Who is alone? On the characteristics of social isolation and loneliness Beate Völker, Henk Flap & Gerald Mollenhorst Dept of Sociology/ICS Utrecht University.

Who is alone?

On the characteristics of social isolation and loneliness

Beate Völker,

Henk Flap & Gerald Mollenhorst

Dept of Sociology/ICS

Utrecht University

QMSS conference, Prague, June 20-24, 2007 2

…so many of us eke out an existence as loveless and unloved atoms – free individuals in an open society, condemned to form part of the great, grey subculture of the lonely.

Robert Brain, 1976:259


Why studying loneliness - from a sociological perspective?

• Social isolation is the counterpart of social integration (or is it not?)

• Sociologists view personal relations as the mortar of society: – Norms and rules about how to behave are transmitted via social

relations.– Social resources are an important means for achieving many

important individual goals. Isolated people probably lack social resources and are therefore disadvantaged

– a society consisting of isolated members will fall apart and is prone to crumble (see Fischer and Phillips 1982:21)

• Actually: consequences of loneliness in a sociological view go further then individual feelings related to loneliness


Research questions

1. What are the social characteristics of those who are lonely?

2. How is loneliness related with a person’s network? • Which network patterns (size and composition) are

associated with feelings of loneliness? • Do those who feel lonely lack in particular strong ties

or also weaker ties? • And do lonely people also lack social resources?

3. Is there a spatial component to loneliness? Does loneliness differ between neighborhoods?


BackgroundAd 1. Many studies have been conducted on psychological

characteristics of loneliness, e.g. relational standards, role socialization, social comparison processes.

Research into socio-economic characteristics of those who feel lonely is scarce.

Ad 2. We do know how networks help to get a better life, but we lack knowledge about how networks make our life miserable. Consequences of relations that are absent are an important part of the study of social networks.

Ad 3. There are studies showing important differences between cultures concerning feelings of loneliness. These differences are largely unexplained.


Background (2)

• Because of selection problems is the relation between social characteristics – networks and loneliness studied best via a longitudinal design:

Individual socio-economic characteristics

Networks

Loneliness


Background (3)

• There are major concerns about meaning and consequences of small networks (see ‘Social isolation in America’, by Smith Lovin, McPherson and colleagues 2006). – Meaning of ‘zeros’ is in particular unclear because only one namegenerating question and no outcome measures are available.


Arguments (1)

1. On individual level characteristics:• As far as studies do exist it is shown that those who

are socially disadvantaged in general do also feel more lonely. Elderly, those with lower education and income and unmarried people are hence expected to suffer more from loneliness.

Arguments (2)2. On network patterns:

• Given the distinction between social and emotional loneliness, network patterns - are not expected to be perfectly associated with feelings of loneliness (see e.g. Weiss, 1973). We do expect however that loneliness is more experienced among those with few or no strong ties and less among those with few or no weaker ties

• Straightforward argument from social resource/social capital theory: small networks provide less social capital than large networks. Yet: 1) benefits of social relations might decrease at the margin; 2) people might be able to produce wellbeing also via other (material) resources. Hence: relation between network and access to all kin of social resources not quite clear


Arguments (3)

3. On the level of the neighborhood:• Communities with few meeting opportunities are

expected to promote higher feelings of loneliness and vice versa

• Disadvantaged communities are expected to promote higher feelings of loneliness and vice versa

• Heterogeneous communities are expected to promote higher feelings of loneliness and vice versa

• Close communities are expected to promote higher feelings of loneliness for those who are not belonging and vice versa


Data

• SSND1: Survey of the social networks of the Dutch (n=1007 respondents; Völker & Flap 1999) – Detailed overview of people’s personal networks, delineated by

the exchange method– Representative sample of 160 neighborhoods in the Netherlands

• SSND2, 2007: second wave among same respondents, n=604; – Many network questions are the same as in 1999– measurement of loneliness according to Jong Gierveld en van

Tilburg (1999)

Measuring loneliness (SSND2)

• There is always someone available on who I can rely with my daily difficulties• I miss a really good friend• I experience a kind of emptiness around myself• There are enough people to lean on in case of problems• I miss companionship • I think my circle of acquaintances is too restricted• I know many others who I can trust completely• There are enough others to whom I feel really close• I miss people around me• Often I feel just left for my own• If I need anybody, I have always friends to talk with

Cronbach’s alpha = .87


• How got current/last job?

• Advice asking/providing in case of problems at work?

• Having trouble with somebody?

• Working together

• Who is your boss?

• How got your current house?

Measuring networks (SSND1 and 2) Exchange method:

name generators – partly standard, partly focusing on own research questions; step 1:

• Who has keys to your house?

• Small repairs in and around the house?

• Visiting

• Discussing personal matters

• Who are your direct neighbors?

