Introduction
Life expectancy at birth has increased substantially for Irish males and females in recent decades, and the country's older people are living longer. While in general Irish older people have comparatively high levels of social contact, more and more live alone or with an elderly partner. The considerable international research on loneliness and social isolation indicates that they are part of the experience of old age and are associated with physical and psychological effects such as depression, insomnia, anxiety, isolation, loss of appetite and distress (DiTommaso and Spinner Reference DiTommaso and Spinner1997). There is, however, limited evidence of the experience of loneliness among older Irish people, and limited understanding in any country of the experience amongst older people of social loneliness and emotional loneliness. The multi-dimensionality of loneliness, as characterised by Weiss's (Reference Weiss1973) typology of social and emotional loneliness, was adopted as the theoretical basis of the study reported in this paper. It measured the experience of social and emotional loneliness among older Irish people, and analysed the risk factors for both types.Footnote 1
Social and emotional loneliness
The sense of loneliness comprises a complex set of feelings that encompass reactions to the loss of intimate and/or social needs (Ernst and Cacioppo Reference Ernst and Cacioppo1999). In this way, it can be construed as the emotional response to the discrepancy between desired and available inter-personal relationships (Walton et al. Reference Walton, Schulz, Beck and Walls1991; Kileen Reference Kileen1998). Weiss, the seminal writer on social and emotional loneliness, cautions that it is misleading to define loneliness as ‘caused by the condition of being alone’; he held loneliness to be a condition caused by ‘being without some definite needed relationship or set of relationships’ (Weiss Reference Weiss1973: 17). In this way, Weiss defined loneliness with reference to its cause and emphasised the role of close and intimate relations. The experience of loneliness appears to be a response to the absence of ‘some particular relational provision’, either the presence of an intimate attachment, such as a meaningful friendship, or a link to a coherent community (Weiss Reference Weiss1973: 17). On this basis, Weiss distinguished two dimensions of loneliness: the ‘loneliness of emotional isolation’ (emotional loneliness), a condition arising out of the loss or absence of a close emotional attachment, and the ‘loneliness of social isolation’ (social loneliness), a condition arising out of the absence of ‘an engaging social network’ (1973: 18–9).
Emotional loneliness is associated with the loss of a spouse, children or a confidant. It is experienced as a sense of pervasive apprehensiveness, involving poor concentration, persistent vigilance to threat or tension, and an inability to organise one's perceptual and emotional energies towards finding a remedy for the loneliness (Weiss Reference Weiss1973). Emotional loneliness also involves a sense of the absence of others in one's environment, and feelings of emptiness. The role of a single emotional attachment (a close friend or a romantic partner) can act as a buffer in staving off loneliness among those at risk (Ernst and Cacioppo Reference Ernst and Cacioppo1999; Victor et al. Reference Victor, Scambler, Bowling and Bond2005). Social loneliness is associated with disrupted linkages to the person's supportive social network. It is characterised by a lack of integration and may be associated with several contributory factors, including instability of residence, infrequent contact with friends, children and siblings, lack of participation in social groups, and a decline in health (Dugan and Kivett Reference Dugan and Kivett1994). Social isolation that marks or leads to the loss of a supportive network may expose the older person to feelings of vulnerability, marginality, tension and boredom (Weiss Reference Weiss, Hojat and Crandell1989). The loneliness of social isolation can be remedied by new social contacts and networks. Nevertheless, while it may be assumed that the way to overcome loneliness arising out of social isolation is to end social isolation, this may not be effective, since some individuals well connected with others still experience loneliness.
The experience and the prevalence of loneliness among older people
Loneliness may be experienced by everyone at some time in their lives and, as such, is a temporary state that dissipates as a person's circumstances change (Lauder, Sharkey and Mummery Reference Lauder, Sharkey and Mummery2004). For some people, however, loneliness is persistent (Ernst and Cacioppo Reference Ernst and Cacioppo1999). The experience of loneliness is found at all ages, from childhood to advanced old age. While it may manifest the ‘lifestyle dissatisfaction’ that is prevalent in industrialised countries (Lauder, Sharkey and Mummery Reference Lauder, Sharkey and Mummery2004), loneliness is evident across cultures and societies, from agrarian to post-industrial. Nevertheless, culture may moderate the extent to which people experience the feeling of being lonely (Rokach et al. Reference Rokach, Orzeck, Moya and Expósito2002).
