Living in single person households and the risk of isolation in later life

Data from the International Social Survey Programme (2001) was used to analyse the social networks of older people and whether living in single person households increased the risk of isolation. When comparing respondents with one or more adult children, there was no significant difference in the likelihood of experiencing familial isolation between people living in single person households and those living in larger households. A majority of those living in single person households had at least regular contact with a sibling, adult child or close friend and participated in a social organisation. Friends compensate to some extent for a lack of support from the family, although in southern and eastern European countries, other relatives appeared to be more important in support networks. People living in single person households were more likely to experience isolation, but this was largely related to advanced age and childlessness. Whilst a very small minority in Japan were living in single person households, they were significantly more likely to be severely isolated than those living in single person households in other countries.


Introduction
The International Social Survey Programme (ISSP) is a continuing annual programme of cross-national social science survey activity.In 2001Á 2002, the agreed standard questions included topics relating to social network contact.The data included accounts of respondents' contacts with relatives and friends, their participation in civic and religious organisations as well as information about where they would be most likely to turn for help if faced with illness.The secondary data analysis in this paper is based on a sub-sample extracted for an international project supported by the Economic and Social Research Council of the UK.The sub-sample of the 2001 ISSP includes individual cases of those aged 50 or over from 18 countries that are members of the Organisation for Economic Co-operation and Development (OECD) (13 of which are European).In some places in the article, the findings are discussed in terms of European regions: where Italy and Spain represent Southern Europe; Britain, France, Germany, Switzerland and Austria represent Western Europe; Denmark, Norway and Finland represent Northern Europe; and Hungary, Poland and Czech Republic represent Eastern Europe.

Background
Concerns around the possible social isolation of older people living in single person households are not new.In the UK, for example, several important post-war studies (Sheldon 1948;Townsend 1957;Tunstall 1966) were carried out which explored the connection between these two variables.Such studies have consistently identified living in single person households as a risk factor for wider social isolation and loneliness (Victor et al. 2004).However, as Victor et al. have pointed out, there have in more recent years been a number of social changes, which may affect the experience of older persons living in single person households, not least the fact that this has become a much common arrangement.Living in a single person household cannot be assumed to lead to social isolation, as older people living without others may have good friendship, neighbourhood and community social networks.Similarly, living with a partner may lead to feelings of social isolation if the relationship is abusive and not of a good psychological quality.

International Journal of Ageing and Later Life
The increasing number of older people living in single person households has often been viewed as a negative development related to a decline in familial solidarity.An interest in the subject of familial obligations towards elders was reawakened towards the end of the 1970s when sociologists started to question assumptions of this simple association (Attias-Donfut & Rozenkier 1996: 51).Such writers have argued a ''myth of abandonment'', through which a stereotype had been created of the lonely older person, neglected by family (Connidis 1983).
A growing government interest in the role of the family in supporting older people has been evident in recent years.This has been fuelled at least in part by a growing interest to maintain a system heavily dependent on informal support due to demographical and political change which have in many states resulted in squeezing public funding in social care for older people (Attias-Donfut & Rozenkier 1996: 51).These issues have been brought to public attention through government departments responsible for healthy ageing agendas, and under the banner of the broader concept of ''social exclusion''.The term ''social exclusion'', first appearing in France in the 1960s, has become official parlance of the European Union (Cavelli et al. 2007: 10).It is a contested notion often used to highlight the multidimensional inter-connected elements of social isolation and material deprivation.However, the majority of national government and EU studies on social exclusion have focused on children, adolescents and those of working age, whilst there has been less interest in the exclusion suffered by older people (Scharf et al. 2004: 83Á 84).Reducing social exclusion among older people has, however, now become a Europe-wide priority (The Council of the European Union 2002).
A recent UK study commissioned by the Social Exclusion Unit has focused on these issues, defining the social exclusion of older people by seven dimensions.The study found living in single person households to be associated with exclusion from social relationships and civic activities.Being female was also associated with social exclusion in cultural and civic activities (but not contact with family and friends).Other risk factors of social exclusion included having no living children, being aged 80 and over, as well as non-White ethnicity, having poor health, low income, rented accommodation, no car and no telephone (Barnes et al. 2006).Kharicha et al. (2007) found living in single person households to increase the risk of social isolation even when controlling for age, sex, income and educational attainment.Holmen & Furukawa (2002), however, found that with increasing age, ''having a good friend to talk to'' and ''being subjectively healthy'' to be increasingly important in guarding against loneliness, whilst the effect of ''not living in single person households'' remained stable.A study by Scharf et al. (2004) found that the social and material exclusion was related to ethnicity (specific non-White ethnic groups were at greater risk) and age (those aged 75 and over being more likely to be at risk), whilst risk of exclusion was not found to vary significantly by gender.
Research findings are inconclusive over the affect of age on social isolation.Cavelli et al.'s (2007) study in Switzerland, for example, found that family and other social relations remained stable in advanced old age and social activity was only reduced as a result of declining health.Van Tilburg et al. (2004) and Wenger & Burholt (2004) found advancing age to increase the risk of isolation.Jylha ¨( 2004) finds weakening social integration in old age (along with increasing disability) to be a causal factor of increased isolation.Victor et al. (2004), however, found rates of isolation among the older cohort to fall.
If older persons live without others, we cannot assume they are ''lonely''.Cross-cultural studies have found loneliness among older persons to be more frequently reported in more traditionally familycentred countries of southern Europe, despite a lesser prevalence of living in single person households (Andersson & Sundstro ¨m 1996: 23).This, however, also raises issues around definition of the terms ''social isolation '' and ''loneliness'', and to what extent the psychological experience of loneliness is related to external patterns of social contact and engagement.
The degree to which one may feel isolated varies depending on the extent of contact with others, but also with a number of other social and psychological factors including personal temperament.Indeed, it is possible to be lonely without living on your own and vice versa.As found by Wenger & Burholt (2004), caring responsibilities among older persons is one factor that may contribute to loneliness, i.e.where older persons (often women) are themselves the primary carer for their husband or wife and may feel isolated in this role.
Social isolation is defined in this study with reference to three dimensions.These are the number and regularity of reported social contacts with: family relatives (separate to any partner or spouse); friends; and the wider community.The concept of ''severe isolation'' is used to describe those who lack a quantity of contact in all these different social networks.No attempt is made to control for social psychological, selfreported accounts of feelings and personal experiences of isolation and loneliness.Such data was not available to this study.Social isolation is defined and explored with reference to the quantity of contacts that older people have with other people, and measurement is based on the selfreporting of such contacts.It is also beyond this study and dataset to explore a more multifaceted definition of social exclusion and the focus of this study is on the quantity of social contacts.

