You are here

Article Text

Article menu

Download PDFPDF

Other topics
Friends are equally important to men and women, but family matters more for men's well-being
  1. Noriko Cable1,
  2. Mel Bartley1,
  3. Tarani Chandola2,
  4. Amanda Sacker3
  1. 1Department of Epidemiology and Public Health, University College London, London, UK
  2. 2The Cathie Mash Centre for Census and Survey Research, University of Manchester, Manchester, UK
  3. 3Institution for Social and Economic Research, University of Essex, Colchester, UK
  1. Correspondence to Dr Noriko Cable, Senior Research Fellow, Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London WC1E 6BT, UK; n.cable{at}ucl.ac.uk

Abstract

Background People with larger social networks are known to have better well-being; however, little is known about (1) the association with socio-demographic factors that may predict the size and composition of social networks and (2) whether the association with well-being is independent of pre-existing psychological health or socio-demographic factors.

Methods The authors used information collected from 3169 men and 3512 women who were born in Great Britain in 1958. First, age on leaving full-time education, partnership and employment status at age 42 were used to predict the size and composition of cohort members' social networks at age 45 using ordered logistic regression. Second, using multiple linear regression, the associations between social network size by composition (relatives and friends) and psychological well-being at age 50 were assessed, adjusting for socio-demographic factors and psychological health at age 42.

Results Not having a partner and staying in full-time education after age 16 was associated with a smaller kinship network in adults. Having a smaller friendship network at age 45 was associated with poorer psychological well-being among adults at age 50, over and above socio-demographic factors and previous psychological health. Additionally, having a smaller kinship network was associated with poorer psychological well-being among men.

Conclusions Having a well-integrated friendship network is a source of psychological well-being among middle-aged adults, while kinship networks appear to be more important for men's well-being than for women's. These relationships are independent of education, material status and prior psychological health.

  • Cohort studies
  • social networks
  • well-being
  • gender differences
  • alcohol
  • mental health
  • social capital
  • social inequalities

https://doi.org/10.1136/jech-2012-201113

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Berkman and Syme1 showed that stronger links to the community were associated with lower mortality in Alameda County, California. Many studies have followed, examining associations with mortality,2–5 physical health6–8 and mental health.9 ,10 A recent meta-analysis concluded that the effect of social relationships on mortality risk was comparable to that of other established risk factors such as smoking.11

    Studies of the association between social ties and health are not entirely consistent in their findings, however. Bergelt et al 12 did not find a relationship to health outcomes at all. Prospective studies have increased in number, but there are few which have examined long-term associations to establish a temporal order between social networks and mental health.13 ,14 Another issue that has bedevilled research on social networks and health is the possibility of confounding by socioeconomic factors. Although Berkman and Syme1 showed that there was a strong association between social networks and social class in Alameda County, other studies have not taken this factor fully into account.9 ,15 Even fewer studies have examined the relationship of social networks in midlife to previous socioeconomic position or life circumstances. Also some have reported gender differences in the effects of partnership16 and social capital17 on mental health. We cannot be sure that the characteristics of the social network are not acting as markers either of social circumstances or psychological health, or of how they act on health outcomes differently for men and women over the life course.

    The aims of our study are to (1) examine the association between socio-demographic factors and social networks and (2) to examine whether the association with well-being is independent of pre-existing psychological health or confounding socio-demographic factors in British residents who were born during the same week in 1958. This paper addresses five questions:

    1. Do socially advantaged adults have larger social networks than those who are less advantaged?

    2. Does the association of socio-demographic factors with social network size differ according to the composition of the networks (relatives or friends)?

    3. Does having a large social network contribute to subsequent psychological well-being independently of socio-demographic factors?

    4. Does the association of social network size with psychological well-being differ according to the composition of the networks (relatives or friends)?

    5. Does the association of social network size with psychological well-being differ according to gender?

    Methods

    Study population

    We used data from the National Child Developmental Study (NCDS). The NCDS cohort is a prospective birth cohort study in Great Britain, and details of the study are described elsewhere.18 Information collected from the participants at ages 42,19 4520 and 5021 was used for this study.

