Our Apologies! - APA PsycNET Login

6 downloads 405 Views 118KB Size Report
User ID or e-mail: Password: Using your membership number? Try it without dashes. Ajax login status image. Keep me logged in. Forgot your password?
Canadian Journal of Behavioural Science 2011, Vol. 43, No. 1, 40 –51

© 2011 Canadian Psychological Association 0008-400X/11/$12.00 DOI: 10.1037/a0021199

Loneliness, Attachment, and the Perception and Use of Social Support in University Students Stephanie Bernardon, Kimberley A. Babb, Julie Hakim-Larson, and Marcia Gragg University of Windsor This study examined factors related to family, social, and romantic loneliness in 173 undergraduate students (66 males, 107 females). Participants completed measures of attachment style, perceived availability of social support, use of social support coping, and loneliness. Results indicated that participants with greater attachment security reported lower levels of all types of loneliness compared to those with less attachment security, and this was partially mediated by perceived social support but not the use of social support coping. When examining underlying attachment constructs, a more positive model of others was related to less family and social loneliness, and these associations were mediated by greater perceived social support. Model of others also was associated with less romantic loneliness and model of self was associated with less loneliness in all domains, and these relations were partially mediated by perceived social support. Findings are discussed with respect to possible interventions to increase students’ perceptions of available social support and to decrease overall loneliness levels. Keywords: loneliness, attachment, social support, coping, undergraduate

other relationships (DiTommaso & Spinner, 1997). For example, a person may report great family satisfaction but experience loneliness in romantic relationships. This multidimensional approach is especially useful for studying loneliness in emerging adults transitioning to university. New university students often experience a reorganization of their social support networks as they gain more independence from their families and meet new peers at school (Green, Richardson, Lago, & Schatten-Jones, 2001; Kenny & Rice, 1995; Weiss, 1973). This transition can bring about many positive social benefits, but for some students, it may lead to feelings of loneliness with the reduction or loss of an established social support system. When this loneliness occurs in one domain and not another (e.g., a student has a number of new friends at school but misses the support of her family who lives far away), there may still be negative effects on the individual’s well-being. Assessing loneliness using the unidimensional approach may obscure these discrepancies in types of loneliness, thus making it difficult to identify those at risk. Given the importance of studying different types of loneliness in a university student population, the purpose of the present study was to assess family, social, and romantic loneliness in university students and to examine what factors are associated with greater loneliness in these domains. Specifically, we explored the relation of attachment to loneliness and whether perceived social support and the use of social support coping mediates this relation.

Loneliness has been defined as an unpleasant and distressing subjective experience that arises from a qualitative or quantitative deficiency in a person’s relationships (Perlman, 1988). It can vary in frequency and intensity (Russell, 1982) and has been associated with a myriad of aversive consequences, such as reduced life satisfaction (Goodwin, Cook, & Yung, 2001), decreased academic performance and persistence (Nicpon et al., 2006 –2007), and psychological distress (DiTommaso & Spinner, 1997). Popular measures of loneliness, such as the UCLA Loneliness Scale (Russell, Peplau, & Cutrona, 1980), have taken a unidimensional approach to loneliness. This approach views loneliness as a global phenomenon (Russell, 1982) and measures loneliness using a single overall score. Other researchers have suggested that loneliness is more multifaceted than can be reflected in one overall score, and it may be difficult for individuals to conceptualize their loneliness in this global fashion (Killeen, 1998). Earlier multidimensional theories proposed different types of loneliness, such as emotional and social loneliness, that were conceptually distinct (Rubenstein & Shaver, 1982; Weiss, 1973). More recently, DiTommaso and Spinner (1993) proposed a new tripartite theory of loneliness and developed a measure that assesses loneliness in three domains: family, social, and romantic. This approach improves upon other multidimensional theories that consider family and romantic loneliness part of emotional loneliness because the construct of emotional loneliness may be too broad to effectively capture loneliness variations in these different domains. Individuals can report one type of loneliness while being completely satisfied in

Attachment and Loneliness Attachment theory states that early attachment relationships with caregivers help form cognitive frameworks called internal working models that affect individuals’ expectations for security and support in future relationships (Bowlby, 1969; Klohnen & John, 1998). As children venture into the world, such as when transitioning to university, they still rely on family for support, but social and romantic relationships become more salient (Goldberg, 2000; Kenny & Rice, 1995). Social relationships, such as friend-

Stephanie Bernardon, Kimberley A. Babb, Julie Hakim-Larson, and Marcia Gragg, Department of Psychology, University of Windsor. Stephanie Bernardon is now with the Department of Theoretical and Behavioral Sciences at Wayne State University. Correspondence concerning this article should be addressed to Kimberley Babb, Department of Psychology, University of Windsor, 401 Sunset Avenue, Windsor, ON, N9B 3P4. E-mail: [email protected] 40

LONELINESS AND ATTACHMENT

ships, contribute to overall psychological well-being (Corsano, Majorano, & Champretavy, 2006) and serve as a precursor to positive development of romantic relationships (Collins, Welsh, & Furman, 2009; Connolly, Furman, & Konarski, 2000; Furman, Simon, Shaffer, & Bouchey, 2002). Measurement of adult attachment generally distinguishes between those who have developed a secure attachment style and those who have developed forms of insecure attachment. Bartholomew and colleagues (Bartholomew, 1990; Bartholomew & Horowitz, 1991) proposed a model of attachment based on Bowlby’s (1973) work on internal working models. At the core of this model are two underlying attachment constructs: model of self and model of others. Model of self reflects the degree to which individuals feel a sense of self-worth and competence in relationships, and model of others reflects the degree to which individuals feel that relationships with others are positive experiences and actively seek them out (Bartholomew, 1990). Different positive and negative combinations of these models result in four attachment styles that describe individuals’ primary attachment style in their internal working models (Bartholomew & Horowitz, 1991). Those who have a secure attachment style are those with positive models of self and others. In contrast to this adaptive attachment style are three less adaptive insecure attachment styles. Those classified as fearful-avoidant have negative models of self and others. Preoccupied individuals have a negative model of self but a positive model of others. Finally, dismissing-avoidant individuals have a positive model of self but a negative model of others. Given that an individual’s attachment style is related to positive or negative expectations for future relationships, discrepancies between what is expected and what is experienced in relationships may lead to greater loneliness. Studies using unidimensional measures of loneliness have consistently shown that attachment security in adulthood is associated with lower levels of loneliness (e.g., Kafetsios & Sideridis, 2006; Larose, Guay, & Boivin, 2002). However, there is a paucity of research on how attachment is related to different types of loneliness, such as the family, social, and romantic types proposed by DiTommaso and Spinner (1993). Given the typical developmental progression of types of close relationships, we would also expect changes in different types of loneliness corresponding to these domains. For example, adolescent loneliness has been shown to be associated mainly with family relationship deficits, whereas loneliness in university students tends to be associated with peer relationship deficits (Goldenberg & Perlman, 1984). Of the few studies on attachment and different types of loneliness, there is evidence to suggest that secure and insecure attachment styles affect loneliness (e.g., DiTommaso, Brannen-McNulty, Ross, & Burgess, 2003). Even less is known about what factors link attachment and these types of loneliness. An individual’s attachment style may evoke specific cognitive and/or behavioural aspects of social support that, in turn, facilitate the development of subsequent loneliness. This study explored two of these factors—perceived social support and use of social support coping—as potential mediators between attachment and family, social, and romantic loneliness.

Attachment, Perceived Social Support, and Loneliness One cognitive factor that may mediate the relation between attachment and different types of loneliness is perceived social

41

support, which is the perception that social support is available when required (Sarason, Pierce, Shearin, Sarason, & Walze, 1991). Studies have shown that those with secure attachment styles tend to report more perceived social support from friends and family (e.g., Blain, Thompson, & Whiffen, 1993), as well as from romantic partners (e.g., Collins & Feeney, 2004), compared to those with insecure attachment styles. Furthermore, more positive models of self and others have been associated with greater perceived social support from friends and family (Ognibene & Collins, 1998). Perceived social support also has been linked to loneliness. Studies have demonstrated that individuals who reported greater perceived social support also reported less loneliness (e.g., Davis, Morris, & Kraus, 1998; Nicpon et al., 2006 –2007), and in one study, perceived friendship support was found to be the best predictor of lower loneliness scores (Pierce, Sarason, & Sarason, 1991). Some researchers have even suggested that perceived social support may have a more significant effect on loneliness than actual received social support (Pinquart & Sorensen, 2003; Wethington & Kessler, 1986). Taken together, these findings suggest perceived social support is related to both attachment and loneliness; however, little research has examined perceived social support as a mediator between the two. One study that did investigate mediation models found that perceived support mediated the association between avoidant attachment and mental well-being for adults, age 35 to 66 (Kafetsios & Sideridis, 2006). However, this study used the unidimensional UCLA Loneliness Scale (Russell et al., 1980), which does not tap into different domains of loneliness that would be expected to differ in a universityage sample. To our knowledge, no study has investigated perceived social support as a mediator between attachment and DiTommaso and Spinner’s (1993) three-part model of loneliness.

