Development of a Multidimensional Scale of Social Integration in Later Life

Abstract

Social integration is increasingly recognized as key to successful aging; however, definitions and assessments vary greatly across gerontological studies. This study describes the development of and provides preliminary evidence for the multidimensional Social Integration in Later Life Scale (SILLS). A review of previous measures led to the development of a 30-item questionnaire, which was given to a community-based sample of 399 older adults (ages 60–100). Factor analysis was used to determine a four-factor structure that included dimensions of frequency–social ties, frequency–social activities, satisfaction–social ties, and satisfaction–social activities. The overall scale Cronbach’s α was .86 (subscales ranged from .72 to .84), demonstrating good internal consistency. Preliminary results suggest that the SILLS has adequate concurrent and convergent validity. By assessing enacted and perceived integration across social ties and activities, this comprehensive measure is a useful tool for understanding social integration in later life.

Keywords

social integration scale development later life aging

Successful aging is widely promoted within the gerontological literature (Depp & Jeste, 2006). In addition to emphasizing health and cognitive functioning, Rowe and Kahn’s (1997) model of successful aging designates social integration (e.g., meaningful and purposeful social activities) as crucial for successful aging. In recent decades, numerous studies have demonstrated that various aspects of social integration including social support, social participation, and social network composition have a positive influence on the physical and mental health of older adults (Amieva et al., 2010; Arcury et al., 2012; Bath & Deeg, 2005; Ertel, Glymour, & Berkman, 2009; Fothergill et al., 2011; Thomas, 2011; Zunzunegui, Alvarado, Del Ser, & Otero, 2003). Social integration is conceptualized as a multidimensional construct consisting of social engagement across a wide range of social activities and relationships (Arcury et al., 2012; Brissette, Cohen, & Seeman, 2000); yet, there is substantial variability in the way social integration is defined and assessed throughout the adult lifespan. Given the unique social roles of older adults (i.e., retired, widowed), it may not be appropriate to apply the same definitions and measures as those used among younger adults. Moreover, much of the research on social integration among older adults is limited in scope by targeting a specific domain of social integration (e.g., close support relationships, religious involvement, etc.) instead of examining social integration broadly. This study aims to develop a comprehensive measure of social integration that incorporates engagement across a broad array of social domains and is appropriate for assessment among older adults.

Defining Social Integration

Although a limited number of studies explicitly address social integration (Ertel et al., 2008; Mezuk & Rebok, 2008; Zunzunegui et al., 2003), in many studies, terms such as social support, social engagement, and social participation are used interchangeably with social integration. It is important to note that the meanings of these terms differ. Social support is the perceived availability of help, affection, and instrumental aid from significant social partners (e.g., family members, close friends, neighbors, and colleagues; Cohen, 2004). Social support is often described as a function of an individual’s social ties and the level of instrumental and emotional help available (Bath & Deeg, 2005; Ertel et al., 2009), and thus can be measured in terms of the social network (Ajrouch, Blandon, & Antonucci, 2005). Social engagement refers to participation in community-related activities such as getting together with people of the same/different age-group or participating in organizations (Ertel et al., 2009), whereas social participation is a broad term implying social interaction with people other than immediate family members (Bukov, Maas, & Lampert, 2002; Utz, Carr, Nesse, & Wortman, 2002). Social engagement and social participation are often used interchangeably to describe participation in activities involving interacting with other people, community involvement, volunteer activities, and/or leisure activities (Demers, Robichaud, Gelinas, Noreau, & Desrosiers, 2009; Thomas, 2011).

Social integration is a complex construct that encompasses both social support and social engagement/participation. In its broadest terms, social integration has been defined as involvement in a web of social relationships and activities individuals maintain throughout life, including immediate family members and friends as well as formal relationships with other people, groups, and organizations (Ertel et al., 2009; Seeman, 1996). In this study, we find it helpful to employ an ecological systems approach in defining social integration. An ecological systems theoretical framework emphasizes the importance of recognizing the impact of the complex social context on individual development. Bronfenbrenner’s (1979) ecological perspective identifies varying levels of environment that influence development. The microsystem represents the proximal and principal contexts in which development takes place. For older adults, the microsystem may include contexts such as interpersonal relationships (family and friends), neighborhood, community, organizations, and religious institutions. Social integration refers to the extent of contextual immersion primarily within this microsystem. However, it is integral to understand influences across multiple contexts of the microsystem (i.e., not just family, but also community, neighborhood, etc.).

In addition to engagement in an array of relationships and activities, social integration comprises both behavioral and cognitive components (Brissette et al., 2000; Cohen, 2004). The behavioral component of social integration can be quantified in terms of actions such as frequency of engagement with interpersonal relationships or participation in social activities. In contrast, the cognitive component of social integration is more subjective, encompassing perceptions of social engagement. The perception of social support or community involvement may denote a sense of belongingness, centrality, or purpose (i.e., the feeling of being socially integrated). These behavioral and cognitive components represent distinct dimensions of social integration that may be in sync or discordant. Thus, in a synthesis of the complex dimensions highlighted previously, in this study, we define social integration as an individual’s enacted and perceived engagement with social ties (i.e., family, friends, etc.) and social contexts (i.e., community, organizations, work, etc.).

Theoretical Perspectives on Social Integration

Despite the growing body of evidence on the importance of social integration for well-being in later life (Arcury et al., 2012; Berkman, Glass, Brissette, & Seeman, 2000; Ertel et al., 2009), there is a need for further investigation to adequately explicate the nature of social integration in later life. Varying theoretical perspectives on social integration focus on social networks and social roles, extent of social participation, and perceived social support.

