Marital status has been identified as a risk factor for suicide among older adults, especially men, internationally (e.g. Australia, Australian Bureau of Statistics, 2000; Sweden, Ajdacic-Gross et al., Reference Ajdacic-Gross2008; Italy, Masocco et al., Reference Masocco2010). For example, in Australia, widowered men (43.1/100,000) are three times more likely to complete suicide than married men (14.3/100,000), and widowed women (6.2/100,000) are almost twice as likely to suicide than married women (3.5/100,000; Australian Bureau of Statistics, 2000).

Two perspectives (social support and protection, and loss) have been used to explain the relationship between marital status and mental health. First, it is proposed that marriage serves a protective function by providing social support, facilitating social participation, and increasing self-esteem (Hagedoorn et al., Reference Hagedoorn2006; Holmes and Joseph, Reference Holmes and Joseph2011). A recent meta-analysis on age-related changes in social networks indicated that marriage was associated with larger social networks, and that the loss of a spouse was associated with smaller family and personal networks (Wrzus et al., Reference Wrzus, Hanel, Wagner and Neyer2013). Being widowed is a significant predictor of self-reported loneliness among older adults (Savikko et al., Reference Savikko, Routasalo, Tilvis, Strandberg and Pitkala2005). Moreover, Stroebe et al. (Reference Stroebe, Stroebe, Abakoumkin and Schut1996) found that emotional loneliness caused by the loss of a spouse could not be buffered by support from family and friends.

The second perspective used to explain the relationship between marital status and mental health is loss. The surviving spouse can experience a grief reaction to the loss, as well as to the changing roles that accompany their changed marital status (Hagedoorn et al., Reference Hagedoorn2006; Naef et al., Reference Naef, Ward, Mahrer-Imhof and Grande2013). The death of a spouse is usually associated with a period of bereavement, with research suggesting that widows and widowers are most vulnerable in the first six years after the loss (Savikko et al., Reference Savikko, Routasalo, Tilvis, Strandberg and Pitkala2005). MacDonald and Leary (Reference Macdonald and Leary2005) argue that such separation from significant others results in emotional and social pain. In addition, marital relationships contain clearly defined gendered roles which are lost with the death of a spouse. As such, the surviving spouse must adapt and change to new gendered roles such as domestic and social tasks while still grieving the loss of their partner (Calasanti, Reference Calasanti2004).

In summary, it is evident from research findings that for older adults, marriage can help promote well-being through social support and participation, and can have a buffering effect against mental illness. The loss of a spouse can also necessitate a difficult bereavement period and result in the loss of social networks, shared gender roles, and relational opportunities. As yet however, the underlying psychological dimensions that account for these explanations are unknown.

It is proposed here that a further loss, the loss of the psychological experience of sense of belonging, may explain the relationship between marital status and suicide among older adults. Sense of belonging has been defined as a perception of personal involvement in an environment or social system, which allows the person to feel that he or she is an integral component of that environment or system (Hagerty et al., Reference Hagerty, Lynch-Sauer, Patusky, Bouwsema and Collier1992). Sense of belonging refers to feeling valued and important, and as though one fits in (Hagerty et al., Reference Hagerty, Lynch-Sauer, Patusky, Bouwsema and Collier1992). Low levels of belonging have been associated with suicidal ideation among older adults (McLaren et al., Reference McLaren, Gomez, Bailey and Van Der Horst2007). Marriage can be regarded as one system that facilitates a sense of belonging (Kissane and McLaren, Reference Kissane and McLaren2006). It is possible that a sense of belonging is lost or reduced when one loses their spouse.

The Interpersonal Theory of Suicide (Van Orden et al., Reference Van Orden, Witte, Cukrowicz, Braithwaite, Selby and Joiner2010) provides further reasoning for why sense of belonging is likely to explain the relationship between marital status and suicide. The theory proposes that two interpersonal constructs, thwarted belongingness and perceived burdensomeness, contribute to suicidal desire. Thwarted belongingness occurs when the basic human need to belong is unmet. The theory proposes that thwarted belongingness and perceived burdensomeness need to occur simultaneously to produce suicidal desire; however, the presence of just one is enough for passive suicidal ideation. Christensen et al. (Reference Christensen, Batterham, Soubelet and Mackinnon2013) tested the theory using a community sample of 6,133 Australians aged from 28 to 72 years. For the participants aged 68 to 72 years, thwarted belongingness was an independent predictor of suicidal ideation. Further analysis with all ages combined demonstrated that thwarted belongingness was an independent predictor of suicidal ideation for men but not women. Whether thwarted belongingness is a predictor of suicidal ideation among older men, but not older women, was not tested.