Attachment, Use of Social Support Coping, and Loneliness Another behavioural factor that may mediate the relation between attachment and loneliness is the use of social support during coping efforts. Researchers have identified two types of social support coping: instrumental support (using others as a resource for information and problem-solving) and emotional support (using others for emotional comfort; Cutrona, 1990). Because securely attached individuals have a positive view of themselves and others, they typically expect their social environment to be supportive of help-seeking and feel confident in their own ability to seek out the instrumental or emotional support they require (Florian, Mikulincer, & Bucholtz, 1995; Mikulincer, Florian, & Weller, 1993). Conversely, individuals with insecure attachment styles expect others to be unresponsive to their needs and may dismiss seeking social support as a coping option (Hazan & Shaver, 1987; Klohnen & John, 1998). This lack of reliance on social support coping may feed into individuals’ perceived deficits in their social networks and thus result in greater feelings of loneliness. Studies have shown that compared to insecurely attached individuals, securely attached individuals report a greater tendency to seek out social support as a method of coping with stressful life events (e.g., DeFronzo, Panzarella, & Butler, 2001; Florian et al., 1995; Mikulincer et al., 1993; Seiffge-Krenke, 2006; Simpson, Rholes, & Nelligan, 1992). For example, Seiffge-Krenke (2006) found that when coping with relationship stressors, securely attached individuals used their social networks for support, dismissing-

BERNARDON, BABB, HAKIM-LARSON, AND GRAGG

42

avoidant individuals engaged less in support seeking, and preoccupied individuals tended to withdraw from others and from the problem. Loneliness also has been shown to be associated with the use of social support coping. For instance, Larose and colleagues (2002) found that less emotional support seeking was associated with higher levels of loneliness, independent of attachment style differences. Furthermore, using a student and community sample, Rokach (2001) found that participants reported using social support networks as a beneficial strategy for coping with loneliness. Thus, the patterns of findings indicate that the use of social support coping also is associated with both attachment and loneliness, but whether the use of social support coping mediates the association between attachment and different types of loneliness remains to be explored.

was that the association between greater attachment security and lower levels of loneliness, as well as the association between more positive models of self and others and lower levels of loneliness, would be mediated by greater perceived social support and greater use of social support coping. Early attachment relationships with caregivers are thought to contribute to a person’s expectations of available social support and their capability to use this support when needed (Bowlby, 1969). Therefore, these expectations should provide a link between secure attachment and positive outcomes, such as less loneliness.

Method Participants

The Present Study This study extends previous research on loneliness in three ways. First, in contrast to the majority of research that conceptualizes loneliness as a unidimensional construct, we examined factors relating to DiTommaso and Spinner’s (1993) three unique types of loneliness: family, social, and romantic. Second, we investigated whether underlying attachment constructs (model of self and model of others) have different associations with these three types of loneliness. Other researchers (e.g., Diehl, Elnick, Bourbeau, & Labouvie-Vief, 1998; Furman et al., 2002; Ognibene & Collins, 1998) have looked at these constructs as predictors; however, none have looked specifically at how they predict family, social, and romantic loneliness. Finally, we explored whether two cognitive and behavioural factors—perceived social support and use of social support coping—mediate the relation between attachment style and these types of loneliness. Although prior studies have found associations between attachment styles and these three types of loneliness, as well as associations between the two social support variables and other measures of loneliness and attachment, to our knowledge, this study is the first to investigate these specific mediated relations.

Hypotheses We tested four hypotheses in this study. Based on the work by DiTommaso and colleagues suggesting that family, social, and romantic are distinct domains of loneliness (e.g., DiTommaso et al., 2003; DiTommaso & Spinner, 1993), our first hypothesis was that there would be within-subject differences in these types of loneliness. Our second hypothesis was that participants with a primarily secure attachment style would experience lower levels of the different types of loneliness compared to those with a primarily insecure attachment style. Research on attachment and unidimensional measures of loneliness suggests that securely attached individuals generally report less loneliness than insecurely attached individuals (e.g., Larose et al., 2002). Because attachment style is thought to extend to other types of close relationships, we expected that this same pattern of findings would hold for the three types of loneliness, as well. Related to this, our third hypothesis was that a more positive model of self and others would be associated with less loneliness in all domains, compared to a more negative model of self and others. Having positive views about close relationships with others and about one’s worthiness of these relationships are conducive to maintaining good relationship bonds and, therefore, should be associated with less loneliness. Our fourth hypothesis

After receiving ethics clearance, 179 participants were recruited from the Department of Psychology participant pool at a midsize university in southwestern Ontario. Six participants who had more than one outlying value (z-score greater than 2) on key variables in the study were excluded from the analyses. Of the remaining 173 participants, 66 were male and 107 were female (age range ⫽ 18 to 54 years; Mage ⫽ 22.15, SD ⫽ 5.55). There were 80 lower level university students (52 females, 28 males) and 93 upper level university students (55 females, 38 males). Students were primarily single (83%), resided with their parent(s) (47%) or a roommate (32%), and were Caucasian (74%). In addition, the majority of students reported being currently employed (67%) and were involved in one or more social groups (61%).

Measures Background questionnaire. Participants provided demographic information on this questionnaire, such as their age, gender, ethnicity, marital status, employment status, year in school, and living situation (e.g., alone, with family, roommate, etc.). In order to understand students’ group involvement, participants also listed their current social groups/clubs (e.g., church, sports, school clubs, etc.). Attachment style (Relationship Scales Questionnaire, RSQ; Griffin & Bartholomew, 1994a). The RSQ assesses adult attachment styles experienced in close relationships. It consists of 30 statements representing the four attachment styles defined by Bartholomew and Horowitz (1991): secure (e.g., “I find it easy to get emotionally close to others”), fearful-avoidant (hereafter referred to as fearful; e.g., “I worry that I will be hurt if I allow myself to become too close to others”), preoccupied (e.g., “I want to be completely emotionally intimate with others”), and dismissing-avoidant (hereafter referred to as dismissing; e.g., “It is very important for me to feel independent”). Participants respond to these statements using a five-point Likert-type scale ranging from 1 (not at all like me) to 5 (very much like me), and then a mean is created for each of the four attachment style subscales. Higher subscale scores indicate a greater likelihood that the attachment style is representative of the person’s feelings about close relationships. The ranges of scores for the subscales were as follows: 2.13–5.00 (secure), 1.00 – 4.43 (fearful), 1.00 – 4.17 (preoccupied), and 1.00 – 4.50 (dismissing). The original scoring system developed by Griffin and Bartholomew (1994a), which includes 18 of the 30 RSQ items, has