• Open question

Step 2: Characteristics of alters and the relationship ego-alter

• Characteristics of alter:– Sex, age, education, occupation, having a paid job, family

situation , religion, – role relation with ego (15 categories, 3 different roles could be

mentioned)• Characteristics of the relationship ego- alter:

– Degree of intensity, trust and liking (5-point-scale)– Duration of relationship (years/months)– Frequency of contact (6 categories)– Geographical distance (5 km)– Still relation in about 5 years?– Where did you meet first? (13 different settings)– Where do you meet currently?

Measuring network resources:Position Generatorfamily friend acquaintance

Doctor/physician 19 10 20

Cook 16 11 18

Engineer 36 13 15

Manager 34 18 13

Real estate manager

6 6 18

Lawyer 16 11 19

IT expert 28 18 20

Musician/artist 21 17 16

Scientist 19 12 11

Nurse 39 16 20

Machinist 8 3 8

Unskilled worker 15 6 15

Note: in the SSND about 30 different positions have been presented to respondents. The positions are coded with their socioeconomic status and their occupational prestige

Loneliness and network measures

SOCIAL NETWORKS (SSND1 AND 2):• Network size • Composition (e.g. friends, kin, partner, neighbors etc.)• Strength: core ties vs. other ties• Resources: position generator (Lin and Dumin 1986)

LONELINESS:• SSND1: proxy for loneliness: ‘ I would like to have more friends’• SSND2: Loneliness scale

• Note: it is analyzed whether network size and the proxy for loneliness in 1999 influenced participation in the survey; no association has been found

The Survey of the Social Networks of the Dutch (SSND) - Data Collection


Analysis:Loneliness

• Scale range from 11 – 55 (higher values indicate higher feelings of loneliness)

• About two third of the respondents have a value of 22 or lower, that is the lower half of the scale

• About one third feels – according to that criterion – alone

• Note: the real degree of loneliness might be underestimated because the interviews are face-to-face interviews (see De Leeuw, 1992)

Who is alone? Model1: socio-dem.

Char

Model2: added: network

parameters

Model 3: Added: social capital

Sex (male) .498 (.409) -.247 (.406) -.088 (.408)

Age .035 (.020)+ .030 (.019) .026 (.019)

Education -.202 (.091)* -.120 (.090) -.108 (.090)

Being Dutch -2.528 (.702)** -2.047 (.676)** -2.234 (.677)**

Number of children .312 (.163)+ .227 (.165) .266 (.145)+

Being married -.812 (.422)+ -.904 (.452)* -.810 (.421)+

Having a paid job -1.639 (.494)** -1.219 (.496)** -1.196 (.495)*

Degree of urbanization -.008 (.149) -.041 (.146) -.006 (.146)

Core network size 99 -- -.415 (.108)*** -.359 (.111)***

Other network members 99 -- .043 (.051) .061 (.051)


Wish to have more friends ’99 1.823 (.321)** 1.864 (.321)*** 1.811 (.321)***

Resources ’99 -- -- -.045 (.040)

Resources ’07 -- -- -.076 (.044)+

Intercept 24.774 (1.368) 22.842 (1.552) 23.946 (1.594)

Explained (adjusted R2) .055 .16 .17


Who is alone?

See table, column 1 Those who are:• Older, • Lower educated• Not married• Having no paid work and• Foreigners (parents not born in the Netherlands)

feel more lonely

• Note: number of children (as well as having or not having children), rather increases than decreases feelings of loneliness

• No effect of urbanization

What are the network correlates of loneliness ?

Model1: socio-dem.

Char


parameters

Model 3: Added: social capital

Sex (male) .498 (.409) -.247 (.406) -.088 (.408)

Age .035 (.020)+ .030 (.019) .026 (.019)

Education -.202 (.091)* -.120 (.090) -.108 (.090)

Being Dutch -2.528 (.702)** -2.047 (.676)** -2.234 (.677)**


Being married -.812 (.422)+ -.904 (.452)* -.810 (.421)+

Having a paid job -1.639 (.494)** -1.219 (.496)** -1.196 (.495)*






Resources ’99 -- -- -.045 (.040)

Resources ’07 -- -- -.076 (.044)+

Intercept 24.774 (1.368) 22.842 (1.552) 23.946 (1.594)



What are the network correlates of loneliness

See table, column 2 - Core discussion networks in 1999 as well as in 2006 predict

(negatively) loneliness- No effect of weaker relationships on loneliness

- ‘Wish to have more friends’ is also a predictor of loneliness

- Note: further analyses show that in particular partner and friends are the members of the core network whose presence decreases feelings of loneliness; kin and other relationships have less or no impact

- Correlation between core network size in 1999 and 2007 only .27- Average core network size: 2.32 (sd=1.85) and 2.43 (sd=2.21)

How is loneliness related with network resources?Model1: socio-dem.