Thwarted belongingness consists of two dimensions: loneliness and absence of reciprocal care (Van Orden et al., Reference Van Orden, Witte, Cukrowicz, Braithwaite, Selby and Joiner2010). Van Orden et al. specially mention marriage and loss of a spouse in their discussions of thwarted belongingness. They note that marriage is negatively associated with loneliness and that loss of a spouse through death is associated with the absence of reciprocal care. It would appear that being widowed is associated with unmet needs to belong, which in turn, is associated with suicidal ideation. In other words, being widowed is associated with suicidal ideation via thwarted belongingness. This proposition is yet to be tested.

It has also been suggested that marriage as a protective factor may be stronger among men than women (Shye et al., Reference Shye, Mullooly, Freeborn and Pope1995; Corcoran and Nagar, Reference Corcoran and Nagar2010), and that widowhood may be more problematic for men than for women. A meta-analysis revealed that the loss of a spouse is detrimental to well-being and that this effect is somewhat greater for men (Stroebe et al., Reference Stroebe, Stroebe and Schut2001). According to House et al. (Reference House, Landis and Umberson1998), men are less practiced in social relationships and therefore find it more difficult than women to seek out social support. Hence, men are less likely to receive the social support needed through the bereavement process and are more likely to be lonely. Similarly, Bennet et al. (Reference Bennett, Hughes and Smith2005) suggest that the bereavement process is more difficult for men as they are less likely to seek help, are less expressive, and more stoic.

As previously noted, marital relationships contain clearly defined gendered roles which are lost with the death of a spouse. Men are required to take on new domestic roles that are undervalued by society and are not in keeping with masculine roles and identity characteristics. Calasanti (Reference Calasanti2004) found evidence of this difficult identity negotiation where many older males referred to themselves as bachelors rather than widowers as bachelor is a more masculine identification. This shift from traditional and socially valued masculine roles into more feminine roles may be a difficult transition for men, and may result in older men feeling less valued and a lessened sense of belonging.

In summary, there is a small body of evidence that suggests that widowhood may be more difficult for men than women. While social support and gender role theory have been offered as plausible explanations for these gender differences, there is still no widely accepted empirical explanation. We propose that thwarted sense of belonging may be the psychological construct that explains and represents these gender differences.

Little research has investigated sense of belonging as a function of marital status among older adults. Hagerty et al. (Reference Hagerty, Williams, Coyne and Early1996) conducted a study with 379 community college students aged from 18 to 72 years (mean = 26 years). The majority of the sample was single (69%), with 22% married, and the remaining 10% being either separated, divorced, or widowed. Results indicated no difference in sense of belonging scores for men or women. The small number of adults in the widowed, separated, and divorced categories may have influenced the results, as might the relative young age of the sample. Vanderhorst and McLaren (Reference Vanderhorst and McLaren2005) reported a relationship between marital status and sense of belonging among a sample of older adults. Being single (combined sample of widowed, single, and divorced, though the large majority were widowed) was associated with lower levels of belonging.

In summary, suicide is a significant problem among older adults, particularly older men. Marital status is a predictor of suicide among older adults, with loss of a spouse being more strongly associated with suicide among older men than woman. Thwarted belongingness has been associated with loss of a spouse, and the relationship between thwarted belongingness and suicide is stronger among older men than women. It is proposed here that a lack of sense of belonging (or thwarted belongingness) is an explanation for the relationship between marital status and suicidal ideation among older adults, that is sense of belonging mediates the relationship between marital status and suicidal ideation among older adults. Given that there is evidence that thwarted belongingness is a significant predictor of suicidal ideation in men but not women, and that marriage has a stronger protective factor in men than women, it can be speculated that this mediation effect would be moderated by gender, with the mediation being more evident in males than females.