LONELINESS AND ATTACHMENT

been shown to result in relatively low Cronbach’s alpha coefficients because the attachment styles contain two orthogonal dimensions (model of self, model of other; see Griffin & Bartholomew, 1994a, for a more detailed discussion of the reliability of the RSQ). Alternate scoring methods of the RSQ have improved its internal consistency values. In the current study, Ognibene and Collins’ (1998) scoring procedure was adopted. In this scoring procedure, the remaining RSQ items that are consistent with the theoretical descriptions of the four attachment styles (based on comparisons with Bartholomew & Horowitz’s, 1991, attachment style paragraphs) and that increase the alphas are added to their respective attachment style subscales. The Cronbach’s alpha coefficients for the present study were: .63 (secure), .80 (fearful), .76 (preoccupied), and .59 (dismissing). These values are similar to those found by Ognibene and Collins (1998). The RSQ also has demonstrated good convergent validity with other measures of attachment that used different methods (correlations ranging from .34 to .50; Griffin & Bartholomew, 1994b), as well as adequate discriminant validity when compared to the “Big Five” personality dimensions (Griffin & Bartholomew, 1994a). We also created continuous variables of the underlying attachment constructs of model of self and model of others (Bartholomew & Horowitz, 1991; Scharfe & Bartholomew, 1994). Higher scores on these measures represented more positive views of self and others in the context of relationships. Using the guidelines by Ross, McKim, and DiTommaso (2006), the model of self score was derived by summing the preoccupied and fearful attachment styles (negative view of self) and subtracting this from the sum of the secure and dismissing styles (positive view of self). The model of others score was derived by summing the dismissing and fearful attachment styles (negative view of others) and subtracting this score from the sum of the secure and preoccupied styles (positive view of others). Strong test–retest reliability coefficients have been found over an 8-month period for model of self (.81 to .84) and model of others (.72 to .85; Scharfe & Bartholomew, 1994). These variables also have been shown to have very good predictive validity, with model of self being strongly correlated (.77 to .86) with measures of self-concept and model of others being moderately to strongly correlated (.46 to .73) with interpersonal orientation (Griffin & Bartholomew, 1994b). Loneliness (The Social and Emotional Loneliness Scale for Adults - Short Form, SELSA-S; DiTommaso, Brannen, & Best, 2004). The SELSA-S is a 15-item questionnaire that assesses three domains of loneliness in adults: family, social, and romantic. Each domain subscale consists of five statements about feelings of loneliness within the past year, and participants rate the extent of their agreement with these statements on a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). The family loneliness subscale assesses feelings toward family relationships (e.g., “In the last year I felt alone when I was with my family”a). The social loneliness subscale measures feelings toward being part of a social group (e.g., “In the last year I didn’t have a friend(s) who understood me, but I wish I had”a). The romantic loneliness subscale measures the degree to which participants feel they have significant others in their lives (e.g., “In the last year I had an unmet need for a close romantic relationship”a). Mean scores are calculated for each subscale, with higher scores indicating greater feelings of loneliness in that domain. The ranges

43

of scores for the subscales were as follows: 1.00 – 6.40 (family), 1.00 – 6.60 (social), and 1.00 –7.00 (romantic). DiTommaso and colleagues (2004) have found very good internal consistency for the SELSA-S, with Cronbach’s alpha coefficients ranging from .87 to .90. Each subscale on the SELSA-S was significantly correlated with the Revised UCLA Loneliness Scale (ranging from .34 to .73), as well as with their analogues on the full version of the SELSA (ranging from .78 to .85), thus supporting its concurrent validity. Evidence for discriminant validity also was found, such as negative associations between quality of parental relationships and family and social loneliness (ranging from ⫺.18 to ⫺.62), as well as no significant associations between quality of parental relationships and romantic loneliness (see DiTommaso et al., 2004, for additional discussion of the reliability and validity of the SELSA-S). In the present study, the following Cronbach’s alpha coefficients were obtained: .88 (family), .77 (social), and .89 (romantic). Perceived social support (Interpersonal Support Evaluation List, ISEL; Cohen & Hoberman, 1983). The ISEL is a 48-item questionnaire used to measure the perceived availability of social support in college students. Respondents are asked to rate each social support statement (e.g., “I know someone who I see or talk to often with whom I would feel perfectly comfortable talking about any problem I might have adjusting to college life”b) as either probably true or probably false. The ISEL yields four subscales, as well as a total score of perceived social support availability. Only the total scale score was used in the current study, which was a sum of all the items. The range for this total score was 19 to 48, with higher scores indicating a higher level of perceived social support. The ISEL has been reported to have good internal consistency, with a Cronbach’s alpha coefficient of .77 for the ISEL total score, as well as a test-retest correlation of .87. The ISEL has been shown to be positively correlated with the Inventory of Socially Supportive Behaviors (.46) and negatively correlated with social anxiety (⫺.64; Cohen & Hoberman, 1983; see also Cohen, Mermelstein, Kamarch, & Hoberman, 1985, for a review of the ISEL’s reliability and validity). In the present study, the Cronbach’s alpha coefficient for the ISEL total score was .85. Use of social support coping (Coping Orientations to Problems Experienced Scale, COPE; Carver, Scheier, & Weintraub, 1989). The 60-item COPE inventory assesses the way in which participants typically cope with stressful situations by having them rate each type of coping strategy using a four-point Likert-type scale ranging from 1 (I usually don’t do this at all) to 4 (I usually do this a lot). Although the entire measure was given, only two subscales were used in this study to assess how participants used social support as coping mechanisms during stressful a Reproduced from “Measurement and validity characteristics of the short version of the social and emotional loneliness scale for adults,” by E. DiTommaso, C. Brannen, and L. A. Best, 2004, Educational and Psychological Measurement, 64, p. 107. Copyright 2004 by Sage Publications. Reprinted with permission of Sage Publications, Inc. b Reproduced from “Positive events and social supports as buffers of life change stress,” by S. Cohen and H. M. Hoberman, 1983, Journal of Applied Social Psychology, 13, p. 124. Copyright 1983 by Scripta Publishing Co. Reprinted with permission of John Wiley & Sons, Inc.

BERNARDON, BABB, HAKIM-LARSON, AND GRAGG

44

encounters. The first subscale, seeking social support for instrumental reasons (hereafter referred to as instrumental coping), measures the use of social support by asking for advice (e.g., “I try to get advice from someone about what to do”). The second subscale, seeking social support for emotional reasons (hereafter referred to as emotional coping), measures the likelihood of seeking sympathy and moral support (e.g., “I discuss my feelings with someone”). Scores for each type of coping were obtained by summing the scores for the four items in each subscale, with higher scores indicating greater use of that type of coping. Scores for instrumental coping ranged from 6 to 16, and scores for emotional coping ranged from 4 to 16. Carver and colleagues (1989) have reported very good test– retest reliability of the COPE for 6-week and 8-week intervals (from .64 to .76 for instrumental coping and from .77 to .72 for emotional coping, respectively), as well as Cronbach’s alpha coefficients of .75 for instrumental coping and .85 for emotional coping. The COPE also has demonstrated good validity. Adaptive personality traits (e.g., optimism, hardiness) have been shown to be positively associated with the functional coping subscales on the COPE (correlations ranging from .17 to .32) and negatively associated with the less functional coping subscales (correlations ranging from ⫺.21 to ⫺.34; see Carver et al., 1989, for further discussion of reliability and validity of the COPE). For the current study, the following Cronbach’s alpha coefficients were obtained: .72 (instrumental coping) and .87 (emotional coping).

Procedure Participants were brought into a room in small groups of 12 to 20 people. They first read and signed the consent form and then completed the questionnaires. To minimise gender differences in carryover effects, the four measures were counterbalanced within gender, with the demographic questionnaire always presented first. At the completion of the study, participants were given a letter of information about the study, allowed a chance to ask questions, and then were thanked for their participation. All participants received bonus marks for their participation.

Results Overview Following preliminary analyses, the results are presented in four sections. The first section examines within-subject differences in family, social, and romantic loneliness. The second section examines attachment style differences in the three types of loneliness using participants’ primary attachment style (secure, fearful, preoccupied, and dismissing). The third section examines whether perceived social support and two forms of social support coping (instrumental and emotional) mediate the relation between primary attachment category (secure and insecure) and the three types of loneliness. Finally, the fourth section examines the underlying attachment constructs of model of self and model of others as predictors of the three types of loneliness and whether or not these associations are mediated by perceived social support and social support coping.

Preliminary Analyses Skewness. We assessed the skewness of the following variables: family loneliness, social loneliness, romantic loneliness, per-

ceived social support, instrumental coping, and emotional coping. All three types of loneliness showed a significant amount of positive skew, and perceived social support was significantly negatively skewed. Following the guidelines of Tabachnick and Fidell (1996), we transformed these variables using a square root transformation (the perceived social support measure, ISEL total, was both reflected and transformed because of its negative skew), and this resulted in a significant reduction in skewness. These transformed variables were then used for the remaining analyses. The distributions for instrumental and emotional coping were not significantly skewed; therefore, no transformations were warranted. For ease of interpretation, all means presented in this paper are nontransformed means. Covariates. We then tested a number of demographic variables as potential covariates. Bivariate correlations showed no significant associations between the three types of loneliness and age and number of groups to which the participant belonged (rs ⱕ ⫺.12, ps ⱖ .12). Using the three types of loneliness as dependent variables, we also ran independent samples t -tests assessing differences between gender, level in school, employment status, whether or not they lived alone, whether or not they lived at home with their family, and marital status (whether or not they were married or living with a significant other). A difference in marital status was found for romantic loneliness, t(66) ⫽ 7.27, p ⬍ .001. Those who were married or living together reported less romantic loneliness (M ⫽ 1.67, SD ⫽ 0.72) than those who were not (M ⫽ 3.42, SD ⫽ 1.97). No other significant differences were found for any of the other variables (ts ⱕ ⫺1.91, ps ⱖ .06). Thus, only marital status was used as a covariate for all further analyses on romantic loneliness.