    Social network size

    In the age 45 survey, NCDS cohort participants were asked to respond to one of the five categories to approximate the numbers of relatives and friends with whom they meet ‘once a month or more’. The response categories are: none, 1–2, 3–5, 6–10 and more than 10, for both relatives and friends. We assigned ‘none’ to those who answered that they did not have any relatives or friends. These are outcomes for the study questions 1 and 2 and independent variables for the study questions 3–5.

    Psychological well-being

    A total score from the Warwick–Edinburgh Mental Well-Being Scale (WEMWBS) administered at age 50 was used to indicate the level of psychological well-being, which is the study outcome for the study questions 3–5 of this study. This is a 14-item scale with each item measured using a 5-point Likert response addressing hedonic and eudaemonic perspectives of mental health.22 The total score ranges from 14 to 70, with higher scores indicating positive psychological well-being. This scale has a reported internal reliability of 0.91 along with respectable validity and low social desirability bias.22 It is also reported to have a near normal distribution pattern with no ceiling effect22 and we used mean scores obtained from the WEMWBS as advised.23

    Socio-demographic factors

    Age on leaving full-time education, employment status and partnership were all independently included in the model to assess the effects of these factors. All this information was collected at age 42. Age on leaving full-time education was categorised into: aged 16, between ages of 17 and 19 and aged 20 or over. The study participants who were in employment were classified as ‘employed’, and those who were unemployed, in education or training, or were not participating in the labour force at all were classified as ‘not employed’. Study participants were identified as ‘partnered’ if they were living with their partner or spouse at the time of the survey. Those who were not living with a partner or spouse were classified as ‘not partnered’.

    Previous psychological health

    This variable was included to avoid the introduction of selection bias (poor mental health resulting in a reduction of contact with friends or relatives) into the study and was indicated by responses on the Malaise Inventory, administered at age 42. A total score of 7 or more24 was used as a cut-off point to indicate that a study participant was in poor psychological health.

    Statistical analyses

    The distribution of the network size by network composition (relatives and friends), socio-demographic factors and previous psychological health at age 42 was assessed. Differences in mean scores for adult psychological well-being stratified by network size and composition were tested using analysis of variance. Trend and post-hoc tests were also used to examine possible associations between the size of the networks and adult psychological well-being.

    Ordered logistic regression was used to examine the relationship of social network size with age on leaving full-time education, partnership and employment status at age 42.

    Longitudinal relationships between psychological well-being and the size of kinship networks (composed of relatives) and friendship networks (composed of friends) were examined using multiple linear regression. The final model included all factors, including psychological health at age 42. This factor was included to minimise the introduction of selection bias into the results of the study. Men and women were analysed separately, and all analyses were performed using Stata V.12.25

    Results

    In general, friendship networks tended to be larger than kinship networks in men and women (table 1). Just over 15% of the participants reported having no contacts with relatives or no living relatives outside of their household, whereas fewer than 11% of the participants reported that they had no friends. Forty per cent of the men and 32% of the women reported that they had more than six friends whom they saw on a regular basis. The majority of them had left full-time education at the age of 16, were in employment, were partnered and had good psychological health at the age of 42.

    View this table:
    Table 1

    Distribution of the main variables in the analyses

    The relationship between socio-demographic factors and social network size

    Table 2 shows the associations between socio-demographic factors and the size of social networks according to the composition of networks (relatives or friends), using ordered logistic regression. Education and partnership are associated with the size of kinship networks, while education is the only factor associated with the size of friendship networks. Employment status was not significantly associated with the size of either kind of social network in either men or women.