A meta-analysis of research findings on loneliness found that from five to 15 per cent of older adults experience frequent loneliness (Pinquart and Sorensen Reference Pinquart and Sorensen2001). Most international research indicates that the majority of older people are not lonely, with the estimated prevalence ranging from five to 16 per cent. In one English study, older people reported lower levels of loneliness than younger people (Victor, Bowling and Bond Reference Victor, Bowling and Bond2002). Research in the United States, however, seems to indicate that loneliness is quite prevalent among older people in that country, and may be a significant negative factor in their lives. Dugan and Kivett (Reference Dugan and Kivett1994) reported that 21 per cent of rural older Americans experienced ‘much loneliness’; Johnson, Waldo and Johnson (Reference Johnson, Waldo and Johnson1993) reported that 62 per cent of older people in America experience loneliness; while Ryan and Patterson (Reference Ryan and Patterson1987) found that among older Americans' the possibility of being lonely was ranked as a ‘major fear’. The American evidence is at variance, however, with much other international research on the prevalence of loneliness. In Taiwan, just 3.5 per cent of a sample of older rural dwellers reported experiencing a ‘high level’ of loneliness (Wang, Snyder and Kaas Reference Wang, Snyder and Kaas2001). Forbes (Reference Forbes1996) reported that only one-in-10 people experienced serious loneliness in Britain. From a qualitative study of a large sample of older people in Britain, Victor et al. (Reference Victor, Scambler, Bowling and Bond2005) found that only a small minority (7%) of older people reported that they were ‘often’ or ‘always’ lonely. While the percentage of older people reporting loneliness is relatively low, it is possible that the prevalence of loneliness has not decreased in the past 60 years (Victor, Bowling and Bond Reference Victor, Bowling and Bond2002). The prevalence figures for Ireland are generally consistent with those of international studies, although the estimates vary considerably. A major report on health and social services for older people identified that the great majority (90%) of older Irish people were not ‘bothered by loneliness’ (Garavan, Winder and McGee Reference Garavan, Winder and McGee2001). Other studies of older people have identified levels of loneliness ranging from seven per cent ‘feeling persistently lonely’ (Power Reference Power1980), to 14 per cent reporting ‘feelings of loneliness’ (Commission of the European Communities 1993).
Predictors of loneliness in older people
International studies indicate that loneliness is associated with several socio-demographic variables, such as age, social and economic circumstances, living arrangements, social networks, family function, and the quality of social relationships (de Jong-Gierveld Reference de Jong Gierveld1987; Mullins, Elston and Gutkowski Reference Mullins, Elston and Gutkowski1996; Fees, Martin and Poon Reference Fees, Martin and Poon1999; Victor et al. Reference Victor, Scambler, Bowling and Bond2005). Various characteristics of older people, such as their physical health, cognitive integrity, self-esteem and pre-morbid personality are also associated. Other associated factors include social norms and values, the expectations of support in certain types of relationships, and the individual's evaluation of their available social relationships (Lauder, Sharkey and Mummery Reference Lauder, Sharkey and Mummery2004). From an examination of factors associated with loneliness, Victor, Bowling and Bond (Reference Victor, Bowling and Bond2002) referred to ‘risk factors’ for loneliness, and identified broad inter-related categories, including those associated with socio-economic attributes, health resources, material resources, social resources, and social networks. Some of these factors, such as not being married and spending much time alone, appeared to increase older people's vulnerability to loneliness, while others, such as having an educational qualification, appeared to have a protective effect. A large study among older people in Britain reported loneliness to be most likely in specific groups of older people, namely the very old, women, the non-married, those living alone, those lacking material resources (home, car), those lacking an educational qualification, and the physically or mentally frail (Victor et al. Reference Victor, Bowling and Bond2002, Reference Victor, Scambler, Bowling and Bond2005).
