Content uploaded by Audrey U Swift
Author content
All content in this area was uploaded by Audrey U Swift on Aug 25, 2014
Content may be subject to copyright.
http://spr.sagepub.com
Relationships
Journal of Social and Personal
DOI: 10.1177/0265407509106718
2009; 26; 273 Journal of Social and Personal Relationships
U. Swift and Joelle C. Ruthig
Nancy E. Newall, Judith G. Chipperfield, Rodney A. Clifton, Raymond P. Perry, Audrey
longitudinal study
Causal beliefs, social participation, and loneliness among older adults: A
http://spr.sagepub.com/cgi/content/abstract/26/2-3/273
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of:
International Association for Relationship Research
can be found at:Journal of Social and Personal Relationships Additional services and information for
http://spr.sagepub.com/cgi/alerts Email Alerts:
http://spr.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.co.uk/journalsPermissions.navPermissions:
http://spr.sagepub.com/cgi/content/refs/26/2-3/273 Citations
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Causal beliefs, social participation,
and loneliness among older
adults: A longitudinal study
Nancy E. Newall, Judith G. Chipperfield, Rodney A. Clifton,
Raymond P. Perry, & Audrey U. Swift
University of Manitoba
Joelle C. Ruthig
University of North Dakota
ABSTRACT
Loneliness can be seen as a social failure subject to causal
search: Why am I lonely? Why do I lack friends? According to
attribution theory, answers to these questions can influence
emotions, motivation, and behaviours. This study examined
the relationships between various affiliative causal beliefs (i.e.,
beliefs about loneliness and friendship development), social
participation, and loneliness among older adults (72+ years).
Cross-sectional and longitudinal (over five years) results
showed that more strongly endorsing internal/controllable
causal beliefs (i.e., believing that making friends depends on
effort) related to greater social participation. Moreover, greater
social participation related to less loneliness. External/un-
controllable causal beliefs predicted greater loneliness. In fully
addressing loneliness, it may be important to focus on people’s
causal beliefs.
KEY WORDS: attribution theory • causal attributions • health •
loneliness • social participation
Journal of Social and Personal Relationships Copyright © 2009 SAGE Publications
(www.sagepublications.com), Vol. 26(2–3): 273–290. DOI: 10.1177/0265407509106718
This research was supported by a Canadian Institutes of Health Research (CIHR) Canada
Graduate Scholarships Doctoral Award to the first author, and a CIHR operating grant
(MOP-64335) and Mid Career Award in Aging to the second author. Preliminary findings of
this research were presented in a poster at the Annual Scientific and Educational Meeting
of the Canadian Association on Gerontology, Victoria, B.C., Canada, October 2004. Address
correspondence to Nancy E. Newall, Department of Psychology, University of Manitoba,
Winnipeg, Manitoba, Canada, R3T 2N2 [e-mail: N_Newall@umanitoba.ca]. Duncan Cramer
was the Action Editor on this article.
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Loneliness is commonly conceived of as a feeling resulting from perceived
deficits in social relationships (Dykstra & Fokkema, 2007).These deficits can
appear to take on an almost physiological form as lonely individuals often
describe themselves as feeling empty inside (e.g., Schultz & Moore, 1984;
Weiss, 1973). Loneliness is a negative experience in its own right. Further-
more, research focusing on older adults has shown that loneliness is related
to other negative outcomes such as poor health (e.g., Russell, 1996) and
depression (e.g., Cacioppo, Hughes, Waite, Hawkley, & Thisted, 2006).These
findings emphasize the importance of identifying and understanding the
causes of loneliness in older adults in order to develop intervention strat-
egies aimed at reducing loneliness.
The percentage of older adults reporting moderate loneliness appears to
range from 20%–40% (Pinquart & Sorensen, 2001; Weeks, 1994; Wenger &
Burholt, 2004). In considering research which has focused on age differences
in the prevalence of loneliness, it appears that older adults may be no more
lonely than younger adults (e.g., Green, Richardson, Lago, & Schatten-Jones,
2001; Schulz & Moore, 1988). In their meta-analysis of correlates of loneli-
ness in older adults, Pinquart and Sorensen (2001) found that the relation
between age and loneliness appeared to be U-shaped, such that loneliness
decreased with age for the youngest subgroup (age < 60 years), was un-
related to age for the next oldest subgroup (between 60.1–80 years), and
increased with age for the oldest subgroup (age > 80.1). Therefore, the
oldest may be especially vulnerable to loneliness and this possibility under-
scores the importance of studying the development of loneliness.
Loneliness has been studied from various perspectives. The present study
examined a representative sample of older adults to consider whether lone-
liness develops in part from the way people think about their affiliations
with others. In particular, we examined whether older adults’ explanations
about their affiliations with others (e.g., friendship development) predicted
social participation in activities, as well as their current level of loneliness and
their subsequent loneliness five years later. This focus on the role of cogni-
tive factors in understanding loneliness follows past research on the relation-
ship between loneliness and causal beliefs or attributions (e.g., Anderson,
Horowitz, & French, 1983; Peplau & Caldwell, 1978) as well as perceptions
of control or locus of control (e.g., Moore & Schulz, 1987; Solano, 1987).
Causal beliefs and loneliness
Although Weiner’s attribution theory (e.g., Weiner, 1985) has been devel-
oped and applied most extensively in the academic achievement domain
(e.g., Menec et al., 1994; Perry, Hechter, Menec, & Weinberg, 1993; Ruthig,
Perry, Hall, & Hladkyj, 2004), it also offers a particularly useful framework
for studying causal beliefs in the affiliation domain.This theoretical perspec-
tive is based on the premise that people generally want to understand why
events happen in their lives. For example, people who are lonely may seek
to find an explanation for their loneliness or lack of satisfying relationships.