Social Roles and Networks

Involvement in social roles has traditionally been used as a measure of social integration. To define social integration, Thoits (1986) assessed involvement in a range of social relationships such as spouse, child, coworker, and so on. More recently, Cohen, Doyle, Skoner, Rabin, and Gwaltney (1997) used a similar method to develop a scale that assesses active participation across 11 social roles such as spouse, parent, friend, neighbor, and so on. However, the reliance on social roles in defining social integration is problematic, given normative transitions in social roles that occur in later life (i.e., retirement, loss of spouse, moving into senior facilities). Thus, such measures may contain inherent biases when used in later life.

Much of the recent empirical work on social integration is grounded in the assessment of social network size and quality of social relationships (Pillemer, Moen, Wethington, & Glasgow, 2000). Often social integration in late life is defined in terms of network size and frequency of contact with network (Abbott, Bettger, Hampton, & Kohler, 2012; Bath & Deeg, 2005) and thus social networks serve as a proxy for social integration. The hierarchical mapping technique, developed by Antonucci (1986), has been frequently used to measure the structure of social networks. Individuals are asked to identify varying levels of social support partners based on the closeness, importance, and support expected (Ajrouch et al., 2005). From this, network size and composition (e.g., social partners on the basis of relationship such as family or friends) can be calculated. Because the social networks of older adults generally consist of mostly family and friends (Fiori, Smith, & Antonucci, 2007), this method may be limited in its scope. While an understanding of close social relationships is integral, less close relationships are also relevant. For instance, neighbors may not be identified as close social relationships, yet neighbors are often a significant component of older adults’ social engagement and integration mainly due to geographical proximity and years of living in the same neighborhood (Shaw, 2005).

Perceived Social Support

In addition to counting members in the social network, the perceived support from these relationships is frequently measured (Amieva et al., 2010). The Social Support Questionnaire (Sarason, Sarason, Shearin, & Pierce, 1987) incorporates items that assess both the number of social support partners (across specific types of support such as emotional, instrumental, etc.) and the satisfaction with these support relationships. Perceived social support is considered a reflection of cognitive components of social integration, in terms of the individual’s perceptions of their own social involvement and belongingness. Cutrona and Russell’s (1987) Social Provisions Scale specifically examines satisfaction with social relationships, whereas the Duke Social Support Index expands upon this by including both measures of the satisfaction with support as well as perceived social support (Landerman, George, Campbell, & Blazer, 1989). Even more complex is Krause’s (1995) perceived support scale, which encompasses dimensions of instrumental support, emotional support, informational support, satisfaction with support, and negative interactions. A growing body of research documents the implications of perceived social support in later life (Amieva et al., 2010; Berkman et al., 2000; Cacioppo, Hawkley, & Thisted, 2010; Uchino, Cacioppo, & Kiecolt-Glaser, 1996); however, this subjective aspect of social integration is not frequently examined together with more objective dimensions.

Social Participation

In contrast to the focus on social network support, social participation has frequently been examined among older adults; yet often types of community and organizational ties have been examined in isolation (e.g., volunteering, religious involvement) or neglected (e.g., leisure, lifelong learning). Affiliation with various institutions such as churches, clubs, and senior centers is also measured frequently as an aspect of social integration in older adults (e.g., Aday, Kehoe, & Farney, 2006; Rote, Hill, & Ellison, 2013). Involvement in community activities such as volunteering has been linked to better well-being (Bath & Deeg, 2005; Harris & Thoresen, 2005). The Brief Assessment of Social Engagement (BASE) scale assesses engagement across a variety of social contexts (Morgan, Dallosso, Arie, Byrne, & Waite, 1987). The BASE scale consists of items on actual (e.g., attending religious services, activities on holidays, visiting library) and virtual (e.g., writing letter, reading newspapers/magazine) social activities, in effect, assessing social participation. Recent studies have used similar assessments of social engagement as an important dimension of healthy aging (de Leon, Glass, & Berkman, 2003); however, this body of research on social participation often neglects the interpersonal or social support component of social integration.

Social Integration as a Complex Construct

Given these multiple dimensions of social integration, to fully address the role of social integration in late life, it is essential to assess social integration comprehensively—including enacted and perceived interpersonal support and social engagement. In particular, identifying the varying social contexts in which older adults are integrated is important to understanding implications for late life development and aging. Ample studies address specific domains of social integration (Brissette et al., 2000); however, comprehensive measurement of social integration that expands beyond a focus solely on close interpersonal ties or community participation is frequently lacking.

Berkman and Syme (1979) were among the first to measure multiple dimensions of social integration simultaneously. In their Social Network Index (SNI), they included marital status, intimate interpersonal contacts, community group participation, and religious participation. Originally, marital status and intimate contacts were given 4 times the weight of community and religious participation; however, an unweighted version is also used (Seeman, Rodin, & Albert, 1993). This complex measure has been used frequently in research among the general population (Kawachi et al., 1996; Loucks et al., 2006) and less often with older adult populations (Michael, Colditz, Coakley, & Kawachi, 1999). While touted for its conciseness, one potential concern of using the SNI with older populations is its strong reliance on marital status, particularly given the high preponderance of widowhood among older adults and lack of decisive evidence supporting the use of widowhood as a proxy for social integration. Additionally, the SNI relies to a large extent on social network size, which has the potential to be more indicative of family size as opposed to social integration in later life.

More recent studies have initiated measurement of social integration from dual perspectives including both interpersonal and community engagement (Cornwell, Laumann, & Schumm, 2008; Fothergill et al., 2011). Cornwell and Waite (2009) developed a measure of social isolation (that they defined as the inverse of social integration), which incorporated dimensions of disconnectedness and perceived isolation. The social disconnectedness scale consisted of aspects of enacted social integration such as network size and social activities, whereas the perceived isolation scale consisted of perceptions of social integration such as social support. Although this work is comprehensive in its inclusion of multiple domains of social integration, it is somewhat limited in cohesiveness (i.e., draws items from various assessments with differing response scales).

