The Influence of Meaning-making after Spousal Loss on Trajectories of Psychological Distress

Abstract

Stress process researchers note that people actively seek to alter the impact of stressful life events through various coping mechanisms. Spousal bereavement sometimes requires individuals to revise their assumptions about the world and themselves. Meaning-making, comprising the dual processes of searching for and finding meaning, may be employed to alleviate symptoms of grief following spousal bereavement. The current study uses multilevel modeling to examine the influence of searching for and finding meaning on individual growth trajectories of psychological distress using data from 764 widows and widowers from a nationally representative longitudinal study. Results indicate that searching for meaning after spousal loss may be a double-edged sword: it is beneficial when meaning is found, but harmful when meaning remains elusive. Widows and widowers who searched for but did not find meaning exhibited the highest initial levels of psychological distress, but they also had slightly faster declines in psychological distress, becoming more similar to other meaning-making groups over time.

Keywords

bereavement widowhood mental health grief

Introduction

Most American adults are married (U.S. Census Bureau 2011), and marriage is often significantly associated with diminished stress (Lupton and Smith 2003; Pearlin and Johnson 1977) and improved mental health (Horwitz, White, and Howell-White 1996; Mark and Lambert 1998; Simon 2002). However, the corollary to marriage’s potential mental health benefits is the possibility of significant mental health problems following marital bereavement. As Shuchter (1986:170) notes, “virtually everyone whose spouse dies exhibits some signs and symptoms of depression”; even individuals who remain resilient in the face of spousal loss often experience “transient perturbations in normal functioning” (Bonanno 2004:21). Widowhood is a common part of the life course, and some individuals—women in particular—may remain in this social status for long periods of time if not permanently (Wortman and Silver 1990). With over 14 million widows and widowers currently living in the United States, it is important for researchers to understand the health implications of spousal loss (U.S. Census Bureau 2011).

Spousal loss has been considered a prototypical stressful life event. Past research has documented associations between spousal loss and increased depressive symptoms and incidence of major depression (Bennett 1998; Gallagher et al. 1983; Harlow, Goldberg, and Comstock 1991; Zisook et al. 1997, 2004). While psychological distress usually lessens among older adults between 1½ to 2 years after the loss (Harless and Zisook 2003), depressive symptoms may last upwards of 3 years in some cases (Chou and Chi 2000; Lucas 2005; Lucas et al. 2003; Umberson, Wortman, and Kessler 1992).

More recent research has emphasized that widowed persons do not comprise a monolithic group in their responses to spousal loss but instead display a great deal of variation in adjustment to widowhood. Bonanno and colleagues (2002) highlight the resilience that a majority of bereaved individuals display, experiencing only minor fluctuations in well-being and functioning up to 18 months following their loss. However, while a majority of recent widows and widowers are resilient, others experience short-term grief, long-term grief, or even substantial declines in depression immediately following their loss (Bonanno and Kaltman 2001; Bonanno, Westphal, and Mancini 2011; Bonanno, Wortman, and Nesse 2004; Mancini, Bonanno, and Clark 2011). After 18 months, a majority of widowed persons return to previous functioning, exhibiting decreasing levels of depression and grief, with the exception of those experiencing chronic depression or chronic grief (Bonanno et al. 2002).

Social factors such as gender (Chen et al. 1999; Gallagher et al. 1983), social support (Bennett 2009; Dimond, Lund, and Caserta 1987; Walker, MacBride, and Vachon 1977), financial strain (Utz 2002), prior social context (Wheaton 1990), and participation in post-bereavement volunteering (Li 2007) all explain some of the variation in the mental health consequences of spousal loss. This article contributes to the life events literature by examining how meaning-making, or “making sense of loss” (Corr, Nabe, and Corr 2009:230), can help to explain variation in psychological distress after spousal bereavement. We conceptualize meaning-making as a dual process, constituting searching for and finding meaning, and distinguish membership in three meaning-making groups: those who do not search for meaning, those who search for and find meaning, and those who search for but fail to find meaning. Then, we use multilevel modeling to measure the effect of membership on the initial status and rate of change of individual growth trajectories of psychological distress using nationally representative longitudinal data from the Americans’ Changing Lives Survey.

Review of the Literature

The Role of Meaning-making

Life stress literature conceptualizes and operationalizes meaning-making in a variety of ways, from attempts to assess meaning by contextualizing a negative life event to understanding individuals’ cognitive appraisals of the life event (Simon 1997). This article focuses on meaning-making as a dual process of self-reported searching for and finding meaning after a negative life experience. Many people may begin to search for meaning, a reason or an explanation as to why a specific event occurred. They may ask themselves, “Why me?” or “Why my spouse?” (Schwartzberg and Janoff-Bulman 1991). Because traumatic events can result in an existential crisis, shattering an individual’s perception of the world or themselves, the ability to find meaning in a negative life experience can restore a view of the world as orderly, predictable, controllable, fair, and meaningful (Davis et al. 2000; Janoff-Bulman 1992; Simon 1997).

Losses may appear more or less reasonable given the age of the deceased, with “off-time” losses more frequently prompting meaning-making on the part of the bereaved (Davis and Nolen-Hoeksema 2001). Also, sudden, unexpected, or traumatic losses may prompt the search for meaning (Davis et al. 2000). When undergoing an existential crisis, very few bereaved persons attempt to alter their worldview, but instead find a way to reconcile the loss with their existing worldview by “making-sense” of the loss (Janoff-Bulman 1992). Family members facing the death of a loved one may attempt to make sense of loss through a variety of justifications, such as conceptualizing the death as predictable or consistent with the deceased person’s perspective on life or viewing the death as in accordance with religious or spiritual beliefs (Davis and Nolen-Hoeksema 2001). Religious and spiritual beliefs may play an important role in coping with trauma, making it easier to find meaning by providing a framework of beliefs for incorporating negative events (Davis et al. 2000).

Not all individuals undergoing a loss search for meaning. This may be because they rely on avoidant coping strategies such as denial or substance abuse (Updegraff, Silver, and Holman 2008). Alternatively, they may not view the experience of the life event as an existential crisis or they may have what Antonovsky (1979) refers to as a sense of coherence, or a general, enduring view that the world is predictable and everything will work out as well as can reasonably be expected. Thus, while some widowed individuals may search for meaning after their loss, others may not. Similarly, while the search for meaning may be successful for some, others find they are unable to make sense of or find meaning in the death of their spouse.