274 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
According to attribution theory (e.g., Weiner, 1985), the types of explana-
tions that individuals generate can have profound effects on their subsequent
emotions, expectations, behaviours, and motivation. Proposing a cognition–
emotion–action sequence, Weiner (1985) argues that all perceived causes
can be characterized in terms of three dimensions: locus of causality (internal
vs. external), controllability (uncontrollable vs. controllable), and stability
(unstable vs. stable). Each dimension predicts unique emotions, which, in
turn, guide actions.Thus, attribution theory focuses on the underlying dimen-
sions of attributions, rather than specific attributions per se.
Peplau, Perlman, and colleagues proposed that loneliness reflects a dis-
crepancy between individuals’ desired and achieved social networks (Peplau
& Caldwell, 1978; Perlman, 2004). Thus, loneliness can be precipitated by
changes in a person’s achieved social contacts (e.g., widowhood, moving)
or by changes in a person’s desired social contacts (e.g., through social
norms) (Peplau, Russell, & Heim, 1979). Peplau et al. (1979) also argued
that understanding the causes for loneliness can help people regain control
over their social relationships so as to maintain a balance between achieved
and actual social relations. Using Weiner’s theoretical framework, they put
forth that the types of explanations given for loneliness might greatly affect
emotional reactions to loneliness as well as the enactment of coping beha-
viours (Peplau et al., 1979). In terms of emotional reactions, they noted that
embarrassment would follow from people believing that they are lonely
because they are unattractive (an internal and uncontrollable cause). On the
other hand, blaming others for loneliness (external and controllable cause)
might result in feelings of anger towards others. Furthermore, in terms of
coping behaviours, Peplau et al. argued that people may be motivated to
overcome their loneliness only if they perceive that it is due to personally
controllable reasons such as lack of effort. Thus, framed within the discrep-
ancy model of loneliness, high perceived controllability may imply a confi-
dence in the ability to improve actual social networks so that they match
desired social networks (we would like to thank an anonymous reviewer
for this suggestion).
The importance of the controllability dimension in causal attributions is
highlighted by Anderson and his colleagues’ work on loneliness and attri-
butional style among college students (e.g., Anderson & Arnoult, 1985;
Anderson et al., 1983; Anderson & Riger, 1991). Their research shows that
the controllability dimension is most commonly associated with what they
refer to as “problems in living”, namely loneliness, depression, and shyness.
That is, people who are lonely are likely to see their interpersonal failures
as being due to uncontrollable causes. Based on this work, Anderson and
his colleagues developed the Controllability Attributional Model (CAM)
which predicts that attributing failures to uncontrollable causes leads to low
success expectancies, negative affect, and low motivation (Anderson &
Riger, 1991).
However, it should be noted that some researchers have speculated that,
in people who have tried and failed to establish relationships, blaming the
self (i.e., self-blame, making an internal and controllable attribution) may
Newall et al.: Loneliness among older adults 275
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
lead to greater feelings of shame and social withdrawal as compared to
blaming others or blaming circumstances beyond control (making an
external and uncontrollable attribution; see Peplau et al., 1979). Related to
this, Perlman (2004) noted that socially-isolated individuals may avoid
feeling lonely by attributing their isolation to uncontrollable external forces.
Notably, these studies focus on the emotional effects (i.e., negative affect,
intensity of loneliness) of attributions in people who are chronically lonely
or socially isolated. In contrast, our study focuses more on the motivational
effects of general causal beliefs about friendship development and lone-
liness. Thus, this may explain different expected relationships between the
beliefs and loneliness.
The present study
The primary objective of this study was to examine older individuals’ causal
beliefs about affiliation as immediate and long-term predictors of loneliness.
In particular, we examined whether older adults’ explanations about their
affiliations with others (e.g., friendship development) predicted social parti-
cipation and subsequent loneliness. Because attributing interpersonal failures
to uncontrollable causes is associated with loneliness and low motivation in
the affiliative domain (Anderson & Riger, 1991), it was hypothesized that
external/uncontrollable causal beliefs would predict lower social partici-
pation and greater loneliness, whereas internal/controllable affiliative causal
beliefs would predict greater social participation and lower loneliness. This
conforms to the logic that an external/uncontrollable belief (e.g., believing
that friends are made through luck) is de-motivating, resulting in a passive
approach to friendship development or social contacts and, in turn, foster-
ing loneliness. In contrast, a more proactive approach to making friends
may be taken by someone who believes that making friends is caused by
one’s own efforts (a controllable/internal attribution), which could then
lead to less loneliness.
Our approach is an improvement over past studies that have typically
examined the role of causal beliefs and loneliness without considering the
contribution of other predictor variables. In order to determine whether or
not causal beliefs predicted social participation and loneliness above and
beyond various demographic and health variables, we specified structural
equation models in which these background variables, together with causal
beliefs, predicted social participation, and, in turn, social participation pre-
dicted loneliness (see Figure 1). We examined this model using both cross-
sectional and longitudinal data.
Age was included in the models to capture possible age-related factors,
although it may be widowhood or physical incapacity rather than age per se
that influences loneliness (e.g., Perlman, 2004; Pinquart & Sorensen, 2001).
Gender was also included because it has been generally been found that
older women are more likely to be lonely than older men (e.g., Jylha, 2004;
Pinquart & Sorensen, 2001); however, this may be due to an artifact of how
276 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
loneliness is measured, resulting in women’s greater willingness to admit to
feeling lonely (Perlman, 2004) or to factors that covary with gender (e.g.,
women being older in age, living alone, or in poorer health). Education level
was included to account for differences in socioeconomic status that relate
to social isolation (Wenger, Davies, Shahtahmasebi, & Scott, 1996) and
may provide more available resources and social opportunities that could
prevent loneliness (Pinquart & Sorensen, 2001). Living alone, versus living
with others, was also considered because it is an important risk factor for
loneliness in older adults (e.g., De Jong Gierveld, 1987; Havens, Hall,
Sylvestre, & Jivan, 2004). Finally, health was included in the models because
it consistently correlates with loneliness (e.g., Wenger et al., 1996), although
the association may become weaker at older ages (Dykstra, Van Tilburg,
& De Jong Gierveld, 2005; but see Hawkley & Cacioppo, 2007). Recent
research shows that loneliness may directly influence health through its
negative effect on immune response (Pressman et al., 2005; Hawkley &
Cacioppo, 2003), although the nature of the relationship between health
and loneliness is likely reciprocal (e.g., Fees, Martin, & Poon, 1999).
