تغییرات در شرایط مالی خانواده و سلامت جسمی ازدواج و مادران به تازگی طلاق گرفته

Abstract This study investigates how divorce followed by single parenthood undermines the long-term physical health of rural mothers using four waves of survey data collected in Iowa, USA from 336 married and 80 divorced mothers during a 10-year period. Findings generally support the hypothesized pathways in that single-parenthood creates financial difficulties for rural mothers. Furthermore, this financial adversity is linked to self-assessed physical health trajectories that then contribute to change in morbidity. This reflects the developmental course of morbidity during the middle years. Methodologically, this extends existing research on the association between women's marital status and well being by explicitly examining individual trajectories of change in family financial strain and physical health, as well as by examining the dynamic association between both during the middle years.

Introduction Midlife is a time of increasing heterogeneity in health. Although most people remain in good health through their middle years, others are beginning to manifest some of the chronic health conditions associated with age (Robert & House, 1996; Wickrama et al., 2001). For physiological reasons that are becoming increasingly well understood (e.g., Lovallo, 2005), at least some of this increasing heterogeneity may be due to differential exposure to chronically stressful conditions (House et al., 1994; Lovallo, 2005; Marmot et al., 1998). One of the most pervasive and disruptive sources of chronic stress at midlife is divorce: nearly half of all marriages in the USA end in divorce (Goldstein, 1999) and it has been shown to have both mental and physical health consequences (Johnson & Wu, 2002; Wu & Hart, 2002). Divorce as a chronic stressor may be greater for midlife women than midlife men because women are more likely to have had interrupted work careers and fewer independent financial resources (Kitson & Holmes, 1992; Simons and Associates, 1996). Further, they are likely to become the custodial parent with greater child-rearing responsibilities at a time when the children are in their difficult adolescent years. Moreover, for these single mothers, increasing biological vulnerability to chronic illness due to advancing age may set the stage for stronger health effects of chronic stressful conditions (e.g., Wickrama et al., 2001). In this paper, we examine heterogeneity in morbidity in 2001 in a panel of 336 married women and 80 women who divorced between early 1989 and late 1990. All of these women were biological mothers of adolescent children in 1991. Part of our task is to examine the mediating role of post-divorce financial stress in linking divorce to subsequent morbidity. This relationship may not necessarily be straightforward (Lorenz et al., 1997). Although previous research has linked financial stress to morbidity, much of it has been cross-sectional. Even when panel data were available, most studies did not take into account more than two waves of data. This limitation in previous study designs makes it difficult to examine the dynamics of change in financial stress and health (Rogosa, Brandt, & Zimowski, 1982). Post-divorce financial problems can be expected to decrease as the event of divorce recedes into the background and the single mother accommodates to her new economic reality. We know little about how chronic stress, reflected in persistently high absolute levels of post-divorce financial problems over time, competes with relative changes in post-divorce financial problems when influencing health of single mothers. As we shall develop more completely in the theoretical framework, the differential roles of the persistent and changing dimensions of financial stress will be distinguished by using growth curves to model the intercepts and slopes as their empirical indicators.