Research Questions
. Are older people living in single person households at a significantly greater risk of social isolation than those living in a two or more person household? .What influence do the factors of childlessness, age, gender and social class have upon the risk of social isolation among older persons who live in single person households? .What are the differences in social and support patterns between countries by household structure?In particular, do countries with a larger proportion of older people living in single person households also have a larger proportion at risk of isolation?

Method
The ISSP is a random sample drawn from each host country's population.All samples include sample response rates of over 900 and most have final response samples of between 1000 and 2000.The ISSP social network data (2001Á 2002) was collected in 30 countries.We reduced the sample study to 18 countries, by only using those countries that are also members of the OECD (N 0 9942).The final sample included 13 European countries.Our analysis was only applied to a sub-sample of those aged 50 and over.
Single households and risk of isolation Klein & Harkness (2001) indicate that response rates are difficult to estimate precisely and vary considerably between countries taking part in the ISSP owing to different quota procedures used and the use of substitution in some countries (see also Park & Jowell 1997).Data has been analysed with SPSS, using cross-tabulations and Pearson Chi Square significance testing.Significance values when quoted reflect these tests.
Country aggregate data has also been explored using bivariate correlations.

Household Structure and Number of Adult Children by Country
Those living in single person households were more than twice as likely to have no children than those living with one or more other persons.Around a quarter of those living in single person households were childless, which compared to just over 9% of respondents living in a two or more person household.However, as shown in Figure 1, this difference varied between countries, most notably Spain and Poland in which those living in single person households were nearly four times as likely to have no adult children, and in particular Japan, where almost 40% of people living in single person households were childless.

Family Members
The dataset included information on how often respondents see their close family members: sibling; son or daughter (who they have most contact with); mother and father.Data on contact with extended family members is also available and will be discussed briefly.
Adult child.A large majority of the sample (85%) had one or more adult children.However, those in the sample without children were more than twice as likely to live in a single person household (46%) than those who had one or more offspring (21%).However, less than 25% of people living in single person households in the sample were childless.Therefore an important indicator of the extent to which living in single person households may increase the general risk of social isolation, may be frequency of contact with their adult child.This variable refers to contact with the adult child the respondent sees the most (where they have more than one).In the majority (53%) of cases, respondents stated this adult child (the one which they either saw most often or was their lone child) was a daughter.As shown in Table 1, the data showed that, overall, people living in single person households (excluding those with no children) were slightly less likely to see their adult child at least once a month (83%) than those living in a two or more person household (89%).This pattern was true for all countries except for Great Britain which had a relatively high rate of monthly contact with their adult child amongst its people living in single person households (89%), despite, as shown in Figure 2, also having a comparatively high proportion living in single person households (34%).For the sample as a whole, living in single person households was found to make a significant difference to the percentage seeing their adult child on a   14 4 14 7 17 7 19 1 0 20 7 20 9 21 1 1 22 1 1 24 1 0 25 1 3 25 9 26 1 1 27 7 30 1 4 31 12 38 1 0 38 7 40 15   1  When the analysis was conducted by social class, working/lower middle class respondents were more likely to see their adult child once weekly or monthly than upper middle/upper class respondents (p B 0.0005).Likelihood of monthly contact increased with age for those living in single person households but decreased for those living with one or more other persons.With increasing age, therefore, the gap in frequency of contact narrowed between those living in single person households and those living in a two or more person household.The greater proportion of women in the older age groups may be one reason for this, since for all age groups, household types and social classifications, women were more likely than men to see their adult child at least weekly/monthly.
Table 1 shows there is some variation in face-to-face contact between the countries, such as the comparatively high levels of contact evident in the Southern European countries.However, the data shows that the large majority see their adult child at least once a month, and the norm for people living in single person households in all countries (excluding Japan, where to live apart from one's adult child is less common) is to retain a minimum of weekly face-to-face contact.1.4 2.1 2.4 0.8 2.2 2.8 0.6 3.8 3.8 3.3 2.6 4.3 2.2 2.9 4.9 3.1 2.6 5.5 2.4 2.7 3.1 5.7 2.9 3.9 3.9 3.9 7.6 7.1 3.9 3.8 7.5 12.3 6.7 8.6 9.6 11.7 5.8