    View this table:
    Table 3

    Means (SDs) of psychological well-being at age 50 by the size and composition of social networks and gender

    The patterns of association between socio-demographic factors and the size of social networks were also assessed according to the composition of the networks and the gender of the participants. Men who left full-time education between ages 17 and 19 were 45% less likely to have a larger kinship network, and those who left education after age 20 were 60% less likely to have a larger kinship network. A similar trend was seen in the women: those who left full-time education between ages 17 and 19 were 17% less likely to report a larger kinship network, and those who left full-time education at age 20 or beyond were 60% less likely to do so. Having no partner reduced the probability of being in a larger kinship network by 31% for men and 26% for women.

    Education is the only factor associated with the size of the friendship networks; however, the direction of the association for men is opposite to that for women. In men, a significant but relatively small negative association was found between leaving full-time education between ages 17 and 19 and the size of their friendship networks (OR=0.84, 95% CI 0.72 to 0.99). However, for women, education was strongly and positively associated with the size of their friendship networks. Women who left full-time education between ages 17 and 19 increased the odds of having a larger friendship network by 38% and by 74% if they left full-time education after age 20.

    Longitudinal association between social network size and psychological well-being

    Results of an analysis of variance showed significant overall differences in mean WEMWBS scores according to the network size for both relatives and friends in the participants (table 3); however, the effect size was smaller in women. A trend test confirmed an increase in psychological well-being according to the size of kinship networks (men: Z=4.16, p<0.0001; women: Z=2.85, p=0.04) and friendship networks (men: Z=8.16, p<0.0001; women: Z=10.18, p<0.0001), supporting positive and crude associations between the size of social networks and psychological well-being.

    A post-hoc test further showed that detailed relationships between social networks and psychological well-being according to composition of the social networks. Mean scores for the WEMWBS according to the size of kinship networks were only significant (p<0.05) if comparisons were made between the next-to-neighbouring categories, for example, comparing the category ‘none’ against the category ‘3–5’. In contrast, apart from the pair ‘6–10’ and ‘over 10’, paired group mean WEMWBS scores were significantly (p<0.05) different from each other in friendship networks. These findings are the same for men and women.

    Table 4 shows the results of multiple linear regression testing a longitudinal relationship between network size and psychological well-being at age 50. In comparison to the largest network size (those with more than 10 regular contacts), smaller friendship networks at age 45 were associated with lower levels of psychological well-being in both men and women at age 50.

    View this table:
    Table 4

    Linear regression coefficients for models of psychological well-being at age 50 on social network sizes at age 45 stratified by gender

    Having fewer than five friends at age 45 predicted significantly poorer psychological well-being at age 50, even after the effects of socio-demographic factors and previous psychological health were taken into account. In men, psychological well-being at age 50 was lower if they had regular contacts with fewer than 10 relatives at age 45. In women, there was no association of psychological well-being to the size of the kinship networks.

    Psychological well-being were particularly poor among those who were socially isolated (had no relatives or no friends). Compared with those who had the largest (over 10) social networks, men's psychological well-being was 2.3 points lower if they had no relatives and 2.6 points lower if they had no friends. In the case of women, having no friends reduced their psychological well-being at age 50 by almost 4 points compared with those who had more than 10 friends. These effects were independent of the effects of socio-demographic factors and previous psychological health at age 42.

    Difference in the association between social network size and psychological well-being by the composition of networks

    A post-hoc Wald test showed a significant overall association between friendship networks and adult psychological well-being for men (F=12.95, p<0.001) and women (F=18.34, p<0.001). The association between kinship network and psychological well-being was considerably smaller for men (F=3.41, p=0.009) and was not significant for women (F=1.08, p=0.37). Friendship networks appear to have much stronger association with psychological well-being than kinship networks.

    Gender differences in the associations between socio-demographic factors, network size and psychological well-being

    Gender differences in the association of social networks and socio-demographic factors with adult psychological well-being at age 50 were also tested. Interaction terms between all the independent variables and gender were created and added to the final model individually. The associations between the sizes of both kinds of network (relatives and friends) and psychological well-being were not significantly different between men and women.