Results Change in family financial strain and self-assessed global physical health As a preliminary examination, we plotted mean levels of family financial strain and self-assessed poor global health for mothers that were divorced or married in 1991. The plot in Panel A of Fig. 2 shows that divorced mothers had chronically higher levels of family financial strain than did married mothers throughout the study period from 1991 to 2001. However, divorced mothers also had a more dramatic decline in family financial strain than did married mothers. The plot in Panel B shows that divorced mothers had a higher rate of increase in self-assessed poor health from 1991 to 2001. The curves also indicate that changes in financial strain and self-assessed health for mothers were approximately linear over the study period. Mean changes in financial strain and self-assessed poor health from 1991 to 2001 ... Fig. 2. Mean changes in financial strain and self-assessed poor health from 1991 to 2001 for divorced (SPP) and married (IYFP) mothers. Figure options Univariate growth curves To systematically investigate these individual changes, univariate latent growth curve models were estimated separately for family financial strain and self-assessed poor health from 1991 to 1994 within the structural equation framework using covariances (Jöreskog & Sörbom, 1993). Unstandardized coefficients for the univariate growth models are presented in Table 2. The latent growth curve model for family financial strain with three successive measurements showed respectable fit with the data (View the MathML sourceχ(1)2=3.89, p<.05p<.05; GFI=1.00). The average initial level of family financial strain (mean of π0) was 2.74 (p<.05p<.05), and varied significantly (0.72, p<.05p<.05) indicating substantial differences among mothers in beginning levels of financial strain. The average rate of change (mean of π1) was −0.04 (p<.05p<.05) indicating that the family financial strain of mothers decreased significantly (shift in overall mean) from 1991 to 1994. Moreover, the significant variation (0.03, p<.05p<.05) in rates of change (variance of π1) shows that some mothers were above, and others below the average rate of change for the sample as a whole. Thus, the fitted equation for average scores at time (t=0t=0, 1, and 3) for mothers’ family financial strain is: View the MathML sourceMothers'familyfinancialstrain=2.74-0.04t. Turn MathJax on Table 2. Estimates for univariate growth curves for family financial strain and self-assessed poor physical health Initial level Rate of change χ2 (df) AGFI Mean Variance Mean Variance Financial strain 2.74* (62.37) .72* (12.25) −.04* (−3.83) .03* (2.82) 3.89 (1) .98 Self-assessed poor health 2.47* (93.35) .51* (11.18) .07 (7.52) .02* (2.41) 13.65 (1) 1.00 *p<.05p<.05, t- values are in parentheses. Table options The latent growth curve model for self-assessed poor physical health with three successive measurements also showed an acceptable fit with the data (View the MathML sourceχ(1)2=13.65, p<.01p<.01; GFI=1.00). The average initial level of poor physical health (mean of π0) was 2.47 and also varied significantly (0.51, p<.05p<.05) indicating substantial differences among mothers in beginning levels of self-assessed poor physical health. The average rate of change (mean of π1) was 0.07 (p<.05p<.05) indicating that poor physical health increased significantly (shift in overall mean) from 1991 to 1994. Moreover, significant variation (0.02, p<.05p<.05) in rates of change (variance of π1) shows that some mothers were above, and others below the average rate of change for the sample as a whole. Thus, the fitted equation for average scores at time (t=0t=0, 1, and 3) for mothers’ self-assessed physical health is: View the MathML sourceMothers'self‐assessedphysicalhealth=2.47+0.07t. Turn MathJax on The model presented in Fig. 3 contains marital status as a predictor of growth parameters of family financial strain, which in turn predict morbidity of mothers in 2001. This model supports our hypothesis that divorced status increases the level of family financial strain (β=.36β=.36, t=7.91t=7.91). As expected, during the first few years of divorce, mothers experienced a lower rate of increase in family financial strain than did married mothers (β=-.23β=-.23, t=-3.62t=-3.62). Also, findings showed that divorced status followed by single parenthood influenced long-term morbidity directly (β=.17β=.17, t=3.01t=3.01). Initial level and rate of change in family financial strain both contributed to subsequent morbidity (β=.24β=.24, t=2.47t=2.47 and β=.20β=.20, t=4.37t=4.37 respectively). The total effect of divorced status in 1991 on morbidity of mothers in 2001 was .20. Standardized coefficients among single parenthood, growth parameters of ... Fig. 3. Standardized coefficients among single parenthood, growth parameters of financial strain and morbidity (t values are in parentheses). Figure options The model in Fig. 4 presents the results of the theoretical model including a range of control variables. In this model, financial strain trajectories serve as a mediator between divorce and self-assessed poor health, and self-assessed poor health trajectories serve as a mediator between financial strain and subsequent morbidity after controlling for the levels and changes in a range of variables