JP CA NO ES AU FI IT NZ DE PL CZ DK HU CH FR AT GB US Tot
The majority of respondents had other contact (besides visiting) with their adult child at least once a week.In fact, if those who were living in the same household as their adult child are excluded, only 20% of respondents reported to have no form of distance contact with their adult child at least once a week.Most of those (63%) who saw their adult child less than once a week were in remote contact with him or her at least once a week, and 21% at least several times a week.The majority (52%) of those in face-to-face contact less than once a month were also in remote contact at least once a week.As has been found by findings from the previous ISSP dataset on social networks (Finch 1989), contact from a distance does not therefore appear to reflect a straightforward pattern of compensation for lack of face-to-face contact.In fact, there was a strong positive correlation (by country) between the proportion who see their adult child at least weekly and the proportion who have weekly distance contact (r0 0.899, p B0.0005).Thus, those who reported to see their adult child more often were also more likely to be in more frequent contact by remote means.The proportion of those who saw their adult child at least once a week (but didn't live with them) and also kept in touch through another form of contact on a weekly (or more often) basis was particularly high (89%) and in fact most of these (66%) were in remote contact with their adult child at least several times a week.
Japan stands out from the overall pattern since a very low proportion of its respondents were in touch by remote means on a weekly basis.In fact, among those who see their adult child less than once a week in Japan, less than a quarter (24%) were in contact through other means.This suggests that where Japanese older persons do not live with their adult child, they may be less likely to stay in regular contact.This may be to some extent explained by the traditional family pattern, whereby a married daughter may have little contact with adult parents as she would be considered part of her husband's family (Martin & Noriko 1991).
Sibling.Although siblings are less likely than children and spouses to provide instrumental support, they have been found to be important providers of support in times of crisis for older persons in general, and as particularly important sources of help for those who are single, widowed and childless (Campbell et al. 1999).The ISSP data supports this finding International Journal of Ageing and Later Life showing that among those with a living sibling, people living in single person households were significantly more likely to see him/her on a weekly basis than those living in a multiple person household (p 00.006).The difference was, however, more acute between those with and without children with childless respondents significantly more likely to regularly see their brother or sister than those who had one or more adult child (p B 0.0005) (Figure 3).In fact, when childlessness was controlled for, no significant difference was found in contact levels by household type.Higher levels of contact amongst those living in single person households therefore appears to be a confounding factor of childlessness, i.e. since those living in single person households are more likely to have no children.
Greater weekly contact was particularly significant for the childless group, with a greater proportion of childless respondents reporting weekly contact with a sibling in each of the 18 countries.However, this  Single households and risk of isolation difference was only statistically significant for Hungary, New Zealand and Denmark.Nevertheless, as is shown in the chart below (for the whole sample), the data suggests that childlessness, and less frequent contact with an adult child may to a small extent be compensated for by greater contact with a brother or sister.
Distance contact.Among all respondents with a living sibling, 39% were in touch with him or her at least once a week and 69% at least once a month.Seventy-seven percent of those who saw their sibling at least once a week were also in contact by remote means and 26% of those who did not see their sibling at least once a week were in touch through another means.Therefore, nearly half (43%) of those with a sibling were in some form of contact (either remote or face-to-face) at least once a week and the majority (69%) were in touch with, or saw their sibling at least once a month.
Those living in single person households were more likely to have regular distance contact with a brother or sister than those living with others.Forty-six percent of people living in single person households had (at least) weekly contact, compared to 38% of those living with others (p B 0.0005) and 71% were in touch at least once a month compared to 68% of those living in bigger households (p 00.041).Those without children were also more likely to have regular remote contact with their sibling, although this was only statistically significant at a weekly level (p B0.0005).
Regular remote contact declined by age.Women were more likely than men to be in touch with their sibling at least once a week (pB 0.0005) or once a month (p B0.0005), but for both genders contact was greater for those who had no adult children.Contact was also greater for those living in single person households when controlling for gender and childlessness, i.e. for both men and women, with or without children, contact was greater when living in single person households.No significant difference was found by social class.When comparing weekly remote contact with sibling by country, the variation was greater than for adult child, ranging from over half in Italy, Spain and Hungary to only 18% in Japan.

