تفاوت های سن و جنس در عاطفه منفی، آیا نقشی برای تنظیم احساسات وجود دارد؟

Abstract Elderly people report less negative affect than the young, and women report more negative affect than men. This study investigated whether age and gender differences in negative affect could be explained by emotion regulation, measured as defensiveness and rumination, while controlling for the influence of life events. One-hundred-and-ninety-five young (20–35 years) and 302 elderly (70–85 years) men and women completed the Emotional Control Questionnaire-Rehearsal, Marlowe–Crowne Social Desirability Scale, Profile of Mood States, Beck’s Depression Inventory and List of Recent Experiences. Hierarchical regressions with negative affects as dependent variables showed that age was reduced to non-significance when controlling for defensiveness, and gender was reduced to non-significance when controlling for the interaction between age and gender, which in turn was reduced when entering rumination. Life events also influenced the association between age and negative affect. The results indicate that age differences in negative affect are mediated by defensiveness and life events and that when these two influences are accounted for elderly people score higher on sadness. Gender differences in negative affect were due to the young women’s higher scores on negative affect and this is partly explained by rumination.

. Introduction The elderly have often been found to report less negative affect than the young (Gross et al., 1997 and Mroczek and Kolarz, 1998) and women have been found to report more negative affect than men (Costa et al., 1987 and Fujita et al., 1991). These group differences in negative affect could arise from differences in life events and/or emotion regulation strategies. In the present study we tested whether rumination and defensiveness mediated age- and gender differences in negative affect, while controlling for life events. Rumination is defined as conscious, spontaneous, recurrent thoughts and/or images about past negative information. Rumination has shown positive associations with negative affect, like depression, anxiety and anger (Thomsen, under review) and may be viewed as an emotion regulation strategy that increases negative affect through cognitive focus (Gross, 1999). Defensiveness has been measured using social desirability scales (Paulhus, 1984). While early interpretations of the scales were based upon the individual’s tendency to consciously skew their reports in order to create a good impression, later interpretations emphasize the self-deceptive aspect of the scales (Crowne & Marlowe, 1964). In this way, defensiveness can be viewed as an attempt to down-regulate negative affect (Gross, 1999), and has shown inverse correlations with anxiety, sadness, depression and anger (Barrett, 1996, Clark et al., 1998 and Marlowe and Crowne, 1960). 1.1. Age and gender differences in negative affect The elderly have often been found to score lower on negative affect than younger individuals in cross-sectional studies (Gross et al., 1997 and Mroczek and Kolarz, 1998). Longitudinal studies show mixed evidence (Charles et al., 2001 and Costa et al., 1987) possibly depending on the length of the follow-up period. Another commonly noticed age difference in negative affect is that elderly depressed patients more often display “masked” or “atypical” depression, which refers to the finding that the elderly often complain more of somatic problems and sleep disturbances while not reporting sadness (Mulsant and Ganguli, 1999 and Rosenberg et al., 1992). Theorists argue that the elderly often have poorer health than younger people and that this explains the “over-reporting” of somatic symptoms (Mulsant & Ganguli, 1999). Alternatively, the “over-presence” of some symptoms, especially sleep problems, may be an integrated part of a natural aging process (Rosenberg et al., 1992). However, another possibility is that the elderly express fewer affective depressive symptoms because of their generally lower levels of negative