Within-Subject Differences in Loneliness Given DiTommaso and Spinner’s (1993) proposal that loneliness is a multidimensional construct, we examined within-subject differences in students’ levels of family, social, and romantic loneliness. We conducted a repeated-measures ANCOVA, controlling for marital status, and found significant differences between loneliness ratings, F(2, 307) ⫽ 12.58, p ⬍ .001. Post-hoc analyses using the Bonferroni correction indicated that students reported significantly more romantic loneliness (M ⫽ 3.17, SD ⫽ 1.94) compared to both family loneliness (M ⫽ 2.40, SD ⫽ 1.40) and social loneliness (M ⫽ 2.46, SD ⫽ 1.09).

Differences in Loneliness by Primary Attachment Style We then examined differences in participants’ reported levels of loneliness by primary attachment style. Participants’ primary attachment style was the subscale on the RSQ for which they had the highest mean. The majority of participants were classified as secure (n ⫽ 104), with fewer classified as fearful (n ⫽ 15), preoccupied (n ⫽ 17), and dismissing (n ⫽ 37). These frequencies are consistent with other studies that categorized participants by primary attachment style (e.g., Bartholomew & Horowitz, 1991; Diehl et al., 1998; Murphy & Bates, 1997). With primary attachment style as the independent variable, we conducted two separate one-way ANOVAs using family and social loneliness as the dependent variables and one ANCOVA with romantic loneliness, using marital status as a covariate. As seen in Table 1, the secure group reported less family loneliness than those in the preoccupied

LONELINESS AND ATTACHMENT

45

Table 1 Means and Standard Deviations of Loneliness by Primary Attachment Style Primary attachment style Type of loneliness

Secure (n ⫽ 104)

Fearful (n ⫽ 15)

Preoccupied (n ⫽ 17)

Dismissing (n ⫽ 37)

Family Social Romantic

2.06a (1.28) 2.18a (1.07) 2.67a (1.73)

2.55a,b (1.65) 3.07b (0.73) 4.26b (1.93)

3.22b (1.50) 2.93b (0.95) 3.86a,b (1.98)

2.93b (1.31) 2.81b (1.09) 3.81b (2.09)

Note. Numbers reflect the untransformed loneliness ratings. Means in the same row that do not share the same subscript differ significantly at p ⱕ .05.

or dismissing groups. For social loneliness, the secure group reported less loneliness than all of the insecure groups. Finally, the secure group reported less romantic loneliness than the fearful and dismissing groups. To further examine the relation between the four attachment styles and loneliness, and to ensure that the previous findings were not an artefact of the grouping by primary attachment style, we ran bivariate correlations between participants’ continuous scores on each of the RSQ subscales and their scores on the three types of loneliness. For romantic loneliness, we conducted partial correlations, controlling for marital status. As shown in Table 2, the pattern of associations mirrored the ANOVA/ANCOVA results. Higher scores on the secure attachment subscale were significantly associated with lower levels of all types of loneliness, whereas higher scores on the fearful, preoccupied, and dismissing attachment subscales were significantly associated with higher levels of loneliness. Taken together, these findings suggest that secure attachment is associated with less loneliness and insecure attachment is associated with more loneliness. Given this pattern and the small sample sizes of participants in the fearful, preoccupied, and dismissing groups, we conducted the remaining analyses on attachment styles by combining the three insecure groups into one (n ⫽ 69).

Perceived Social Support, Use of Social Support Coping, and Loneliness Correlations within attachment style. Next, we explored the associations between loneliness and the three potential mediators of perceived social support and two forms of social support coping (instrumental coping and emotional coping) within primary attachment category. As shown in Table 3, for participants whose Table 2 Summary of Zero-Order and Partial Correlations Between Continuous Attachment Style Ratings and Loneliness Ratings (n ⫽ 173) Type of loneliness Family Social Romantica a

Secure ⴱⴱ

⫺.39 ⫺.39ⴱⴱ ⫺.38ⴱⴱ

Fearful ⴱⴱ

.36 .41ⴱⴱ .36ⴱⴱ

Preoccupied ⴱⴱ

.45 .32ⴱⴱ .23ⴱ

Dismissing .34ⴱⴱ .20ⴱ .29ⴱⴱ

Marital status was controlled for in the partial correlations for romantic loneliness. ⴱ p ⬍ .01. ⴱⴱ p ⬍ .001.

primary attachment category was secure, greater perceived social support (ISEL total) was associated with less social and romantic loneliness. In addition, greater use of instrumental coping was associated with less social loneliness. For participants whose primary attachment category was insecure, greater perceived social support was associated with less family and social loneliness, and greater use of instrumental coping was associated with less family loneliness. Statistical analyses using the Fisher r to z transformation (Cohen, 1988) showed no significant difference between the secure and insecure groups in the relative size of the correlation between perceived social support and social loneliness. Mediation analyses. Based on the correlational findings, we then assessed whether perceived social support and instrumental coping mediated the relation between attachment security and the three types of loneliness using Baron and Kenny’s (1986) procedure for identifying mediated relations and adjusting for Type I error (alpha criterion ⫽ .0253; Kenny, 2009). Emotional coping was not tested as a mediator because it was not associated with any type of loneliness. In the regression analyses, the predictor was the dichotomous attachment security variable (0 ⫽ insecure, 1 ⫽ secure). For the analyses examining the relation between the predictor and outcome variables, greater attachment security was found to be associated with less family loneliness (␤ ⫽ ⫺.32, t ⫽ ⫺4.34, p ⬍ .001), less social loneliness (␤ ⫽ ⫺.35, t ⫽ ⫺4.84, p ⬍ .001), and, controlling for marital status, less romantic loneliness (␤ ⫽ ⫺.31, t ⫽ ⫺4.53, p ⬍ .001). Analyses examining the relation between the predictor and proposed mediators indicated that greater attachment security was associated with higher levels of perceived social support (␤ ⫽ ⫺.48, t ⫽ ⫺7.18, p ⬍ .001) and greater use of instrumental coping (␤ ⫽ .18, t ⫽ 2.37, p ⬍ .05). Controlling for marital status, significant associations also were found for attachment security as a predictor of both perceived social support (␤ ⫽ ⫺.48, t ⫽ ⫺7.23, p ⬍ .001) and instrumental coping (␤ ⫽ .18, t ⫽ 2.33, p ⬍ .05). Standardised beta values in the regressions for perceived social support are in a negative direction because ISEL scores were reflected during the transformation to reduce skewness. To test the mediation effect, each of the proposed mediators was added to the regression analyses already containing the attachment security variable. When the variable of perceived social support was added to the regressions, results indicated that perceived social support was associated with less loneliness in all domains: family (␤ ⫽ .27, t ⫽ 3.38, p ⫽ .001), social (␤ ⫽ .40, t ⫽ 5.26, p ⬍ .001), and romantic (␤ ⫽ .28, t ⫽ 3.65, p ⬍ .001). However, the associations between attachment security and the three types of

BERNARDON, BABB, HAKIM-LARSON, AND GRAGG

46

Table 3 Summary of Zero-Order and Partial Correlations for Loneliness, Perceived Social Support, and Social Support Coping as a Function of Attachment Security (n ⫽ 173)

1. 2. 3. 4. 5. 6.

Family loneliness Social loneliness Romantic lonelinessa Perceived social supportb Instrumental coping Emotional coping

1

2

3

4

— .17 .14 .41ⴱⴱⴱ ⫺.24ⴱ ⫺.15

.25ⴱⴱ — ⫺.12 .36ⴱⴱ ⫺.11 ⫺.18

.02 .15 — .23 ⫺.08 ⫺.11

.15 .38ⴱⴱⴱ .30ⴱⴱ — ⫺.24ⴱ ⫺.27ⴱ

5

6

⫺.09 ⫺.20ⴱ ⫺.09 ⫺.22ⴱ — .64ⴱⴱⴱ

⫺.17 ⫺.18 ⫺.09 ⫺.20ⴱ .63ⴱⴱⴱ —

Note. Correlations for secure participants (n ⫽ 104) are presented above the diagonal, and correlations for the insecure participants (n ⫽ 69) are presented below the diagonal. a Marital status was controlled for in the partial correlations for romantic loneliness. b Because of the transformation to the perceived social support measure (ISEL total), higher scores indicate less perceived social support. ⴱ p ⱕ .05. ⴱⴱ p ⱕ .01. ⴱⴱⴱ p ⱕ .001.

loneliness also remained significant. Therefore, we did not find complete mediation. To assess whether these indirect relations indicated partial mediation, we conducted the Sobel test (Sobel, 1982). For all three types of loneliness, the Sobel test was significant (family: z ⫽ ⫺3.06, p ⬍ .01; social: z ⫽ ⫺4.28, p ⬍ .001; romantic: z ⫽ ⫺3.25, p ⫽ .001). Thus, the lower reported levels of loneliness by participants who were securely attached were partially mediated by their perceiving more social support. When instrumental coping was added to the regression analyses already containing attachment security, no significant associations were found between instrumental coping and all three types of loneliness, after adjusting for Type I error (alpha criterion ⫽ .0253): family (␤ ⫽ ⫺.15, t ⫽ ⫺1.98, p ⫽ .05); social (␤ ⫽ ⫺.16, t ⫽ ⫺2.21, p ⬍ .05); romantic (␤ ⫽ ⫺.08, t ⫽ ⫺1.18, p ⬎ .05). Thus, instrumental coping was not a mediator of the relation between attachment security and loneliness.