The Search for Meaning

Most research on searching for and finding meaning after loss comes from a number of small studies of specialized populations. Due to small sample sizes and issues of generalizability, the prevalence of searching for and finding meaning differs across studies. For example, when asked in interviews whether they searched for a reason for their loss, 80 percent of bereaved persons in the Harvard Bereavement Study indicated that they had (Parkes and Weiss 1983).¹ Using data from a cross-sectional study of 94 adults coping with the loss of either a spouse or a child from a motor vehicle accident four to seven years prior, Davis and colleagues (2000) found that 70 percent of bereaved spouses and 79 percent of bereaved parents reported that they had searched for meaning (but only 40 percent had found meaning by that time). Schwartzberg and Janoff-Bulman (1991) found that 90 percent of undergraduates dealing with the loss of a parent searched for meaning, whereas only 50 percent were able to answer the question of “Why me?” or “Why my parent?” when interviewed approximately three years after the death.

There have been a number of prospective panel studies conducted on coping with loss and grief. Although these studies also confront issues of small sample sizes and lack of generalizability, they follow bereaved individuals over time and thus offer more insight into the timeframe of searching for and finding meaning. Nevertheless, it should be noted that these studies vary greatly regarding the extent to which searching for meaning continues over time (Park 2010). Among a sample of 124 bereaved parents whose infant died of SIDS,² Davis and colleagues (2000) reported that two to four weeks after the loss, 14 percent of parents did not search for meaning at all, 68 percent reported that they were still continuing to search for meaning, and 18 percent were no longer searching for meaning. The authors concluded that parents who did not find meaning by two to four weeks after their loss were not likely to develop a meaningful account of the event. Among the parents who reported searching for meaning during their first interview, only 42 percent found meaning in the 18 months following the death of their child (Davis et al. 2000).

Using data from the Stanford Bereavement Project,³ Davis, Nolen-Hoeksema, and Larson (1998) found that 68% of the sample reported having made sense of the death and another 10 percent reported having made “some sense” of the death six months after their loss. The researchers concluded that those who were unable to make sense of their loss after six months were generally unable to ever make sense of it. Lastly, in a study of bereaved parents whose child died by accident, homicide, or suicide, respondents were asked “have you searched for meaning in your child’s death as well as in your own life?” (Murphy, Johnson, and Lohan 2010:387). Approximately 12 percent reported finding meaning 12 months after the loss and 57 percent reported finding meaning five years after the loss. Thus, slightly under a majority did not find meaning up to five years after the loss (Murphy et al. 2010).

The Changing Lives of Older Couples Study (CLOC) is one of the only representative, prospective studies of bereavement. The original CLOC is a Detroit-area study of noninstitutionalized, English-speaking married individuals who lived in a household where the husband was 65 years of age or older. Researchers asked a subset of CLOC respondents who experienced the death of a spouse after their initial interview to participate in a follow-up study, surveying them 6, 18, and 48 months after their loss. In the 6-month follow-up interview, researchers asked the respondents, “During the past month, have you found yourself searching to make sense of or find some meaning in your husband/wife’s death?” and “Have you made any sense of or found any meaning in your husband/wife’s death?” Bonanno et al. (2004) combined these questions, assigning participants to one of three categories: not searching for meaning, searching for but not finding meaning, or searching for and finding meaning. Using this categorization, they found that among the bereaved sample (N = 185), 71 percent reported not searching for meaning, 14 percent reported searching for but not finding meaning, and 15 percent reported searching for and finding meaning 6 months after their loss. The estimates derived from the CLOC study of searching for meaning are much lower than those reported previously, but this may be due to the fact that many of the other studies had relatively small sample sizes and examined specialized populations for whom the loss was sudden, unexpected, or off-timed (Davis et al. 2000; Murphy et al. 2010; Schwartzberg and Janoff-Bulman 1991). As previously noted, these types of losses often provoke a search for meaning by the bereaved (Davis and Nolen-Hoeksema 2001).

Past research findings challenge clinical assumptions about coping with loss, namely, that people undergoing loss inevitably search for meaning and that most people are able to find meaning over time (Davis et al. 2000). The prevalence of searching for meaning varies by study, with estimates as low as 30 percent (Bonanno et al. 2004) and as high as 80 percent (Parkes and Weiss 1983). Moreover, a sizable proportion of bereaved individuals who have searched or continue to search for meaning report that they have not made sense of their loss (Davis et al. 2000).

Meaning-making and Mental Health

Some clinicians assume that meaning-making is an essential, critical step in adjustment to loss (Davis et al. 2000), but past research findings have been inconsistent in this regard. Meaning-making purportedly diminishes distress, acting as a buffer against complicated grief, which consists of feelings of distance, numbness, or shock (Currier, Holland, and Neimeyer 2006; Davis et al. 1998). Bennett and Vidal-Hall (2000) describe how many widows in their sample felt “numb” when their spouses passed away. One person described being “in a complete fog” (Bennett and Vidal-Hall 2000:422). Among bereaved persons actively trying to make sense of their loss, finding meaning is associated with lower levels of mental distress and depression (Davis et al. 1998, 2000; Murphy et al. 2010). Davis and colleagues (1998) reported that finding meaning within 6 months of a loss event predicted lower levels of mental distress at 6 and 13 months post-loss, even after controlling for initial levels of distress. Such results exemplify why Holland, Currier, and Neimeyer (2006) identify meaning-making as the strongest predictor of adjustment to bereavement and others view meaning-making as essential to adjusting to stressful life events (Park 2010).

Although most studies on meaning-making find that it is protective against poor mental health for those who search for meaning, more recent findings from the CLOC study cast doubt on the beneficial effect of finding meaning in loss. Bonanno and colleagues (2004) examined the relationships between trajectories of depression and reports of meaning-making among individuals who had been widowed for 18 months. Chronic grievers (characterized by heightened levels of depression through the 18 months following their loss) were most likely to report having searched for and found meaning. Those who did not search for meaning at all were most likely to be resilient (characterized by little change in mental health following the loss).

Indeed, the mere search for meaning may itself increase anxiety, and the relationship of searching for meaning to adjustment may be contingent on whether or not meaning is found. Davis and colleagues (2000) found that parents losing a child to SIDS who did not search for meaning were less emotionally stressed throughout the entire 18-month study period. Searching for meaning in a potentially traumatic event may reflect a persistent fixation on the event, or rumination, and can jeopardize successful adjustment to loss (Bonanno et al. 2011; Park 2010). Indeed, the benefit of actually finding meaning for mental health may be partially due to decreased rumination (Michael and Snyder 2005). Thus, several researchers argue that those who do not attempt to make sense of their loss are better off than those who do (for examples, see Park 2010). In short, the extent to which meaning-making is associated with improved mental health may be dependent on its course. Searching for meaning following the death of a loved one but failing to find it may be worse for mental health than not searching for meaning at all.

