Benefit finding and well-being in older adults: The utility of the General Benefit Finding Scale.

The research aim was to test the General Benefit Finding Scale (GBFS) in a cross-sectional survey of adults aged 55 years and older and to assess to what extent it relates to mental well-being, perceived social support, health and personality. Participants (n=341) completed a questionnaire which included demographic questions, the GBFS, and measures of mental well-being, perceived social support, activity levels, and personality. Benefit finding was higher in older adults and correlated positively with mental well-being and perceived support. The impact of the psychosocial factors investigated on benefit finding levels was small (9.2%) with mental well-being, sex and agreeableness being significant predictors. The study provides additional support for the use of the GBFS and suggests that perceived social support, particularly from friends, is especially important for older adults.

1. Introduction

Health has been described as “not merely the absence of disease or infirmity” but as a “state of complete physical, mental and social well-being” (WHO, 1947) with a steady acceptance that national subjective well-being scores tend to be strongly correlated with objective health indicators (Morrison & Bennett, 2012), reflecting the quality of life in a society. They have been included in recent United Nations Development Program reports and will also be considered as part of British policy in the future (Diener, Oishi, & Lucas, 2015).

Over the past twenty years or so, attention has shifted from an emphasis on research into stressors and coping to looking at “adversarial growth” (Tennen & Affleck, 2002) or “benefit finding” (as cited in Cassidy, McLaughlin, & Giles, 2014, p. 268). There is increasing evidence that some people can actually derive some benefit in the face of adversity. Links have been found between benefit finding and health (Helgeson, Reynolds, & Tomich, 2006) in those suffering chronic or life threatening diseases. Furthermore, benefit finding in times of trauma has been linked to lower levels of depression and higher levels of well-being (Affleck, Tennen, Croog, & Levine, 1987; Bower, Kemeny, Taylor, & Fahey, 1998; McGregor & Antoni, 2009).

Benefit finding refers to an individual’s perceptions that they have derived some positive effects from challenging experiences, including illness, and traumatic events. It is generally defined as consisting of a number of dimensions including greater acceptance of one’s life situation, a strengthening of family bonds experience of growing psychologically as a person, affirmation of relationships, a sense of greater empathy with others, and a reappraisal of one’s life and reprioritization of goals (Cassidy et al., 2014; Helgeson et al., 2006).

Social and psychological resources have been found to mediate the relationship between stressors and benefit finding (Cassidy et al., 2014), but most research has focused on those who have experienced serious trauma, or serious or chronic disease. The 28-item General Benefit Finding Scale (GBFS) was devised Cassidy et al. (2014) to measure benefit finding levels in response to everyday hassles and provide a step in the direction of a deeper understanding of the resources required for preventative interventions to improve positive psychological health (Cassidy et al., 2014).

Although the sample in this study was impressively large, it was a very specific cohort - of university students with a mean age of 23.4 years, and the authors suggest the measures need to be tested in other populations. More people in the Western world are living longer (and birth rates are falling). In Ireland alone it is predicted that the number of people over 65 years will double over the next 30 years and those over the age of 80 will quadruple (Department of Health, 2015). This demographic change, described as a “ticking time bomb” (Wheatley, 2013), is not only likely to have financial consequences, but more attention will be required to ensure that individuals age in as healthy, productive and happy a way as possible.

In terms of positive psychological health for all ages, factors such as social support, mental well-being and physical activity are important correlates. Social support has been related to lower levels of anxiety, mortality and depression with the personality traits of extraversion, neuroticism and agreeableness more likely to be related to perceived social support (Allemand, Schaffhuser, & Martin, 2015). Personality traits, particularly conscientiousness and neuroticism, are associated with health and well-being (Deary, Weiss, & Batty, 2010; Goodwin & Friedman, 2006; Weiss, Bates, & Luciano, 2008).

The relationship between benefit finding and stress is far from consistent (Kleim & Ehlers, 2009) and indeed, the authors of the GBFS found that age, amongst other things, may be a factor in determining the linearity of the relationship. The trajectory of well-being in older adults was found to follow an inverted U-shape (Baird, Lucas, & Donnellan, 2010), with a similar trajectory, associated with self-reported health status and chronic illness, also recently found between optimism and age (Chopik, Kim, & Smith, 2015).