Model of Self, Model of Others, and Loneliness Correlational analyses. In order to understand which aspects of the underlying constructs of attachment are associated with loneliness, we also examined model of self and model of others as predictors of loneliness. First, we ran correlations to examine the relation between these two underlying constructs, the three types of loneliness, and the three potential mediators. As before with romantic loneliness, we conducted partial correlations, controlling for marital status. Table 4 summarises these correlations. As expected, higher scores on both model of self and model of others were significantly associated with less loneliness in all three domains. Higher scores on model of self and model of others also were significantly associated with higher levels of perceived social support. Greater use of instrumental coping was significantly associated with higher scores on model of self, but not with model of others. In contrast, greater use of emotional coping was significantly associated with higher scores on model of others, but not with model of self. Mediation analyses. We then explored whether perceived social support and the use of social support coping were mediators of the relation between underlying attachment constructs and loneliness. The predictor variables were the continuous variables of model of self and model of others. Based on the correlational

analyses, for the mediation analyses on model of self we assessed perceived social support and instrumental coping as possible mediators, and for model of others we assessed perceived social support and emotional coping as possible mediators. As before, we used the Baron and Kenny (1986) procedure for identifying mediated relations, followed by the Sobel test (Sobel, 1982) for assessing the significance of indirect relations. Regression analyses examining the association between the predictor and outcome variables indicated that model of self was found to be associated with less family loneliness (␤ ⫽ ⫺.41, t ⫽ ⫺5.92, p ⬍ .001), less social loneliness (␤ ⫽ ⫺.43, t ⫽ ⫺6.23, p ⬍ .001), and less romantic loneliness (␤ ⫽ ⫺.30, t ⫽ ⫺4.31, p ⬍ .001), after controlling for marital status. Testing the association between the predictor and potential mediators for the mediation analyses on family and social loneliness, model of self was associated with higher levels of perceived social support (␤ ⫽ ⫺.41, t ⫽ ⫺5.79, p ⬍ .001) but not with instrumental coping (␤ ⫽ .16, t ⫽ 2.07, p ⬍ .05, not significant after adjusting for Type I error). When marital status was controlled for in the analyses on romantic loneliness, model of self also was associated with higher levels of perceived social support (␤ ⫽ ⫺.43, t ⫽ ⫺6.10, p ⬍ .001) but not with instrumental coping (␤ ⫽ .15, t ⫽ 1.89, p ⬎ .05). Therefore,

Table 4 Summary of Zero-Order and Partial Correlations for Underlying Attachment Constructs, Loneliness, Perceived Social Support, and Social Support Coping (n ⫽ 173) Measure Family loneliness Social loneliness Romantic lonelinessa Perceived social supportb Instrumental coping Emotional coping a

Model of self ⴱⴱ

⫺.41 ⫺.43ⴱⴱ ⫺.31ⴱⴱ ⫺.41ⴱⴱ .16ⴱ .09

Model of others ⫺.25ⴱⴱ ⫺.29ⴱⴱ ⫺.34ⴱⴱ ⫺.43ⴱⴱ .11 .29ⴱⴱ

Marital status was controlled for in the partial correlations for romantic loneliness. b Because of the transformation to the perceived social support measure (ISEL total), higher scores indicate less perceived social support. ⴱ p ⬍ .05. ⴱⴱ p ⱕ .001.

LONELINESS AND ATTACHMENT

only perceived social support was tested as a mediator for all types of loneliness. When the variable of perceived social support was added to the regressions already containing model of self, perceived social support predicted less loneliness in all domains: family (␤ ⫽ .23, t ⫽ 3.11, p ⬍ .01), social (␤ ⫽ .36, t ⫽ 5.09, p ⬍ .001), and romantic (␤ ⫽ .29, t ⫽ 3.95, p ⬍ .001). However, the associations between model of self and the three types of loneliness also remained significant. Significant Sobel tests (family: z ⫽ ⫺2.76, p ⬍ .01; social: z ⫽ ⫺3.86, p ⬍ .001; romantic: z ⫽ ⫺3.32, p ⬍ .001) indicated the lower reported levels of loneliness by participants who had greater model of self scores were partially mediated by their perceiving more social support. Model of others, the second underlying attachment construct predictor variable, also was found to be significantly associated with less family loneliness (␤ ⫽ ⫺.25, t ⫽ ⫺3.40, p ⫽ .001), less social loneliness (␤ ⫽ ⫺.29, t ⫽ ⫺3.93, p ⬍ .001), and less romantic loneliness (␤ ⫽ ⫺.32, t ⫽ ⫺4.74, p ⬍ .001), controlling for marital status. Figure 1 shows a path model using the standardised regression coefficients of the analyses in which we tested perceived social support as a mediator of the relation between model of others and both family and social loneliness. Participants with higher scores on model of others had higher levels of perceived social support. Additionally, perceived social support was significantly associated with less family and social loneliness. When the variable of perceived social support was added to the regression analysis already containing model of others, the direct associations between model of others and both family and social loneliness were no longer significant. For romantic loneliness, model of others also predicted perceived social support (␤ ⫽ ⫺.43, t ⫽ ⫺6.20, p ⬍ .001), and when perceived social support was added to the regression already containing model of others, it was significantly associated with less romantic loneliness (␤ ⫽ .27, t ⫽ 3.76, p ⬍ .001). However, the direct association between model of others and romantic loneliness remained significant. All Sobel tests were significant (family: z ⫽ ⫺3.33, p ⬍ .001; social: z ⫽

(-.25*) -.12

Family loneliness .31*

Model of others

-.43*

Perceived social support

47

⫺4.19, p ⬍ .001; romantic: z ⫽ ⫺3.21, p ⫽ .001). Thus, the lower levels of family and social loneliness reported by those with higher scores on model of others was mediated by their perceiving more social support. Perceived social support partially mediated the relation between model of others and romantic loneliness. For the analyses on emotional coping as a potential mediator, model of others was associated with greater use of emotional coping (␤ ⫽ .29, t ⫽ 3.96, p ⬍ .001), even when controlling for marital status (␤ ⫽ .29, t ⫽ 3.94, p ⬍ .001). When emotional coping was entered into the regression analyses already containing model of others, a significant association was found between emotional coping and social loneliness (␤ ⫽ ⫺.17, t ⫽ ⫺2.26, p ⬍ .05) but not for family loneliness (␤ ⫽ ⫺.16, t ⫽ ⫺2.13, p ⬍ .05, not significant after adjusting for Type I error) or romantic loneliness (␤ ⫽ ⫺.07, t ⫽ ⫺0.97, p ⬎ .05). The direct association between model of others and social loneliness also remained significant. A nonsignificant Sobel test for the indirect relation between model of others and social loneliness (z ⫽ ⫺1.86, p ⬎ .05) indicated that emotional coping was neither a complete nor a partial mediator of the association between model of others and social loneliness.

Discussion The purpose of this study was to examine family, social, and romantic loneliness in university students and to investigate possible mediators of the relation between attachment and these types of loneliness. As anticipated, students reported differences in their experiences of family, social, and romantic loneliness. Specifically, they reported significantly higher levels of romantic loneliness compared to family and social loneliness. These differences in levels of loneliness provide further support for the multidimensional approach to measuring loneliness. Greater attachment security and more positive models of self and others were associated with less loneliness in all domains. Perceived social support, but not social support coping, mediated the relation between model of others and family and social loneliness, but only partially mediated the relation between model of others and romantic loneliness. Perceived social support also partially mediated the relation between attachment security and all types of loneliness, as well as the relation between model of self and all three types of loneliness. These results are discussed below in light of past research and the need to continue to refine the constructs underlying the study of attachment styles in adulthood, social support, and loneliness.