Previous studies provide a solid foundation from which to work but often rely on relatively small sample sizes and/or cross-sectional data or provide little generalizability due to their reliance on data from one specific location. The current study uses multilevel modeling to examine the influence of searching for and finding meaning on individual growth trajectories of psychological distress. The data are from 764 ever-widowed men and women from a nationally representative longitudinal study. Unlike previous research studies, this sample encompasses individuals who have been widowed recently, as well as those who have been widowed many years prior. In keeping with previous research from the CLOC study, we model three meaning-making categories, including: those who do not search for meaning, those who search for meaning without finding it, and those who search for and find meaning after spousal loss. We examine how individuals change over time after bereavement (intraindividual change) and the extent to which meaning-making can explain differences between individuals (interindividual differences) in growth parameters.

Methods

Participants

The Americans’ Changing Lives (ACL) study is a longitudinal panel study conducted by the Survey Research Center at the University of Michigan, with data collected in 1986, 1989, 1994, and 2001/2002. The ACL is designed to facilitate an understanding of the role of numerous psychosocial and behavioral factors in the maintenance of health and functioning as people age (House, Lantz, and Herd 2005). Pertinent to the present study, the ACL provides a wealth of information on stressful life events, such as spousal bereavement, meaning-making, and health outcomes. Additionally, the Americans’ Changing Lives Survey offers longitudinal data on a large number of individuals who have experienced widowhood at some point during their lives. Thus, we can use it to assess the influence of searching for and finding meaning in widowhood across a larger timeframe from among individuals who are recently bereaved to those who have been bereaved for decades.

The ACL sample includes persons who were 25 years of age or older and living in the continental United States at baseline. The first wave consists of face-to-face interviews with a nationally representative cross-section of Americans, oversampling blacks (2:1) and those 60 years of age or older (2:1). With an individual response rate of 68 percent and a household response rate of 70 percent, Wave 1 includes a sample size of 3,617. Wave 2 (N = 2,867) and Wave 3 (N = 2,562) both have a survivor response rate of 83 percent.

For the multilevel analysis presented in this study, we converted the data into a person-period data set, in which each individual may have multiple records, contributing up to three records of data (one for each wave they participated in). Instead of time representing wave of data collection, year, or age of respondent, time is measured as months since the bereavement event.⁴ In Wave 1, 804 respondents reported that they had ever been widowed. We then excluded respondents who had missing information on the primary independent variables of interest, the year in which they were widowed, or who had more than three indicators of psychological distress missing at Wave 1. Of these 764 respondents, 580 participated in Wave 2 (115 were nonresponders and 69 were deceased), and 446 participated in Wave 3 (90 were nonresponders and 228 were deceased). Of the possible 1,790 person-period records, 79 records had more than three indicators of psychological distress missing in either Wave 2 or Wave 3, leaving a sample size of 1,711 person-period records.

In an effort to understand the influence of sample attrition on the results, we estimated multinomial logistic regression models predicting sample attrition using variables included in the models presented in the following analysis (see Appendix A). Ever-widowed individuals who were lost to follow-up (nonparticipants) at Wave 2 were significantly older than the rest of the sample. In addition, sample members who died between Wave 1 and Wave 2 were more likely to be older, male, unmarried, and widowed for longer than those who remained in the sample. Compared to those who participated in the third wave of data collection, sample members who had died by Wave 3 were more likely to be male and to be unmarried.

Analytic Strategy

This study fits individual growth models (using PROC MIXED in SAS 9.2) to assess change in psychological distress over time, with time measured as months since the death of a spouse. The number of months was the most accurate, estimable measure of time and is a continuous variable, ranging at baseline from 2 to 769 months (64 years) with a mean of 179 months (15 years).⁵ We use a multilevel model where the level 1 submodel describes how each person changes over time while the level 2 submodel relates interindividual differences in change to a set of predictor variables (Singer and Willet 2003). This type of multilevel analysis allows us to ascertain how psychological distress changes over time (intraindividual change) and to test the extent to which meaning-making can predict differences between individuals in their growth parameters.

Equation 1 presents the level 1 submodel demonstrating intraindividual change. Y_ij represents the dependent variable for individual i at time j, π_0i is individual i’s true initial status, π_1i is individual i’s true rate of change during the study period, and r_ij is the portion of individual i’s outcome that is unpredicted at time j (Singer 1998; Singer and Willet 2003).

Y_{ij} = π_{0 i} + π_{1 i} TIM E_{i j} + r_{ij}, where r_{ij} ~ N (0, σ^{2}) .

Equations 2 and 3 model interindividual differences in change. β₀₀ and β₁₀ are the intercepts that represent the average initial status and rate of change in the population. COVAR represents a predictor variable of interest. β₀₁ and β₁₁ represent the effect of the predictor variable on changes in the initial status or rate of change of the trajectory. μ_0i and μ_1i represent the portion of initial status and rate of change unexplained in the level 2 model (Singer 1998; Singer and Willet 2003).

\begin{array}{l} π_{0 i} = β_{00} + β_{01} COVA R_{i} + μ_{0 i} \\ (\begin{matrix} μ_{0 i} \\ μ_{1 i} \end{matrix}) ~ N (\begin{matrix} 0 \\ 0 \end{matrix}), [\begin{matrix} τ_{00} & τ_{01} \\ τ_{10} & τ_{11} \end{matrix}] \\ π_{1 i} = β_{10} + β_{11} COVA R_{i} + μ_{1 i} . \end{array}

Substituting the level 2 submodel into the level 1 submodel leads to the composite model, presented in Equation 4 (Singer 1998; Singer and Willet 2003).

Y_{i j} = β_{00} + β_{10} TIM E_{i} + β_{01} COVA R_{i} + β_{11} (COVA R_{i} \times TIM E_{ij}) + μ_{0 i} + μ_{1 i} TIMEij + r_{i j} .