A number of Irish studies have contributed to the body of knowledge about loneliness and its correlates, including its association with social isolation. Whelan and Vaughan (Reference Whelan and Vaughan1982) reported that the experience of loneliness was related to the quality and not the quantity of social contacts. Of the 10 per cent of older Irish people in a recent study who reported being ‘bothered by loneliness’, four-in-ten spent an average of 10 to 14 hours alone each day (Garavan, Winder and McGee Reference Garavan, Winder and McGee2001). Horkan and Woods (Reference Horkan and Woods1986) found that those who identified living alone as a major disadvantage in their lives experienced a degree of loneliness, the major associated factors being absence of company and the desolation following bereavement. Fahey and Murray (Reference Fahey and Murray1994) reported that being widowed and living alone had the strongest association with the feeling of loneliness, far exceeding that of being never married. They also concluded that the quantity of social contact may have little bearing on the experience of loneliness, and that a single strong bond may be more important than multiple weak social relationships. From an analysis of data on social contacts and the experience of feeling lonely, Fahey and Murray (Reference Fahey and Murray1994) found that the sense of loneliness was unrelated to the levels of contact with family members.
Loneliness may have implications for the health of those who experience it (Forbes Reference Forbes1996). It is an important predictor of welbeing among older people, for several researchers have demonstrated a relationship between loneliness and health status (Holmén and Furukawa Reference Holmén and Furukawa2002). Health status can be both a predictor and a consequence of loneliness, and loneliness may act as a precipitant of declines in mental and physical health. For example, physical disabilities and mobility problems are associated with increased loneliness, and loneliness may increase depression, alter sleep patterns, and disturb appetite (Tijhuis et al. Reference Tijhuis, de Jong-Gierveld, Feskens and Kromhout1999). To the extent that these effects adversely impact on an individual's ability to interact with others, reduced physical or mental health may exacerbate social isolation and loneliness (Kileen Reference Kileen1998). By the same token, perceived loneliness can mediate perceived physical health. A study of a sample of older Americans in the southern States demonstrated that feelings of loneliness decreased older people's evaluation of their physical wellbeing; in other words, loneliness mediated self-assessed health (Fees, Martin and Poon Reference Fees, Martin and Poon1999).
The literature suggests that loneliness is a complex, multi-dimensional phenomenon with two principal components, social and emotional loneliness. Loneliness is prevalent among the entire population, and international empirical research indicates that although it is a widely held stereotype that old age is associated with social isolation and loneliness, only five to 15 per cent of older adults report frequent loneliness (Pinquart and Sorensen Reference Pinquart and Sorensen2001). The literature indicates that loneliness has numerous correlates, some of which are predictive and others consequential. No model of the predictors of loneliness among older adults in Ireland has previously been published. This study addresses the gap.
Method
Study design
The study design was a cross-sectional national telephone survey of loneliness in older people conducted in 2004. There is no sampling frame in Ireland that specifically identifies older people; therefore respondents were randomly contacted by landline telephone using a technique known as Random Digital Dialling. This approach was used to ensure accurate sample coverage and to reduce sampling error. The use of telephone surveys has been found to be effective in reassuring respondents about a survey as well as clarifying points that an individual may not understand (Dillman Reference Dillman2000). Furthermore, the availability since 1978 of free landline telephone rentals for older people in Ireland means that phone possession rates are very high. Following in-depth training, the telephone calls were undertaken by experienced female interviewers.Footnote 3 A check-back number was given to respondents to verify the credentials of the interviewer prior to the interview. Participants were also provided with the Senior Helpline telephone number (O'Shea Reference O'Shea2006).Footnote 4
Measures
The predictors of loneliness were identified from previous studies of loneliness in old age and informed by our broad knowledge of social conditions in Ireland. The predictor variables included socio-demographic characteristics, health characteristics and social network characteristics. The socio-demographic characteristics were age, gender, marital status (married, never married, widowed, separated/divorced), area of residence (urban/rural), education level (no formal education to university level), and self-assessed net income per week (in Euros). The health variables included the respondents' subjective self-reports of their overall health, eyesight and hearing on five-point scales ranging from (1) poor to (5) excellent. To measure contact with family, friends and neighbours, proximity to children and relatives and involvement in social and religious groups, social network measures were used from the Network Assessment Instrument (Wenger Reference Wenger1994).