affect. Carstensen’s socioemotional selectivity theory may account for the age differences in negative affect, since it suggests that elderly individuals learn to maximize positive emotions while minimizing negative emotions to increase present well-being (Carstensen, Isaacowitz, & Charles, 1999). Thus, the elderly are thought to have more efficient emotion regulation than younger individuals, and this has been supported in self-report studies (Gross et al., 1997). Carstensen et al. (1999) emphasize efficient emotion regulation as a maturational process thus indicating a positive development of emotion regulation strategies. However, the elderly may use emotion regulation strategies, which are generally thought of as less adaptable, like defensiveness, and studies have found that elderly people score higher on defensiveness (Dijkstra et al., 2001 and Ray, 1988). In summary, age differences in negative affect and defensiveness have often been found and studies also show that the elderly experience fewer life events, although lists of life events may be biased towards events in early adulthood, like divorce and job change (Pearlin & Skaff, 1996). However, it is not known whether age differences in defensiveness and life events may explain age differences in negative affect. We therefore tested whether age differences in negative affect were mediated by defensiveness and/or life events. To the best of our knowledge no previous studies have investigated this question. Women are often found to report more negative affect than men, especially sadness and anxiety (Costa et al., 1987 and Fujita et al., 1991), whereas gender differences in anger are less clear-cut (Bartz, Blume, & Rose, 1996). In a similar vein, the prevalence of depression (Nolen-Hoeksema, 1990) and most anxiety disorders (Reich, 1986) in women is approximately twice that in men. Thus, there seems to be strong evidence for gender differences in sadness and anxiety. Nolen-Hoeksema has in a number of studies (Nolen-Hoeksema, 1990) found that women ruminate more in response to depressive mood, thus exacerbating the mood. Studies also indicate that gender differences in depressed mood may be explained by gender differences in rumination (Butler and Nolen-Hoeksema, 1994 and Roberts et al., 1998). These studies, however, only sampled psychology students, i.e. a young population, and only addressed gender differences in sadness, without controlling for life events. However, since rumination has often been found to be associated with anxiety (Thomsen, under review), gender differences in anxiety may also be explained by rumination. In summary, gender differences in sadness, anxiety and rumination are a common finding and studies suggest that rumination explains the gender differences in sadness. However, it is unknown whether rumination also explains the gender differences in anxiety and whether the effect of rumination is independent of life events. In the present study we therefore included a representative sample of both young and elderly women and men, a measure of life events and also tested whether rumination mediated gender differences in anxiety. To the best of our knowledge no previous studies have investigated these issues

3. Results The means and standard deviations of the young and elderly women and men can be seen in Table 1. A series of two-way ANOVAS with age group and gender as independent variables and the POMS scales, BDI, LRE, ECQ-R and MCSD as dependent variables were calculated in order to test for main and interaction effects. There were significant effects of gender on the ECQ-R, BDI scores and POMS depression and anxiety scales (F(1, 493) range 8.10–15.22, p < 0.01) where the women scored higher on all measures. There were also significant effects of age on LRE, MCSD, BDI somatic and affective-cognitive symptoms and POMS anger and anxiety scales (F(1, 493) range 4.94–162.80, p < 0.01) with the elderly scoring higher on MCSD and BDI somatic symptoms but scoring lower on LRE, BDI affective-cognitive symptoms and POMS anger and anxiety scales. These