Dependent Variable

The 11-item Center for Epidemiologic Studies Depression Scale (CES-D), developed by Kohout and colleagues (1993), is a depression symptom index and is used here to measure level of psychological distress. It is both reliable (Cronbach’s α = .81) and closely associated with the original 20-item CES-D scale (r = .95). Respondents indicated how often in the past week they: “felt depressed,” “felt everything was an effort,” “sleep was restless,” “were happy,” “felt lonely,” “felt people were unfriendly,” “enjoyed life,” “didn’t feel like eating,” “felt sad,” “felt people disliked them,” and “couldn’t get going.” Response categories were: (1) hardly ever, (2) some of the time, and (3) most of the time. We reverse coded positive items such that higher scores indicate higher levels of psychological distress (ranging from 11 to 33). As previously mentioned, respondents were not included if they had missing data for more than three indicators of psychological distress. Respondents with missing information on less than three indicators of psychological distress were assigned the sample mean for the particular indicator that was missing.

Predictor Variables

The main independent variables of interest are dummy variables representing membership in three distinct meaning-making groups: (1) those who did not search for meaning (no search), (2) those who searched for meaning but did not make sense of their loss (search–no sense), and (3) those who searched for meaning and made sense of their loss (search-sense). We created these groups using two retrospective questions ascertaining the respondent’s degree of searching and degree of finding meaning after widowhood. Searching for meaning was an ordinal variable, measured by the following question: “Some people have said that they find themselves searching to make sense or find some meaning in their loved one’s death. Have you ever done this since your [husband’s/wife’s] death?” Response categories included: (1) no, never; (2) yes, rarely; (3) yes, somewhat; (4) yes, frequently; and (5) yes, all the time. Finding meaning was an ordinal variable, measured by the following retrospective question: “Have you made any sense or found any meaning in your [husband’s/wife’s] death?” Response categories included: (1) no, not at all; (2) yes, a little; (3) yes, some; (4) yes, quite a bit; and (5) yes, a great deal. These two questions closely correspond to other single-item questions used in several studies of meaning-making (Bonanno et al. 2004; Davis et al. 1998; Kim et al. 2011; Murphy et al. 2010) and have shown considerable utility in predicting complicated grief (Currier et al. 2006), mental distress (Davis et al. 1998; Murphy et al. 2010), marital satisfaction, physical health (Murphy et al. 2010), and personal strength (Kim et al. 2011). If respondents had reported that they never searched for meaning (code 1: no, never) we coded them as no search. If they reported that they had searched for (response categories 2-5) but not found meaning (code 1), then we assigned them to the search–no sense category. Finally, if respondents reported that they had searched for (response categories 2-5) and found meaning (response categories 2-5), we assigned them to the search-sense category. We used this classification system because a majority of the respondents reported having never searched (code 1) for meaning and additional analyses showed similar mental health outcomes for those who did not search for meaning regardless of whether some meaning was reported.⁶ Moreover, this was the same classification system used by Bonanno and colleagues (2004).

Background and Control Variables

The study’s models include controls for several demographic and death-related variables. These variables may be associated with the independent and dependent variables of interest, and we control for them to ensure that the relationships between meaning-making and psychological distress are not spurious. These factors include age at loss, gender, race, marital status, education, income, religious importance, and whether the death was expected. Age at loss is a continuous variable indicating the age (in years) at which the respondent lost his or her spouse. This variable has a mean of 55 years and standard deviation of 15 years. Gender (1 = female, 0 = male), race (1 = black, 0 = not black), and marital status (1 = married, 0 = not married) are dichotomous variables. Researchers measured education as a continuous variable indicating highest year of education completed. The ACL originally measured total family income (in dollars) as an ordinal variable with 10 response categories. We transformed income into a continuous variable, assigning each respondent the median value for his or her chosen income category and then logging these income estimates. Religious importance is an ordinal variable measured using the question, “In general, how important are religious or spiritual beliefs in your day-to-day life?” Response categories range from (1) not at all important to (5) very important. Respondents answered the question, “Was [his/her] death totally unexpected or did you expect it for some time?” Researchers measured whether the death was expected as a dichotomous variable (1 = expected, 0 = not expected).

Results

Table 1 presents the ever-widowed sample by meaning-making category. Approximately 60 percent of the sample belong in the no search category, while roughly equal proportions fit the designations of search–no sense (19 percent) and search-sense (21 percent). While the groups appear very similar on most demographic factors, those in the no search category are more likely to be black, more likely to have expected the death of their spouse, and, on average, have the lowest income and greatest time (in months) since bereavement.

Table 1.

Means (and percentages) by meaning-making category, Americans’ Changing Lives (ACL) ever-widowed sample (N = 764), 1986.

	No search	Search–no sense	Search-sense
	(N = 452)	(N = 148)	(N =164 )
Meaning-making factors
Ever search for meaning	1.00	3.14	2.97
Made sense of death	1.92	1.00	3.04
Demographic factors
Age at loss	56.46	52.53	53.75
Female (percentage)	80.31	79.73	83.54
Black (percentage)	37.61	26.35	32.32
Currently married (percentage)	15.49	17.57	15.24
Education (years)	9.53	10.28	10.63
Income	$9,817.81	$14,172.51	$11,875.24
Religious importance	3.60	3.61	3.71
Death-related factors
Time since death (months)	184.88	172.26	169.53
Death expected (percentage)	55.97	42.57	49.39
Psychological distress Wave 1	15.58	16.76	15.88

Table 2 presents the parameter estimates from nested multilevel models predicting trajectories of psychological distress. The level 1 component (as shown in Equation 1), known as the individual growth model, represents intraindividual change (i.e., individual change in psychological distress measured after spousal bereavement). The level 2 components (as shown in Equations 2 and 3) allow us to ascertain interindividual variation in the growth parameters. Model 1 represents the unconditional means model, which predicts that the average psychological distress across all individuals i and all occasions j is 15.753. Calculating the intraclass correlation (ρ = 0.47) leads us to conclude that 47 percent of variation in psychological distress is due to differences across individuals. Model 2 is the unconditional growth model, where the only predictor is time (measured as months since bereavement). This model estimates two growth parameters: the initial status (16.284) and slope (–0.003) of the growth trajectory. Again, calculating the intraclass correlation (ρ = 0.535) leads us to conclude that a little more than half of the variation in psychological distress is due to interindividual variation. Thus, the two unconditional models suggest equal amounts of within-person and between-person variation.

Table 2.

Parameter estimates (and standard errors) of the fixed and random effects of the multilevel modeling of psychological distress, Americans’ Changing Lives (ACL; 1986-1994).