These studies have advanced the measurement and understanding of social integration, yet there is still a need for a cohesive, multidimensional measure of social integration that is sensitive to normative transitions of later life. This study seeks to build upon previous work and develop a measure that combines interpersonal and community contexts of social integration into a comprehensive and balanced measure of enacted and perceived social integration that is unique to the lives of older adults.

Study Objectives

This study assesses social integration in older adults from a multidimensional perspective. The first study objective was to develop a measure of social integration that encompasses older adults’ involvement in social ties (interpersonal relationships with family, friends, and neighbors) and social contexts (community organizations, recreation and leisure activities, religious organizations, educational and volunteer opportunities) by incorporating both objective and subjective assessments. The second objective was to determine evidence for concurrent and convergent validity based on associations with demographic characteristics and similar constructs theoretically linked to social integration.

Research Design

Procedure

The data for this study are drawn from the Social Integration and Aging Study (collected by Fuller-Iglesias in 2013) a project approved by the Institutional Review Board of North Dakota State University. The Social Integration and Aging Study is a community based survey of social integration and well-being among older adults in a small metropolitan area in the Midwest. The study consisted of a written survey assessing social integration across various dimensions, social support network structure and quality, health and well-being (such as functioning, physical and psychological health), and demographic characteristics. The time burden to complete the survey was approximately 30 min. Individuals over aged 60 (N = 399) were recruited from senior-focused organizations and clubs including senior centers, community dining and meal-delivery programs, retirement and assisted living communities, and activity clubs and groups. Approximately 68% of the sample was recruited via mail; the response rate for mailed surveys was 34%, although this is likely an underestimate due to lack of response due to mobility, morbidity, and mortality of the targeted population. An additional 32% of the sample was recruited in person, with an estimated 70% participation. As an incentive, participants who returned the completed survey were entered into a drawing for grocery store gift certificates.

Sample

The demographic characteristics of the sample are shown in Table 1. All of the respondents were above the age of 60 years (range 60–100) with a mean age of 80.2 years old. The sample consisted of 286 (71.7%) women and 113 (28.3%) men. The majority of respondents reported at least high school degree (89%). A great majority of the sample (99%) identified their race as White Caucasian. Most participants were widowed (45%) or married (38%), although 9% were divorced/separated and 7% were never married. Most participants were retired (80%).

Table 1.

Demographic Characteristics of the Sample.

	Total (N = 399)
	Mean	SD	Range
Age	80.2	8.6	60–100
Years of education	12.9	2.5	0–17+
	%	N
Female	71.7	286
Caucasian	98.7	392
Married	38.3	153
Widowed	45.4	181
Retired	79.7	318

Measures

Multidimensional Social Integration in Later Life Scale (SILLS)

A battery of items assessed the frequency of and satisfaction with various social integration related activities. These items were developed based on synthesizing previous empirical and theoretical work. Items were designed to target domains of social ties (interpersonal relationships with family, friends, and neighbors; e.g., Antonucci, 1986; Berkman & Syme, 1979; Cutrona & Russell, 1987; Krause, 1995; Landerman et al., 1989; Sarason et al., 1987) and social contexts (community organizations; recreation and leisure activities; religious organizations; educational, employment, and volunteer opportunities; e.g., Cornwell & Waite, 2009; Demers et al., 2009; Loucks et al., 2006; Morgan et al., 1987; Utz et al., 2002). The frequency of involvement in each item was assessed on a 5-point Likert-type scale with responses of never (1), rarely (2), occasionally’ (3), frequently (4), and very frequently (5). The original 22 frequency items are denoted in Table 2. The satisfaction with participation in social integration activities was assessed on a 5-point Likert-type scale ranging from very dissatisfied (1) to very satisfied (5). The original 8 satisfaction items are denoted in Table 4. The subscale and overall scale formation and computation are described in the Results section.

Table 2.

Results of Two- and Three-Factor Exploratory Factor Analyses for Frequency of Social Engagement.

	Two-factor		Three-factor
	1	2	1	2	3
How often do you . . .
1. Get together with family		0.63	0.62
2. Speak to family on the phone		0.73	0.74
3. Get together with friends		0.74	0.73
4. Speak to friends on the phone		0.72	0.73
5. Visit with your neighbors		0.59	0.58
6. Attend meetings of a group, club, or organization	0.57			0.42	0.44
7. Attend a community event	0.73			0.57	0.50
8. Attend a religious service	0.48				0.62
9. Volunteer	0.72			0.70
10. Go on an outing (e.g., museum, movie, play)	0.70			0.68
11. Play social games	0.38				0.69
12. Attend a lecture or seminar	0.70			0.75
13. Interact with family or friends via Internet	0.28			0.53
Eigenvalue/rotated sum of squares loadings	2.9	2.7	3.7	3.3	1.6
% Variation explained	23	21	18	16	8
Excluded items
Due to low saliency:	Due to low communality:
14. Use Internet to make new friends	18. Visit a library (0.18)
15. Attend a skill building	19. Get together with coworkers (0.21)
16. Attend an educational class	20. Visit a gym (0.28)
17. Participate in political events	21. Visit a senior center (0.31)
	22. Eat out at a restaurant (0.35)

Note. Cross-loadings within <0.10 are presented. Only bolded items were included in the final solution.

Social network size

Total network size was calculated based on the prompt “Think about the people who are important to you right now or who provide personal support to you.” Respondents were allowed to list up to 20 people. This instrument was created based on multiple common network inventories (e.g., Antonucci, 1986; Brissette et al., 2000). Social network size was a count of the total number of social support ties identified.