The instruments used to measure loneliness conceive the state as either indivisible and uni-dimensional (Allen and Oshagan Reference Allen and Oshagan1995; Russell Reference Russell1996; Cramer and Barry Reference Cramer and Barry1999), or as multi-dimensional with various components or types (Weiss Reference Weiss1973; DiTommaso and Spinner Reference DiTommaso and Spinner1993, Reference DiTommaso and Spinner1997). The published multi-dimensional typologies generally conceptualise loneliness as either emotional or social (Weiss Reference Weiss1973). The instruments most commonly used by researchers include the University of California Los Angeles (UCLA) Loneliness Scale (Russell Reference Russell1996), the Social and Emotional Loneliness Scale for Adults (SELSA) (DiTommaso and Spinner Reference DiTommaso and Spinner1993Reference DiTommaso and Spinner1997), the de Jong-Gierveld Loneliness Scale (de Jong-Gierveld Reference de Jong Gierveld1987), the Differential Loneliness Scale (Schmidt and Sermat Reference Schmidt and Sermat1983), and the Loneliness Rating Scale (Scalise, Ginter and Gerstein Reference Scalise, Ginter and Gerstein1984). In developing a questionnaire for this study, a number of aspects of the named instruments were taken into consideration, including the validity of the instrument for older people, the comprehensibility of the items, and the time required for completion. Taking these factors into consideration, the most suitable instrument was identified as the short form of the Social and Emotional Loneliness Scale for Adults (SELSA-S) (DiTommaso, Brannen and Best Reference DiTommaso, Brannen and Best2004). SELSA-S was derived from the original 37-item multi-dimensional instrument that measures the constructs of both social and emotional loneliness. It further divides emotional loneliness into subscales that measure ‘family loneliness’ and ‘romantic loneliness’ (DiTommaso and Spinner Reference DiTommaso and Spinner1993, Reference DiTommaso and Spinner1997; Cramer, Ofusu and Barry Reference Cramer and Barry2000).
Following psychometric testing and comparison of the original SELSA with other loneliness measures, including the UCLA Loneliness Scale, Cramer and Barry (Reference Cramer and Barry1999) concluded that it was superior to other loneliness measures because of the multi-dimensional ratings. One major disadvantage, however, is the instrument's length, with the 37 items and socio-demographic questions. To counter this problem, DiTommaso, Brannen and Best (Reference DiTommaso, Brannen and Best2004) subsequently developed and tested the short-version with 15 items, comprising three five-item subscales on social loneliness, family loneliness and romantic loneliness. Each item is rated on a seven-point scale from ‘strongly disagree’ to ‘strongly agree’. The advantages of SELSA-S over other loneliness measures has been summarised as its ability ‘to pinpoint the specific nature … of loneliness’ (DiTommaso, Brannen and Best Reference DiTommaso, Brannen and Best2004: 101). It has been suggested that the UCLA loneliness scale, by comparison, principally measures social loneliness: ‘[it] might represent a useful global index of loneliness, [but] it heavily emphasizes social loneliness and virtually ignores family loneliness’ (Cramer and Barry Reference Cramer and Barry1999: 500). Furthermore, when DiTommaso, Brannen and Best (Reference DiTommaso, Brannen and Best2004) used the SELSA-S, they found that romantic, family and social loneliness were independent constructs and, more specifically, there were indications that romantic loneliness was predicted by the lack of an intimate relationship but not by predictors of family or social loneliness, establishing the discriminant validity of this particular subscale. Four items exemplify those in the SELSA-S:
• I really belong in my family (family loneliness)
• My family really cares about me (family loneliness)
• I have a partner who gives me the support and encouragement I need (romantic loneliness)
• I have friends that I can turn to for information (social loneliness)
A psychometric study of SELSA-S found that, when used with university students, spouses of military personnel and psychiatric patients, the internal reliability of the three scales was high (Cronbach's alpha ranged from 0.87 to 0.90) (DiTommaso, Brannen and Best Reference DiTommaso, Brannen and Best2004). The validity of the SELSA-S was ascertained through exploratory and confirmatory factor analysis by the same study.