Attachment Style Differences in Loneliness .43* Social loneliness

(-.29*) -.11

Figure 1. Path models with standardised coefficients showing that perceived social support mediated the relation between model of others and both family and social loneliness. Values in parentheses represent the coefficients for the unmediated (direct) relation between the predictor and outcome variables. ⴱp ⱕ .001.

As predicted based on the work of DiTommaso and colleagues (DiTommaso et al., 2003, DiTommaso & Spinner, 1997), those with a secure attachment style reported lower levels of family, social, and romantic loneliness as compared to those with insecure attachment styles. Individuals with insecure attachment reported greater loneliness in these three domains, regardless of their specific attachment style. It is interesting to note that the fearful group was the only insecure attachment type that did not differ significantly from the secure subgroup in reported feelings of family loneliness. Both secure and fearful types may report low family loneliness levels, but for different reasons. Feelings of loneliness

48

BERNARDON, BABB, HAKIM-LARSON, AND GRAGG

within the family would be expected to be low for secure individuals because of their positive models of self and others. In contrast, fearful individuals have a negative view of both self and others that may result in perceiving family relationships as unworthy of their investment; therefore, they may minimise any feelings of family loneliness. The secure group also did not differ significantly from the preoccupied group in reported feelings of romantic loneliness. Both secure and preoccupied individuals have a positive model of others. Perhaps having a more positive model of others makes an individual more willing to invest emotionally in their romantic partners and leads to lower feelings of romantic loneliness. These nonsignificant findings raise intriguing future research questions; however, a more likely explanation is low statistical power because of the small sizes of our insecure groups. Though we feel our study’s distribution of attachment styles was representative of our population, researchers may want to prescreen participants for their attachment styles in order to achieve a balanced group design for future studies of loneliness differences amongst the insecure styles. Although we interpret these nonsignificant differences between the secure and the two smaller insecure groups with caution, our overall attachment style results convey an important finding that attachment security appears to be a key correlate of lower levels of loneliness.

Differences in Loneliness and Social Support by Underlying Attachment Constructs: Models of Self and Others Consistent with prior research (e.g., Furman et al., 2002; Ognibene & Collins, 1998), those with positive models of self and others reported lower levels of loneliness in all domains, as well as greater perceived social support, compared to those with more negative models of self and others. It is interesting to note that our correlational analyses suggested that model of self and model of others were related to differential use of social support coping strategies. Model of self was associated with the use of instrumental coping, but not the use of emotional coping, suggesting that those with a more positive view of themselves may be more confident in seeking problem-solving advice as a form of social support coping. Model of others was associated with the use of emotional coping, but not with instrumental coping, suggesting that those who generally view close relationships with others as desirable may be more comfortable with seeking out emotional comfort from others. These findings have important implications for future research with other well-being outcomes that could be related to social support coping, such as depression.

Perceived Social Support and Social Support Coping as Mediators Attachment styles formed early in life are thought to be resistant to change by the time individuals reach adulthood. Young adults with insecure internal working models of attachment may be at risk for problems with their psychosocial well-being. In our research, we investigated potentially modifiable features of the social support environment (perceived social support and social support coping) that might mediate the relation between attachment style and feelings of loneliness. We found partial support for our mediation hypothesis, in

that perceived social support partially mediated the relation between secure attachment and all three types of loneliness. When examining the associations between loneliness and the potential mediators of perceived social support, instrumental coping, and emotional coping within each attachment style category (secure and insecure), we found some interesting patterns. For both the secure and insecure groups, perceived social support was related to less social loneliness. This association did not significantly differ in strength between the two groups. This suggests that the perceived availability of social support is related to less social loneliness in university students regardless of the security of attachment. Other correlations differed between groups. For secure individuals, perceived social support was associated with less romantic loneliness, but it was associated with less family loneliness for insecure individuals. Use of instrumental coping also showed attachment style differences, in that it was related to less social loneliness in the secure group but less family loneliness for the insecure group. These differences may reflect a developmental trend. Given that family relationships serve as precursors to later peer and romantic relationships (e.g., Collins et al., 2009; Kenny & Rice, 1995), the stronger associations between social support factors and less social and romantic loneliness for the secure group may indicate that they have more mature types of attachment relationships in adulthood compared to those in the insecure group. Despite the significant correlations with instrumental coping, the mediational test results suggested that only perceived social support partially mediated the relations between attachment and family, social, and romantic loneliness. This suggests that attachment security is related to lower levels of loneliness not only through direct associations with the relationship foundations contained within individuals’ attachment styles but also through indirect associations with the individuals’ general perception of the availability of social support. We further refined our analyses by examining mediation effects using participants’ models of self and others as predictor variables in order to explore what aspects of attachment are related to social support and loneliness. As before, both forms of social support coping did not act as mediators. For model of self, perceived social support partially mediated the association with all types of loneliness. This suggests that a positive view of the self has both a direct and indirect pathway to reduced loneliness. For the indirect pathway, a positive view of one’s worthiness to receive social support may predispose individuals to perceive their social networks as available to provide this support when needed, and this leads to lower levels of loneliness. Perceived social support was then found to fully mediate the relation between model of others and family and social loneliness but only partially mediate the relation with romantic loneliness. Attachment develops from early primary caregiver relationships (i.e., family) and extends to friends (i.e., social) and, ultimately, to a significant other (i.e., romantic). If young adults have a positive model of others, they are likely to perceive a greater availability of social support in social domains that are well familiar to them, such as family and friends, and this can lead to less loneliness in these domains. Also at this time, a major developmental task for young adults is to develop romantic intimacy while maintaining family and friendship ties that already have a core foundation. Consummate love, the highest form of relationship satisfaction, is achieved through experiences of intimacy, passion, and commitment; passionate love, which is more characteristic of romantic relationships in early adulthood, is not as fully developed (Furman & Wehner, 1997; Sternberg, 1986).

LONELINESS AND ATTACHMENT

The relation between secure attachment and relationship satisfaction is thought to be mediated by feelings of intimacy and commitment (Madey & Rodgers, 2009). Consummate love may provide these feelings of intimacy and commitment that create a stronger overall perception of social support, whereas with passionate love, it may not yet be clear whether one’s romantic partner will provide the same base of social support as do family and social relationships. As our participants were mostly single and within the typical university age range, it is likely that their relatively limited experience with romantic partners is reflected in our results. The path between attachment, perceived social support, and romantic loneliness may not yet be strong enough to produce the mediated effect found for the other two types of loneliness.

Limitations and Future Directions In considering these results, it is also important to acknowledge limitations of the study. The reliance on self-report measures to assess attachment styles can only capture the consciously held feelings of the individual. These measures may not capture other aspects of attachment, which might have a more profound effect on an individual’s style of relating to others. Thus, future research could include interactive tasks that include an examination of attachment-related emotional processes and behaviours within family, peer, and romantic settings. For example, research on emotional reciprocity might be useful in clarifying the nature of the relation between attachment style and loneliness because it examines the extent to which emotionally positive or negative utterances in interactions are reciprocated by the interaction partner (e.g., Lindsey, MacKinnon-Lewis, Campbell, Frabutt, & Lamb, 2002). A sense of belonging in a relationship would likely be fostered when positive emotional reciprocity predominates. Another potential limitation is the age distribution of our sample. We used all participant data because social network changes that occur during university have the potential to be experienced by all students, regardless of age. In fact, no age or university level differences in loneliness were found. Furthermore, 89% of students were within the age range of emerging adulthood (18 to 25 years); therefore, our findings are likely characteristic of university students at this developmental stage. However, the inclusion of older students had the potential to affect our results in other ways. Older students were more likely to be married. Given our developmental explanation for the partial mediation findings between model of other and romantic loneliness, this difference in marital status might have inflated the strength of this indirect path. This does not seem to be the case, at least in our sample. Even with the participants older than 25 removed from analyses, perceived social support remained a partial mediator of the relation between model of others and romantic loneliness. This is not to say that developmental differences do not exist. Future directions for research on attachment and loneliness should include a developmental perspective, considering that younger and older adults may differ in their relationship needs and perceptions, as well as in their use of social support and feelings of loneliness (e.g., Gierveld & Dykstra, 2008). Given the strong evidence in this study that perceived social support provides a link between attachment and loneliness, future studies also would benefit from exploring more nuanced aspects of perceived social support. For example, there may be different pathways from social support to different types of loneliness

49

depending upon who is the provider of support. Furthermore, Davis and colleagues (1998) have distinguished between global support (a sense of support) and domain-specific support (a history of receiving support from particular individuals) and found that these two types had different relations with measures of loneliness. Group belonging was not associated with the different types of loneliness in the current sample; however, the nature of a person’s social networks should not be discounted. Future studies should examine factors, such as the length of group belonging (e.g., short-term vs. long-term commitment), the type of group belonging (e.g., social club vs. political organisation), the quality of group belonging (e.g., in person vs. over the Internet), as well as whether or not an individual’s family members also are part of the group. These differences may play a factor in the different types of loneliness experienced by individuals.