		I	II	III	IV	V
Fixed effects
Initial status	Intercept	15.753***	16.284***	20.645***	21.048***	21.739***
		(0.119)	(0.208)	(1.086)	(1.084)	(1.132)
	Search-sense				−0.939**	−1.891**
					(0.347)	(0.630)
	No search				−0.952**	−1.864***
					(0.298)	(0.535)
Rate of change	Intercept		−0.003**	−0.004***	−0.004***	−0.007***
			(0.001)	(0.001)	(0.001)	(0.002)
	Search-sense					0.005†
						(0.003)
	No search					0.004*
						(0.002)
Variance components
Level 1	Within-person	7.903***	7.801***	7.763***	7.757***	7.755***
		(0.361)	(0.360)	(0.358)	(0.358)	(0.358)
Level 2	Initial status	6.950***	8.958***	8.277***	7.894***	7.776***
		(0.577)	(1.532)	(1.412)	(1.392)	(1.378)
	Rate of change		0.000†	0.000*	0.000*	0.000†
			(0.000)	(0.000)	(0.000)	(0.000)
	Covariance		−0.009	−0.011*	−0.010†	−0.010†
			(0.006)	(0.006)	(0.006)	(0.006)
Goodness-of-fit statistics
−2 log likelihood		9,200.8	9,186.9	9,104.5	9,093.6	9,089.0
Bayesian Information Criterion		9,220.7	9,226.7	9,197.5	9,199.9	9,208.5

Note. This analysis was derived from data of 764 bereaved respondents, who contributed 1,711 observations. Models 3, 4, and 5 control for the effects of age at death, gender, race, education, income, and religiosity and whether the death was expected.

†

p < .10. *p < .05. **p < .01. ***p < .001.

Models 3 through 5 represent nested conditional growth curve models. Model 3 controls for the effect of a number of baseline factors, including age at bereavement (in years), gender, race, marital status, education, income, religiosity, and whether the death was expected. (See Appendix B for complete table including all parameter estimates.) Model 4 incorporates two binary variables representing meaning-making membership (search–no sense is the reference category) and assesses the influence of meaning-making on change in the initial status of the trajectory. The parameter estimates for search-sense and no search are both statistically significant, negative, and of a similar magnitude. Individuals who search for and find meaning after spousal loss have lower initial psychological distress compared to those who search for and do not find meaning; individuals who do not search for meaning after spousal loss also display lower initial psychological distress compared to those who search for but fail to find meaning. Model 5 incorporates interaction terms between meaning-making category and time (No Search × Time/Search-Sense × Time) to examine the influence of meaning-making on change in the slope of the trajectory. Similar to the previous model, parameter estimates for the effect of search-sense and no search on initial status continue to be statistically significant, negative, and similar in magnitude. The parameter estimates for the effect of these two variables on rate of change (i.e., slope) are positive and also similar in magnitude. Thus, individuals in the search–no sense category have a significantly higher initial level of psychological distress but also exhibit a slightly steeper decline over time in psychological distress compared to those in the no search and search-sense categories. The level 2 (interindividual) variance in the initial status of psychological distress declines 13 percent from Model 2 to Model 5. The level 1 and level 2 variances in initial status are statistically significant, and therefore, potentially explainable residual variation remains at both the intra- and interindividual levels.

For greater ease of interpretation, Figure 1 represents the individual growth trajectories by the three meaning-making categories. The horizontal axis, time in months since bereavement, ranges from 0 to 240 because 75 percent of the sample observations were within this range at baseline. Those in the search–no sense category have a higher initial level of psychological distress, but a steeper slope than those in the no search and search-sense groups. Thus, the trajectories for the three groups appear to be converging slowly over time. Individuals in the search-sense and no search categories are similar in terms of both initial psychological distress and their rate of change in distress over time.

Figure 1.

Predicted growth trajectories of psychological distress by meaning-making category, Americans’ Changing Lives (ACL) 1986-1994.

Discussion

The purpose of this study was to examine the influence of meaning-making on individual growth trajectories of psychological distress among men and women who have undergone widowhood. Results indicate that meaning-making influences the initial status and rate of change of individuals’ psychological distress over time. Those who do not search for meaning and those who search for and find meaning after spousal loss appear to have lower levels of initial psychological distress than those who search for but do not find meaning. Thus, searching for meaning may be a double-edged sword; it can be beneficial when meaning is found but harmful in the absence of found meaning.

There are two possible explanations for finding a relationship between meaning-making and mental health: (1) social causation (meaning-making affects psychological distress) or (2) social selection (individuals who are highly distressed are more likely to search for or find meaning after their loss). Cross-sectional associations between meaning-making and mental health may be due to either social causation or social selection; thus, longitudinal data were needed to further disentangle the issue. If meaning-making were predictive of changes in or trajectories of mental health over time, this would provide some additional support for social causation. There are substantial theoretical reasons for expecting social causation. Social-cognitive models of coping and adjustment posit that meaning-making—when successful—will play a central role in the process of dealing with loss and trauma, reinforcing self-worth and fundamental beliefs about how the world works (Davis and Nolen-Hoeksema 2001; Epstein 1973; Janoff-Bulman and Frantz 1997; Parkes 1998). Additionally, crisis theory provides a possible answer for why failure to find meaning may be detrimental to mental health. Events that shock one’s sense of fairness, justice, or expectations may trigger a crisis, with unresolved events leading to depression (Reynolds and Turner 2008). Moreover, other recent bereavement research is consistent with a social causation explanation for the relationship between meaning-making and mental health. Among the recently bereaved, Bonanno and colleagues (2004) found that individuals suffering from chronic depression—highly depressed before and after bereavement—showed no clear relationship to later reports of searching for or finding meaning. In addition, Davis and colleagues (1998) found that among bereaved individuals actively searching for meaning, finding meaning within 6 months of a loss was predictive of lower mental distress after 6 and 13 months even after controlling for initial levels of distress.