Methods
Participants
Aging in Manitoba (Canada) study. Participants were community-dwelling
individuals who took part in the Aging in Manitoba (AIM) longitudinal
studies that have been ongoing for over 35 years. The initial 1971 survey was
conducted by the Manitoba Provincial Department of Health and Social
Development in order to identify the needs of older persons (Mossey,
Havens, Roos, & Shapiro, 1981). A random sample of Manitobans (65+
years) was drawn from the Manitoba Health population registry, and strat-
ified by place of residence (i.e., community or personal care home) and
region. A total of 4,803 individuals participated in the initial 1971 AIM
study. Using similar techniques, two additional cross-sectional samples of
older individuals (60+ years) were selected in 1976 (N = 1,302) and again in
1983 (N = 2,877). Since then, there have been several follow-up data collec-
tion waves (in 1990, 1996, 2001, 2005, and 2006) in which sociodemographic,
health, and psychosocial information has been obtained via face-to-face
Newall et al.: Loneliness among older adults 277
Social Participation Loneliness
Sociodemographics
Health
Causal Beliefs
FIGURE 1
Theoretical model
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
interviews in the participants’ homes. A loneliness scale was included in the
interview schedule for the first time in 1996. Our study complements other
Aging in Manitoba research studies on the social and health factors related
to loneliness and social isolation in older Manitobans (e.g., Havens et al.,
2004).
Obtaining representative samples of older Manitobans was important to
the objectives of the AIM study. The 1971 sample was representative in
terms of age and gender of the Manitoba population aged 65 and over
(Mossey et al., 1981). Furthermore, the AIM 1990 sample was comparable
to the older Manitoba population in 1990 in terms of gender and marital
status (Chipperfield, Havens, & Doig, 1997) and generally representative in
terms of age, although the oldest age category (90+ years) was over repre-
sented and the youngest age category (70–74 years) was underrepresented
(Chipperfield et al., 1997).
Sample selection. The present study’s cross-sectional sample (N = 1243)
was taken from the AIM 1996 sample of 1,868 individuals to include only
those respondents who: (1) were community-dwelling (i.e., not living in a
personal care home); and (2) had complete data for loneliness and causal
beliefs (the main variables of interest). Participants living in a personal care
home (n = 255) and/or who had incomplete data (n = 370) were excluded,
leaving a final sample of 1,243 participants. Data were incomplete due to
proxy respondents completing the interview who did not answer the subjec-
tive questions (n = 132), participants being interviewed over the phone and
given a shorter interview (n = 6) and due to participants refusing to answer
the questions or providing invalid (e.g., don’t know) responses (n = 232).
Some of these latter participants with missing data had required consider-
able help from a proxy respondent to complete the interview (n = 77).
The longitudinal (i.e., AIM 1996–2001) sample included individuals (N =
688) from the 1996 cross-sectional sample who: (1) also participated in the
2001 study; and (2) had completed the AIM 2001 loneliness scale. Out of the
1,243 individuals from the 1996 cross-sectional data, 798 also participated in
AIM 2001. The reasons for non-participation in AIM 2001 included death,
re-location to another province, hospitalization or being too ill, and refusal.
Of the individuals who had completed both surveys, 110 people had incom-
plete 2001 data, leaving a total of 688 individuals in the longitudinal sample.
Data were incomplete due to proxy respondents completing the interview
(n = 106), and participants being interviewed over the phone and given a
shorter interview (n = 4). Not surprisingly, given the reasons for attrition,
those individuals in the longitudinal sample (N = 688), compared to those
who participated in 1996 only (N = 555), were younger (M ages = 78.9 vs.
82.3 yrs, t(1241) = 11.6, p < .01) and were healthier in 1996, as indicated by
a smaller number of health conditions (M = 3.6 vs. 4.3, t(1241) = 4.9, p < .01).
Measures
Table 1 describes the measures used for the cross-sectional and longitudi-
nal samples. Note that, although all the variables were measured at Time 1
in 1996, loneliness was also measured at Time 2 in 2001.
278 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Sociodemographics. Sociodemographic variables included age (in years),
and gender. Education level, or years they had completed in school, was
included to account for differences in socioeconomic status.A living arrange-
ments variable differentiated between people who lived alone and those who
lived with one or more persons. Living arrangements closely corresponded
with marital status. In the cross-sectional sample, 96% (575/598) of married
individuals lived with one or more individuals. Of the not married (widowed,
n = 534; separated/divorced, n = 28; or single, n = 83), only 16% (101/645)
lived with others.
Newall et al.: Loneliness among older adults 279
TABLE 1
Description of study measures for the cross-sectional and longitudinal samples
# of
Measures Anchors items MSDRange
Age (yrs) – 1 80.45 5.43 72–98
(78.93) (4.68) (72–95)
Gender 1 = men 1 1.57 .50 –
2 = women (1.59) (.49)
Education (yrs) – 1 9.23 3.15 0–27
(9.42) (2.95) (0–21)
Independence (IADL) 0 = needs help 12 9.70 1.97 2–12
1 = yes, can do (10.20) (1.55) (2–12)
General perceived health 1 = poor 1 2.42 .73 1–4
4 = excellent (2.34) (.69) (1–4)
Health conditions 0 = no 21 3.94 2.55 0–17
1 = yes (3.62) (2.43) (0–13)
Social participation 0 = no 14 5.05 2.11 0–13
1 = yes (5.35) (2.09) (0–13)
Living arrangements 1 = lives alone
2 = lives with 1 or 1 1.54 .50 –
2 = more persons (1.56) (.50)
Causal belief – effort 1 = total disagreement 1 4.90 1.90 1–7
7 = total agreement (5.04) (1.82) (1–7)
Causal belief – context 1 = total disagreement 1 3.13 2.10 1–7
7 = total agreement (3.08) (2.10) (1–7)
Causal belief – luck 1 = total disagreement 1 3.74 2.05 1–7
7 = total agreement (3.72) (2.07) (1–7)
Loneliness (1996) 0 = no 11 2.76 2.58 0–11
1 = yes
Loneliness (2001) 0 = no 11 – – –
1 = yes (2.68) (2.79) (0–11)
Notes.Numbers shown in brackets represent the longitudinal sample values.All measures taken
in 1996, except loneliness assessed in 1996 and 2001.