Conclusions and Implications Studying loneliness from a multidimensional approach is important, especially when considering the implications for development. Taken together, our results demonstrate several key findings. Security of attachment, as characterized by more positive models of self and others, appears to be a protective factor against loneliness. Additionally, perceived social support seems to play a key role in buffering against feelings of loneliness, whereas we did not find evidence that use of social support during coping efforts provides this same benefit. Our results support the view that believing that there is someone to turn to in times of both happiness and sorrow may prevent feelings of despair and decrease subsequent loneliness (Baumeister & Leary, 1995). The benefits of perceiving that social support is available, whether or not that support is actually used, has important implications for psychological research and professionals working with universitylevel students. To reduce family loneliness levels, professionals working with families can educate them on the importance of family interactions, such as fostering family traditions (e.g., Friese et al., 2002), and the importance of maintaining a supportive family environment. To reduce social loneliness levels, campus outreach programs may need to encourage both face-to-face student interactions through social gatherings, as well as make use of online social networking, which is emerging as a way for students to engage in informal interactions (e.g., Madge, Meek, Wellens, & Hooley, 2009). As young adults gain experience in romantic relationships, they also can gain a sense of confidence in the availability of social support provided by their partner (Shulman & Scharf, 2000), which may buffer against feelings of romantic loneliness. At the individual level, clinicians can focus therapeutic efforts on helping the individual recognise how many people can provide support to them, thereby increasing the salience of their perceived social support. Furthermore, although attachment style may not be changeable, clinicians can work with those who have insecure attachment styles to change the cognitive manifestations of these styles by increasing self-efficacy beliefs and redirecting maladaptive cognitions about relationships with others. Universities also can contribute to this by establishing programs to increase self-efficacy through experiences of mastery, modelling, and verbal persuasion. In conclusion, we have demonstrated that family, social, and romantic are distinct dimensions of loneliness and have different associations with attachment styles, perceived social support, and

BERNARDON, BABB, HAKIM-LARSON, AND GRAGG

50

use of social support coping. Studying loneliness as a multidimensional construct, rather than a unidimensional construct, not only helps us to understand the complex nature of loneliness, but also opens up possibilities for future research into other mechanisms that can help to reduce loneliness and increase well-being.

Re´sume´ Cette e´tude visait a` examiner des facteurs lie´s a` la solitude familiale, sociale et amoureuse apre`s de 173e´tudiants de premier cycle (66 hommes, 107 femmes). Les participants ont comple´te´ des questionnaires portant sur le style d’attachement, la disponibilite´ perc¸ue de soutien social, l’utilisation du soutien social et la solitude. Les re´sultats ont indique´ que les participants avec une plus grande se´curite´ d’attachement ont des niveaux plus faibles de tous les types de solitude comparativement a` ceux avec une se´curite´ d’attachement plus faible, cet effet e´tant partiellement module´ par la perception de soutien social, mais pas par l’utilisation du soutien social. En examinant les construits d’attachement sous-jacents, un mode`le positif d’autrui s’est ave´re´ lie´ a` moins de solitude familiale et sociale, ces associations e´tant module´es par une plus grande perception de soutien social. Le mode`le d’autrui e´tait aussi associe´ a` moins de solitude amoureuse et le mode`le de soi e´tait associe´ avec moins de solitude dans tous les domaines, ces relations e´tant partiellement module´es par la perception de soutien social. Les re´sultats sont discute´s a` la lumie`re d’interventions potentielles pour augmenter les perceptions de soutien social disponible chez les e´tudiants et diminuer la solitude de fac¸on globale. Mots-cle´s : solitude, attachement, soutien social, adaptation, e´tudiant de premier cycle

References Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 31, 1173–1182. doi:10.1037/0022–3514.51.6.1173 Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social and Personal Relationships, 7, 147–178. doi: 10.1177/0265407590072001 Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61, 226 –244. doi:10.1037/0022–3514.61.2.226 Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529. doi:10.1037/0033–2909.117.3.497 Blain, M. D., Thompson, J. M., & Whiffen, V. E. (1993). Attachment and perceived social support in late adolescence: The interaction between working models of self and others. Journal of Adolescent Research, 8, 226 –241. doi:10.1177/074355489382006 Bowlby, J. (1969). Attachment and loss, Vol. 1: Attachment. New York, NY: Basic Books. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. New York, NY: Basic Books. Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267–283. doi:10.1037/0022–3514.56.2.267 Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Cohen, S., & Hoberman, H. M. (1983). Positive events and social supports as buffers of life change stress. Journal of Applied Social Psychology, 13, 99 –125. doi:10.1111/j.1559 –1816.1983.tb02325.x

Cohen, S., Mermelstein, R., Kamarch, T., & Hoberman, H. M. (1985). Measuring the functional components of social support. In I. G. Sarason & B. R. Sarason (Eds.), Social support: Theory, research, and applications (pp. 73–94). Bonas, France: Dordrecht Publishing. Collins, N. L., & Feeney, B. C. (2004). Working models of attachment shape perceptions of social support: Evidence from experimental and observational studies. Journal of Personality and Social Psychology, 87, 363–383. doi:10.1037/0022–3514.87.3.363 Collins, W. A., Welsh, D. P., & Furman, W. (2009). Adolescent romantic relationships. Annual Review of Psychology, 60, 631– 652. doi:10.1146/ annurev.psych.60.110707.163459 Connolly, J., Furman, W., & Konarski, R. (2000). The role of peers in the emergence of heterosexual romantic relationships in adolescence. Child Development, 71, 1395–1408. doi:10.1111/1467– 8624.00235 Corsano, P., Majorano, M., & Champretavy, L. (2006). Psychological well-being in adolescence: The contribution of interpersonal relations and experience of being alone. Adolescence, 41, 341–354. Cutrona, C. E. (1990). Stress and social support-in search of optimal matching. Journal of Social and Clinical Psychology, 9, 3–14. Davis, M. H., Morris, M. M., & Kraus, L. A. (1998). Relationship-specific and global perception of social support: Association with well-being and attachment. Journal of Personality and Social Psychology, 74, 468 – 481. doi:10.1037/0022–3514.74.2.468 DeFronzo, R., Panzarella, C., & Butler, A. C. (2001). Attachment, support seeking, and adaptive inferential feedback: Implications for psychological health. Cognitive and Behavioral Practice, 8, 48 –52. doi:10.1016/ S1077-7229(01)80043–2 Diehl, M., Elnick, A. B., Bourbeau, L. S., & Labouvie-Vief, G. (1998). Adult attachment styles: Their relations to family context and personality. Journal of Personality and Social Psychology, 74, 1656 –1669. doi:10.1037/0022–3514.74.6.1656 DiTommaso, E., Brannen, C., & Best, L. A. (2004). Measurement and validity characteristics of the short version of the social and emotional loneliness scale for adults. Educational and Psychological Measurement, 64, 99 –119. doi:10.1177/0013164403258450 DiTommaso, E., Brannen-McNulty, C., Ross, L., & Burgess, M. (2003). Attachment styles, social skills and loneliness in young adults. Personality and Individual Differences, 35, 303–312. doi:10.1016/S01918869(02)00190 –3 DiTommaso, E., & Spinner, B. (1993). The development and initial validation of the social and emotional loneliness scale for adults (SELSA). Personality and Individual Differences, 14, 127–134. doi:10.1016/0191– 8869(93)90182–3 DiTommaso, E., & Spinner, B. (1997). Social and emotional loneliness: A re-examination of Weiss’ typology of loneliness. Personality and Individual Differences, 22, 417– 427. doi:10.1016/S0191-8869(96)00204 – 8 Florian, V., Mikulincer, M., & Bucholtz, I. (1995). Effects of adult attachment style on the perception and search for social support. The Journal of Psychology, 129, 665– 672. Friese, B. H., Tomcho, T. J., Douglas, M., Josephs, K., Poltrock, S., & Baker, T. (2002). A review of 50 years of research on naturally occurring family routines and rituals: Cause for celebration? Journal of Family Psychology, 16, 381–390. Furman, W., Simon, V. A., Shaffer, L., & Bouchey, H. A. (2002). Adolescents’ working models and styles for relationships with parents, friends, and romantic partners. Child Development, 73, 241–255. doi: 10.1111/1467– 8624.00403 Furman, W., & Wehner, E. H. (1997). Adolescent romantic relationships: A developmental perspective. In S. Shulman & W. A. Collins (Eds.), Romantic relationships in adolescence: Developmental perspectives (pp. 21–36). San Francisco, CA: Jossey-Bass. Gierveld, J. D. J., & Dykstra, P. A. (2008). Virtue is its own reward? Support-giving in the family and loneliness in middle and old age. Aging & Society, 28, 271–287. doi:10.1017/S0144686X07006629