Char


parameters

Model 3: Added: social

capital

Sex (male) .498 (.409) -.247 (.406) -.088 (.408)

Age .035 (.020)+ .030 (.019) .026 (.019)

Education -.202 (.091)* -.120 (.090) -.108 (.090)

Being Dutch -2.528 (.702)** -2.047 (.676)** -2.234 (.677)**


Being married -.812 (.422)+ -.904 (.452)* -.810 (.421)+

Having a paid job -1.639 (.494)** -1.219 (.496)** -1.196 (.495)*






Resources ’99 -- -- -.045 (.040)

Resources ’07 -- -- -.076 (.044)+

Intercept 24.774 (1.368) 22.842 (1.552) 23.946 (1.594)



How is loneliness related with network resources, i.e. social capital?

See table, column 3 - Resources predict loneliness (negatively) at both points

of measurements- Correlation between resources in 1999 and 2007 is .50- If resources are separately in the analysis, coefficients of

resources in 2006 are higher (b=-.102; se=.035) than those in 1999 (b=-.079; se=.035)


lonely

no Yes

Core network

yes 65.3 34.7

No 50.0 50.0

lonely

No Yes

Core network > 1

Yes 68.0 32.0

No 59.8 40.2

N=519

OR:1.88

OR:1.43

Relation between core network size in 1999 and loneliness in 2006

N=597


lonely

No Yes

N of weaker network relations

Many 63.4 36.6

Few 63.5 36.5

OR:0.998

Relation between number of weaker ties in 1999 and loneliness in 2006

N=597

Multinomial logistic regression on having a core network and loneliness:

• Having/not having a core network and feeling lonely/ or not:– Compared to those who have a core network and do not

feel lonely (the happy ones): • those without a core/and who do feel lonely have

fewer resources, are more often male, and lower educated.

• those without a core and who feel not lonely have also fewer resources, are lower educated but live in rural areas

• those with a core network nevertheless feel lonely have often a good job, are Dutch, and live in larger cities

Multinomial logistic regression on having many core and other network members and

loneliness:

• Results are largely the same if the dependent variable is calculated with size of core network (relatively large vs. relatively small), with one exception: those who have children, have often a small core network and feel lonely

• If the dependent variable is calculated using size of the network without core ties, results show – a clear effect of having children on feeling not lonely and having

larger networks compared to all categories – women are more often those with a small network but without

feelings of loneliness as well as those with as good job, and those who are married

No effect of social resources in the multinomial analyses


Is there a spatial dimension of loneliness?

• Roughly 11% of the variation in loneliness is due to neighborhood differences

• Data have been enriched with neighborhood information obtained from the CBS ‘kerncijfers wijken en buurten’

• Analysis explains roughly 70% on the neighborhood level and 20% on the individual level

the spatial dimension of loneliness

Meeting

Length of residence -.076 (.054)~

Facilities for children 1.315 (.643)*

Cultural facilities -.484 (.443)

Price of houses -.003 (.001)**

N of house-owners -1.969 (.868)**

Prop. of foreigners -.050 (.032)+

Average age .074 (.054)

Heterogeneity Homogeneity of family situation

-1.193 (.504)**

Closure Embeddedness of neighborhood

2.517 (.947)**

Disadvantaged

Note: it is controlled for individual as well as network characteristics

• In a multilevel regression analysis the following results have been established: individual loneliness is lower in neighborhoods where– people have been living there for long – few facilities for children exist (??)– the neighborhood is rich (owners and expensive

houses)– neighborhood homogeneity with regard to household

composition is high – and embeddedness is high

• If loneliness in neighborhoods is added to the analysis many effects disappear, because the effect is that strong


Conclusion (1)• As to the social characteristics of loneliness: those who are older,

lower educated and not from the Netherlands feel more lonely• Loneliness is not related to weaker ties, but before all to core

discussion ties• Number of core ties is a good predictor of loneliness• Yet: There are also people with few/no core relations (and other

relationships) who do not feel lonely and vice versa. Females, those in rural areas, with children, and no job feel often not lonely, although they have small networks. Dutch, with good jobs living in cities feel more often lonely although they do have a network.

• Composition of core: presence of friends and partner make for little feelings of loneliness

• Effect of having children on loneliness deserves more attention• Effect of network at t1 is stronger than effect on t2. Both processes -

selection as well as effect of networks on loneliness – seem to be relevant, yet the latter one is stronger


Conclusion (2)

• Social resources are also related to feelings of loneliness, yet to a lesser degree than core ties. Social resources are in particular relevant at t2.

• There is also a clear spatial dimension to loneliness: length of residence, embeddedness in the neighborhood, degree of disadvantage and neighborhood composition all explain individual loneliness are of importance

Thanks for your attention!More information?

Mail to: [email protected]

Who is alone? On the characteristics of social isolation and loneliness Beate Völker, Henk Flap & Gerald Mollenhorst Dept of Sociology/ICS Utrecht University.

Documents

feelings of loneliness

consequences of loneliness

social resources

social relations

qmss conference

study of social networks

lonely lack

loneliness beate vlker