Resilience and an active engagement in life are amongst a range of psychosocial factors considered to be important for healthy ageing (Hicks & Conner, 2014). Loss of a partner or retirement can result in loneliness, a risk factor for Alzheimer’s disease (Wilson et al., 2007) and changes in social interaction, such as social engagement or emotional support, have repercussions on levels of life satisfaction (Huxhold, Fiori, & Windsor, 2013). As age increases, so does the propensity for chronic disease, disability, cognitive decline, and bereavement, with consequent risks to positive health (McKee & Schüz, 2015). Key risks for health in older adults include social isolation, reduced physical activity, and poor mental health (WHO, 2012). In contrast, social integration / support, physical activity, personality, and benefit finding are protective in terms of mental health (Cassidy et al., 2014).

The aim of this study was to assess if the factor structure of the GBFS would be evident in a sample of older adults in response to everyday stressors, and to explore if benefit finding would be related to well-being, support, personality and physical activity.

2. Method

2.1. Sample and Procedures

Following ethical approval, permission was granted to recruit adults, aged years and older, living in the community and of good cognitive health. This allowed assessment of the potential effect of ill or deteriorating health/physical activity on benefit finding levels. Quantitative data collection consisted of an online, selfadministered questionnaire. To protect participants’ identities, no identifying information was present in either case.

Power analysis using G*Power suggested for a medium effect size and power of .95 that a minimum of 138 participants was required. To allow for non-responders, a sample size of approximately 200 participants were targeted. Participants were recruited by various measures. The Aer Lingus Active Retirement group (over 3000 members), was approached and agreed to host the survey. The survey was also hosted on a number of other Facebook pages and various community members were asked to distribute the link to anybody aged 55 years or older. Paper-based surveys were distributed locally and to relatives and friends of the researcher. All were returned anonymously to her address. All participants were provided with an information sheet and advised that their participation was entirely voluntary and that their responses would be kept confidential. Completion of the questionnaire was taken as implicit consent.

A total of 352 responses were received - 19 were paperbased and the remainder online. No differences were observed between the online and paper-based responses. Eleven participants had missing data and were removed from the dataset leaving a total of 341 participants (241 females and 100 males). Ages ranged from 55 years to 83 years with a mean age of 62.7 (sd=6.6). Over two thirds (68.2%) were married, 12.5% were single and the remainder were widowed or divorced (9.9% and 9.3% respectively). The vast majority (99.4%) lived in their own home.

2.2 Measures

Benefit Finding: The 28-item General Benefit Finding Scale (GBFS), which incorporates six different factors, was used to ascertain levels of benefit finding in relation to general life stress. The authors found it to have good internal reliability reporting Cronbach’s Alpha for the six factors as follows - acceptance (𝛼 = .86), family bonds (𝛼 = .76), personal growth (𝛼 = .81), relationships (𝛼 = .83), empathy (𝛼 = .80) and reprioritisation (𝛼 = .82) (Cassidy et al., 2014). Cronbach’s Alphas in the current data are shown in Table 1. Personality: Big Five Personality constructs were assessed using the BFI-10 - short form (Rammstedt & John, 2007). This short, one minute, 10-item measure has been found to have significant levels of validity and reliability with mean intercorrelations when tested on two different cultures (Rammstedt & John, 2007) of .11 and mean Cronbach’s Alpha reported as (𝛼 = .75).