    In contrast, there were significant gender differences in education and partnership with psychological well-being. These gender differences remained significant in the fully adjusted model. Linear prediction of psychological well-being showed that women who left education between the ages of 17 and 19 (figure A1) or were without a partner (figure A2) had higher psychological well-being than their male counterparts in the same response categories.

    Figure A1
    Figure A1

    Predicted psychological well-being score at age 50 by education status and gender (based on fully adjusted model).

    Figure A2
    Figure A2

    Predicted psychological well-being score at age 50 by partnership status at age 42 and gender (based on fully adjusted effects).

    Discussion

    Kinship and friendship networks had significant longitudinal relationships with men's psychological well-being, while only friendship network was longitudinally associated with women's psychological well-being. To our knowledge, this is the first population study to have longitudinally examined the relationship between the size and composition of social networks and mid-adult psychological well-being, independently of socio-demographic factors and prior psychological health.

    Berkman and Syme1 showed that the effect of social networks on mortality was related to individual socioeconomic status. Extending their work, we further examined how socio-demographic factors are associated with social network size, according to the network composition.

    We did not expect to find a negative association between education and the size of kinship networks. It is possible that large kinship networks composed of siblings can be a proxy for early social disadvantages.26 However, we would only be able to test this assumption directly if we had information about individual network members' biological or social relationships to the cohort participants.

    For women only, education was positively associated with the size of friendship networks in our study. Employment status was not significantly associated with the size of social networks among either men or women. These findings are somewhat similar to Moore's27 study, which found an association between education and social networks and no significant associations between employment and networks of kin or friends.

    Consistent with a previous study, showing that the effects of social relations on psychological health were gender specific,28 we also found that the positive longitudinal association between kinship network size and psychological well-being at age 50 was limited to men only. The size of the kinship networks, composed by relatives who were not living with the cohort participants, was associated with their partnership status. It is possible that negative social exchanges within women's social ties might have reduced any positive effects from kinship networks built upon their partnership as these have been found to be related to depression.29 The association between partnership status and psychological well-being was present among men only. However, further analyses showed that values of psychological well-being were actually higher in women, especially those who were not partnered, when compared with men. Therefore, the partnership status of women is unlikely to have modified the effects of their kinship networks. We instead suggest that kinship networks may have a different relationship to psychological well-being for men and women regardless of network size.

    In this study, information about social networks is limited to composition and size as were unable to include other characteristics of social networks such as homogeneity of the network members or physical distance.30

    The previously reported link between friendship networks and psychological well-being among older adults31 and the participants in the NCDS32 are cross sectional. The longitudinal data available for this analysis allowed us to distinguish social networks according to both their composition (kinship or friendship) and their size. We adjusted for the effects of education, employment status, marital status and previous psychological health, which might have influence both network size and psychological well-being. Our finding, which shows significant independent longitudinal associations between large friendship networks and psychological well-being among middle-aged adults, adds new empirical support to the existing research evidence.

    Conclusions

    A significant association between having a friendship network and adult well-being was found, which was independent of employment, education and partnership among middle-aged adults. Men benefit from having an extended kinship network, which is associated with partnership formation. Education was positively associated with the size of women's friendship networks. Policies to support women in pursuit of further education can in turn assist in reducing the well-known gender gap in mental health through women's friendship networks.

    What is already known on this subject

    Previous studies indicated a cross-sectional link between social networks and health. The link between social networks and mortality is suggested to be independent of socioeconomic status.

    What this study adds

    Middle-aged adults psychologically benefit from being well integrated into friendship networks. Partnership is associated with large kinship networks, yet only men psychologically benefit from having an extended kinship network. These findings are over and above socio-demographic factors and previous psychological health.

    View this table:
    Table 2

    OR and 95% CI in parentheses for the ordered logistic regression models of network size by socio-demographic factors, stratified by network composition (relatives and friends) and gender

    Acknowledgments

    We are grateful to the Centre for Longitudinal Studies, Institute of Education for the use of these data and to the UK Data Archive and Economic and Social Data Service for making them available. However, they bear no responsibility for the analysis or interpretation of these data.