Isolation from Family Contact
Although the frequency of face-to-face contact with close family members was greatest in Southern and Eastern Europe, the similarities in level of International Journal of Ageing and Later Life contact are perhaps most interesting with the majority of respondents from each of the 18 countries seeing an adult child or sibling at least once a week, and 89% of the whole sample reporting to have either (or both) faceto-face or distance contact with an adult child or sibling at least once a month.The remaining 11% of respondents reporting no contact with either of these relatives during an average month are, for the purpose of this analysis here defined as ''isolated''.
As shown in Figure 4, the proportion seeing either an adult child or sibling less than once a month, was smallest in Italy (1.9%) and largest in Great Britain (8.7%).There was no significant difference in the likelihood of seeing an adult child or sibling at least once a month by number in household (one or more) or by marital status.Only 4% of all respondents in the total sample lived in a single person household and fitted the criteria of the ''isolated'' family contact variable (2.8% did not have an adult child or sibling).
When the ''isolated'' group was compared with the majority group by age, gender and class, it was found that men were more likely to be isolated, since 13% of men reported no monthly contact with an adult child or sibling, which compared to 9% of women (p B0.0005).Middle to upper class respondents were less likely to be isolated.However, this is largely due to the lower proportion having no living relative.The likelihood of being in the isolated group significantly increased with age for women, from 8% of those aged below 70 to 14% of those aged 70 and above.
If contact with extended family members (i.e.cousins, brothers and sisters-in-law, and nieces and nephews) are taken into account, the proportion with no monthly contact reduces to just 5.5% of the whole sample.Almost half of those with no monthly contact with a sibling or child had some contact with one of these extended relatives.However, people living in single person households were less likely to have seen an extended family member in the last four weeks (pB 0.0005).This was the case even when controlling for age, gender and childlessness, and this pattern was reflected in each of the 18 countries in the sample.The lowest proportion with no monthly contact with a close or extended family relative was amongst Italian respondents (2.5%) and the highest amongst those from Great Britain (11.8%).The majority of total respondents (67%) and over half of people living in single person households (58%) had some monthly contact with at least one of these extended family members.

Friends
The questionnaire asks respondents about numbers of close friends in several categories, including those within the same locality, from the workplace and other friends.It also asks about frequency of contact with closest friend.These two measurements of friendship ties are discussed below.Number of close friends.As shown in Figure 5, the average reported number of total close friends ranged from 5.1 in Hungary to 12.6 in Norway.In contrast with family contact levels, average number of close friends was found to be lowest in Southern (7.4) and Eastern European countries (7.0).In fact, there was a significant negative association between number of non-work based friends and weekly family contact (adult child and sibling) by country (r 0Á0.654, p 00.003).Thus, respondents in countries with an average greater level of family contact tended to report on average a smaller number of close friends in their network.Previous research has found a greater likelihood of older persons living in single person households or with only a spouse, to include more friends and neighbours in their social networks (Wenger 1989).The ISSP data shows little difference in overall average number of close friends between those living in single person households or in a two person household (11.4) and those in a 3' person household (11.7), although those living in a one or two person household had a slightly greater average number of non-work based friends than those in 3' person households.Those living in single person households had on average a slightly smaller number of close friends, but childless respondents living in single person households 1.8 0.4 1.2 2.3 0.5 1.7 1.8 1.2 1.4 2.8 2.4 2.4 1.9 1.9 1.5 2.8 1.9 2.9 1.6 1.9 2.2 2.8 2.5 4.4 3.1 3.6 5.3 4.4 3.6 4.3 4.5 4.3 5.2 5.6 5.5 6.1 6.2 4.2 1.4 2.6 2.3 1.7 2.4 2.5 2.4 2.2 3.2 2.8 2.6 2.6 3.8 3.0 3.1 2.9 3.5 3. The average number of close friends declined with age.However, this appears to be primarily related to the inclusion of workplace friendships held by those of working age.If total friends are compared, those aged below 70 had on average 11.8 close friends which declined to 9.4 among those aged 70 and over.However, the difference was very slight if workplace friends were omitted, with an average of 9.4 for those aged below 70 and 9.2 for the older age group.And those aged 70 and over reported to have more close friends living in their neighbourhood.The most notable decline of average number of friends was in the age group those aged 80 and over (7.7).
In terms of gender and social class, there was a reversal in the pattern identified for family contact where higher levels were found among women and those from the lower social class groups.When average numbers of close friends were compared, those in the higher strata (middle to upper class) reported to have a greater number of close friends in each category, than those in the lower social classes (working to lower middle class).Male respondents on average reported to have a greater number of close friends than women even when controlling for age, marital status, social class, household type and childlessness.
When comparing the number of average friends by country or groups of countries, issues around the cultural interpretation of such a subjective concept should be considered (Hollinger & Haller 1990: 114).There may also be gender differences in interpretation of this concept.Another issue related to this is that simply counting the number of reported friends may also not be an accurate measurement of the strength of non-kin networks because the size of the network gives no information as to the actual level of contact or support received (Keating et al. 2003: 118).Although information of this type is not fully provided in the ISSP dataset, it does include an additional measurement of the strength of friendship ties: frequency of contact with closest friend.
Frequency of contact with close friend.There was no significant association between number of close friends and frequency of contact with closest friend.But, countries in which respondents had on average more frequent family contact also tended to report more frequent contact with their International Journal of Ageing and Later Life closest friend. 1 The higher frequency of best friend contact may indicate that respondents in the more ''family-centred'' countries, may be more likely to have a smaller number of more intense friendships.Indeed, Southern and Eastern Europeans were the most likely to see both their closest friend and their adult child or sibling on a daily basis. 2 However, those from the Southern and Eastern European countries were also on average most likely to report having no close friend.
People living in single person households were more likely to have reported frequent face-to-face contact with their closest friend.Over half (52%) of those who lived alone saw their closest friend at least once a week, which compared to 44% of those in a two or more person households (p B0.0005).In fact as is shown in Figure 6 below, in all countries except Italy, people living in single person households had higher levels of weekly contact with a best friend than people living in larger households, and the difference was significant regardless of age and sex.A similar pattern was identified when comparing contact by marital status in that the unmarried were more likely to have regular contact with their close friend (except in Italy).Women were more likely to have weekly contact with their best friend than were male respondents (pB 0.0005), Single households and risk of isolation whilst middle to upper class respondents were less likely to see a close friend on a weekly basis than those classified as working to lower middle class.Employment status made little difference to levels of weekly contact.