Table 1

Items, factor loadings and psychometric data for the new scale

Total % Variance =67.1	Factor loadings
Acceptance: 𝛼=.79 Eigenvalue =4.10 % Variance =14.7
Led me to be more accepting of things.	.66
Taught me how to adjust to things I cannot change.	.71
Helped me take things as they come.	.75
Given me a more realistic set of expectations.	.61
Taught me to be patient.	.61
Family bonds: 𝛼=.85 Eigenvalue =3.45 % Variance =12.3
Brought my family closer together.	.78
Made me more sensitive to family issues.	.68
Helped me appreciate my family more.	.82
Made me more aware of what my family means to me.	.75
Growth: 𝛼=.90 Eigenvalue =3.43 % Variance =12.2
Made me a more effective person.	.67
Taught me how to cope more effectively.	.77
Helped me become a stronger person.	.80
Taught me I can handle most things.	.79
Led me to deal better with problems.	.74
Helped me to grow emotionally and spiritually.	.69
Relationships: 𝛼=.72 Eigenvalue =3.01 % Variance =10.7
Helped me become more aware of the support available from others.	.69
Helped me realize who my real friends are.	.70
Led me to feel more positive about others.	.60
Led me to meet people who have become some of my best friends.	.42
Empathy: 𝛼=.86 Eigenvalue =2.75 % Variance =9.8
Made me more compassionate to those in similar situations.	.81
Made me more sensitive to the needs of others.	.73
Made me care more about others.	.54
Made me closer to people I care about.	.45
Taught me that everyone has a right to be valued.	.56
Reprioritisation: 𝛼=.90 Eigenvalue =2.04 % Variance =7.3
Led me to place less emphasis on material things.	.76
Led me to live my life more simply.	.81
Led me to change my priorities in life.	.83
Helped me become more focused on real priorities.	.75

In this data the Cronbach Alphas were: extraversion (𝛼 = 0.82), conscientiousness (𝛼 = 0.76), neuroticism (𝛼 = 0.73), openness (𝛼 = 0.71), and agreeableness (𝛼 = 0.70).

Social Support: The short version of the Lubben Social Network Scale (LSNS-6; J. E. Lubben, 1988) is a six item measure which assesses the size, type, closeness and frequency of social contacts in adults thereby measuring levels of social isolation. The scale differentiates between family and non-family but does not differentiate between friends and neighbours. It was constructed initially for use with older adults (aged 65 years and above) although it has been successfully utilized in both research and practice settings for adults of all ages, some as young as 18 (Honeycutt, Nasser, Banner, Mapp, & DuPont, 2008; Howarter & Bennett, 2013; McConkey & Leavey, 2013; Momtaz, Haron, Ibrahim, & Hamid, 2014). High levels of reliability and consistency have been reported (J. Lubben et al., 2006) with Cronbach’s Alpha values ranging from 𝛼 = 0.84 − 0.89 for the family subscale and (𝛼 = 0.80 − 0.82) for the friends subscale. In the current study Cronbach Alphas were friends (𝛼 = 0.84) and family (𝛼 = 0.81).

Activity Levels: Level and intensity of participants’ activity levels were assessed using the Rapid Assessment of Physical Activity (RAPA; Topolski et al., 2006). This short 9-item measure was devised using a cohort of older adults aged 51-92. With test-retest reliability scores ranging between (𝛼 = 0.65) (Vega-López, Chavez, Farr, & Ainsworth, 2014) and (𝛼 = 0.67) (Silva, Queirós, Alvarelhão, & Rocha, 2014), it has been deemed to be a valid and easy-to-use measure of physical activity in older adults (Topolski et al., 2006). In the current data the Cronbach’s Alpha was .94.

Mental Well-Being: The Warwick-Edinburgh Mental Well-Being Scale (WEMWBS) to assess mental wellbeing has been found to be psychometrically robust and suitable for use for those over 16 years of age (Tennant et al., 2007). In an effort to reduce the overall number of questions, the short 7-item version was used (SWEMWBS). This version has been found to be unidimensional and largely free of bias, and the level of reliability has been reported as (𝛼 = 0.85) (Stewart-Brown et al., 2009). The Cronbach Alpha in the current study was .86.

Demographic questions: The demographic questions included gender, age, and relationship status.

Ethical issues: Following careful consideration, the University of Ulster Filter Committee deemed that any ethical issues had been identified and addressed and were not a cause for concern.