    References

    1. ↵
      1. Berkman LF,
      2. Syme SL
      . Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents. Am J Epidemiol 1979;109:186–204.
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. House JS,
      2. Robbins C,
      3. Metzner HL
      . The association of social relationships and activities with mortality: prospective evidence from the Tecumseh Community Health study. Am J Epidemiol 1982;116:17.
      OpenUrl
    3. ↵
      1. Kaplan GA,
      2. Salonen JT,
      3. Cohen RD,
      4. et al
      . Social connections and mortality from all causes and from cardiovascular disease: prospective evidence from eastern Finland. Am J Epidemiol 1988;128:10.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Berkman LF,
      2. Melchior M,
      3. Chastang JF,
      4. et al
      . Social integration and mortality: a prospective study of French employees of electricity of Frans-Gas of France The GAZEL Cohort. Am J Epidemiol 2004;159:167–74.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Pinquart M,
      2. Duberstein PR
      . Associations of social networks with cancer mortality: a meta-analysis. Crit Rev Oncol Hematol 2010;75:122–37.
      OpenUrlCrossRefPubMed
    6. ↵
      1. Melchior M,
      2. Berkman LF,
      3. Niedhammer I,
      4. et al
      . Social relations and self-reported health: a prospective analysis of the French Gazel cohort. Soc Sci Med 2003;56:1817–30.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Rosengren A,
      2. Wilhelmsen L,
      3. Orth-Gomér K
      . Coronary disease in relation to social support and social class in Swedish men. Eur Heart J 2004;25:56–63.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Everson-Rose SA,
      2. Lewis TT
      . Psychosocial factors and cardiovascular diseases. Annu Rev Pub Health 2005;26:469–500.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Mueller DP
      . Social networks: a promising direction for research on the relationship of the social environment to psychiatric disorder. Soc Sci Med 1980;14A:147–61.
      OpenUrl
    10. ↵
      1. Achat H,
      2. Kawachi I,
      3. Levine S,
      4. et al
      . Social networks, stress and health-related quality of life. Qual Life Res 1998;7:735–50.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Holt-Lunstad J,
      2. Smith TB,
      3. Layton JB
      . Social relationships and mortality risk: a meta-analytic review. PLoS Med 2010;7:e1000316.
      OpenUrlCrossRefPubMed
    12. ↵
      1. Bergelt C,
      2. Prescott E,
      3. Gronbaek M,
      4. et al
      . Social ties and risk for cancer–a prospective cohort study. Acta Oncol 2009;48:1010–18.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Henderson AS
      . Interpreting the evidence on social support. Soc Psychiatry 1984;19:49–52.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Brugha TS,
      2. Weich S,
      3. Singleton N,
      4. et al
      . Primary group size, social support, gender and future mental health status in a prospective study of people living in private households throughout Great Britain. Psychol Med 2005;35:705–14.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Hammer M
      . ‘Core’ and ‘extended’ social networks in relation to health and illness. Soc Sci Med 1983;17:405–11.
      OpenUrlCrossRefPubMed
    16. ↵
      1. Fuhrer R,
      2. Stansfeld SA,
      3. Chemali J,
      4. et al
      . Gender, social relations and mental health: prospective findings from an occupational cohort (Whitehall II study). Soc Sci Med 1999;48:77–87.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Berry HL,
      2. Welsh JA
      . Social capital and health in Australia: an overview from the household, income and labour dynamics in Australia survey. Soc Sci Med 2010;70:588–96.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Power C,
      2. Elliott J
      . Cohort profile: 1958 British birth cohort (National Child Development Study). Int J Epidemiol 2006;35:34–41.
      OpenUrlFREE Full Text
    19. ↵
      University of London. Institute of Education. Centre for Longitudinal Studies, National Child Development Study: Sweep 6, 1999–2000 [computer file]. 2nd Edition. Joint Centre for Longitudinal Research, [original data producer(s)]. Colchester, Essex: UK Data Archive [distributor], August 2008. SN: 5578, http://dx.doi.org/10.5255/UKDA-SN-5578-1