Social network index (SNI)

Berkman and Syme’s (1979) SNI is based on marital status, interpersonal involvement, community involvement, and religious participation. An unweighted version was used (Loucks et al., 2006; Seeman at al., 1993) and scoring was calculated as follows: Married (no = 0, yes = 1); social network size (0–5 = 0, 6+ = 1); attends meetings of a group, club, or organization (no = 0, yes = 1); and frequent participation in religious meetings or services (no = 0, yes = 1). Scores were then summed across the four categories, with possible scores ranging from 0 (i.e., isolated) to 4 (i.e., socially connected).

Demographic characteristics

Age was a continuous variable that ranged from 60 to 100. Gender was coded as 0 (male) and 1 (female). Education level was included as an indicator of socioeconomic status (SES); the number of years of education was a continuous variable ranging from no formal education (0 years) to a graduate degree (17+). Participants identified their marital status as married/living with partner, widowed, divorced/separated, and never married. For the sake of analyses, marital status was defined as married (1) or not married (0).

Results

Exploratory Factor Analysis (EFA)

Scale development and analysis of fit were conducted in multiple steps. First, in order to explore the factor loading structure and attempt to reduce the number of social integration frequency items, an EFA was conducted in SPSS Version 22 (IBM, 2013b) on the 22 original frequency items (See Table 2). To begin, we examined the means and standard deviations (SD) of each item to identify items with low saliency and variability, as defined by having a mean lower than 2.0 (i.e., indicating most older adults rarely or never participated in these activities) and a SD lower than 1.0 (i.e., identifying low variability). The following 4 items were identified for deletion and subsequently dropped from further analysis: use Internet to make new friends, attend a skill building class, attend an educational class, and attending political events. The remaining 18 items were examined with EFA.

Three criteria were used to determine how many factors should be accepted. First, factors were only accepted if Eigenvalues were greater than 1. Six factors met these criteria in the initial exploratory analysis having Eigenvalues ranging from 5.1 to 1.0 and representing 60% of the variance in the scale items. Second, the scree plot was observed visually to determine the point at which the scree curve leveled parallel to the x-axis. The scree plot observation suggested a fit of three factors. Finally, a parallel analysis was conducted in SPSS (IBM, 2013b) using procedures described by O’Connor (2000) in order to determine the number of components to retain. Results from the parallel analysis suggested the retention of three factors. Given the results of these criteria, a three factor solution was explored further.

After conducting a factor analysis limited to three factors for the remaining 18 items, we first examined item communalities to determine whether any items had low communalities warranting deletion from the analyses. Items with very low communalities (<0.35) were eliminated 1 item at a time, with a new factor analysis conducted after each deletion, before determining whether to eliminate additional variables. During this process, the following 5 items were deleted: visit a library (0.18), get together with coworkers (0.21), visit a gym or fitness center (0.28), visit a senior center (0.31), and go out to eat at a restaurant (0.34). After these deletions, an examination of the scree plot suggested a two-factor solution. We thus conducted a parallel analysis with just the remaining 13 items and confirmed that a two-factor solution would now represent the best fit. The factor structures for the two- and three-factor solutions are presented in Table 2.

The remaining 13 items were explored with a factor analysis limited to two factors using Varimax orthogonal rotation. For interpretation of the rotated factors, salient loadings were defined as values greater than 0.40 and those lower than 0.40 were eliminated from the scale. Deletion was completed one step at a time, starting with the lowest loading, retesting to ensure remaining items remain nonsignificant, and then proceeding with nonsignificant loadings, following standard procedure (Kline, 2011). The following 2 items were deleted with this procedure: play social games (0.37) and interact with family or friends via Internet (0.28). Finally, within these 11 remaining items, we carefully examined the items to determine face validity. We concluded that from a theoretical standpoint, the item attend a lecture or seminar was more specified (i.e., less broad) than the other social activity items and had potential overlap with multiple other items (going on an outing; attending a meeting of a group, club, or organization; and attending a community event). Thus, the decision was made to delete the item attend a lecture or seminar for the sake of parsimony and face validity. The remaining scale consisted of 10 items equally represented across two factors.

Confirmatory Factor Analysis (CFA)

Social integration frequency subscales

CFAs for the frequency items were conducted on the two-factor solution determined by the EFA, as well as for a parsimonious one-factor solution. CFA was conducted for the 10 remaining frequency items in AMOS Version 22 (IBM, 2013a) using fit statistics to determine the structure of one- and two-factor solutions (see Table 3). The frequency of social integration one-factor and two-factor models both demonstrated good fit indices (one-factor: χ²/df = 1.85, comparative fit index [CFI] = 0.99, and root mean squared error of approximation [RMSEA] = 0.05; two-factor: χ²/df = 1.44, CFI = 0.99, and RMSEA = 0.03). The fit of the two-factor model was chosen as the best fitting model based on model fit, but also because it best matched theoretically driven distinctions in levels of social integration (i.e., the distinction between close interpersonal ties and participation in social activities). Therefore, the two-factor model was retained as the final solution for the frequency of social integration component of the scale. Table 4 presents the loadings, factor intercorrelation, and factor αs of the final solution. Factor 1 was defined by items that were related to older adults’ involvement in ongoing community activities (i.e., attending meetings of clubs or organizations, religious services, community events, volunteering, and going on an outing); we labeled this factor “frequency–social activities.” Factor 2 included items that represented the frequency of involvement with close interpersonal relationships (i.e., get together with family/friends, speak on the phone with family/friends, and visit with neighbors); we labeled this factor “frequency–social ties.” The items in each factor were averaged to create a mean subscale. The alphas for each subscale demonstrated good internal consistency as follows: social activities α = .72 and social ties α = .75.

Table 3.

Fit of the One- and Two-Factor Confirmatory Factor Analyses Models for Frequency and Satisfaction.