Social Participation
Participation in social and community groups has been shown to be an important indicator of non-kin network size (Burholt et al. 2003: 14) and of relational inclusion (Cavelli et al. 2007).In order to measure social participation amongst the ISSP respondents, a binary variable of participation (do/do not participate) was used in relation to participation in a political party, church or religious organisation, sports group, charitable organisation, neighbourhood group or other organisation.
The pattern of social participation by country contrasted with that of family contact in that respondents from Eastern and Southern European countries were less likely to participate in a social group or organisation than other respondents.There was a strong negative correlation by country between weekly close family contact and the percent that participate by country (r 0Á0.823, p B0.0005).This suggests some support to the theory that close-knit family ties may militate against community linkages (Keating et al. 2005: 29).
However, the overall participation rate was lower among people living in single person households than those living in a two or more person household (p B0.0005) even when controlling for gender.There was some variation between the countries, with Great Britain, for example, having higher participation rates among those living in single person households (p 00.022).There was little difference in participation rates between the 50Á 64 (59%) and 65Á 79 (58%) age groups, but level of participation declined (48%) amongst the oldest age (80') group (p B0.0005).
For all age groups, married persons had higher participation rates, than the unmarried, although participation rates were higher for the ''never married'' than for the ''widowed, divorced or separated''.Although there were some country differences in this pattern, overall, men (61%) were more likely to participate than women (55%) even when controlling for marital status and age.Middle to upper class respondents were significantly more likely to participate (p B0.0005) even when controlling for age, gender and household type and childlessness.
Although there are issues around the subjectivity of the concept of close friendship which may have an influence upon the number of friends respondents included in their network, as Hollinger & Haller (1990) note the proportion reporting they have no close friend is likely to be a significant measure of isolation.When asked how about frequency of contact with closest friend, 17% reported to have no close friend, and this group was more likely not to participate in any social organisation (p B 0.0005).About 13.4% of all respondents neither had a close friend nor participated in any social organisation.
Under both definitions (i.e.whether including those with no close friends or having less than monthly contact) Eastern Europeans were most likely to be socially isolated.There were, however, differences between the Eastern European countries, with Hungarians in particular being likely to report having both no close friends (a finding that was also noted by Hollinger and Haller) and no participation (32%), followed by Poles (23%), whilst those from Czech Republic were much less likely to do so (12%).Under both definitions, the Southern European region also had higher levels of ''social isolation'' than those from the other European and non-European countries in the sample, whilst the lowest rates of social isolation were found in Australia, New Zealand and North America.
If ''social isolation'' is here defined as those reported to neither have monthly contact with a close friend nor any participation in a social organisation, people living in single person households were no more likely to be socially isolated, whilst childless respondents were slightly more likely to be in this group (p0 0.005).The risk of isolation increased with age with 30% of those aged 80 and over in the socially isolated group compared to 23% of those aged 65Á79 and 19% of those in the youngest age group (pB 0.0005).Working and lower middle class respondents were over two and a half times more likely to be isolated when using this definition than middle to upper class respondents (p B0.0005) and the difference remained even when controlling for age and gender.