    20. ↵
      University of London. Institute of Education. Centre for Longitudinal Studies, National Child Development Study: Biomedical Data, 2002–2004: Special Licence Access [computer file]. Colchester, Essex: UK Data Archive [distributor], January 2009. SN: 5594, http://dx.doi.org/10.5255/UKDA-SN-5594-1
    21. ↵
      University of London. Institute of Education. Centre for Longitudinal Studies, National Child Development Study: Sweep 8, 2008–2009: First Deposit [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], March 2010. SN: 6137.
    22. ↵
      1. Tennant R,
      2. Hiller L,
      3. Fishwick R,
      4. et al
      . The Warwick-Edinburgh Mental Well-being Scale (WEMWBS): development and UK validation. Health Qual Life Outcomes 2007;5:63.
      OpenUrlCrossRefPubMed
    23. ↵
      1. Stewart-Brown S,
      2. Janmohamed K
      . Warwick-Edinburgh Mental Well-being Scale (WEMWBS): User Guide Version 1. In: Parkinson J , ed. Edingburgh: NHS Health Scotland, 2008:1–32.
    24. ↵
      1. Rodgers B,
      2. Pickles A,
      3. Power C,
      4. et al
      . Validity of the Malaise Inventory in general population samples. Soc Psychiatry Psychiatr Epidemiol 1999;34:333–41.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      StataCorp. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP, 2011.
    26. ↵
      1. Schoon I,
      2. Sacker A,
      3. Bartley M
      . Socio-economic adversity and psychosocial adjustment: a developmental-contextual perspective. Soc Sci Med 2003;57:1001–15.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Moore G
      . Structural determinants of men's and women's personal networks. Am Soc Rev 1990;55:726–35.
      OpenUrlCrossRef
    28. ↵
      1. Fuhrer R,
      2. Stansfeld SA
      . How gender affects patterns of social relations and their impact on health: a comparison of one or multiple sources of support from “close persons”. Soc Sci Med 2002;54:811–25.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Cheng ST,
      2. Li KK,
      3. Leung EM,
      4. et al
      . Social exchanges and subjective well-being: do sources of positive and negative exchanges matter? J Gerontol B Psychol Sci Soc Sci 2011;66:708–18.
      OpenUrlPubMedWeb of Science
    30. ↵
      1. Berkman LF
      . Assessing the physical health effects of social networks and social support. Annu Rev Pub Health 1984;5:413–32.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Fiori KL,
      2. Antonucci TC,
      3. Cortina KS
      . Social network typologies and mental health among older adults. J Gerontol B Psychol Sci Soc Sci 2006;61:P25–32.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Ford E,
      2. Clark C,
      3. Stansfeld SA
      . The influence of childhood adversity on social relations and mental health at mid-life. J Affect Disord 2011;133:320–7.
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Press release

      Press release

      Files in this Data Supplement:

    Footnotes

    • Funding This study is a project from the International Centre for Lifecourse Studies in Society and Health, supported by the Economic and Social Research Council (RES-596-28-0001).

    • Competing interests None.

    • Ethics approval The National Child Developmental Study (NCDS) data are in the public domain, not requiring researchers to obtain ethical approval for the academic use of the data. Ethical approvals from the London Multi-Centre Research Ethics Committees were obtained prior to the data collection by the PI's of the National Childhood Developmental Study at the Centre for Longitudinal Studies. An application to use the Biomedical Survey data (the age 45 data) was made to the UK Data Archive, and a special licence to use theses data was granted. Ethical approvals of the sweeps 6 (age 42) and 8 (age 50) of the NCDS were approved by the London Multi-Centre Research Ethics Committees. The ethical approval of the NCDS biomedical survey (age 45) was obtained from the South-East Multi-Centre Research Ethics Committee.

    • Provenance and peer review Not commissioned; externally peer reviewed.