Frequency	χ²/df	NFI	CFI	TLI	GFI	RMSEA	BCC
One-factor	1.85	0.97	0.99	0.96	0.99	0.05	110.0
Two-factor	1.44	0.96	0.99	0.98	0.98	0.03	94.8
Satisfaction
One-factor	3.48	0.97	0.98	0.93	0.97	0.08	104.4
Two-factor	1.65	0.99	0.99	0.98	0.97	0.04	86.1

Note. NFI = normed fit index; CFI = comparative fit index; TLI = Tucker–Lewis index; GFI = goodness-of-fit index; RMSEA = root mean squared error of approximation; BBC = Browne-Cudeck Criterion.

Table 4.

Final Confirmatory Factor Analyses From Social Integration Frequency and Satisfaction Solutions.

	Frequency
	Community	Social ties
How often do you . . .
1. Get together with family		0.45
2. Speak to family on the phone		0.46
3. Get together with friends		0.82
4. Speak to friends on the phone		0.63
5. Visit with your neighbors		0.56
6. Attend meetings of a group, club, or organization	0.68
7. Attend a religious service	0.48
8. Attend a community event	0.67
9. Volunteer	0.56
10. Go on an outing (e.g., museum, movie, play)	0.56
Factor intercorrelation	.55
α	.72	.75
	Satisfaction
	Community	Social ties
How satisfied are you with your . . .
1. Relationships with close family members		0.57
2. Relationships with extended family members		0.62
3. Relationships with friends		0.68
4. Relationships with neighbors		0.76
5. Involvement in recreation/leisure activities	0.76
6. Participation in social gatherings	0.82
7. Involvement in or connection to your community	0.76
8. Participation in religious or spiritual activities	0.64
Factor intercorrelation	.73
α	.81	.84

Social integration satisfaction subscales

Eight items that assessed satisfaction with social ties and social activities were available. Because the content of these 8 items aligned with the frequency items, a CFA was conducted in AMOS (IBM, 2013a) using fit statistics to determine whether these satisfaction items fit one- and two-factor solutions consistent with those for the frequency items (See Table 3). The satisfaction with social integration one-factor model demonstrated questionable fit (χ²/df = 3.48, CFI = 0.98, and RMSEA = 0.08), whereas the two-factor model demonstrated good fit (χ²/df = 1.65, CFI = 0.99, and RMSEA = 0.04). The two-factor model was thus chosen as the best fitting model based on fit, but also because it best matched the distinction between interpersonal ties and social participation. Therefore, the two-factor model was retained as the final solution for the satisfaction with social integration component. Table 4 presents the loadings, factor intercorrelation, and αs of the final solution. Factor 1 was defined by items that were related to older adults’ satisfaction with social activities (i.e., involvement in leisure activities, social gatherings, connection to community, and religious activities); we labeled this factor “satisfaction–social activities.” Factor 2 included items that represented the satisfaction with interpersonal relationships (i.e., close family members, extended family members, friends, and neighbors); we labeled this factor “satisfaction–social Ties.” The items in each factor were averaged to create a mean subscale. The alphas for each subscale demonstrated good internal consistency as follows: social activities α = .81 and social ties α = .84.

Composite scale

Each of the four subscales were then summed to create an overall social integration score, which demonstrated very good reliability (α = .86). The final solution, labeled the Social Integration in Later Life Scale (SILLS), consisted of 18 items, with two 5-item subscales for frequency of social integration and two 4-item subscales for satisfaction with social integration.

Evidence of Scale Validity

To examine validity, Pearson’s correlation coefficients were conducted between the SILLS overall scale and subscales, demographic characteristics, and similar social integration constructs. Results are presented in Table 5.

Table 5.

Correlation Matrix of Demographic Characteristics, Social Integration Constructs, and SILLS Subscales and Composite Scores.

Variable	1	2	3	4	5	6	7	8	9	10
1. Age	—
2. Gender	−.05	—
3. Education	−.18	.05	—
4. Married	−.25	−.26	.01	—
5. Network size	−.18	.18	.23	.01	—
6. SNI	−.15	.03	.16	.50	.44	—
7. Frequency–social ties	−.04	.20	.05	.03	.22	.32	—
8. Frequency–social activities	−.15	.09	.19	.11	.21	.61	.43	—
9. Satisfaction–social ties	−.19	.13	−.02	.03	.17	.22	.54	.29	—
10. Satisfaction–social activities	.07	.04	.14	.06	.11	.30	.33	.39	.54
11. SILLS composite	−.02	.15	.12	.07	.24	.45	.72	.67	.75	.76

Note. Significant two-tailed correlations are in boldface. Correlations between .10 and .12 are significant at the p < .05 level. Correlations between .13 and .15 are significant at the p < .01 level. Correlations .16 and above are significant at the p < .001 level. Designation of strength of correlations: weak r < .3, moderate .3 < r < .6, and strong r > .6.

Concurrent validity

Scores on the SILLS should differ based on personal characteristics that are theoretically predictive of social integration. We assessed concurrent validity by examining the correlation of the SILLS to salient demographic characteristics (i.e., age, gender, education level, and marital status) that have frequently been identified as predictive of social integration (Berkman et al., 2000). Age was not correlated with the overall SILLS but was negatively correlated with the frequency–social activities (r = –.15, p < .01) and satisfaction–social ties (r = –.19, p < .001) subscales, such that even within this 60+ sample, older individuals reported lower levels of participation in social activities and satisfaction with social ties. Gender was correlated with the overall SILLS (r = 0.15, p < .01) with women reporting higher levels of social integration than men, although it seems this gender difference reflects significant gender differences in social ties subscales (frequency: r = .20, p < .001; satisfaction: r = .13, p < 0.01), but not the social activities subscales. Years of education was positively correlated with the overall SILLS (r = .12, p < .05); however, higher education was only correlated with the social activities subscales (frequency: r = .19, p < .001; satisfaction: r = .14, p < .01), but not the social ties subscales. The correlation between marital status and the composite SILLS was not significant, although married individuals more frequently engaged in social activities (r = .11, p < .05). In the case of marital status, the lack of more significant correlations may indicate an overreliance on marital status in the determination of social integration in previous research. With the exception of marital status, these correlations demonstrated significant effects in the expected directions and thus indicated good concurrent validity.