Consistent with social-cognitive models of coping, crisis theory, and recent bereavement research, this article provides some support for the social causation explanation. Our questions pertaining to meaning-making were retrospective, asking the respondent at baseline if he or she ever searched for or found meaning after widowhood. Although retrospective measures are prone to recall bias, they provide the opportunity to anchor answers to the past and assess temporality beyond what is possible with purely cross-sectional data. As measured, meaning-making was significantly predictive of initial baseline status, with those who did not search and who searched for and found meaning exhibiting lower levels of psychological distress than those who searched but did not find meaning. While it is possible that significant relationships between the retrospective measures of meaning-making and initial levels of psychological distress are attributable to social selection, meaning-making predicted the rate of change in psychological distress over time. Interestingly, it is those who search for but do not finding meaning that have the steepest declines in psychological distress across time, although they report higher initial levels of distress. It may be that failing to find meaning soon after a loss event causes heightened levels of grief and/or psychological distress, but, as the age-old adage goes, time heals all wounds. People who search for but fail to find meaning may display steeper declines in psychological distress over time—perhaps because their higher initial levels of psychological distress leave more room for improvement. Evidence suggests that over time, bereaved individuals focus less and less on coping and restoration, instead working toward developing a new identity (Caserta and Lund 2007; Richardson and Balaswamy 2001; Stroebe and Schut 2010). The failure to find meaning, then, may be introducing a lag effect whereby the process of developing a new identity does not begin until later. However, once begun, the process may be particularly liberating, moderating the influence of the meaning-making process on psychological distress.

In our sample, 60 percent of the widowed individuals did not search for meaning after their loss; only 21 percent reported searching for and finding meaning and 19 percent reported searching for but not finding meaning. While these estimates are similar to those found using data from the CLOC study (Bonanno et al. 2004), they are very different from other research that used smaller samples of select populations. That a majority of widows and widowers do not search for meaning after their loss and exhibit similar levels of psychological distress compared to those who search for and successfully find meaning further reinforces the notion that searching for meaning may not be therapeutic for all. These findings challenge the assertion that a majority of bereaved individuals undergo a search for meaning and that meaning-making is a necessary step in adjustment to loss.

Our findings highlight the importance of incorporating meaning-making into studies of stressful life events. While this study focuses on whether or not individuals searched for and found meaning, the content of meaning may also have important implications for mental health outcomes. Meaning-making (as well as the meaning content) may be important to consider as a potential mechanism in understanding how other social and psychological factors influence individual variation in response to life events. For example, meaning-making may represent a potential pathway between personality traits or religiosity and adjustment to bereavement.

This study has several limitations that deserve discussion. First, although highly predictive of psychological distress, only two items were used to measure meaning-making in this study. While it is not ideal that our assessment of searching for and finding meaning rests on only two variables, these exact two items have been used in previous studies to measure meaning-making and have been strongly related to other measures of mental health and well-being (Bonanno 2004; Currier et al. 2006; Davis et al. 1998; Kim et al. 2011; Murphy et al. 2010). Second, this study measured meaning-making at one point in time. Future analyses should examine whether changes in meaning-making affect changes in mental health. For example, does transitioning from searching but failing to find meaning in loss to no longer searching for meaning provide psychological benefits or harm to the bereaved person? Third, this research models an average growth trajectory across all respondents. Other analytical methods may be used to model multiple, different trajectories of psychological distress across time and examine the relationship between various trajectories of psychological distress and meaning-making membership. Lastly, many of the variables included in our models were not predictive of psychological distress. There still remained a large amount of unexplained variance in the growth parameters that may be explained in future research by the inclusion of other time-invariant or time-varying variables.

Nevertheless, this study improves on research regarding meaning-making and mental health by: (1) incorporating a relatively large sample of widows and widowers from a nationally representative survey of U.S. men and women, (2) using three waves of longitudinal data to examine the effect of meaning-making on trajectories of psychological distress, and (3) using the heterogeneity of the sample to study both recent and long-term widows and widowers. Searching for meaning after spousal loss may be important for some, but it is not important for all. In fact, an unsuccessful search for meaning may result in poorer mental health outcomes. This leads us to conclude that researchers cannot study the effect of searching for meaning in spousal loss without ascertaining whether meaning was found, and vice versa. Future research should more deeply explore the reasons why some individuals search for and/or make sense of the death of their spouse and others do not. Finally, researchers could also explore the extent to which searching for and finding meaning in loss affects physical health outcomes, including self-rated health, morbidity, and mortality.

Footnotes

Appendix

Table B.

Parameter estimates (and standard errors) of the fixed and random effects of the multilevel modeling of psychological distress, Americans’ Changing Lives (1986-1994).

		I	II	III	IV	V
Fixed effects
Initial status	Intercept	15.753***	16.284***	20.645***	21.048***	21.739***
		(0.119)	(0.208)	(1.086)	(1.084)	(1.132)
	Search-sense			—	−0.939**	−1.891**
					(0.347)	(0.630)
	No search			—	−0.952**	−1.864***
					(0.298)	(0.535)
	Age at loss			−0.024*	−0.019^†	−0.018^†
				(0.011)	(0.011)	(0.011)
	Female			−0.082	−0.074	−0.057
				(0.311)	(0.308)	(0.308)
	Black			−0.095	−0.022	−0.013
				(0.266)	(0.266)	(0.266)
	Married			−0.336	−0.301	−0.262
				(0.342)	(0.341)	(0.341)
	Education (years)			−0.173***	−0.169***	−0.168***
				(0.035)	(0.035)	(0.035)
	Income (logged dollars)			−0.133***	−0.140***	−0.140***
				(0.031)	(0.031)	(0.031)
	Religious importance			−0.007	−0.004	−0.014
				(0.172)	(0.171)	(0.171)
	Death expected			0.059	0.108	0.117
				(0.238)	(0.237)	(0.237)
Rate of change	Intercept		−0.003**	−0.004***	−0.004***	−0.007***
			(0.001)	(0.001)	(0.001)	(0.002)
	Search-sense					0.005†
						(0.003)
	No search					0.004*
						(0.002)
Variance components
Level 1	Within-person	7.903***	7.801***	7.763***	7.757***	7.755***
		(0.361)	(0.360)	(0.358)	(0.358)	(0.358)
Level 2	Initial status	6.950***	8.958***	8.277***	7.894***	7.776***
		(0.577)	(1.532)	(1.412)	(1.392)	(1.378)
	Rate of change		0.000†	0.000*	0.000*	0.000†
			(0.000)	(0.000)	(0.000)	(0.000)
	Covariance		−0.009	−0.011*	−0.010†	−0.010†
			(0.006)	(0.006)	(0.006)	(0.006)
Goodness-of-fit statistics
−2 log likelihood		9,200.8	9,186.9	9,104.5	9,093.6	9,089.0
Bayesian Information Criterion		9,220.7	9,226.7	9,197.5	9,199.9	9,208.5

Note. This analysis was derived from data from 764 bereaved respondents, who contributed 1,711 observations. This table presents the parameter estimates and standard errors for the effects of age at death, gender, race, education, income, religiosity, and whether the death was expected in Model 3, 4, and 5.