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Functional status. Participants’ functional status (independence) was meas-
ured by asking whether or not they were independently capable of per-
forming 12 specific instrumental activities of daily living (IADL; e.g.,
light housework, laundry, and food preparation). Based on similar IADL
measures (e.g., Lawton & Brody, 1969), a composite score was created by
summing the items so that higher scores reflected greater independence
(cross-sectional sample: alpha = .75, skewness = 1.13, kurtosis = 1.43).
Health status. Two measures were used to create a health status latent
construct. First, individuals’ self-rated health was assessed by asking them
to rate their health compared to other people who were their own age. This
measure has been shown to predict objective health status (Bailis, Segall,
& Chipperfield, 2003), mortality (Menec, Chipperfield, & Perry, 1999), and
health care use (Mossey et al., 1981). Responses ranged on a 5-point scale
(1 = bad; 2 = poor; 3 =fair; 4 =good; 5 = excellent). This measure was subse-
quently reverse coded and the small number of “bad” responses were re-
categorized to “poor”, resulting in a 4-point scale ranging from 1 = excellent
to 4 = poor.
As a second measure of health, individuals were asked whether they
currently had, or were still feeling, the after-effects of 21 specific health
problems (e.g., heart and circulation problems, arthritis). These items were
summed to create a composite measure, with higher scores indicating
poorer health (cross-sectional sample: alpha = .61, skewness = .84, kurtosis
= .95). Because the two health measures both theoretically tap into people’s
health, and because of their strong empirical association (r = .44), we spec-
ified the two observed health measures as a latent measure in the structural
equation models.
Social participation. In the present study, respondents’ social activity parti-
cipation was measured by tallying the number of social activities they had
participated in during the past week. In particular, respondents were asked
whether they had performed any of 14 activities (e.g., visiting family, visiting
friends, playing sports or games, doing church-related activities, doing com-
munity volunteer work). Affirmative responses were summed to create a
measure of social activity participation (cross-sectional sample: alpha = .56,
skewness = .39, kurtosis = -.03).
Causal beliefs for affiliation. Participants’ causal beliefs for affiliation were
assessed using three items from the Multidimensional Multiattributional
Causality (MMC) scales (Lefcourt, Von Baeyer, Ware, & Cox, 1979). The
original MMC scales were intended to measure causal beliefs (attributions
for outcomes) in the affiliation and achievement domains and items were
classified as: internal/unstable items (Effort subscale); external/stable items
(Context subscale); external/unstable items (Luck subscale); internal/stable
items (Ability subscale). Note that the original MMC scales included items
that varied in only two dimensions (internal/external and stable/unstable) as
identified in Weiner’s early work (Weiner et al., 1971). Thus, the scales did
280 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
not originally include items to tap into the third causal dimension of con-
trollability (i.e., whether events are perceived to be controllable or uncon-
trollable) that was introduced much later by Weiner (1985). Nonetheless, it
was possible to classify the items based on the controllability dimension.
Due to restricted space in the interview schedule, only one item from each
of the four affiliation subscales was included in the 1996 interview schedule,
and only three of those four items were selected for the present study. Note
that we excluded this fourth item that corresponded to the ability subscale
because it was ambiguous in terms of whether people perceive social skills
to be innate (internal/stable/uncontrollable) or modifiable (internal/unstable/
controllable). Weiner (1983) discussed this potential pitfall of researchers
erroneously concluding ability attributions to be innate. Specifically, for the
purposes of the present study, respondents were asked whether they agreed
with statements assessing: (1) Effort: “Do you believe that loneliness comes
from not trying to be friendly?” (internal/unstable/controllable); (2) Context:
“Over your life have you found that no matter what you do, some people
just don’t like you?” (external/stable/uncontrollable); and (3) Luck: “Over
your life, have you found that making friends has largely been a matter of
having the right breaks?” (external/unstable/uncontrollable).
Loneliness. Loneliness was measured in 1996 and again in 2001 using De
Jong Gierveld and Kamphuis’s (1985) 11-item loneliness scale. Five of the
scale’s items measure feelings of belonging (e.g., “I can call on my friends
whenever I need them”), and six items measure emotional loneliness or
missing relationships (e.g., “I miss having a really close friend). Based on
participant agreement with the statements (no; more or less; and yes), a
composite measure was created by summing item scores such that higher
scores reflected greater loneliness. The items were also dichotomized (0 =
no, 1 = more or less; yes), as is typically done with this scale, so that the
potential range was 0–11 (e.g., Dykstra et al., 2005). For both the cross-
sectional and the longitudinal samples, respectively, the scale had good reli-
ability (AIM 1996 (N = 1,243): alpha = .77, skewness = .99, kurtosis = .43;
AIM 2001 (N = 688): alpha = .83, skewness = .99, kurtosis = .20). If scores
of 3 and above reflect loneliness (Lauder, Mummery, Jones, & Caperchione,
2006), then 45% of the 1996 sample could be considered as being lonely.
In addition to the De Jong Gierveld and Kamphuis scale, participants
were categorized based on their self-reported loneliness in 1996 (1 = not
lonely; 2 = moderately lonely; 3 = severely lonely; 4 = extremely lonely;
M = 1.35; SD = .53). Past studies suggest that such 1-item measures may
underreport loneliness (Perlman, 2004; Ernst & Cacioppo, 1999); neverthe-
less, it is of value to obtain a sense of how many participants categorize
themselves as being lonely. In 1996, 1% of participants described them-
selves as “severely” or “extremely” lonely, 32% as “moderately” lonely, and
67% as “not lonely”.