Analysis
Descriptive statistics were used to profile the sample and the reported levels of social, family and romantic loneliness. Differences in levels of loneliness between males and females were ascertained by the independent sample t-test. Predictors of social and emotional (family and romantic) loneliness were ascertained through three hierarchical multiple regression models. Predictors were entered into the regression models if they correlated with social or emotional loneliness. The variables were entered in the social and family models in the sequence: socio-demographic, health and social/family attributes. The variables in the romantic model were entered in the sequence: socio-demographic, health and marital attributes. All categorical variables (gender, marital status, area of residence, access to transport, attendance at church/religious services, involvement in social groups) were converted to dummy variables for the regressions.
Results
Response rate and sample characteristics
A total of 9,711 calls were made to randomly-dialled landline numbers, and 874 people aged 65 or more years were contacted, of whom 191 declined to take part, so there were 683 completed interviews giving a response rate of 78 per cent. The majority of the respondents were female. The age of the respondents ranged from 65 to 99 years (mean 73.5, standard deviation 7.1) (see Table 1). The profile of the sample was broadly similar to that of older people in Ireland in 2002, but there were significant under-representations of women aged 75–79 years (16.8% versus 21.2%) and aged 85 or more years (8.0% versus 11.1%). Single males were also significantly under-represented (12.9% versus 20.3%) (Central Statistics Office 2002). The majority of older people attended church services regularly or occasionally (90%) and one-half were involved in community social groups. Less than one-half (43.8%) of the respondents reported access to a car, and just over one-in-10 reported no access to any mode of transport (public or private). Sixty per cent of the sample resided in urban areas.
Table 1. Gender, age structure and marital status of the study sample and of the population in Ireland aged 65 or more years
Sources: Authors' survey and Central Statistics Office 2002.
Social and emotional loneliness
The levels of loneliness reported for each of the three subscales ranged from ‘1’ (absence of loneliness) to ‘7’ (severe loneliness). The overall loneliness scores were low, with the means ranging from 1.9 (family loneliness) to 3.2 (romantic loneliness), with social loneliness scoring 2.1 (Table 2). Men and women had slightly different loneliness scores in relation to social loneliness and family loneliness, but these differences were small and not statistically significant (p>0.05). However, the women were significantly more likely to be romantically lonely than male respondents (t=4.9, degrees of freedom=654, p<0.001).
Table 2. Average loneliness scores by gender
Notes: SD standard deviation. Each type of loneliness was self-assessed on a scale from ‘1’ (absent) to ‘7’ (severe).
Predictors of social loneliness
Hierarchical multiple regression was conducted to determine the linear combination of variables that best predicted the development of social loneliness. Predictor variables were entered in three steps. At Step 1, age, gender, net income per week, residence, marital status (married, not married, widowed, separated/divorced) and level of education were entered. At Step 2, the health variables (overall health, eyesight and hearing) were added, and at Step 3, the social contact variables were added (contact with friends and neighbours, access to transport, attendance at church and social groups, and provision of care to a dependent spouse or relative). It was found at Step 1 that age (older) and income per week (lower) had a significant influence on social loneliness, and explained 10 per cent of the variance. At Step 2, poor health and poor hearing in conjunction with increasing age and low income had a significant impact on loneliness, and accounted for 13 per cent of the variance. At the final step, age (older), living in a rural setting, poorer health in conjunction with lack of contact with friends had a significant impact on social loneliness. Income and poor hearing as predictors disappeared when the social variables were added. The largest impact on the development of social loneliness was the lack of contact that older people had with friends. Overall, 53 per cent of the variance was explained (Table 3), and social loneliness was associated with greater age, poorer health, living in a rural setting and lack of contact with friends. At the first two steps, relatively low income was a predictor of loneliness.
Table 3. Hierarchical multiple regression models of social loneliness
Notes: R 2=0.10 for Step 1; ΔR 2=0.13 for Step 2; ΔR 2=0.53 for Step 3.
Significance levels: * p<0.05, ** p<0.01, *** p<0.001.
Predictors of family loneliness
A second series of hierarchical multiple regressions were run to determine the linear combination of variables that best predicted the development of family loneliness. Step 1 included the socio-demographic variables, Step 2 added the health variables, and Step 3 added the social and family contact variables that correlated with family loneliness. The initial model established that place of residence (living in an rural setting), gender (being male), income per week (lower income) and marital status (widowed) explained nine per cent of the variance. The health variables added to the second model had no significant impact on family loneliness, but at Step 3 residence, gender, income (lower), being widowed, having no access to transport, infrequent contact with children and relatives, and caring for a spouse or relative at home were significant predictors. Income as a predictor disappeared when the social variables were added. The greatest predictor of family loneliness was limited contact with children and relatives. The final model explained 23 per cent of the variance (Table 4).