The aim of the current study was to test a mediation model in which sense of belonging mediated the relationship between marital status and suicidal ideation, and how this mediation model is moderated by gender. The study involved cross-sectional data and involved older adults. Although it has been recommended that tests of mediation utilize longitudinal data (Maxwell and Cole, Reference Maxwell and Cole2007), we used cross-sectional data. Relative to longitudinal studies, cross-sectional studies require less funds and resources, and can be completed in far less time. Our primary reason to proceed with cross-sectional data (also used extensively in most previous studies of mediation) was to obtain exploratory data that could potentially provide valuable insights into the usefulness of conducting longitudinal studies in the future. Additionally, although the use of cross-sectional data will not enable causal inferences to be made, the results would have important implications for intervention. For instances, if such data show that sense of belonging mediates the relationship between marital status and suicidal ideation, an appropriate intervention could be to directly facilitate higher sense of belonging among individuals. Thus, despite its limitations in terms of explaining causality, tests of mediation utilizing cross-sectional data could provide valuable insights for intervention.

It was hypothesized that sense of belonging would mediate the marital status–suicidal ideation relationship, such that being widowed would be associated with lower levels of sense of belonging, and lower levels of sense of belonging would, in turn, be associated with higher levels of suicidal ideation. In addition, it was expected that this mediation relationship will be moderated by gender, such that the relationship between marital status and sense of belonging, and between sense of belonging and suicidal ideation, would be stronger for older men than older women.

Method

Participants

A convenience sample of 676 Australian adults aged between 65 and 98 years (M age = 75.13, SD = 8.40) participated in the current study. The sample consisted of 286 men (65–94 years, M age = 75.33 years, SD = 8.10 years) and 390 women (65–98 years, M age = 77.98 years, SD = 8.62 years). As shown in Table 1, the majority of the men were married, lived in their own home, and had completed secondary school or equivalent (trade certificate). Just over half of the women were married, most lived in their own home, and had completed secondary school or equivalent. Participants were recruited at a range of venues, including social and sporting clubs, community groups, and assisted-living facilities for older adults.

Table 1.

Relationship status and highest level of education for participants

Materials

A participant information letter described the research and invited individuals to participate in a study about “social relationships and mental health among older adults.” The statement provided the details of resources available to participants who may be experiencing distress, including a free 24-hour telephone counseling service and the contact information of the principal researcher.

The demographics questionnaire consisted of questions regarding the participants’ gender, age, highest education level attained, and relationship status.

The psychological subscale of the Sense of Belonging Instrument (Hagerty and Patusky, Reference Hagerty and Patusky1995) was used to measure sense of belonging. The 18 items assess the extent to which an individual feels valued and important, and the extent to which they feel they fit in their environment (e.g. “I often wonder if there is any place on earth where I really fit in” and “I feel like a square peg trying to fit into a round hole”). Participants responded to each statement by using a 4-point scale, ranging from 1 (strongly disagree) to 4 (strongly agree). The total score range from 18 to 72, with higher scores indicating a greater sense of belonging (Hagerty and Patusky, Reference Hagerty and Patusky1995). Research has indicated that the SOBI-P is a valid and reliable measure with a test–retest reliability correlation of 0.84 over an 8-week period and an α coefficient of 0.93 (Hagerty and Patusky, Reference Hagerty and Patusky1995). Cronbach's α values for the current samples were very high (α = 0.94 for men and women).

The suicide subscale of the General Health Questionnaire (Goldberg and Hillier, Reference Goldberg and Hillier1979) consists of four items regarding suicidal thoughts over the past few weeks. The participants answered two items, “felt that life isn't worth living” and “found yourself wishing you were away from it all,” using the options 0 = not at all, 1 = no more than usual, 2 = rather more than usual, and 3 = much more than usual, and two items, “thought of the possibility that you might do away with yourself” and “found the idea of taking your own life kept coming into your mind,” using the options, 0 = definitely not, 1 = I do not think so, 2 = has crossed my mind, and 3 = definitely has. Higher scores reflected increasing intensity of suicidal ideation. Internal consistency was acceptable for the current samples (men, α = 0.76; women, α = 0.70).

Procedure

The University's Human Research Ethics Committee approved the study protocol. Participants were recruited in person. At various venues (e.g. assisted living facilities, social and community clubs), the principal researcher gave a brief presentation about the study. Older adults interested in participating in the study approached the researcher, and those who wished to take part were given a questionnaire package. The questionnaires were completed by participants at the time and returned to the researcher. A total of 800 questionnaires were distributed, with 676 usable questionnaires being returned, indicating a response rate of 85%.