Analytical approach. Structural equation modelling using AMOS (Arbuckle,
1995) tested the hypothesized cross-sectional and longitudinal models.
Newall et al.: Loneliness among older adults 281
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Coefficients were estimated using the maximum likelihood method and
the goodness of fit was assessed by Chi-square, the comparative fit index
(CFI), the adjusted goodness of fit index (AGFI), and the root mean square
error of approximation (RMSEA). Both the CFI and AGFI indices range
from 0 to 1, with values close to 1 indicating good fit (Byrne, 2001), and,
more specifically, values at or above .95 recommended for the CFI (Hu
& Bentler, 1999). RMSEA values at or below .06 indicate good fit (Hu &
Bentler, 1999).
Results
Bivariate correlations for the variables are presented in Table 2. Focusing
on the 1996 cross-sectional sample, as expected, being older was associated
with being female, living alone, having less education, less functional capa-
bility, poorer health, and participating in fewer social activities. As past
research suggests, a number of age-related variables, rather than simply age,
relates to loneliness (e.g., Jylha, 2004). For this reason we were interested
in examining age in the context of other social and health variables in the
structural equation models. Also, as expected, greater social activity parti-
cipation was correlated with participants’ functional capability and health
(in particular their perceived health) and with loneliness.
Structural equation modelling results
Cross-sectional results. Indices suggested good fit for the cross-sectional
model, χ
2
= 48.47, p < .001, CFI = .97, AGFI = .95, and RMSEA = .06. Table
3 shows all the estimated coefficients between the variables. Although not
provided in Table 3, the estimated coefficients between all of the demo-
graphic, health, and causal belief variables were specified in the models
(Figure 1). In general, the results corresponded to theoretical expectations.
Focusing first on social participation, significant predictors included
younger age, higher education level, greater functional independence, and
greater endorsement of the effort causal belief. Turning to loneliness, signi-
ficant predictors included lower education level, poorer health, living alone,
as well as less endorsement of the effort belief, and greater endorsement of
the context and luck beliefs. Although the magnitude of the relationships
between the causal beliefs and loneliness could be considered small, these
magnitudes exceed most of the effects of the other established demo-
graphic and health predictors in the model.
In addition, as expected, greater social participation was associated with
less loneliness, even after controlling for the demographic, health, and
causal belief variables. This is important because we can also examine how
variables may affect loneliness indirectly through social participation. That
is, age and functional status, while not having a direct effect on loneliness,
do appear to indirectly relate to loneliness through greater social participa-
tion. Moreover, education level and endorsement of the effort causal belief
282 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Newall et al.: Loneliness among older adults 283
TABLE 2
Correlations between study variables for the cross-sectional (N = 1,243) and longitudinal (N = 688) samples
Variables 123456789101112
1. Age –
2. Gender .07* –
(.07)
3. Education –.17** .07* –
(–.16**) (.09*)
4. Independence (IADL) –.37** –.18** .05 –
(–.26**) (–.26**) (–.02)
5. General perceived health .01 .07* –.09** –.33** –
(.02) (.05) (–.09*) (–.32**)
6. Health conditions .11** .07* –.03 –.34** .44** –
(.08*) (.07) (–.03) (–.34**) (–.46**)
7. Social participation –.14** –.04 .11** .24** –.19** –.07* –
(–.06) (–.06) (.12**) (.20**) (–.14**) (–.06)
8. Living arrangements –.27** –.35** .01 .09** –.01 –.06* .01 –
(–.27**) (–.40**) (.03) (.11**) (–.04) (–.09*) (.04)
9. Causal belief – effort –.04 –.03 –.05 .09** –.07* –.07* .13** .08** –
(.03) (–.02) (–.06) (.03) (–.08*) (–.07) (.12**) (.06)
10. Causal belief – context –.05 –.12** –.04 –.02 .03 .07* –.03 –.03 –.03 –
(–.03) (–.08*) (–.09*) (–.03) (.03) (.07) (–.07) (–.06) (–.04)
11. Causal belief – luck .08* –.05 –.08* –.05 .04 .10** –.03 .03 .13** .13** –
(.06) (–.02) (–.08*) (–.08**) (.06) (.10*) (–.02) (–.01) (.11**) (.22**)
12. Loneliness 1996/ .06* .00 –.11** –.17** .19** .25** –.17** –.12** –.22** .18** .14** –
12. (Loneliness 2001) (.07*) (.07) (–.04) (–.11**) (.09*) (.16**) (–.18**) (–.12**) (–.13**) (.05) (.11**)
Notes. Correlations without parentheses are from the cross-sectional sample and correlations in parentheses are from the longitudinal sample. All variables are from 1996 except
Loneliness measured in 1996 and 2001. *p ≤ .05; **p ≤ .01.
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
are both associated indirectly, through social participation, and directly to
loneliness.
In summary, for the cross-sectional results, causal beliefs predicted lone-
liness in the expected directions even after controlling for the sociodemo-
graphic, social, and health variables. Moreover, the internal/controllable
belief was directly associated with loneliness, as well as indirectly associated
with loneliness through its association with greater social participation.
Longitudinal results. The longitudinal version of the model in Figure 1 was
tested only with those participants who completed both the 1996 and 2001
data collections. Moreover, all of the predictor variables, including social
participation, were from the 1996 data collection whereas the loneliness
variable was from the 2001 data collection.
Indices suggested good fit for the longitudinal model, χ
2
= 15.58, p = .08,
CFI = .99, AGFI = .97, and RMSEA = .03. Note that although they are not
shown in Table 4, for theoretical reasons the estimated coefficients between
all of the demographic, health, and causal belief variables were specified in
the models (Figure 1). Similar to the cross-sectional sample results, greater
education level, greater functional independence, and greater endorsement
of the effort causal belief in 1996 Time 1 were associated with greater social
participation in 1996 (Table 4). Unlike the cross-sectional results, age was
not associated with social participation. Note that because greater social
participation is, in turn, associated with less loneliness five years later in 2001,
this means that education level, functional status, and effort beliefs can be
seen as having an indirect effect on loneliness through social participation.