Convergent validity

The SILLS should be significantly associated with previously used constructs of social integration. To examine convergent validity, we calculated the correlation between social network size, the SNI, and the SILLS. These were selected in order to determine the relationship of the SILLS composite scale and subscales with measures that have been previously demonstrated as suitable measures of social integration, yet may be limited for use in later life due to reliance on constructs that transition in later life (e.g., marital status, network size). The SILLS was positively correlated with network size (r = .24, p < .001), indicating a weak association, and with the SNI (r = .45, p < .001), indicating a moderate association. We anticipated that the SILLS would be more closely related to the SNI and less closely related to social network size. Our expectation was that social network size would be weakly correlated to the SILLS, given that it is a theoretically distinct construct that serves as an imperfect proxy for social integration. These significant correlations are consistent with our expectations and demonstrate adequate convergent validity.

Unique characteristics of the SILLS

In examining the unique sociodemographic correlates of each of the three social integration constructs (i.e., social network size, SNI, and SILLS), we identified the unique characteristics of the SILLS. The SILLS was not correlated with age, whereas network size and the SNI were both negatively correlated with age. This suggests that the structure of the SILLS avoids the age bias that we suggest may be inherent when marital status and social network are key dimensions of social integration. Only the SILLS and network size were associated with gender. Women’s higher social integration as measured by the SILLS is consistent with theorized gender differences in interpersonal relationships in later life. All three social integration constructs were correlated with education level, indicating SES distinctions across social integration constructs. Only the SNI was associated with marital status. Although it is often suggested that married individuals are more socially integrated than widowed or single individuals, these data suggest this may not be the case. It is important to note that marital status is often used as a proxy for social integration or used in the calculation of social integration, as is the case with the SNI. It may be the case that although marital status is predictive of social integration, marital status should not be considered a proxy for or component of late life social integration.

The SILLS represents the sum of four subscales encompassing the frequency of and satisfaction with older adults’ integration in social activities and social ties. Network size was significantly, but weakly correlated with each of those subscales. The SNI was significantly correlated to each of the subscales, although the strength of correlations varied greatly with a weak correlation to the satisfaction–social ties subscale and a strong correlation to the frequency–social activities subscale. These correlations suggest, as expected, that network size is indicative of interpersonal relationships, whereas the SNI is more indicative of integration in community and productive activities. On the other hand, this evidence indicates the SILLS incorporates both of these integral components of social integration.

Discussion

This study describes the development of and provides preliminary evidence for the reliability and validity of the multidimensional SILLS. The SILLS is a measure of social integration, unique to later life, that contains 18 questions about the frequency and satisfaction of involvement in social ties and social activities. An analysis of the scale’s psychometric properties suggests it has good reliability (internal consistency = .86) and validity (concurrent and convergent). Additional research is needed to confirm these attributes as well as to extend the understanding of the psychometric properties of the SILLS. For instance, examination of test–retest reliability and predictive validity is warranted in the future.

We acknowledge that there are some limitations to this study. The SILLS distinguishes between frequency and satisfaction with social integration; however, because items are self-reported, the frequency items are somewhat subjective. Similarly, despite the added strength of the satisfaction items, they do not fully address perceived support. This measure does not indicate the value that an older adult places on the various dimensions of social integration, and thus there may be a mismatch in terms of desired social integration (i.e., low frequency of interacting with friends, but high satisfaction with friend relationships). Moreover, it is important to note that this measure does not assess social network, social roles, or relationship quality.

Because our aim was to develop a measure of social integration that would be unique to later life, this scale may not be applicable for younger age-groups. While all of the included items are applicable across adulthood, a notable omission is involvement in the workplace. Future work may consider testing a “work” item that may coincide with the “volunteer” items. Moreover, in the initial questionnaire, items related to online engagement were included; however, due to low variability, none of these items were included in the final solution. Although increasing numbers of older adults are showing interest in new technologies (i.e., e-mail, social networking) to keep in touch with families, friends, and other close members and to seek out information (Cotten, Ford, Ford, & Hale, 2014), the extent of online engagement was quite limited among the current sample. Future research should continue to explore older adults’ use of technology as well as other unique and new avenues to maintain their social integration. While this study is a first step in analyzing the SILLS, we note that the current sample is a convenience sample and thus is limited in its representation of the overall older adult population. Moreover, although the study sample demonstrated good age diversity, it lacked greatly in ethnic and to a lesser extent gender diversity. Thus, the SILLS necessitates further testing with more representative as well as targeted diverse samples. An important future step is to examine links between the SILLS and well-being to further determine the role of social integration for health in later life (Berkman et al., 2000).