Severe Isolation
The influences of various social characteristics upon the likelihood of being ''isolated'' from family or friendship contact and social participation have been discussed above.This section now explores the occurrence of severe isolation, which is here defined by a combination of ''isolation'' in all three dimensions.
Only a very small proportion of respondents (3.5%) in the international sample could be defined as ''severely isolated'' in terms of reporting no monthly contact (face-to-face or remote) with either a close relative (sibling or adult child), a close friend or any involvement in social groups and organisations.This however ranged from under 1% in Italy to 6.5% in Great Britain.Nearly a quarter (23%) were found to be isolated in two of the three dimensions.
Overall, a greater risk of severe isolation (all three dimensions), was apparent for men than women (pB 0.0005) although the gap narrowed in the older age groups.Being unmarried increased the risk of severe isolation more for men than for women.Those from rural areas were slightly more likely to be severely isolated than those from suburban/ urban areas (p 00.040).For all types of areas and for both genders, working/lower middle class respondents were more likely to be severely isolated than those defined as middle/upper class.Those on a relatively low household income 3 were also slightly more likely to be in the severely isolated group (p 00.004).
Risk of severe isolation increased with age, with those aged 80 and over more than twice as likely to be in this category than younger respondents (p B0.0005).Those with no adult children, however, were eight times more likely to be characterised by all three aspects of isolation (pB 0.0005).The data suggests that the risk of isolation is not increased by living in single person households per se, but due to the confounding factors of greater likelihood of advanced age, and primarily of childlessness among those living in single person households.
Few respondents were ''isolated'' in all three dimensions, or even in the two dimensions of apparent weak family and weak non-kin ties.A large majority (almost 90%) of the sample had either at least above average contact with sibling, adult child, close friend or social participation.However, the data does not appear to suggest that those who are isolated in one dimension significantly compensate in another.Those who were isolated in family contact were in fact more likely to also be isolated in terms of social contact and participation (p B0.0005).This was true for all countries in the sample, except Italy.There may, however, be some degree of substitution for low family contact through friendship networks as those with below average frequency of family contact, reported to have a larger number of friends on average (p B0.0005).
When comparing countries by regional groupings, no significant differences were found in the overall risk of severe isolation.However, the Japanese were much more likely to be severely isolated if living in single person households.Risk of severe isolation among people living in single person households was also relatively high among Eastern Europeans and almost one in five childless Eastern Europeans were severely isolated, which compared with 13% of the sample population as a whole.