Similar to the cross-sectional results, better health, greater endorsement
of the effort causal belief, and less endorsement of the luck causal belief in
1996 Time 1 were directly related to less loneliness in 2001 (Table 4). Unlike
284 Journal of Social and Personal Relationships 26(2–3)
TABLE 3
SEM coefficients for the cross-sectional model (N = 1,243)
Endogenous variables
Exogenous variables 1996 Social participation 1996 Loneliness 1996
Age –.06+ –.02
Gender (1 = men; 2 = women) –.01 –.05+
Education level .09** –.08**
Independence (IADL) .17** .02
Poorer health –.08+ .28**
Living arrangements –.04 –.10**
Causal belief-effort .11** –.19**
Causal belief-context –.02 .13**
Causal belief-luck –.01 .11**
Social participation – –.09**
R
2
.09 .20
+p = .06–.08; *p ≤ .05; **p ≤ .01.
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
the cross-sectional results, endorsement of the causal belief of context and
education level in 1996 were not directly associated with loneliness in 2001,
and living with others in 1996 was associated with loneliness in 2001.
In summary, the longitudinal results replicated to a great extent the cross-
sectional results. This is not surprising given the strong association between
the 1996 and 2001 loneliness ratings (r = .46). As expected, results showed
that strongly endorsing the statement that loneliness stems from not being
friendly (effort belief) was related to less loneliness at both measurement
points separated by five years (Table 3 and 4). Similarly, strongly endorsing
the uncontrollable/external causal belief of luck was related to greater lone-
liness at both measurement points. However, no significant association was
found between the external/uncontrollable causal belief of context and lone-
liness in 2001. Thus, this belief was only related to loneliness in the imme-
diate context, and not over the 5-year time span of the longitudinal study.
Discussion
This study examines the experience of loneliness among a representative
sample of older adults. Consistent with past research, suggesting that 20–40%
percent of older adults report occasional or moderate loneliness (Pinquart
& Sorensen, 2001; Wenger & Burholt, 2004), our study found that, based on
the single-item self-reported loneliness measure, about 33% of participants
described themselves as “severely” or “moderately” lonely.This implies that
for many people loneliness is a common experience in later life, underscoring
the need to understand how it evolves and how to mitigate its effects.
Our study provides some insight into the development and maintenance
of loneliness. In particular, we showed that beliefs that new relationships
Newall et al.: Loneliness among older adults 285
TABLE 4
SEM path coefficients for the longitudinal model (N = 688)
Endogenous variables
Exogenous variables 1996 Social participation 1996 Loneliness 2001
Age .00 .03
Gender (1 = men; 2 = women) –.02 .02
Education level .11** .00
Independence (IADL) .18** .02
Poorer health –.03 .14*
Living arrangements .00 –.08*
Causal belief-effort .11** –.10**
Causal belief- context –.05 .00
Causal belief-luck .01 .10**
Social participation – –.15**
R
2
.07 .09
+p = .06–.08; *p ≤ .05; **p ≤ .01.
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
are forged through internal/controllable causes, like effort, are associated
with greater social participation and less loneliness, whereas beliefs that
making friends is due to the context or to luck are associated with greater
loneliness. Moreover, the predictive value of endorsing these causal beliefs
about affiliation was demonstrated both immediately and over five years
even after accounting for the effects of other sociodemographic and health
variables.
Causal beliefs, social participation, and loneliness
The association found between endorsing internal/controllable affiliative
causal beliefs (i.e., effort) and being less lonely suggests that these types
of beliefs are, in fact, protecting people from loneliness. Moreover, effort
beliefs related to less loneliness both directly and indirectly through greater
social participation. This evidence is consistent with attribution theory that
posits that the explanations individuals give for events can profoundly
influence their expectations, emotions, and behaviours. Perceiving that a
social failure, such as loneliness, stems from a controllable reason, for
example, might at least provide individuals with a sense of control. Such a
sense could lead to subsequent efforts to ameliorate their social situation
and to bring into line their actual versus desired social networks.
In contrast, endorsing external/uncontrollable beliefs was related to
greater loneliness which may imply that uncontrollable attributions for
friendship erode feelings of control for individuals about creating or main-
taining relationships. This may lead people to have low motivation and
effort in the affiliative domain which, in turn, leads to greater loneliness.
In this way, an uncontrollable attribution, such as luck, could foster a
“passive”, rather than an “active”, approach to creating and maintaining
friends. Of note, our results suggested that strongly endorsing the context
or luck beliefs was not related to social participation. Thus, the results
suggest that social participation is not part of the mechanism through which
external/uncontrollable causal beliefs are associated with greater loneliness.
In addition, our findings are not consistent with the speculation that
internal/controllable causal beliefs result in self-blame thereby fostering
greater loneliness. Nor do they support the thinking that loneliness may be
avoided by attributing loneliness to external/uncontrollable forces. These
issues could be addressed more directly, however, in research focusing on
individuals’ feelings of self-blame surrounding their loneliness and their
perceptions of control. For example, given theory which emphasizes the
adaptiveness of a correspondence between control perceptions or control
strategies and the objective potential of control in a given situation (e.g.,
Heckhausen & Schulz, 1998), it may be that our results would have been
different had we examined beliefs among people living in institutions
offering less objective potential for control over social relationships. That
is, it is possible that if a person has little objective potential to control their
social relations, then making external/uncontrollable attributions for lone-
liness may be more adaptive than making internal/controllable attributions.
Based on the potential to change individuals’ causal beliefs through inter-
ventions, these findings have important implications. For example, enabling
286 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
people to become aware of how their perceptions affect their behaviours
and emotions might help them to deal more effectively with their loneliness.