Table 4. Hierarchical multiple regression of the factors predicting family loneliness
Notes: R 2=0.09 for Step 1; ΔR 2=0.10 for Step 2; ΔR 2=0.23 for Step 3.
Significance levels: * p<0.05, ** p<0.01, *** p<0.001.
Predictors of romantic loneliness
The final series of models examined the impact of marital status on romantic loneliness. As before, the initial model included the socio-demographic variables (age, net income per week, and level of education), Step 2 added the health variables (overall self-rated health), and Step 3 added the marital status binary variables (married, never married, widowed, separated or divorced). The initial model indicated that age (older), low income and being female explained 28 per cent of the total variance. At Step 2, poor health in conjunction with increasing age, being female and lower income had a significant impact on loneliness, and the model accounted for 29 per cent of the variance. At the final step, greater age, low income and marital status had a significant impact on romantic loneliness. Older people who had never married, or who were divorced, separated or widowed were significantly more romantically lonely than those who were married. Gender as a predictor was insignificant when the marital status variables were added. The strongest association with romantic loneliness was with being widowed, followed by never having married and finally being separated or divorced. The final model explained 56 per cent of the variance (Table 5). Overall, romantic loneliness most associated with greater age, lower income and being widowed, single or separated/divorced.
Table 5. Hierarchical multiple regression of the factors predicting romantic loneliness
Notes: R 2=0.28 for Step 1; ΔR 2=0.29 for Step 2; ΔR 2=0.56 for Step 3.
Significance levels: * p<0.05, ** p<0.01, *** p<0.001.
Discussion
Social and emotional loneliness
The overall level of loneliness reported by the study sample was low, especially of its social and family components. Most respondents had large social networks and frequent interactions with relatives, friends or neighbours. The majority attended church regularly and one-half attended community clubs or social groups either regularly or occasionally. The principal finding, that the majority of older people are not lonely, replicated the evidence from other national and international studies (Forbes Reference Forbes1996; Victor et al. Reference Victor, Scambler, Bowling and Bond2005). The Living in Ireland Survey demonstrated that older Irish people are unlikely to live alone and that both older Irish men and women have high levels of social contact (Layte, Fahey and Whelan Reference Layte, Fahey and Whelan1999). However, loneliness was experienced by some older people and several factors were identified that were associated with the experience of social and emotional loneliness.
Four independent factors were associated with the experience of social loneliness: greater age, poor overall health, living in a rural area and, in particular, lack of contact with friends. Social loneliness is mediated and alleviated by the quality of the person's social network, with friends being a more important protection than contact with neighbours, number of children, or being involved with the church (Pinquart and Sorensen Reference Pinquart and Sorensen2001). Living in a rural area predicts social loneliness in Ireland. It may be that patterns of social interaction vary by location and that older people living in rural settings have fewer daily interactions than those living in urban areas – both the low density of population and scarce public transport could play a role – and that such living circumstances exacerbate social isolation and in turn social loneliness (Layte, Fahey and Whelan Reference Layte, Fahey and Whelan1999). Participation in social activities has been found to result in significantly larger social networks and reduced feelings of loneliness among older people (Moorer and Suurmeijer Reference Moorer and Suurmeijer2001).
The factors that were associated with family loneliness were more complex and included being male, having a lower income, living in a rural area, being widowed, having limited contact with children and relatives, having to care for a spouse or relative at home and having limited access to transport. The last was a particularly strong predictor of family loneliness, especially for older people living in rural communities. From previous studies, the factors that facilitate family and social contacts include good transport (National Council for the Elderly 1996), while conversely the absence of transport can reduce the older person's opportunities for interaction, which can in turn contribute to loneliness (Holmén and Furukawa Reference Holmén and Furukawa2002). For the majority (82%) of the respondents to the authors' survey, the nearest relative lived within five miles, but the urban dwellers were more likely than rural residents to have a sibling living between one and five miles away. The growing urbanisation of Ireland increasingly leaves the older residents of rural areas at risk of social loneliness as a result of the out-migration of their younger kin (NUI Maynooth/University College Dublin/Teagasc 2005).