Statistical analysis: Confirmatory factor analysis using the AMOS programme was used to test the factor structure of the scale. One-sample t-tests were used to compare the mean scores in this sample to the means in the original study. Hierarchical Multiple Regression Analysis (HMRA) was used to test the partial correlations between benefit finding, well-being, support and physical activity.

3. Results

The first aim of this study was to test the psychometric properties of the General Benefit Finding Scale (GBFS) in an older adult population. The scale was initially developed and tested in a student sample. To test the factor structure a Confirmatory Factor Analysis was applied using the Amos programme on SPSS. In testing model fit there is still some debate about cut off values for goodness of fit. For this study we used the review by Hooper, Coughlan, and Mullen (2008), and specifically the following cut offs. A non-significant chi-square or a chi-square to degrees of freedom ratio of 3:1 or less (CMin/Df), a Comparative Fit Index (CFI) of .95 or greater, an Incremental Fit Index (IFI) of .95 or greater, a Root Mean Square Error of Approximation (RMSEA) of .08 or less (ideally as close to .05 as possible), and a Standardised Root Mean Square Residual (SRMR) of less than .08. The initial model had a 𝜒2 of 50.39 with 8 degrees of freedom (𝑝 < .001), with a CFI =.95, an IFI =.95, a RMSEA =.12, and a SRMR =.03. The modification index indicated that the model fit could be improved by allowing reprioritization to correlate with both acceptance and relationships. The new model thus produced had a 𝜒2 of 14.95 with 6 degrees of freedom (𝑝 < .05), CFI =.99, an IFI =.99, a RMSEA =.06 (CI: .02-.08), and SRMR =.02. Although the 𝜒2 is significant this may be because of the large sample size and the relative / normed 𝜒2 (𝜒2:df) is 2.4:1 which is lower than the accepted cut off of 3:1 recommended by Kline et al. (2005). Overall the model is a good fit.

We used the set of criteria proposed by Lamping et al. (2002) and summarised by Smith et al. (2005) to set out some of the psychometric properties of the data (see Table 2). The data used in this study was based on a cross sectional survey so testretest reliability was not tested. However, the items and factors perform well on all the other dimensions as shown in Table 2.

Table 2

Psychometric criteria and scale performance

Property	Criteria for acceptability	Performance of scale
Item analysis/ reduction	All items should have factor loadings> 0.30	Factor loadings range 0.45 to 0.83
Item analysis/ reduction	Missing data < 5%	No missing data
Item analysis/ reduction	Inter-item correlations < 0.75	Inter-item correlations range 0.35 to 0.68
Item analysis/ reduction	Item total correlations ≥ 0.25	Item total correlations range 0.45 to 0.76
Item analysis/ reduction	Maximum endorsement frequency < 80%	Maximum endorsement frequency = 72.3%
Item analysis/ reduction	Minimum adjacent endorsement > 10%	Minimum adjacent endorsement = 18.9%
Acceptability	Skewness values < 1	Maximum skewness 0.89
Acceptability	Missing data < 5%	No missing data
Reliability	Cronbach’s Alpha > 0.70	Cronbach’s Alphas range 0.72 to 0.90
Reliability	Item total correlations ≥ 0.20	Item total correlations range 0.45 to 0.76

The next aim was to compare levels of benefit finding in the current sample with those in the original sample; one-sample t-tests were performed on benefit-finding in general, and on each of the individual variables within benefit-finding. Contrary to expectation, on all tests, the current sample scored significantly higher than the original sample (𝑝 = 0.001) - see Table 3 for details. Benefit finding in general, t(340) = 13.4, 𝑝 < .001, growth, t(340) = 11.9, 𝑝 < .001), reprioritisation, t(340) = 12.99, 𝑝 < .001, empathy, t(340) = 19.07, 𝑝 < .001, family bonds, t(340) = 6.19, 𝑝 < .001, acceptance, t(340) = 3.89, ?? < .001 and relationships, t(340) = 8.35, 𝑝 < .001.