Moreover, increasing practitioners’ awareness of these connections may
help them to provide guidance for their patients. At the same time, it is
important to note that we do not mean to suggest that individuals are solely
responsible for their loneliness.As acknowledged earlier, to understand and
to address the complex phenomenon that is loneliness requires multiple
perspectives. Indeed, our results point to the importance of adults being
given resources and opportunities for social activities and interaction as
they grow older. What we are suggesting is that designing interventions to
address loneliness among older adults may be facilitated by considering the
role that general causal beliefs play in loneliness.
Demographic, health, and loneliness
The present study replicated past research in finding a health–loneliness
relationship (e.g., Fees et al., 1999; Russell, 1996). Our results showed that,
even after accounting for other variables such as gender and education,
poorer health was associated with greater loneliness. This study also repli-
cated past research (e.g., Jylha, 2004) in showing that no direct relationship
between age and loneliness exists, after controlling for other age-related
variables. In the cross-sectional analyses, that increasing age predicted less
social participation suggests one possible intervening variable between age
and loneliness. Participants’ level of functionality or independence was also
indirectly related to loneliness through social participation. That is, parti-
cipants who were more independent participated in more social activities,
which, in turn, related to less loneliness.
Strengths, limitations, and conclusions
There were several strengths of the study, including the opportunity to
longitudinally examine loneliness among a large, representative sample of
older adults that included those who were aged 85 years and older. Access
to longitudinal data enabled us to consider the replicability of results estab-
lished in cross-sectional analyses. We have greater confidence in the gener-
alizability of these results because, unlike past studies using convenience
samples, our sample was initially selected to reflect the population of older
adults in the province of Manitoba, Canada.
Some limitations of the study should also be noted. In particular, we had
available only single items to assess different types of affiliative causal
beliefs. A preferable method would have been to have participants offer
reasons for their lack of friends and greater loneliness and subsequently to
have them rate these causes on the dimensions of internality, controllabil-
ity, and stability. Another limitation is that research questions relating to
causality cannot be definitively answered in this study; it is only possible to
determine how concepts vary in relation to others and not strictly whether
one variable caused changes in another variable. For example, although it
was argued that causal beliefs predicted social participation, it is also
Newall et al.: Loneliness among older adults 287
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
possible that the reverse holds true. Although the present study does not
preclude this possible interpretation, the study findings are consistent with
attribution theory and with the Controllability Attributional Model, which
propose that cognitions influence motivation and behaviours (Anderson &
Riger, 1991; Weiner, 1985).
To conclude, our findings provide insights into how the causal dimensions
of internality/externality and controllability relate to loneliness for older
adults. In addition to replicating past research, which shows a relation be-
tween greater loneliness and uncontrollable causal attributions (Anderson
& Riger, 1991), our research provides a more nuanced picture of the
relative importance of causal beliefs for social participation and loneliness.
In particular, the findings point to the value of focusing not only on how
older adults’ health and social activities relates to loneliness, but also on
the potentially modifiable beliefs people hold, how these can be changed,
and the effects these changes will likely have on people’s loneliness in the
future.
REFERENCES
Anderson, C. A., & Arnoult, L. H. (1985). Attributional style and everyday problems in living:
Depression, loneliness, and shyness. Social Cognition, 3, 16–35.
Anderson, C. A., Horowitz, L. M., & French, R. D. (1983). Attributional style of lonely and
depressed people. Journal of Personality and Social Psychology, 45, 127–136.
Anderson, C. A., & Riger, A. L. (1991). A controllability attribution model of problems in
living: Dimensional and situational interactions in the prediction of depression and loneli-
ness. Social Cognition, 9, 149–181.
Arbuckle, J. L. (1995). Amos for windows. Analysis of moment structures. Chicago: Small
Waters.
Bailis, D. S., Segall, A., & Chipperfield, J. G. (2003). Two views of self-rated general health
status. Social Science and Medicine, 56, 203–217.
Byrne, B. (2001). Structural equation modeling with Amos: Basic concepts, applications, and
programming. London: Lawrence Erlbaum Associates.
Cacioppo, J. T., Hughes, M. E., Waite, L. J., Hawkley, L. C., & Thisted, R. A. (2006). Loneliness
as a specific risk factor for depressive symptoms: Cross-sectional and longitudinal analyses.
Psychology and Aging, 21(1), 140–151.
Chipperfield, J. G., Havens, B., & Doig, W. (1997). Method and description of the Aging in
Manitoba project: A 20-year longitudinal study. Canadian Journal on Aging, 16(4), 606–625.
De Jong Gierveld, J. (1987). Developing and testing a model of loneliness. Journal of Person-
ality and Social Psychology, 53, 119–128.
De Jong Gierveld, J., & Kamphuis, F. (1985).The development of a Rasch-type loneliness scale.
Applied Psychological Measurement, 9, 289–299.
Dykstra, P. A., & Fokkema, T. (2007). Social and emotional loneliness among divorced and
married men and women: Comparing the deficit and cognitive perspectives. Basic and
Applied Social Psychology, 29(1), 1–12.
Dykstra, P. A., Van Tilburg, T. G., & De Jong Gierveld, J. (2005). Changes in older adults’ lone-
liness: Results from a seven-year longitudinal study. Research on Aging, 27(6), 725–747.
Ernst, J. M., & Cacioppo, J. T. (1999). Lonely hearts: Psychological perspectives on loneliness.
Applied and Preventive Psychology, 8(1), 1–22.
Fees, B. S., Martin, P., & Poon, L. W. (1999). Model of loneliness in older adults. Journals of
Gerontology: Psychological Sciences and Social Sciences, 54B(4), 231–239.
288 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Green, L. R., Richardson, D. S., Lago, T., & Schatten-Jones, E. C. (2001). Network correlates
of social and emotional loneliness in young and older adults. Personality and Social Psychol-
ogy Bulletin, 27, 281–288.
