who were more pessimistic on the achievement composite score reported more depressive symptoms (r=−0.25, P<0.08). However, these correlations were masking some very interesting effects. As shown in Table 1, when the composite affiliation score of the ASQ-OAV was entered into a diathesis-stress APV model on the prediction of depressive symptoms at six month follow-up, the interaction term of explanatory style ∗ life events added significantly to the predictive power of the model, F(1,46)=4.09, P<0.05. The direction of this interaction effect was opposite that predicted. As shown in Fig. 1, we tested the direction by calculating predicted values for six-month follow-up BDI score using the final regression coefficients at one standard deviation above and below the mean for affiliation composite score (one SD above for optimists and one SD below for pessimists) and number of life events (none or several). If pessimism is a risk factor for depressive symptoms in the face of life events, then the more pessimistic participants who experienced life events should have the highest predicted value for six-month follow-up BDI score. As shown in Fig. 1 however, the interaction was in the opposite direction. It was the more optimistic participants who experienced negative life events over the six months who then went on to have the highest predicted values of BDI score at six-month follow-up. Optimists with no life events had the lowest predicted BDI scores, while pessimists had intermediate scores regardless of life events. There was no effect of achievement composite score or total composite score. Predicted values for depressive symptoms at six-month follow-up in optimists and ... Fig. 1. Predicted values for depressive symptoms at six-month follow-up in optimists and pessimists. Figure options A similar effect again emerged at one-year follow-up. At that time, both composite affiliation and achievement scores were uncorrelated with BDI score at the follow-up (respectively, r=−0.16 and −0.19, both n.s.). Composite ASQ score was also uncorrelated with BDI score at one-year follow-up (r=−0.21, n.s.). As shown in Table 2, the six-month follow-up interaction effect was replicated, though only at the level of a trend, F(1,40)=3.7, P<0.07. Table 2. Test of diathesis-stress effects of explanatory style and life events on depressive symptoms at 6-month follow-up Entry order Predictor Cumulative R2 F-value df pr 1 Time 1 depression 0.50 49.90a (1,49) 0.71 2 Life events over 6 months 0.53 2.17 (1,48) 0.21 3 Explanatory style for affiliation items 0.53 0.03 (1,47) 0.03 4 Life events ∗ explanatory style 0.57 4.09b (1,46) 0.29 a P<0.01. b P<0.05. Table options Fig. 2 shows that the effect at one year was similar in direction to the six-month follow-up effect; however, one interesting difference at one-year follow-up was that there was a main effect of life events (Table 3). That is, on average, participants who experienced more life events over the year had more depressive symptoms at the end of the year. Optimists who had experienced several negative life events over the year were doing worse than all other groups. It is notable that the predicted BDI scores for optimists with few or no negative life events at one-year follow-up are extremely nondepressive. While this is not an independent replication because it includes those events from the first six-month period as well, it does suggest that the result is fairly robust across a long time period and was not just a spurious finding from the six-month follow-up. Predicted values for depressive symptoms at one-year follow-up in optimists and ... Fig. 2. Predicted values for depressive symptoms at one-year follow-up in optimists and pessimists. Figure options Table 3. Test of diathesis-stress effects of explanatory style and life events on depressive symptoms at 1-year follow-up Entry order Predictor Cumulative R2 F-value df pr 1 Time 1 depression 0.22 12.64a (1,43) 0.46 2 Life events over year 0.31 5.57b (1,42) 0.35 3 Explanatory style for affiliation items 0.32 0.44 (1,41) −0.10 4 Life events ∗ exp. style 0.38 3.70c (1,40) 0.29 a P<0.01. b P<0.05. c P<0.07. Table options 3.2. Hypothesis 2: predictive pessimism Our hypothesis concerning predictive pessimism was most concerned with the prediction of events and changes in depressive symptoms over a short time period; and indeed, we found support for our hypothesis at one-month follow-up. A higher number of events predicted over the next month was correlated with higher BDI score at baseline (r=0.28, P<0.02) and one-month follow-up (r=0.42, P<0.001). To test for a unique effect of negative predictions on time 2 depressive symptoms, we computed the partial correlation between depressive symptoms at time 2 and number of negative life events predicted for the month, while partialling out the effects of time 1 depressive symptoms and actual life events experienced over the month from both variables. The partial correlation was significant (r=0.30, P<0.05), with those participants who predicted negative life events over the month reporting more depressive symptoms than those who did not, independent of original symptoms and actual life stressors experienced. We conducted several t-tests to further investigate the differences between those participants who predicted negative life events versus those who predicted none. Compared to participants who predicted at least one negative event, participants who predicted no life events at the initial assessment reported the occurrence of fewer negative life events during the previous year (t(69)=2.53, P<0.05), and fewer life events over the month between times 1 and 2 (t(21)=−2.50, P<0.05). There was indeed a strong positive correlation between number of predicted life events over the month and number of actual life events over the month (r=0.42, P<0.001). However, it is important that the relationship between number of negative predictions and depressive symptoms at one-month follow-up remained strong even after the effects of intervening life events were partialled out, as shown in the partial correlation above. In other words, this effect was not merely about accuracy, as a correlation remained between gloomy predictions and depressive symptoms one month later even after controlling for the actual life stressors experienced during the month. There appears to be a specific quality to negative predictions that may put people at risk for depressive symptoms above and beyond accurately believing that bad events will happen in the near future. After finding support for our notion of predictive pessimism as a risk factor for later depressive symptoms, we tested whether it was also symptomatic of depression, as conceptualized by Alloy and Ahrens (1987). Interestingly, we found no support for their basic premise that depressed people make more pessimistic future predictions than do nondepressed people, as there was no simple difference in prediction of future negative life events between participants classified as depressives (BDI score of nine or above at baseline: see Alloy & Ahrens, 1987, for a justification of this cut-off, N=26) and nondepressives (N=45), t (69)=−1.35, ns. However, there was a correlation between depressive symptoms and number of life events predicted (r=0.28, P<0.05). Thus, while people with more depressive symptoms tended to make more negative future predictions, this did not seem to differentiate participants classified as depressives and nondepressives using a traditional, though very liberal, classification technique. Finally, we determined whether predictive pessimism related to depressive symptoms over time periods longer than a month. We had only asked for predictions over the initial month of the study, so we were able only to assess whether these predictions related to depressive symptoms beyond that month of prediction. The relevant partial correlation was not significant for six month follow-up (r=0.06, n.s.), nor for the one-year follow-up (r=0.02, n.s.), indicating that this measure may be predictive only over short, focused and specific time periods.