Table 3

Differences between males and females and between current sample and original sample on benefitfinding

Variable	Male	Female	Current	Original
	(𝑛=100)	(𝑛=241)	Sample	Sample
	Mean (Sd)	Mean (Sd)	Mean (𝑀+𝐹)	Mean (𝑀+𝐹)
Benefit-finding	20.99(3.77)	22.13(3.44)	3.63	3.2
Empathy	3.78(0.84)	4.05(0.69)	3.97	3.2
Family Bonds	3.54(0.87)	3.76(0.88)	3.70	3.4
Relationships	3.27(0.77)	3.50(0.71)	3.43	3.1
Acceptance	3.39(0.75)	3.47(0.67)	3.45	3.3
Growth	3.49(0.77)	3.64(0.77)	3.60	3.1
Reprioritisation	3.53(0.98)	3.71(0.91)	3.65	3.0

Next, two separate hierarchical multiple regression analyses (HMRA) were undertaken to explore partial correlations between the study variables. In the first analysis well-being was entered as the dependent variable (Table 4). On the first step age and sex were entered as predictor variables. On the second step the personality dimensions (extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience) were entered. On the third step support was added and on the fourth step we entered physical activity. Finally, benefit finding was entered. Overall the model accounted for a small but significant percentage of variance in well-being (𝑅2 = .196, F (10, 330) = 8.047, 𝑝 < .001). The individual significant partial correlations were extraversion(𝛽 = .142, 𝑝 < .01, neuroticism (𝛽 = −.153, 𝑝 < .01), support from friends (𝛽 = .348, 𝑝 < .001), and benefit finding (𝛽 = .141, 𝑝 < .01).

Table 4

Partial correlations with well-being from Hierarchical Multiple Regression

	B	SE B	𝛽	𝑃
		??2 = .007, F (2, 338) = 1.211, 𝑝 = .299
Sex	−.577	.494	−.063	.244
Age .	034	.034	.054	.324
		𝑅2 = .079, F (5, 333) = 5.762, 𝑝 < .001
Sex	−.433	.488	−.047	.376
Age	.031	.034	.049	.364
Extraversion	.271	.118	.129	.022
Agreeableness	.193	.129	.088	.135
Conscientiousness	.095	.124 .	047	.447
Neuroticism	−.390	.127	−.170	.002
Openness	−.118	.116	−.059	.309
		𝑅2 = .082, F (1, 332) = 32.594, 𝑝 < .001
Sex	−.585	.467	−.064	.211
Age .	029	.032	.047	.361
Extraversion	.303	.113	.145	.008
Agreeableness	.358	.126	.164	.005
Conscientiousness	.254	.122	.126	.038
Neuroticism	−.375	.121	−.163	.002
Openness	−.033	.112	−.016	.769
Support from friends	.154	.027	.324	.000
𝑅2 = .01, F (1, 331) = 3.892, 𝑝 < .05
Sex	−.521	.466	−.057	.265
Age	.032	.032	.051	.321
Extraversion	.285	.113	.136	.012
Agreeableness	.293	.130	.134	.025
Conscientiousness	.177	.127	.088	.165
Neuroticism	−.368	.120	−.160	.002
Openness	−.078	.114	−.039	.491
Support	.178	.029	.375	.000
Physical activity	.107	.054	.144	.048
		𝑅2 = .018, F (1, 330) = 7.558, 𝑝 < .01
Sex	−.707	.467	−.078	.131
Age	.026	.032	.042	.407
Extraversion	.297	.112	.142	.008
Agreeableness	.241	.130	.110	.065
Conscientiousness	.180	.126	.089	.155
Neuroticism	−.374	.119	−.163	.002
Openness	−.075	.113	−.037	.507
Support	.165	.030	.348	.000
Physical activity	.098	.054	.133	.068
Benefit finding .	982	.357	.141	.006

The second HMRA followed the same format except that well-being was removed and benefit finding entered as the dependent variable (Table 5). Again, the model accounted for a small but significant percentage of variance in benefit finding (𝑅2 = .092, 𝐹 (9, 331) = 4.913, 𝑝 < .01). The individual significant partial correlations were sex (𝛽 = .145, 𝑝 < .01), agreeableness (𝛽 = −.172, 𝑝 < .01), and support from friends (𝛽 = .189, 𝑝 < .001). Females scored significantly higher on benefit finding than males (t (339) = 2.699, 𝑝 < .01).