The highest reported prevalence of loneliness was for romantic loneliness. The predictors among the study sample included greater age, lower income and particularly being widowed, single or separated/divorced. Of those who were widowed, slightly over 80 per cent were women, and the mean reported duration of widowhood was 15.3 years, indicating that many older people live for an extended period after spouse bereavement (Department of Health and Children 2001). Having a spouse provides an important source of both social and emotional support and the absence or loss of a partner has long been recognised as associated with increased loneliness among the very old (Tijhuis et al. Reference Tijhuis, de Jong-Gierveld, Feskens and Kromhout1999; Victor, Bowling and Bond Reference Victor, Bowling and Bond2002). The finding that romantic loneliness was the most prevalent type of loneliness is consistent with Weiss's (Reference Weiss1973) contention that loneliness can result from the lack of an intimate relationship.
Age as a predictor of loneliness
Greater age has been shown to be a predictor of both social and romantic loneliness among older people. With increasing age, there is of course an increasing risk of losing partners and friends, which may reduce the number of meaningful relationships and thus increase the prevalence of loneliness (Walton et al. Reference Walton, Schulz, Beck and Walls1991; Fahey and Murray Reference Fahey and Murray1994; O'Leary, O'Cinnéide and Staunton Reference O'Leary, O'Cinnéide and Staunton2004). Increasing age was not found, however, to predict family loneliness. Family support has been identified as protective against loneliness (Victor et al. Reference Victor, Scambler, Bowling and Bond2005). In general, family networks remain strong in Ireland, and provide much of the emotional contact that people experience. Over 60 per cent of the respondents in the telephone survey had three or more children, and over 80 per cent had at least one child alive.
The finding that social and romantic loneliness were related to greater age is consistent with previous reports that have found loneliness to be relatively frequent among the very old (Holmén et al. Reference Holmén, Ericsson, Andersson and Winblad1992; Holmén, Ericsson and Winblad Reference Holmén, Ericsson and Winblad2000; Dugan and Kivett Reference Dugan and Kivett1994; Tijhuis et al. Reference Tijhuis, de Jong-Gierveld, Feskens and Kromhout1999). By contrast, the survey of loneliness among older people in Great Britain by Victor et al. (Reference Victor, Scambler, Bowling and Bond2005) found that the ‘oldest old’ (aged 85 or more years) had a comparatively low risk of loneliness. The results from the present study suggest that the discrepancy may lie in the stronger age effect for social and romantic loneliness than for family support and contacts, and indicates that even those older people who are close to and supported by their family can still experience forms of loneliness. Thus, changes in loneliness over time are not a linear function of time (or chronological age), but rather related to life events and transitions, as in marital status and health. Most particularly, age-related critical life events such as widowhood tend to engender emotional loneliness.
Marital status as a predictor of loneliness
Marital status was not identified as a predictor of social loneliness although it did predict emotional loneliness. This finding is consistent with that of Dykstra and de Jong Gierveld (Reference Dykstra and de Jong Gierveld2004) ; they also found no difference between married and not-married individuals in the experience of social loneliness. Weiss (Reference Weiss1973) theorised that social loneliness, as distinct from emotional loneliness, arises from the absence of an engaging social network. The majority of older people in this study, whether married, widowed, single or separated, had networks of family and friends with which they were socially engaged. This social engagement could protect against social loneliness (Victor, Bowling and Bond Reference Victor, Bowling and Bond2002). However, being single, divorced or separated, and particularly being widowed, were identified as predictors of emotional loneliness (family and romantic loneliness). This finding matches Weiss's (Reference Weiss1973) contention that emotional loneliness is experienced when there is a lack of a close and intimate relationship. This lack of, or loss of, a close personal relationship was evident in that the impact of widowhood on emotional loneliness was greater than that of being separated and divorced. As Dykstra and de Jong Gierveld (Reference Dykstra and de Jong Gierveld2004: 149) suggested, this is a consequence of the ‘emotional losses’ of widowhood impacting to a greater extent on individuals than the emotional loss associated with separation or divorce.