†

p < .10. *p < .05. **p < .01. ***p < .001.

Notes

References

Antonovsky

Aaron

. 1979. Health, Stress and Coping. San Francisco: Jossey-Bass.

Bennett

Kate M.

1998. “Longitudinal Changes in Mental and Physical Health among Elderly, Recently Widowed Men.” Mortality 3:265-73.

Bennett

Kate M.

2009. “Gender Difference in Bereavement Support for Older Widowed People.” Bereavement Care 28:5-9.

Bennett

Kate M.

Vidal-Hall

Steph

. 2000. “Narratives of Death: A Qualitative Study of Widowhood in Later Life.” Ageing and Society 20:413-28.

Bonanno

George A.

2004. “Have We Underestimated the Human Capacity to Thrive after Extremely Averse Events?” American Psychologist 59:20-8.

Bonanno

George A.

Kaltman

Stacey

. 2001. “The Varieties of Grief Experience.” Clinical Psychology Review 21:705-34.

Bonanno

George A.

Westphal

Maren

Mancini

Anthony D.

2011. “Resilience to Loss and Potential Trauma.” Annual Review of Clinical Psychology 7:511-35.

Bonanno

George A.

Wortman

Camille B.

Lehman

Darrin R.

Tweed

Roger G.

Haring

Michelle

Sonnega

John

Carr

Deborah

Nesse

Randolph M.

2002. “Resilience to Loss and Chronic Grief: A Prospective Study from Preloss to 18-Months Postloss.” Journal of Personality and Social Psychology 83:1150-64.

Bonanno

George A.

Wortman

Camille B.

Nesse

Randolph M.

2004. “Prospective Patterns of Resilience and Maladjustment during Widowhood.” Psychology and Aging 19:260-71.

10.

Caserta

Michael

Lund

Dale A.

2007. “Toward the Development of an Inventory of Daily Widowed Life (IDWL): Guided by the Dual Process Model of Coping with Bereavement.” Death Studies 31:505-35.

11.

Chen

J. H.

Bierhals

A. J.

Prigerson

H. G.

Kasl

S. V.

Mazure

C. M.

Jacobs

1999. “Gender Differences in the Effects of Bereavement-related Psychological Distress in Health Outcomes.” Psychological Medicine 29:367-80.

12.

Chou

Kee-Lee

Chi

Iris

. 2000. “Stressful Events and Depressive Symptoms among Old Women and Men: A Longitudinal Study.” International Journal of Aging and Human Development 51:275-93.

13.

Corr

Charles A.

Nabe

Clyde M.

Corr

Donna M.

2009. Death & Dying, Life & Living. 6th ed. Belmont, CA: Wadsworth.

14.

Currier

Joseph M.

Holland

Jason M.

Neimeyer

Robert A.

2006. “Sense-making, Grief, and the Experience of Violent Loss: Toward a Mediational Model.” Death Studies 30:403-28.

15.

Davis

Christopher G.

Nolen-Hoeksema

Susan

. 2001. “Loss and Meaning: How Do People Make Sense of Loss?” American Behavioral Scientist 44:726-41.

16.

Davis

Christopher G.

Nolen-Hoeksema

Susan

Larson

Judith

. 1998. “Making Sense of Loss and Benefiting from the Experience: Two Construals of Meaning.” Journal of Personality and Social Psychology 75:561-74.

17.

Davis

Christopher G.

Wortman

Camille B.

Lehman

Darrin R.

Silver

Roxane Cohen

. 2000. “Searching for Meaning in Loss: Are Clinical Assumptions Correct?” Death Studies 24:497-540.

18.

Dimond

Margaret

Lund

Dale A.

Caserta

Michael S.

1987. “The Role of Social Support in the First Two Years of Bereavement in an Elderly Sample.” The Gerontologist 27:599-604.

19.

Epstein

Seymour

. 1973. “The Self-concept Revisited: Or a Theory of a Theory.” American Psychologist 28:404-16.

20.

Gallagher

Dolores E.

Breckenridge

James N.

Thompson

Larry W.

Peterson

James A.

1983. “Effects of Bereavement on Indicators of Mental Health in Elderly Widows and Widowers.” Journal of Gerontology 38:565-71.

21.

Harless

Kelly

Zisook

Sidney

. 2003. “Bereavement and Depression in Late Life.” Pp. 109-32 in Depression in Later Life: A Multidisciplinary Psychiatric Approach, edited by Ellison

J. M.

Verma

New York, NY: Marcel Dekker.

22.

Harlow

Siobán D.

Goldberg

Evelyn L.

Comstock

George W.

1991. “A Longitudinal Study of the Prevalence of Depressive Symptomatology in Elderly Widowed and Married Women.” Archives of General Psychiatry 48:1065-68.

23.

Holland

Jason M.

Currier

Joseph M.

Neimeyer

Robert A.

2006. “Meaning Reconstruction in the First Two Years of Bereavement: The Role of Sense-making and Benefit-finding.” OMEGA 53:175-91.

24.

Horwitz

Allen V.

White

Helen Raskin

Howell-White

Sandra

. 1996. “Becoming Married and Mental Health: A Longitudinal Analysis of a Cohort of Young Adults.” Journal of Marriage and the Family 58:895-907.

25.

House

James S.

Lantz

Paula M.

Herd

Pamela

. 2005. “Continuity and Change in the Social Stratification of Aging and Health over the Life Course: Evidence from a Nationally Representative Study from 1986 to 2001/2002 (Americans’ Changing Lives Study).” Journal of Gerontology, Series B: 60(Special Issue 2):S15-26.

26.

Janoff-Bulman

Ronnie

. 1992. Shattered Assumptions: Towards a New Psychology of Trauma. New York: Free Press.

27.

Janoff-Bulman

Ronnie

Frantz

Cynthia McPherson

. 1997. “The Impact of Trauma on Meaning: From Meaningless World to Meaningful Life.” Pp. 91-106 in The Transformation of Meaning in Psychological Therapies: Integrating Theory and Practice, edited by Brewin

C. R.

Hoboken, NJ: John Wiley and Sons, Inc.

28.

Kim

Su Hyun

Kjervik

Diane

Belyea

Michale

Choi

Eun Sook

. 2011. “Personal Strength and Finding Meaning in Conjugally Bereaved Older Adults: A Four-year Prospective Analysis.” Death Studies 35:197-218.

29.

Kohout

Frank J.

Berkman

Lisa F.

Evans

Denis A.