Statistical analysis

Simple mediation and moderated mediation analyses were conducted using PROCESS for SPSS (Hayes, Reference Hayes2013). The models for simple mediation and moderated mediation analyses were models 4 and 58 respectively in PROCESS. These models are shown in Figure 1. As shown, for the simple mediation, there was direct path from marital status (coded: 1 = married, 2 = widowed) to suicidal ideation, and also an indirect path through sense of belonging. For the moderated mediation model, gender (coded: 1 = male, 2 = female) was specified to influence the path from marital status to sense of belonging, and also sense of belonging to suicidal ideation. As age (Bertolote and Fleischann, Reference Bertolote and Fleischmann2002), place of residence (own home or assisted living facility; Ron, Reference Ron2004) and highest level of education (Wiktorsson et al., Reference Wiktorsson, Runeson, Skoog, Östling and Waern2010) have been related to suicide among older adults, age, highest education attainted (coded: 1 = primary school, 2 = secondary school, and 3 = university), and residence (coded: 1 = assisted-living facility, 2 = own home) were treated as covariates. The number of bootstrap samples for bias-corrected bootstrap confidence intervals (CI) was 10,000, and a bootstrapped 95% CI was used to infer significance. Significance is supported if the CI does not include zero (Hayes, Reference Hayes2013).

Figure 1.

Conceptual path model of the simple mediation model and moderated mediation model tested in the study.

Results

Descriptive statistics

Descriptive statistics can be seen in Table 2. For men, widowers scored significantly lower on sense of belonging and significantly higher on suicidal ideation than married men. For women, widows scored significantly higher on suicidal ideation than married women.

Table 2.

Mean scores and SDs for married and widowed men and women

**p < 0.01; ***p < 0.001.

Simple mediation

As we used cross-sectional data, we first tested if an alternate model with suicidal ideation mediating the relationship between marital status and sense of belonging was tenable. In this analysis, the β (95% CI) values of the direct effect and indirect effect of marital status on sense of belonging were 1.32 (−2.70, −0.05) and 0.72 (−1.43, −0.01), respectively. As will be noticed, the 95% CI for the indirect effects included zero, thereby indicating no support for this mediation model. Thus, the alternative mediation model was not supported.

Next, we tested our hypothesized mediation model where sense of belonging mediated the relationship between marital status and suicidal ideation. The β (95% CI) values of the direct effect and indirect effect of marital status on suicidal ideation were 0.19 (−0.06, −0.43) and 0.22 (0.11, −0.35), respectively. The 95% CI for the direct effect included zero, whereas the indirect effect did not include zero. The normal theory test for indirect effects (Sobel test) was significant (β = 0.22, z = 3.85, p < 0.001). Taken together, the findings are supportive of mediation by sense of belonging on the relationship between marital status and suicidal ideation. Table 3 shows the results of the simple mediation by sense of belonging on the relationship between marital status and suicidal ideation. As shown, the regression of sense of belonging on marital status was significant and negative. As married was coded 1 and widowed was coded 2, the findings indicated being widowed was associated with lower levels of sense of belonging. Consistent with support for mediation, the table also shows that when suicidal ideation was regressed on marital status, sense of belonging, and the covariates concurrently, the coefficient for suicidal ideation was not significant, whereas the regression of sense of belonging was significant and negative. The latter finding indicates that higher sense of belonging was associated with lower suicidal ideation.

Table 3.

Results for the simple mediation by sense of belonging on the marital status–sense of belonging relationship

***p < 0.001.

SE = standard error.

Moderated mediation

Table 4 shows the results of the moderated mediation analysis with gender as a moderator in the relationships between marital status and sense of belonging, and sense of belonging and suicidal ideation. As shown, the relation between sense of belonging and suicidal ideation was not moderated by gender (as indicated by the non-significant gender × sense of belonging interaction). However, the relation between marital status and sense of belonging was moderated by gender (as indicated by significant gender × marital status interaction). Thus, there was a moderating effect by gender on the indirect relationship between marital status and suicidal ideation. The relation between marital status and sense of belonging was significant for men as its bootstrapped 95% CI did not include zero (0.20, −0.67). In contrast, this relation was not significant for women as its bootstrapped 95% CI included zero (−0.03, −0.25).

Table 4.

Results for the moderated mediation analysis with gender as the moderator

*p < 0.05; ***p < 0.001.

SE = standard error.