including per capita family income, poor mental health, occupational status and number of children. Findings showed that the levels of family financial strain and self-assessed poor physical health, as well as the rates of change in family financial strain and self-assessed poor physical health were systematically associated (β=.11β=.11, t=2.41t=2.41; and β=.22β=.22, t=2.33t=2.33, respectively). Also as expected, the level of financial strain influenced the rate of change in poor physical health (β=.24β=.24, t=2.57t=2.57). That is, the level of financial strain interacts with time when influencing poor physical health. The negative influence of the level of poor physical health on rate of change in poor physical health is evidence for the regression to the mean phenomena. Maximum likelihood estimation of the theoretical model with all the control ... Fig. 4. Maximum likelihood estimation of the theoretical model with all the control variables. Figure options As expected, both the level and rate of change in self-assessed health contributed to morbidity in 2001, even after controlling for morbidity in 1994 (β=.31β=.31, t=3.80t=3.80; and β=.24β=.24, t=2.22t=2.22, respectively). Our findings also demonstrate that divorce contributes directly to morbidity of mothers in 2001 independent of measured financial strain and assessed poor physical health (β=.10β=.10, t=2.11t=2.11). This suggests that our hypothesized indirect paths do not completely capture the association of divorce at the beginning of the study (1991) with morbidity in 2001. The hypothesized direct paths from divorce to self-assessed health were not significant (not shown in Fig. 4). The model explained moderate proportions of variance in levels of financial strain (26%) and self-assessed health (14%), changes in financial strain (11%) and self-assessed health (38%), and morbidity in 2001 (52%). During the study period, 16 married mothers divorced, and 27 divorced mothers recoupled. As noted previously, we incorporated three orthogonal contrasts associated with initial marital status and changes in marital status in the sample to accommodate for these changes in marital status. The first contrast, “divorced vs. married in 1991” compared divorced mothers with married mothers at the beginning of the study. The association of this dichotomous variable with an outcome variable corresponds to the differences between divorced and married mothers in 1991 in relation to that outcome. For example, the association of this contrast with financial strain in 1991 is .23 indicating that on average, financial strain levels are .23 standard deviation units higher for divorced mothers than for married mothers in 1991. The second contrast, “married to divorced” compared mothers who changed from married to divorced with mothers who remained married from 1991 to 1994. Associations with this contrast correspond to differences between those two groups of mothers. Findings show that there were no significant differences between these two groups in relation to outcome variables examined in this study. The third contrast, ‘divorced to married’, compared mothers who changed from divorced to married with mothers who remained single from 1991 to 1994. Mothers who changed from divorced to married showed a greater increase in per capita family income from 1991 to 1994 than did mothers who remained single (β=.12β=.12, p<.05p<.05). Because the three contrasts were constructed to be orthogonal (i.e., independent), correlations among the contrasts were fixed at zero. These findings show that divorce associates with self-assessed health indirectly through both the level and changes in poor mental health. Divorced status is associated with higher initial levels of poor mental health (β=.21β=.21, t=4.26t=4.26). As in the case of financial strain, divorced mothers experienced a lower rate of increase in depression than did married mothers during the first few years following divorce, (β=-.14β=-.14, t=-2.82t=-2.82). The model also controlled for education level, age, number of children and occupational status (not shown in Fig. 4). The level of family financial strain (β=-.08β=-.08, t=2.13t=2.13) and per capita family income (β=.17β=.17, t=2.31t=2.31) were both significantly influenced by education, whereas age was only associated with per capita family income (β=.11β=.11, t=2.10t=2.10). The number of children in 1991 significantly influenced the level of financial strain in 1991 (β=.21β=.21, t=2.34t=2.34) whereas change in number of children significantly influenced change in physical health (β=.10β=.10, t=1.97t=1.97). Initial level of occupational status and change in occupational status were not significantly associated with any study variables. Finally, we estimated the model with morbidity in 1994 as the outcome variable controlling for morbidity in 1991 using the same covariate controls from 1991 to 1994. In addition, we allowed morbidity in 1991 to correlate with other model constructs. Findings show that both the level in 1991 (β=.36β=.36, t=4.46t=4.46) and change in self-assessed health (β=.30β=.30, t=2.18t=2.18) through 1994 contribute to morbidity in 1994 after controlling for morbidity in 1991. All the other paths in the model did not show any changes in statistical significance from those in the model predicting morbidity in 2001.