The findings in this study are contrary, however, to Dykstra and de Jong Gierveld's (Reference Dykstra and de Jong Gierveld2004) report that never-married older people (and especially women) have low vulnerability to emotional loneliness. Among our sample, never-married men and women reported similar romantic loneliness scores to those who were widowed. This finding indicates the importance of close personal attachment relationships and highlights the buffering effect of a single emotional attachment in alleviating loneliness. Although women had significantly higher levels of romantic loneliness than men, when marital status was controlled, gender was not a factor in the prevalence of romantic loneliness.
Gender as a predictor of loneliness
Men and women have been found to differ in their levels of isolation and need for social support, with women being more concerned with intimacy and men more in need of friendship ties (Allen and Oshagan Reference Allen and Oshagan1995). Holmén, Ericsson and Winblad (Reference Holmén, Ericsson and Winblad2000) found that women reported both social and emotional loneliness significantly more than men. In this study, however, when the socio-demographic and health attributes were controlled, the prevalence of social and romantic loneliness was similar for men and women. In fact, being a widowed male was a stronger predictor of family loneliness than being a widowed female. This may be due to the focal point that women play in marriage in initiating and sustaining family contacts. There is evidence that men rely on their spouse to initiate and organise family functions and that this may account for the relatively frequent family loneliness among widowed men (Dykstra and de Jong Gierveld Reference Dykstra and de Jong Gierveld2004; Stevens and Westerhof Reference Stevens and Westerhof2006).
Conclusions
The experience of loneliness among older people is complex and is influenced by many factors including increasing age, marital status changes, income, health, contacts with friends and family, and type and place of residence. The respondents least likely to report high levels of loneliness were married, had regular contact with friends and family, relatively good incomes, and access to transport. It has also been shown that older people experience different types of loneliness, and that romantic loneliness is most prevalent. The relatively high level of romantic loneliness identifies the importance of close, intimate relationships throughout life. Unlike social loneliness, romantic loneliness cannot be easily remedied through engagement with a social network; among older people this form of loneliness may persist regardless of whether one is widowed or single. This study has demonstrated that the quality of social and family relations may not buffer an older person from the experience of romantic loneliness. The differentiation of social and emotional loneliness follows from ‘the notion that loneliness differentially manifests itself depending on the context within which an individual's needs are unmet’ (Heinrich and Gullone 2006: 700). Another pertinent finding concerns the role of income as a factor in both family and social loneliness through its role in helping older people socialise and maintain contact with family and friends. Given that on an international comparison the basic social welfare pension in Ireland is low (National Council on Ageing and Older People 2004), this finding supports the case for a review of older people's pensions and benefit levels.
In relation to the study's methods, it has been shown that a large telephone survey of community dwelling older people is not only feasible but also effective, especially in a country with a high penetration of landlines (or wired handsets). Furthermore, we argue that the use of the composite SELSA-S measure to identify the social and emotional aspects of loneliness enabled a deeper understanding of the experience of multi-dimensional loneliness than univariate measures would have permitted. There were limitations to the study, not least in the representativeness of the sample – only community dwelling older people with access to a landline telephone were included. Therefore the results are not generalisable to older people living in care or without access to a telephone. However, overall the sample matched, in most respects, the population of older people living in Ireland.
In conclusion, strategies that are used to address loneliness need to take account of the complexity and the individuality of the experience, of its precipitating and perpetuating factors, and of the subjective meaning of the experience. Specific strategies should encompass the individual older person's own desire to communicate, the promotion of good health, the opportunity to meet people with a shared interest, background or experience, support from family members and the provision of a transport system, especially in rural areas. It has been shown overall that loneliness is not indivisible or unidimensional but a multifaceted and complex experience that can affect older people both socially and emotionally.
Acknowledgements
The authors acknowledge the support of the National Council on Ageing and Older People in Ireland for commissioning the study. We also thank Dr Clare Wenger of the University of Bangor, Wales for permission to use the Network Assessment Instrument, and are grateful for the information and support received from Dr Enrico DiTommaso, University of New Brunswick, concerning the Social and Emotional Loneliness Scale for Adults. We especially acknowledge the collaboration and help of all the older people who participated in this study; without their openness and willingness to engage, the study would not have been possible.