Support Networks
The ISSP questionnaire asks some key questions about who the respondent would turn in three hypothetical scenarios which give an indication of sources of support accessed in three different areas: emotional; practical; and financial.The questionnaire asks who the respondent would turn to supposing s/he: (a) were ill for a few days and needed some practical help; (b) were a bit down or depressed and wanted someone to talk to; and (c) needed to borrow a large sum of money.This section will compare the reported primary support networks of those living in single person households and/or have no children with other respondents, and it will also consider the impact of country, age, gender and social class.However, for want of space, this section will focus primarily on the first variable (who to turn to when ill).
Thus far, we have discussed data relating to the social networks of the participants and how these differ between people living in single person households and those living in larger households, as well as looking at differences based on other social criteria.However, one problematic issue with such social network analysis is that information regarding the size of networks and frequency of contact, tell us little about how that equates to actual support.As Walker has argued: ''One truism of network analysis is that mere presence of a tie between two people does not equate with the provision of support'' (Walker et al. 1993cited in Keating et al. 2005: 25).Therefore, although we have shown, for example, how people living in single person households had on average more frequent contact with a Single households and risk of isolation close friend, this tells us little about the role of that friendship in terms of supporting that person in everyday life.And in particular, such social network data gives us little clue about the potential caring capacity of informal ties should need arise, and the extent to which family and friends may ''rally round'' in a crisis (Cavelli et al. 2007: 23;Keating et al. 2003).
Social network analysis has found that older people tend not to mobilise the whole of their social networks for support but rather turn to their ''core network'', which is usually constituted of close kin (Keating et al. 2005;Phillips et al. 2000;Phillipson et al. 2001).Frequency of contact with close family gives one indication of the strength of these potential support networks.However, as Hollinger & Haller (1990) argue, the reduced frequency of face-to-face family contacts in Northern and Western European countries does not necessarily imply a reduction of support.As Hollinger and Haller's study of previous ISSP data found, informal support networks continue to be much more important than formal sources of support in all countries (except for borrowing money).Indeed, the 2001 ISSP data shows that formal services would only be turned to in the first instance in a very small number of cases if the respondents were ill (although more common in Western Europe).
As was found in relation to the 1986 ISSP module on social networks (Finch, 1989: 87), first impressions are that the data on sources of support, reveal more similarities between the countries than differences.Past research has found older persons to prefer to receive support from spouses than from children, other family members or friends and neighbours (Ajrouch et al. 2001).This universal importance of the role of the spouse is indeed most noticeable in the ISSP data.The data suggests that respondents are most likely to turn first to a spouse for practical and emotional support in each of the 18 countries in the sample.Nevertheless, as (Finch, 1989: 101) found in regard to Italy, respondents from Southern and Eastern Europe and Austria, were less likely to look to a spouse and more likely to state they would turn to an adult child than were other respondents.In addition, when first and second choices were combined, adult children were more commonly selected than a spouse or partner in Southern and Eastern Europe.
As shown in Figure 7, non-kinship informal contacts were more important as primary providers of emotional than practical support.In fact, 18% reported they would turn to a friend first when depressed which compared with 16% who stated they would turn to a daughter or son.Again, Southern and Eastern Europeans were the least likely to select a friend (as a first or second choice), whilst non-Europeans were the most likely.Although non-kin ties may be less dependable than close family in terms of providing more intensive practical and care support, their importance in terms of contributing to the mental health and well-being of older persons in particular should not be under-estimated.Indeed, as research in the UK has suggested, the importance of friendship ties in the ''personal communities'' of older people may be increasing (Phillipson et al. 2001).The importance of support from friends was particularly evident for people living in single person households without children.Nearly half of this group (43%) said they would turn first to a friend or other non-kin informal contact (e.g.neighbour or colleague) for support when either ill or depressed.Those with more regular contact with a close friend were more likely to turn to a close friend for support and were more likely to have a larger number of close friends on average.This suggests  that stronger non-kin social ties are more likely to translate into support.It is important, however, to better understand the role of non-kin ''weak ties'', as there has been little research around their importance in the lives of older people (Keating et al. 2005: 29).
In the case of illness, people living in single person households were most likely to turn to an adult child for support than to friends and other relatives, although a daughter (29%) was more than twice as likely to be called upon as a son (14%).However, where the lone person had no adult child to turn to, a friend, neighbour or colleague was reported to be either a first or second choice contact (44%) followed by other relatives (37%) Á although the reverse was true in Southern Europe where extended kin appeared to have a greater role in providing support.Childless respondents living with one or more other persons were more than twice as likely to turn first to another relative, than to a friend, perhaps due in part to the greater likelihood of co-residence with an extended family member.
When controlling for marital status, the greater likelihood of childless respondents turning to a close relative, is strongly influenced by cultural differences, particularly noticeable among those from Southern Europe.Among Southern Europeans who were married without children, only 60% reported they would turn first to their spouse (compared with 86% overall) when ill, and over a quarter (26%) to another relative (compared to less than 6% of the sample as a whole).Over three quarters of lone unmarried Southern Europeans with an adult child stated they would turn to him/her (52% to a daughter) and only 7% to a friend.In contrast, almost a quarter (23%) of Western Europeans turned first to a friend and just over half to a daughter or son (51%).People living in single person households with no adult child were also most likely to turn to another relative in Southern Europe (48%), whilst Western Europeans were the most likely to turn to an informal non-kin contact (48%).North Americans in this group were the most likely to say they would turn to no-one (13%), whilst Japanese (19%) and Northern Europeans (13%) were the most likely to state they would turn to formal services or pay someone to help.
Few persons, whether living in single person households or with others, stated they had no-one to turn to for practical (1.5%), emotional (6%) or financial (13%) help.However, the likelihood of having no-one to turn to, was greater for people living in single person households.This was especially true for those in the ''severely isolated'' group, 15% of whom stated they would have no-one to turn to when ill, 27% if depressed and 44% if in need of a large loan.The likelihood of having no-one to turn to differed little between those who were severely isolated (i.e. in all three dimensions) and had an adult child or not (since by definition they had little contact with their adult child).Severely isolated persons also tended to be more likely to turn to formal services for help.In terms of gender, there was little difference in the proportion stating they would have no-one to turn to, despite the greater preponderance of widowhood among women.Even among married persons it is notable that females were more likely to turn to someone other than their spouse for support than males.In terms of emotional support it was particularly evident that married women were more than four times as likely to turn to a daughter and more than twice as likely to turn to another relative or close friend (p B0.0005).As has been found in past research, the data suggests that support networks may differ less for women by marital status than for men (Scott & Wenger 1995: 160).Those who had no informal contact to turn to (i.e.stated they would turn either to no-one or a formal service) when either ill or when depressed were more than twice as likely to be aged 80 or over (15%) than under 80 (7%).Women over 80 were nearly twice as likely to state they had no-one to turn to when ill than men in this age group (p 0 0.018), but there was no difference between the genders in terms of the emotional support variable.In all the age groups, males were significantly less likely to state they would turn to an informal contact to borrow a large sum of money even when controlling for household type (p B0.0005).
Despite the greater likelihood of having no-one to turn to, the large majority of people living in single person households, with or without adult children could state at least two informal contacts they could turn to.As Keating et al. note, the question thus remains as to ''whether the nonmarried have smaller support networks because of the absence of members with normative obligations or larger networks because, in the absence of a spouse or children, they invest more in diverse supportive relationships'' (Keating et al. 2003: 119).The ISSP data does suggest though, that childless unmarried respondents compensate at least to some extent for lack of support from offspring by turning to other relatives (particularly siblings) and friends (with those living in single person households depending to a greater extent on friends).The ISSP data also shows, however, that the make-up of support networks varies between countries.There are some broad similarities such as how, across the board most were primarily dependent on a spouse and few people stated they would have no-one to turn to for support, or that they would turn to formal services.However, the most notable difference suggested by the data is a lesser dependence on spousal support in Southern Europe even where there are no adult children to turn to.Although married Eastern Europeans and Japanese were also more likely to turn to an adult child, childless married persons were much more likely to be dependent on a spouse than in Southern Europe.This was particularly evident in Italy, where there was no significant difference by household type in having an informal contact to turn to for practical or emotional support (this was also surprisingly true of Norway where despite high dependence on spousal support in a two or more person households, very few stated they would turn to formal services or have no-one to turn to if they were ill).
The greater likelihood among Southern Europeans of married persons in a two or more person household, turning to someone other than a spouse, may be in part related to a greater likelihood of co-residence with another family member.The proportion by country turning to an adult child for support, and the proportion having their offspring living with them is, for example, strongly correlated, particularly when including only married persons (r 00.837, p B0.0005).However, the greater proportion turning to another family member in Southern Europe (particularly Italy) and also (to a lesser extent) in Eastern Europe and Japan, even when married and coresiding with an adult child, is perhaps indicative of differences in closeness of extended kinship ties (meaning the person requiring care is less dependent on a partner), and differing cultural expectations concerning the role of spouse, children and other kin (Attias-Donfut et al. 2005;Hollinger & Haller 1990)