main effects were qualified by significant interaction effects on the BDI affective-cognitive symptoms, BDI and POMS depression and anger scales (F(1, 493) range 4.03–6.05, p < 0.05). The means in Table 1 indicates that the interaction may be due to the larger gender differences among the young participants. Table 1. Age- and gender differences in rumination, defensiveness, life events and negative affect Variable The young participants The elderly participants Women Men Women Men M(SD) M(SD) M(SD) M(SD) LRE 7.30(4.59) 7.05(3.92) 2.60(2.52) 3.31(3.55) ECQ-R 20.41(7.82) 17.40(7.07) 18.54(6.80) 17.73(7.44) MCSD 17.78(4.80) 17.79(5.32) 23.57(3.93) 23.16(5.15) BDI af. 4.62(6.31) 2.27(2.98) 2.80(3.25) 2.34(3.98) BDI som. 2.18(2.54) 1.21(1.48) 2.86(2.00) 2.34(2.08) BDI 7.08(8.68) 3.56(3.99) 6.37(5.10) 5.56(6.03) POMS dep. 4.32(5.09) 2.08(3.64) 2.77(3.70) 2.36(3.68) POMS anx. 5.52(4.67) 4.05(3.03) 3.48(3.26) 2.94(3.33) POMS ang. 2.97(3.17) 2.19(2.66) 1.46(2.11) 1.65(2.55) Table options The LRE, ECQ-R and MCSD all showed the predicted associations with the negative affect measures for both the young and the elderly participants, but generally the correlations were significantly stronger for the young participants, especially for the LRE and the ECQ-R (see Table 2). Table 2. Person’s correlations between LRE, ECQ-R and MCSD and the negative affect measures for the total sample, the elderly participants and the young participants Negative affect measure Total sample The elderly participants The young participants LRE ECQ-R MCSD LRE ECQ-R MCSD LRE ECQ-R MCSD BDI total 0.24 0.49 −0.28 0.17* 0.42* −0.33 0.41 0.57 −0.36 BDI af. symptoms 0.31 0.49 −0.35 0.19* 0.41* −0.31 0.38 0.58 −0.38 BDI somatic symptoms 0.12 0.37 −0.13 0.10* 0.34 −0.27 0.42 0.45 −0.26 POMS depression 0.28 0.45 −0.30 0.19* 0.34* −0.25 0.36 0.57 −0.36 POMS anxiety 0.33 0.45 −0.39 0.16* 0.38* −0.28 0.35 0.53 −0.39 POMS anger 0.32 0.40 −0.36 0.19* 0.27* −0.26 0.33 0.53 −0.37 All correlations significant at the 0.01 level, except the correlation between LRE and BDI somatic symptoms for the elderly participants. The asterisks show significant differences between the correlations for the young and elderly participants at the 0.05 level. Table options 3.1. Age and gender differences in negative affect When conducting mediation analysis four steps are usually followed: (1) establishing a link between the independent variable and the dependent variable, (2) establishing a link between the independent variable and the mediator variable, (3) establishing a link between the mediator variable and the dependent variable and (4) testing if the mediator variable reduces or eliminates the link between the independent variable and the dependent variable (Baron & Kenny, 1986). From the above results, it can be seen that steps 1–3 are fulfilled; (1) there are age and gender differences in negative affect (2) there are age and gender differences in the mediator variables, i.e. LRE, MCSD and ECQ-R and (3) LRE, MCSD and ECQ are correlated with negative affect. In order to test the fourth step of the mediation model we conducted a series of hierarchical multiple regressions entering each of the negative affects as the dependent variable. To control for sociodemographic variables we entered educational level, income and “living alone” as a proxy for marital status. At the second step we entered age group (1 = young, 2 = elderly) and gender. In order to test if the interactions between age and gender accounted for the gender and age differences in negative affect and if LRE, MCSD and ECQ mediated the interaction effect we entered the interaction term by calculating a dummy variable for the young women (as the results from Table 1 indicated that the interaction effect was driven by the young women) and this variable was entered at the third step (0 = all other groups, 1 = young women). ECQ, MCSD and LRE were entered at the fourth, fifth and sixth step respectively in order to allow us to disentangle which of the variables accounted for age, gender and interaction effects. If