The development of the SILLS adds to the body of empirical research on social integration by addressing the distinct context of later life and by fully incorporating interpersonal and community level dimensions of social integration within a cohesive measure. When compared to other complex measures of social integration (Cornwell & Waite, 2009; Loucks et al., 2006), the SILLS is unique in that it incorporates objective and subjective assessments of social integration and benefits from its lower reliance on marital status and network size. The combination of enacted and perceived social integration in the SILLS brings together these distinct lines of research permitting a more comprehensive understanding of the behavioral and cognitive dimensions of social integration. Moreover, the SILLS is uniquely valuable due to its sensitivity to normative transitions in marital status and changes in network size that are associated with increased age. The SNI has consistently documented value, especially in predicting health outcomes (Berkman et al., 2000), and is particularly useful due to its conciseness. Yet, we assert that the SILLS allows for greater variability and provides a fuller picture of older adults’ social integration regardless of marital status. We believe the SILLS, as a holistic measure of social integration that assesses enacted and perceived integration within social ties and activities, serves as a useful tool for measuring social integration in later life.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Abbott

K. M.

Bettger

J. P.

Hampton

Kohler

H. P.

(2012). Exploring the use of social network analysis to measure social integration among older adults in assisted living. Family & Community Health, 35, 322–333. doi:10.1097/FCH.0b013e318266669f

Aday

R. H.

Kehoe

G. C.

Farney

L. A.

(2006). Impact of senior center friendships on aging women who live alone. Journal of Women & Aging, 18, 57–73. doi:10.1300/J074v18n01_05

Ajrouch

K. J.

Blandon

A. Y.

Antonucci

T. C.

(2005). Social networks among men and women: The effects of age and socioeconomic status. Journals of Gerontology Series B-Psychological Sciences and Social Sciences, 60, S311–S317. doi:10.1093/geronb/60.6.S311

Amieva

Stoykova

Matharan

Helmer

Antonucci

T. C.

Dartigues

J. F.

(2010). What aspects of social network are protective for dementia? Not the quantity but the quality of social interactions is protective up to 15 years later. Psychosomatic Medicine, 72, 905–911. doi:10.1097/PSY.0b013e3181f5e121

Antonucci

T. C.

(1986). Hierarchical mapping technique. Generations, 10, 10–12.

Arcury

T. A.

Grzywacz

J. G.

E. H.

Saldana

Nguyen

H. T.

Bell

R. A.

… Quandt

S. A.

(2012). Social integration and diabetes management among rural older adults. Journal of Aging and Health, 24, 899–922. doi:10.1177/0898264312449186

Bath

P. A.

Deeg

(2005). Social engagement and health outcomes among older people: Introduction to a special section. European Journal of Ageing, 2, 24–30. doi:10.1007/s10433-005-0019-4

Berkman

L. F.

Glass

Brissette

Seeman

T. E.

(2000). From social integration to health: Durkheim in the new millennium. Social Science & Medicine, 51, 843–857. doi:10.1016/s0277-9536(00)00065-4

Berkman

L. F.

Syme

(1979). Social networks, host resistance, and mortality: A nine-year follow-up study of Alameda county residents. American Journal of Epidemiology, 109, 186–203.

10.

Brissette

Cohen

Seeman

(2000). Measuring social integration and social network. In Cohen

Underwood

Gottlieb

(Eds.), Social support measurement and interventions: A handbook for health and social scientists (pp. 53–85). New York, NY: Oxford University Press.

11.

Bronfenbrenner

(1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press.

12.

Bukov

Maas

Lampert

(2002). Social participation in very old age: Cross-sectional and longitudinal findings from BASE. Journals of Gerontology Series B: Psychological Sciences & Social Sciences, 57B, P510. doi:10.1093/geronb/57.6.P510

13.

Cacioppo

J. T.

Hawkley

L. C.

Thisted

R. A.

(2010). Perceived social isolation makes me sad: 5-year cross-lagged analyses of loneliness and depressive symptomatology in the Chicago Health, Aging, and Social Relations Study. Psychology and Aging, 25, 453. doi:10.1037/a0017216

14.

Cohen

(2004). Social relationships and health. American Psychologist, 59, 676. doi:10.1037/0003-066X.59.8.676

15.

Cohen

Doyle

W. J.

Skoner

D. P.

Rabin

B. S.

Gwaltney

J. M.

Jr . (1997). Social ties and susceptibility to the common cold. Journal of the American Medical Association, 277, 1940–1944. doi:10.1001/jama.1997.03540480040036

16.

Cornwell

Laumann

E. O.

Schumm

L. P.

(2008). The social connectedness of older adults: A national profile. American Sociological Review, 73, 185–203. doi:10.1177/000312240807300201

17.

Cornwell

E. Y.

Waite

L. J.

(2009). Measuring social isolation among older adults using multiple indicators from the NSHAP study. Journals of Gerontology B Social Sciences, 64B, i38–i46. doi:10.1093/geronb/gbp037

18.

Cotten

S. R.

Ford

Hale

T. M.

(2014). Internet use and depression among retired older adults in the United States: A longitudinal analysis. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 69, 763–771. doi:10.1093/geronb/gbu018

19.

Cutrona

C. E.

Russell

D. W.

(1987). The provisions of social relationships and adaptation to stress. Advances in Personal Relationships, 1, 37–67.

20.

de Leon

C. F. M.

Glass

T. A.

Berkman

L. F.

(2003). Social engagement and disability in a community population of older adults - The New Haven EPESE. American Journal of Epidemiology, 157, 633–642. doi:10.1093/aje/kwg028

21.

Demers

Robichaud

Gelinas

Noreau

Desrosiers

(2009). Coping strategies and social participation in older adults. Gerontology, 55, 233–239. doi:10.1159/000181170

22.

Depp

C. A.

Jeste

D. V.

(2006). Definitions and predictors of successful aging: A comprehensive review of larger quantitative studies. American Journal of Geriatric Psychiatry, 14, 6–20. doi:10.1097/01.JGP.0000192501.03069.bc

23.

Ertel

K. A.

Glymour

M. M.

Berkman

L. F.

(2008). Effects of social integration on preserving memory function in a nationally representative US elderly population. American Journal of Public Health, 98(7), 12–15.

24.

Ertel

K. A.

Glymour

M. M.

Berkman

L. F.

(2009). Social networks and health: A life course perspective integrating observational and experimental evidence. Journal of Social and Personal Relationships, 26, 73–92. doi:10.1177/0265407509105523

25.