Conclusion
People living in single person households were no less likely to see their adult child than those living in a two or more person household where the respondent's adult child did not reside.People living in single person International Journal of Ageing and Later Life households were more likely to have frequent face-to-face contact with a sibling, although this was due to the confounding factor of childlessness, i.e. childless respondents were more likely to have frequent contact with a sibling and people living in single person households were more likely to be childless than those living with a spouse and/or others.In addition, those living in single person households were more likely to have frequent distance contact with a sibling and no less likely to have distance contact with an adult child when controlling for age.People living in single person households were, however, less likely to have seen another extended family member in the previous four weeks.Overall, those living in single person households were more likely to be isolated from family contact when all the different family relationships were considered together.However, there was no significant difference when controlling for childlessness.Thus, people living in single person households were at a greater risk of familial isolation in terms of lack of contact with an adult child or sibling but this is mainly due to the greater likelihood of having no adult child, and the limited extent to which sibling contact compensates for this.
There was no evidence that people living in single person households were any more likely to be socially isolated on the other dimensions of friendships and social participation.They were no less likely to have a smaller number of friends when controlling for age, and in fact were more likely to have frequent contact with a close friend.They were less likely to participate in social organisations overall, but when comparing the oldest group (over 80s) there was no significant difference.For many people with below average family contact, social contact seems to compensate to a limited extent.Friends also appear to play a more important role in the support networks of people living in single person households, especially among those with no adult child.Friendship ties may compensate for the lack of close family support available to people who live in single person households and do not have children.However, there remained a small minority of people living in single person households who were severely isolated in both the family and the social sphere and stating they would have no-one to turn to when in need of support.
Overall, people living in single person households were no more likely to experience severe isolation (i.e.combined social and familial isolation).There were, however, some country differences which should be emphasised, most notably the different pattern evident in Japan, where people living in single person households were at a much greater risk of severe isolation even when controlling for childlessness.
For the unmarried, in particular, the data also reveals the important role of non-spousal kinship ties in providing practical and emotional support, particularly in Southern and Eastern Europe, whilst non-kinship ties tended to be relied upon to a greater extent elsewhere.Again Japan differed from this pattern, clustering with Southern and Eastern Europe on the variable showing type of support preferred when ill, but being top of the ''league table'' in turning to a friend for emotional support.
This study raises some concerns about the small sub-section of severely isolated people living in single person households who were cut off from contact with both a family member, or friend, as well as having no participation in social organisations.Although living in single person households in itself does not appear to increase the risk of severe isolation, the combination of factors associated with living in single person households, such as advanced age (and worsening health), lower social class, low income and most importantly, childlessness, means that those living in single person households are likely to be at a greater risk.The fact that the data shows the small number of older people who are living in single person households and severely isolated to be more likely to have ''no-one to turn to'' for practical, emotional or financial support, remains a cause for concern.This study therefore highlights the need for policy initiatives to target this most isolated of groups and to find ways of supporting and developing their social networks.Befriending schemes through which volunteers visit isolated older people in their homes, are one example of a practical intervention which aims to address this need and combat social exclusion (Fyvie-Gauld & de Podesta 2007).Notes 1.When correlating weekly contact for seeing adult child or sibling (valid percentage) against weekly contact with closest friend r0 0.626, p 0 0.005.2. The Japanese also scored highly on daily contact with close relative, but were much less likely to see a close friend daily.3.However, it should be noted that household income data is relative to each specific country and not therefore strictly comparable Á i.e. ''low income'' persons are computed as those whose household income falls within the bottom third of the distribution for each country.

Corresponding Author
Laura Banks, University of Brighton, SSPARC (Social Science Policy and Research Centre), School of Applied Social Science, Mayfield House, Falmer, Brighton BN1 9PH, UK.Email: l.c.banks@bton.ac.uk

Figure 1 .
Figure 1.Proportion with no adult children by household structure (living alone/2' household).

Figure 2 .
Figure 2. Bar chart showing proportion living alone, by contact with adult child.

Figure 4 .
Figure4.Proportion of all respondents having infrequent (less than monthly) contact with an adult child or sibling, by household type and country.n0 9942 (includes those with no living relative and missing cases as no contact).

Figure 5 .
Figure 5. Average number of close friends by type and country.
IT FR PL DE DK GB AT FI CH JP CA NZ US AU NO Total Single households and risk of isolation had in fact more friends on average than childless persons living in a two or more person household.

Figure 6 .
Figure 6.Proportion of respondents who see their closest friend at least once a week by household type (living alone/2' household) and country.

Figure 7 .
Figure 7. Bar chart showing who the respondent would turn to first for support, by household type (living alone/2' household).