the interaction effect, LRE, MCSD and/or ECQ mediate the link between age, gender and negative affect entering these variables into the regression should reduce the betas of age, gender and/or the interaction. The regressions with R2, beta and significance levels for the different variables at each step can be seen in Table 3a, Table 3b, Table 3c, Table 3d and Table 3e, note that the 1. step and the sociodemographic variables are not shown (data for the POMS depression scale not shown as they mirror results on BDI affective-cognitive symptoms). Table 3a. Regression using POMS anger as the dependent variable Variable Step 2 Step 3 Step 4 Step 5 Step 6 Age group −0.21⁎⁎ −0.12(⁎) −0.12(⁎) −0.04 0.04 Gender 0.04 −0.04 −0.04 −0.03 −0.01 Interaction 0.15(⁎) 0.08 0.10 0.08 ECQ-R 0.39⁎⁎ 0.31⁎⁎ 0.29⁎⁎ MCSD −0.21⁎⁎ −0.18⁎⁎ LRE 0.20⁎⁎ Model F(5, 449) = 3.68 ⁎⁎ F(6, 448) = 3.67 ⁎⁎ F(7, 447) = 15.30 ⁎⁎ F(8, 446) = 15.52 ⁎⁎ F(9, 445) = 16.10 ⁎⁎ Adj. R2 0.03 0.03 0.18 0.20 0.23 (⁎) p < 0.10. ⁎⁎ p < 0.01. Table options Table 3b. Regression using POMS anxiety as the dependent variable Variable Step 2 Step 3 Step 4 Step 5 Step 6 Age group −0.24⁎⁎ −0.16⁎ −0.16⁎ −0.07 0.01 Gender 0.10⁎ 0.03 0.03 0.04 0.06 Interaction 0.14(⁎) 0.07 0.08 0.06 ECQ-R 0.43⁎⁎ 0.34⁎⁎ 0.32⁎⁎ MCSD −0.22⁎⁎ −0.20⁎⁎ LRE 0.19⁎⁎ Model F(5, 449) = 7.35 ⁎⁎ F(6, 448) = 6.68 ⁎⁎ F(7, 447) = 22.18 ⁎⁎ F(8, 446) = 22.44 ⁎⁎ F(9, 445) = 22.42 ⁎⁎ Adj. R2 0.07 0.07 0.25 0.27 0.30 (⁎) p < 0.10. ⁎ p < 0.05. ⁎⁎ p < 0.01. Table options Table 3c. Regression using BDI affective-cognitive symptoms as the dependent variable Variable Step 2 Step 3 Step 4 Step 5 Step 6 Age group −0.13⁎ 0.00 −0.01 0.07 0.15⁎ Gender 0.11⁎ −0.01 0.00 0.00 0.03 Interaction 0.22⁎⁎ 0.14(⁎) 0.15⁎ 0.13(⁎) ECQ-R 0.47⁎⁎ 0.40⁎⁎ 0.38⁎⁎ MCSD −0.19⁎⁎ −0.17⁎⁎ LRE 0.21⁎⁎ Model F(5, 449) = 5.26 ⁎⁎ F(6, 448) = 5.72 ⁎⁎ F(7, 447) = 25.65 ⁎⁎ F(8, 446) = 24.83 ⁎⁎ F(9, 445) = 25.31 ⁎⁎ Adj. R2 0.05 0.06 0.28 0.30 0.33 (⁎) p < 0.10. ⁎ p < 0.05. ⁎⁎ p < 0.01. Table options Table 3d. Regression using BDI somatic symptoms as the dependent variable Variable Step 2 Step 3 Step 4 Step 5 Step 6 Age group 0.20⁎⁎ 0.27⁎⁎ 0.27⁎⁎ 0.32⁎⁎ 0.41⁎⁎ Gender 0.14⁎⁎ 0.08 0.08 0.08 0.11(⁎) Interaction 0.12 0.05 0.06 0.04 ECQ-R 0.37⁎⁎ 0.32⁎⁎ 0.30⁎⁎ MCSD −0.14⁎ −0.11⁎ LRE 0.21⁎⁎ Model F(5, 449) = 7.33 ⁎⁎ F(6, 448) = 6.50 ⁎⁎ F(7, 447) = 17.36 ⁎⁎ F(8, 446) = 16.15 ⁎⁎ F(9, 445) = 16.98 ⁎⁎ Adj. R2 0.07 0.07 0.20 0.21 0.24 ⁎ p < 0.05. (⁎) p < 0.10. ⁎⁎ p < 0.01. Table options Table 3e. Regression using BDI total as the dependent variable Variable Step 2 Step 3 Step 4 Step 5 Step 6 Age group 0.02 0.15⁎ 0.14⁎ 0.22⁎⁎ 0.31⁎⁎ Gender 0.12⁎ 0.00 0.01 0.01 0.04 Interaction 0.23⁎⁎ 0.14(⁎) 0.15⁎ 0.13(⁎) ECQ-R 0.48⁎⁎ 0.41⁎⁎ 0.39⁎⁎ MCSD −0.19⁎⁎ −0.17⁎⁎ LRE 0.22⁎⁎ Model F(5, 449) = 3.72 ⁎⁎ F(6, 448) = 4.45 ⁎⁎ F(7, 447) = 24.78 ⁎⁎ F(8, 446) = 24.02 ⁎⁎ F(9, 445) = 24.82 ⁎⁎ Adj. R2 0.03 0.04 0.27 0.29 0.32 (⁎) p < 0.10. ⁎ p < 0.05. ⁎⁎ p < 0.01. Table options Concerning the age differences in negative affect the regressions generally show that entering the interaction term reduces or even reverses the negative betas for the age effect, except for BDI somatic symptoms and BDI total where the interaction term increases the betas. The age effect is further reduced or reversed by entering MCSD and LRE indicating that these variables mediate age differences in negative affect. Again, the exception is BDI somatic symptoms and BDI total where the betas are further increased by entering MCSD and LRE. Thus, the results indicate that age differences in negative affect are in part due to the young women scoring higher than the other three groups. Defensiveness and life events also appear to mediate the age differences and for the measures of sad affect it appears that when controlling for MCSD and LRE the elderly score higher than the young. The gender differences appeared to be due to the interaction term, i.e. only the young women scored higher on negative affect. The interaction term was reduced by entering ECQ-R indicating that the interaction was partly mediated by rumination.