Havens, B., Hall, M., Sylvestre, G., & Jivan, T. (2004). Social isolation and loneliness: Differences
between older rural and urban Manitobans. Canadian Journal on Aging, 23(2), 129–140.
Hawkley, L. C., & Cacioppo, J. T. (2003). Loneliness and pathways to disease. Brain, Behavior
and Immunity, 17(Suppl. 1), S98–105.
Hawkley, L. C., & Cacioppo, J. T. (2007). Aging and loneliness: Downhill quickly? Current
Directions in Psychological Science, 16(4), 17–191.
Heckhausen, J., & Schulz, R. (1998). Developmental regulation in adulthood: Selection and
compensation via primary and secondary control. In J. Heckhausen and C. S. Dweck (Eds.),
Motivation and self-regulation across the life span (pp. 50–77). New York: Cambridge
University Press.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.
Jylha, M. (2004). Old age and loneliness: Cross-sectional and longitudinal analyses in the
Tempere Longitudinal Study on Aging. Canadian Journal on Aging, 23, 157–168.
Lauder, W., Mummery, K., Jones, M., & Caperchione, C. (2006). A comparison of health beha-
viours in lonely and non-lonely populations. Psychology, Health and Medicine, 11(2), 233–245.
Lawton, M. P., & Brody, E. M. (1969).Assessment of older people: Self-maintaining and instru-
mental activities of daily living. The Gerontologist, 9, 179–186.
Lefcourt, H. M., von Baeyer, C. L., Ware, E. E., & Cox, D. J. (1979). The multidimensional-
multiattributional causality scale: The development of a goal specific locus of control scale.
Canadian Journal of Behavioural Science, 11, 286–304.
Menec, V. H., Chipperfield, J. G., & Perry, R. P. (1999). Self-perceptions of health: A prospec-
tive analysis of mortality, control, and health. Journals of Gerontology: Series A: Biological
Sciences and Medical Sciences, 54B(2), 85-P93.
Menec, V. H., Perry, R. P., Struthers, C. W., Schonwetter, D. J., Hechter, F. J., & Eichholz, B.
(1994). Assisting at-risk college students with attributional retraining and effective teaching.
Journal of Applied Social Psychology, 24, 675–701.
Moore, D., & Schultz, N. R. (1987). Loneliness among the elderly:The role of perceived respon-
sibility and control. Journal of Social Behavior and Personality, 2, 215–224.
Mossey, J. M., Havens, B., Roos, N. P., & Shapiro, E. (1981). The Manitoba Longitudinal Study
on Aging: Description and Methods. The Gerontologist, 21, 551–558.
Peplau, L. A., & Caldwell, M. A. (1978). Loneliness: A cognitive analysis. Essence: Issues in the
Study of Ageing, Dying, and Death, 2, 207–220.
Peplau, L. A., Russell, D., & Heim, M. (1979). The experience of loneliness. In I. H. Frieze,
D. Bar-Tal, & J. S. Carroll (Eds.), New approaches to social problems: Applications of attri-
bution theory (pp. 53–78). San Francisco, CA: Jossey-Bass.
Perlman, D. (2004). European and Canadian studies of loneliness among seniors. Canadian
Journal on Aging, 23, 181–188.
Perry, R. P., Hechter, F. J., Menec, V. H., & Weinberg, L. E. (1993). Enhancing achievement
motivation and performance in college students: An attributional retraining perspective.
Research in Higher Education, 34, 687–723.
Pinquart, M., & Sorensen, S. (2001). Risk factors for loneliness in adulthood and old age: A
meta-analysis. Advances in Psychology Research, 19, 111–143.
Pressman, S. D., Cohen, S., Miller, G., E., Barkin, A., Rabin, B. S., & Treanor, J. J. (2005). Lone-
liness, social network size, and immune response to influenza vaccination in college fresh-
man. Health Psychology, 24, 297–306.
Russell, D. (1996). UCLA Loneliness Scale (version 3): Reliability, validity and factor
structure. Journal of Personality Assessment, 66(1), 20–40.
Ruthig, J. C., Perry, R. P., Hall, N. C., & Hladkyj, S. (2004). Optimism and attributional re-
training: Longitudinal effects on academic achievement, test anxiety, and voluntary course
withdrawal. Journal of Applied Social Psychology, 34(4), 709–730.
Newall et al.: Loneliness among older adults 289
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
Schultz, N. R., & Moore, D. D. (1984). Loneliness: Correlates, attributions, and coping among
older adults. Personality and Psychology Bulletin, 10, 67–77.
Schultz, N. R., & Moore, D. D. (1988). Loneliness: Difference across three age levels. Journal
of Social and Personal Relationships, 5(3), 275–284.
Solano, C. H. (1987). Loneliness and perceptions of control: General traits versus specific attri-
butions. Journal of Social Behavior and Personality, 2, 201–214.
Weeks, D. J. (1994). Review of loneliness concepts, with particular reference to old age. Inter-
national Journal of Geriatric Psychiatry, 9(5), 345–355.
Weiner, B. (1983). Some methodological pitfalls in attributional research. Journal of Educa-
tional Psychology, 75, 530–543.
Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psycho-
logical Review, 92(4), 548–573.
Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S., & Rosenbaum, R. M. (1971). Perceiving the
causes of success and failure. Morristown, NJ: General Learning Press.
Weiss, R. S. (1973). Loneliness: The experience of social and emotional isolation. Cambridge,
USA: MIT Press.
Wenger, G. C., & Burholt, V. (2004). Changes in levels of social isolation and loneliness among
older people in a rural area: A twenty-year longitudinal study. Canadian Journal on Aging,
23(2), 115–127.
Wenger, G. C., Davies, R., Shahtahmasebi, S., & Scott, A. (1996). Social isolation and loneli-
ness in old age: Review and model refinement. Ageing and Society, 16, 333–358.
290 Journal of Social and Personal Relationships 26(2–3)
at University of Manitoba Libraries on November 2, 2009 http://spr.sagepub.comDownloaded from