Cornoni-Huntley

Joan

. 1993. “Two Shorter Forms of the CES-D Depression Symptoms Index.” Journal of Aging and Health 5:179-93.

30.

Yunqing

. 2007. “Recovering from Spousal Bereavement in Later Life: Does Volunteer Participation Play a Role?” Journal of Gerontology: Series B 62:S257-66.

31.

Lucas

Richard E.

2005. “Time Does Not Heal All Wounds: A Longitudinal Study of Reaction and Adaptation to Divorce.” Psychological Science 16:945-50.

32.

Lucas

Richard E.

Clark

Andrew E.

Georgellis

Yannis

Diener

. 2003. “Reexamining Adaptation and the Set Point Model of Happiness: Reactions to Changes in Marital Status.” Journal of Personality and Social Psychology 84:527-39.

33.

Lupton

Joseph

Smith

James P.

2003. “Marriage, Assets, and Savings.” Pp. 129-52 in Marriage and the Economy: Theory and Evidence from Industrialized Societies, edited by Grossbard-Shechtman

Cambridge, UK: Cambridge University Press.

34.

Mancini

Anthony D.

Bonanno

George A.

Clark

Andrew E.

2011. “Stepping Off the Hedonic Treadmill: Individual Differences in Response to Major Life Events.” Journal of Individual Differences 32:144-52.

35.

Mark

Nadine F.

Lambert

James David

. 1998. “Marital Status Continuity and Change among Young and Midlife Adults: Longitudinal Effects on Psychological Well-being.” Journal of Family Issues 19:652-86.

36.

Michael

Scott T.

Snyder

C. R.

2005. “Getting Unstuck: The Roles of Hope, Finding Meaning, and Rumination in the Adjustment to Bereavement among College Students.” Death Studies 29:435-58.

37.

Murphy

Shirley A.

Johnson

L. Clark

Lohan

Janet

. 2010. “Finding Meaning in a Child’s Violent Death: A Five-year Prospective Analysis of Parent’s Personal Narratives and Empirical Data.” Death Studies 27:381-404.

38.

Park

Crystal L.

2010. “Making Sense of the Meaning Literature: An Integrative Review of Meaning Making and Its Effects on Adjustment to Stressful Life Events.” Psychological Bulletin 136:257-301.

39.

Parkes

Colin M.

1998. “Bereavement as a Psychosocial Transition: Processes of Adaptation to Change.” Journal of Social Issues 44:53-65.

40.

Parkes

Colin M.

Weiss

Robert S.

1983. Recovery from Bereavement. New York: Basic Books.

41.

Pearlin

Leonard I.

Johnson

Joyce S.

1977. “Marital Status, Life Strains, and Depression.” American Sociological Review 42:704-15.

42.

Reynolds

John R.

Turner

R. Jay

. 2008. “Major Life Events: Their Personal Meaning, Resolution, and Mental Health Significance.” Journal of Health and Social Behavior 49:223-37.

43.

Richardson

Virginia E.

Balaswamy

Shantha

. 2001. “Coping with Bereavement among Elderly Widowers.” OMEGA 43:129-44.

44.

Schwartzberg

Steven S.

Janoff-Bulman

Ronnie

. 1991. “Grief and the Search for Meaning: Exploring the Assumptive Worlds of Bereaved College Students.” Journal of Social and Clinical Psychology 10:270-88.

45.

Shuchter

Steven R.

1986. Dimensions of Grief. San Francisco: Jossey-Bass, Inc.

46.

Simon

Robin W.

1997. “The Meanings Individuals Attach to Role Identities and Their Implications for Mental Health.” Journal of Health and Social Behavior 38:256-74.

47.

Simon

Robin W.

2002. “Revisiting the Relationship among Gender, Marital Status, and Mental Health.” American Journal of Sociology 107:1065-96.

48.

Singer

Judith D.

1998. “Using SAS PROC MIXED to Fit Multilevel Models, Hierarchical Models, and Individual Growth Models.” Journal of Educational and Behavioral Statistics 24:323-55.

49.

Singer

Judith D.

Willet

John B.

2003. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York: Oxford University Press.

50.

Stroebe

Margaret

Schut

Henk

. 2010. “The Dual Process Model of Coping with Bereavement: A Decade on.” OMEGA 61:273-89.

51.

Umberson

Deborah

Wortman

Camille B.

Kessler

Ronald C.

1992. “Widowhood and Depression: Explaining Long-term Gender Differences in Vulnerability.” Journal of Health and Social Behavior 33:10-24.

52.

Updegraff

John A.

Silver

Roxane Cohen

Holman

E. Alison

. 2008. “Searching for and Finding Meaning in Collective Trauma: Results from a National Longitudinal Study of the 9/11 Terrorist Attacks.” Journal of Personality and Social Psychology 95:709-22.

53.

U. S. Census Bureau. 2011. “Population: Marital Status and Living Arrangements.” Retrieved February 4, 2011 (http://www.census.gov/compendia/statab/cats/population/marital_status_and_living_arrangements.html).

54.

Utz

Rebecca L.

2002. “The Economic Consequences of Widowhood: Does Income Loss Impact Psychological Well-being?” Paper presented at the annual meeting of the Gerontological Society of America, November, Boston, Massachusetts.

55.

Walker

Kenneth N.

MacBride

Arlene

Vachon

Mary L. S.

1977. “Social Support Networks and the Crisis of Bereavement.” Social Science & Medicine 11:35-41.

56.

Wheaton

Blair

. 1990. “Life Transitions, Role Histories, and Mental Health.” American Sociological Review 55:209-23.

57.

Wortman

Camille B.

Silver

Roxane C.

1990. “Successful Mastery of Bereavement and Widowhood: A Life Course Perspective.” Pp. 225-64 in Successful Aging: Perspectives from the Social Sciences, edited by Bates

P. B.

Bates

M. M.

New York: Cambridge University Press.

58.

Zisook

Sidney

Paulus

Martin

Shuchter

Stephen R.

Judd

Lewis L.

1997. “The Many Faces of Depression Following Spousal Bereavement.” Journal of Affective Disorders 45:85-95.

59.

Zisook

Sidney

Rush

A. John

Albala

Ari

Alpert

Jonathan

Balasubramani

G. K.

Fava

Maurizio

Husain

Mustafa

Sackem

Harold

Trivedi

Madhukar

Wisniewski

Stephen

. 2004. “Factors that Differentiate Early vs. Late Onset of Major Depression Disorder.” Psychiatry Research 129:127-40.