4. Discussion

At the outset, it was expected that benefit finding levels would be lower in the older sample, given the adverse health consequences and sometimes negative changes in social interaction that aging can bring. Contrary to expectations, the older adults in this study scored higher than the university student sample on general benefit finding levels and on every individual element of the benefit finding scale. Both a curvilinear and linear relationship between post-traumatic stress disorder and post-traumatic growth was found in a recent metaanalysis by Shakespeare-Finch and Lurie-Beck (2014), with the type of event and the age of the individual believed to possibly affect the linearity of the relationship.

Longitudinal research has found that aging effects may differ depending on the cohort examined (Jivraj, Nazroo, Vanhoutte, & Chandola, 2014; Tampubolon, 2015). Recent findings have noted generational differences, with increased dissatisfaction and lower subjective well-being in younger cohorts (Baird et al., 2010). Those who participated in this study were aged 55 years and older and therefore from the Baby Boomer generation. Recent evidence has shown that Baby Boomers are significantly more likely to rate their health as “excellent” than those in the following Generation X cohort (Pilkington, Taylor, Hugo, & Wittert, 2014) and that the greater the emotional support provided by both parents of Baby Boomers, the more positive the impact on their emotional well-being in later life (Poon & Knight, 2013). The university students studied in the Cassidy et al. (2014) research, however, were from the Millennial generation, recently described as being “highly optimistic” and living in an age of “soaring expectations and crushing realities” (Twenge, 2014, p3)

Cassidy et al. (2014) found that benefit finding grows within a framework of psychological and social resources. However, with significant social and economic changes since the Baby Boomer generation were children, we may be witnessing the emergence of cohort effects in terms of benefit-finding. If a cohort effect does exist in terms of benefit-finding, then it is possible that our focus may be in the wrong place - other factors such as changing parenting techniques, changes in social interactions, or a more individualistic society may have resulted in a greater number of less happy, less resilient people, less able to find benefit in adverse situations. This could possibly explain the unexpected results regarding the higher benefit finding levels found in the older sample. The cross-sectional nature of this study, however, limits the ability to draw any conclusions on whether a cohort effect exists.

Future research could look at these generational effects more closely and also extend the research to cross cultural analysis. Generational effects no doubt play a part in the health and well-being of individuals through changing environments and social conditions. This is further complicated by the growth of multi-cultural societies and is likely to be differentially distributed across other cultures and societies.

The psychometric properties performed well in this study with good levels of reliability and internal consistency shown and the criteria suggested by Lamping et al. (2002) were clearly met, similar to the original study. When testing for goodness of fit, it was found that allowing reprioritisation to correlate with acceptance and relationships improved the model substantially. This concurred with the original study where these factors were also allowed to correlate and, once again, the GBFS was found to be a good fit for the data when tested on this different, older sample.

The total benefit finding score was significantly positively related to well-being, to agreeableness and to all three elements of the support received from friends. Of the Big Five personality types examined, the only significant inverse correlation found was between neuroticism and the acceptance element of benefit finding.

The big surprise, with respect to the psychosocial factors, was that no correlation was found between benefit finding total score and social support from family. This contradicts other recent research (Boehmer, Glickman, Winter, & Clark, 2014; Brand, Barry, & Gallagher, 2016; Dunn, Occhipinti, Campbell, Ferguson, & Chambers, 2011; Măirean, 2016) but does lend support to a recent systematic review which found that evidence for the influence of social support on benefit-finding levels was inconclusive (Pascoe & Edvardsson, 2013). This finding suggests that, in terms of benefit finding, the perceived support from friends may be more important than that of family members and concurs with Huxhold, Miche, and Schüz (2013)) who found that, as people age, social interaction with friends becomes more important than with family members, and can serve to protect against some of the negative effects of aging. When considering the findings in this study, several possible limitations must be taken into account. Firstly, benefit-finding is reliant on self-report measures. Since the relationship between self-reported growth and actual growth has been found by some to be small (Frazier et al., 2009) and, in fact, only “fairly accurate” among certain individuals (Gunty et al., 2011, p65), the suggestion that benefitfinding may have illusory aspects (Zoellner & Maercker, 2006) may need to be investigated further.