LONELINESS AND ATTACHMENT Goldberg, S. (2000). Attachment and development. New York, NY: Oxford University Press Inc. Goldenberg, S., & Perlman, D. (1984). Social relations and loneliness during adolescence. Unpublished manuscript, University of British Columbia, Division of Family Science, Vancouver, Canada. Goodwin, R., Cook, O., & Yung, Y. (2001). Loneliness and life satisfaction among three cultural groups. Personal Relationships, 8, 225–230. doi: 10.1111/j.1475– 6811.2001.tb00037.x 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 Psychology Bulletin, 27, 281–288. doi: 10.1177/0146167201273002 Griffin, D., & Bartholomew, K. (1994a). Metaphysics of measurement: The case of adult attachment. In K. Bartholomew & D. Perlman (Eds.), Advances in personal relationships, Vol. 5: Attachment processes in adulthood (pp. 17–52). London, England: Jessica Kingsley. Griffin, D., & Bartholomew, K. (1994b). Models of the self and other: Fundamental dimensions underlying measures of adult attachment. Journal of Personality and Social Psychology, 67, 430 – 445. doi: 10.1037/0022–3514.67.3.430 Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52, 511–524. doi:10.1037/0022–3514.52.3.511 Kafetsios, K., & Sideridis, G. D. (2006). Attachment, social support, and well-being in young and older adults. Journal of Health Psychology, 11, 863– 876. doi:10.1177/1359105306069084 Kenny, D. A. (2009, November 15). Mediation. Retrieved from http:// davidakenny.net/cm/mediate.htm Kenny, M. E., & Rice, K. G. (1995). Attachment to parents and adjustment in late adolescent college students: Current status, applications, and future considerations. The Counseling Psychologist, 23, 433– 456. doi: 10.1177/0011000095233003 Killeen, C. (1998). Loneliness: An epidemic in modern society. Journal of Advanced Nursing, 28, 762–770. Klohnen, C. E., & John, P. O. (1998). Working models of attachment: A theory-based prototype approach. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 115–140). New York, NY: The Guilford Press. Larose, S., Guay, F., & Boivin, M. (2002). Attachment, social support, and loneliness in young adulthood: A test of two models. Personality and Social Psychology Bulletin, 28, 684 – 693. doi:10.1177/ 0146167202288012 Lindsey, E. W., MacKinnon-Lewis, C., Campbell, J., Frabutt, J. M., & Lamb, M. E. (2002). Marital conflict and boys’ peer relationships: The mediating role of mother-son emotional reciprocity. Journal of Family Psychology, 16, 466 – 477. doi:10.1037/0893–3200.16.4.466 Madey, S. F., & Rodgers, L. (2009). The effects of attachment and Sternberg’s triangular theory of love on relationship satisfaction. Individual Differences Research, 7, 76 – 84. Madge, C., Meek, J., Wellens, J., & Hooley, T. (2009). Facebook, social integration and informal learning at university: “It is more for socialising and talking to friends about work than for actually doing work.” Learning, Media and Technology, 34, 141–155. Retrieved from http:// dx.doi.org/10.1080/17439880902923606 Mikulincer, M., Florian, V., & Weller, A. (1993). Attachment styles, coping strategies, and posttraumatic psychological distress: The impact of the gulf war in Israel. Journal of Personality and Social Psychology, 64, 817– 826. doi:10.1037/0022–3514.64.5.817 Murphy, B., & Bates, G. W. (1997). Adult attachment styles and vulnerability to depression. Journal of Personality and Individual Differences, 22, 835– 844. doi:10.1016/S0191-8869(96)00277–2 Nicpon, M. F., Huser, L., Blanks, E. H., Sollenberger, S., Befort, C., & Robinson Kurpius, S. E. (2006 –2007). The relationship of loneliness

51

and social support with college freshmen’s academic performance and persistence. Journal of College Student Retention, 8, 345–358. Ognibene, T. C., & Collins, N. L. (1998). Adult attachment styles, perceived social support and coping strategies. Journal of Social and Personal Relationships, 15, 323–345. doi:10.1177/0265407598153002 Perlman, D. (1988). Loneliness: A life-span, family perspective. In R. M. Milardo (Ed.), Families and social networks (pp. 190 –220). Newbury Park, CA: Sage. Pierce, G. R., Sarason, I. G., & Sarason, B. R. (1991). General and relationship-based perceptions of social support: Are two constructs better than one? Journal of Personality and Social Psychology, 61, 1028 –1039. doi:10.1037/0022–3514.61.6.1028 Pinquart, M., & Sorensen, S. (2003). Risk factors for loneliness in adulthood and old age: A meta-analysis. In S. P. Shohov (Ed.), Advances in psychology research (Vol. 19, pp. 111–143). Hauppauge, NY: Nova Science. Rokach, A. (2001). Strategies of coping with loneliness throughout the lifespan. Current Psychology: Developmental, Learning, Personality, Social, 20, 3–18. Ross, L. R., McKim, M. K., & DiTommaso, E. (2006). How do underlying “self” and “other” dimensions define adult attachment styles? Canadian Journal of Behavioural Science, 38, 294 –310. doi:10.1037/cjbs2006016 Rubenstein, C., & Shaver, P. (1982). The experience of loneliness. In L. A. Peplau & D. Perlman (Eds.), Loneliness: A sourcebook of current theories, research, and therapy (pp. 206 –223). New York, NY: Wiley. Russell, D. (1982). The measurement of loneliness. In L. A. Peplau & D. Perlman (Eds.), Loneliness: A sourcebook of current theory, research, and therapy (pp. 81–104). New York, NY: Wiley. Russell, D., Peplau, L. A., & Cutrona, C. E. (1980). The revised UCLA loneliness scale: Concurrent and discriminant validity evidence. Journal of Personality and Social Psychology, 39, 472–480. doi:10.1037/0022–3514.39.3.472 Sarason, B. R., Pierce, G. R., Shearin, E. N., Sarason, I. G., & Waltz, J. A. (1991). Perceived social support and working models of self and actual others. Journal of Personality and Social Psychology, 60, 273–287. doi:10.1037/0022–3514.60.2.273 Scharfe, E., & Bartholomew, K. (1994). Reliability and stability of adult attachment patterns. Personal Relationships, 1, 23– 43. doi:10.1111/j .1475– 6811.1994.tb00053.x Seiffge-Krenke, I. (2006). Coping with relationship stressors: The impact of different working models of attachment and links to adaptation. Journal of Youth and Adolescence, 35, 25–39. doi:10.1007/s10964-005–9015-4 Shulman, S., & Scharf, M. (2000). Adolescent romantic behaviors and perceptions: Age- and gender-related differences, and links with family and peer relationships. Journal of Research on Adolescence, 10, 99 –118. doi:10.1207/SJRA1001_5 Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992). Support seeking and support giving within couples in an anxiety-provoking situation: The role of attachment styles. Journal of Personality and Social Psychology, 62, 434 – 446. doi:10.1037/0022–3514.62.3.434 Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological methodology (pp. 290 –312). Washington, DC: American Sociological Association. Sternberg, R. J. (1986). A triangular theory of love. Psychological Review, 93, 119 –135. doi:10.1037/0033-295X.93.2.119 Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York, NY: Harper Collins. Weiss, S. R. (1973). Loneliness: The experience of emotional and social loneliness. London, England: The MIT Press. Wethington, E., & Kessler, R. C. (1986). Perceived support, received support, and adjustment to stressful life events. Journal of Health and Social Behavior, 27, 78 – 89. doi:10.2307/2136504

Received June 4, 2009 Revision received June 4, 2010 Accepted June 10, 2010 䡲