Fiori

K. L.

Smith

Antonucci

T. C.

(2007). Social network types among older adults: A multidimensional approach. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 62, P322–P330. doi:10.1093/geronb/62.6.P322

26.

Fothergill

K. E.

Ensminger

M. E.

Robertson

Green

K. M.

Thorpe

R. J.

Juon

H. S.

(2011). Effects of social integration on health: A prospective study of community engagement among African American women. Social Science & Medicine, 72, 291–298. doi:10.1016/j.socscimed.2010.10.024

27.

Harris

A. H. S.

Thoresen

C. E.

(2005). Volunteering is associated with delayed mortality in older people: Analysis of the longitudinal study of aging. Journal of Health Psychology, 10, 739–752. doi:10.1177/1359105305057310

28.

IBM. (2013a). IBM AMOS for Windows, version 22.0. Armonk, NY: IBM Corp.

29.

IBM. (2013b). IBM SPSS Statistics for Windows, version 22.0. Armonk, NY: IBM Corp.

30.

Kawachi

Colditz

G. A.

Ascherio

Rimm

E. B.

Giovannucci

Stampfer

M. J.

Willett

W. C.

(1996). A prospective study of social networks in relation to total mortality and cardiovascular disease in men in the USA. Journal of Epidemiology and Community Health, 50, 245–251. doi:10.1136/jech.50.3.245

31.

Kline

R. B.

(2011). Principles and practice of structural equation modeling. New York, NY: Guilford Press.

32.

Krause

(1995). Negative interaction and satisfaction with social support among older adults. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 50, P59–P73. doi:10.1093/geronb/50B.2.P59

33.

Landerman

George

L. K.

Campbell

R. T.

Blazer

D. G.

(1989). Alternative models of the stress buffering hypothesis. American Journal of Community Psychology, 17, 625–642. doi:10.1007/BF00922639

34.

Loucks

Sullivan

D’Agostino

Larson

Berkman

Benjamin

(2006). Social networks and inflammatory markers in the Framingham Heart Study. Journal of Biosocial Science, 38, 835–842. doi:10.1017/S0021932005001203

35.

Mezuk

Rebok

G. W.

(2008). Social integration and social support among older adults following driving cessation. Journals of Gerontology Series B: Psychological Sciences & Social Sciences, 63B, S298–S303. doi:10.1093/geronb/63.5.S298

36.

Michael

Colditz

Coakley

Kawachi

(1999). Health behaviors, social networks, and healthy aging: Cross-sectional evidence from the Nurses’ Health Study. Quality of Life Research, 8, 711–722. doi:10.1023/A:1008949428041

37.

Morgan

Dallosso

H. M.

Arie

Byrne

E. J.

Jones

Waite

(1987). Mental health and psychological well-being among the old and the very old living at home. The British Journal of Psychiatry, 150, 801–807. doi:10.1192/bjp.150.6.801

38.

O’Connor

B. P.

(2000). SPSS and SAS programs for determining the number of components using parallel analysis and Velicer’s MAP test. Behavior Research Methods, Instrumentation, and Computers, 32, 396–402. doi:10.3758/BF03200807

39.

Pillemer

Moen

Wethington

Glasgow

(2000). Social integration in the second half of life. Baltimore, MD: Johns Hopkins University Press.

40.

Rote

Hill

T. D.

Ellison

C. G

. (2013). Religious attendance and loneliness in later life. The Gerontologist, 53, 39–50. doi:10.1093/geront/gns063

41.

Rowe

J. W.

Kahn

R. L.

(1997). Successful aging. The Gerontologist, 37, 433–440. doi:10.1093/geront/37.4.433

42.

Sarason

I. G.

Sarason

B. R.

Shearin

E. N.

Pierce

G. R.

(1987). A brief measure of social support: Practical and theoretical implications. Journal of Social and Personal Relationships, 4, 497–510. doi:10.1177/0265407587044007

43.

Seeman

T. E.

(1996). Social ties and health: The benefits of social integration. Annals of Epidemiology, 6, 442–451. doi:10.1016/s1047-2797(96)00095-6

44.

Seeman

T. E.

Rodin

Albert

(1993). Self-efficacy and cognitive performance in high-functioning older individuals MacArthur Studies of Successful Aging. Journal of Aging and Health, 5, 455–474. doi:10.1177/089826439300500403

45.

Shaw

B. A.

(2005). Anticipated support from neighbors and physical functioning during later life. Research on Aging, 27, 503–525. doi:10.1177/0164027505277884

46.

Thoits

P. A.

(1986). Multiple identities: Examining gender and marital status differences in distress. American Sociological Review, 51, 259–272. doi:10.2307/2095520

47.

Thomas

P. A

. (2011). Trajectories of social engagement and limitations in late life. Journal of Health and Social Behavior, 52, 430–443. doi:10.1177/0022146511411922

48.

Uchino

B. N.

Cacioppo

J. T.

Kiecolt-Glaser

J. K.

(1996). The relationship between social support and physiological processes: A review with emphasis on underlying mechanisms and implications for health. Psychological Bulletin, 119, 488–531. doi:10.1037/0033-2909.119.3.488

49.

Utz

R. L.

Carr

Nesse

Wortman

C. B.

(2002). The effect of widowhood on older adults’ social participation: An evaluation of activity, disengagement, and continuity theories. The Gerontologist, 42, 522–533. doi:10.1093/geront/42.4.522

50.

Zunzunegui

M.-V.

Alvarado

B. E.

Del Ser

Otero

(2003). Social networks, social integration, and social engagement determine cognitive decline in community-dwelling Spanish older adults. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 58B, S93–S100. doi:10.1093/geronb/58.2.S93