Surprisingly, only 9.2% of the variance in benefitfinding in the current study could be attributed to the significant psychosocial factors investigated. Perhaps other psychosocial factors may be more important. For example, neither religiosity nor ethnicity were considered, yet lower levels of religiosity have been associated with lower levels of benefit-finding (Kleim & Ehlers, 2009; Sim, Lee, Kim, & Kim, 2015; Tsai, El-Gabalawy, Sledge, Southwick, & Pietrzak, 2015; Urcuyo, Boyers, Carver, & Antoni, 2005) and ethnicity may be a potential predictor of post-traumatic growth (Helgeson et al., 2006; Kleim & Ehlers, 2009).

The cross-sectional nature of this study is perhaps the greatest limitation. Much of the research into benefit finding has been cross-sectional (Helgeson et al., 2006) and can therefore only give a reflection of the situation at a particular moment in time. It does not allow for any interpretation of the relationships found or causal claims, and any suppositions made within this paper are moot until satisfactory longitudinal analysis is conducted.

5. Conclusion

This research paper does, however, add to the literature both on benefit-finding and on older adults. It has provided additional support for the use of the GBFS and has demonstrated that it is a reliable measure and a good fit for the data across different populations.

The findings regarding the effect of perceived social support from friends on benefit finding could have useful applied benefits for those involved in the promotion and maintenance of health in older people. Longitudinal research into aging and subjective well-being in later life has found that quality of life is negatively affected over time by depression, functional limitations and low levels of perceived social support (Zaninotto, Falaschetti, & Sacker, 2009). Subjective well-being is also adversely affected by isolation and loneliness in older adults (Rafnsson, Shankar, & Steptoe, 2015). Supporting independent living and enhancing the quality of people’s social network can help to minimise much of the age effects on quality of life (Zaninotto et al., 2009). The finding that the perceived support from friends appears to have a greater impact on benefit-finding than that from family means that opportunities in the future to improve and maintain older adults’ benefit-finding levels may, in fact, be greater. For a variety of reasons, the availability of supportive relatives cannot be relied upon; however, it may be easier for those who are involved with older adults to work instead on ensuring that a wide network of friends and social acquaintances is maintained.

Table 5

Partial correlations with benefit finding from hierarchical regression analysis

	B	SE B	𝛽	𝑃
		𝑅2 = .028, F (2, 338) = 4.913, 𝑝 < .01
Sex	.193	.070	.148	.006
Age	.008	.005	.085	.114
		𝑅2 = .019, F (5, 333) = 1.362, 𝑝 = .238
Sex	.195	.071	.149	.007
Age	.005	.005	.060	.272
Extraversion	−.014	.017	−.045	.432
Agreeableness	.048	.019	.152	.012
Conscientiousness	−.008	.018	−.026	.675
Neuroticism	.005	.019	.015	.788
Openness	−.006	.017	−.020	.735
		𝑅2 = .020, F (1, 332) = 7.008, 𝑝 < .01
Sex	.184	.071	.141	.010
Age	.005	.005	.059	.275
Extraversion	−.011	.017	−.038	.510
Agreeableness	.059	.019	.189	.002
Conscientiousness	.004	.019	.012	.846
Neuroticism	.006	.018	.019	.740
Openness	.000	.017	.001 .988
Support from friends	.011	.004	.159	.009
		𝑅2 = .003, F (1, 331) = 1.203, 𝑝 = .273
Sex	.190	.071	.145	.008
Age	.006	.005	.061	.258
Extraversion	−.013	.017	−.043	.454
Agreeableness	.054	.020	.172	.007
Conscientiousness	−.003	.019	−.010	.882
Neuroticism	.007	.018	.020	.715
Openness	−.004	.017	−.012	.836
Support	.013	.004	.189	.004
Physical activity	.009	.008	.085	.273