The co-development of perceived support and the Big Five in middle and older adulthood

Abstract

The current study examined whether relationships also influence personality trait development during middle and older adulthood, focusing on the individual’s perception of support from the relationship partner. Using data from the Health and Retirement Study (n = 20,422; mean age = 65.9 years), we examined the longitudinal relationships between Big Five personality trait levels and perceived support from children, family, friends, and spouses. Results found that participants who reported more positive social support and lower negative support also tended to score higher on conscientiousness, extraversion, agreeableness, and openness to experience, but lower on neuroticism. Moreover, changes in positive support across relationship partners coincided with trait changes over time, in the form of more positive support was associated with seemingly adaptive changes on the Big Five. Findings are discussed with respect to identifying social influences on personality development in adulthood.

Keywords

adult development older adulthood personality trait development social support

Contrary to the more normative fluctuations on personality traits evident during the adolescent and emerging adult years, it appears that personality trait development during middle and later adulthood are better characterized by stability rather than change, both when considering mean-level (Roberts, Walton, & Viechtbauer, 2006) and rank-order (Roberts & DelVecchio, 2000) stability. That said, research consistently points to the capacity for individual differences in trait change during these developmental periods (e.g., Lucas & Donnellan, 2011; Specht, Egloff, & Schmukle, 2011), though the factors implicated in these changes remain largely unclear. When considering potential catalysts for change, it has been suggested that researchers should revisit the possibility that social relationships play a role in shaping personality trait development in adulthood (Neyer, Mund, Zimmerman, & Wrzus, 2014).

Social relationships and personality development

Multiple longitudinal studies have now demonstrated the potential for personality traits and social relationship variables, such as quality and support, to develop in tandem. For instance, research has demonstrated that the Big Five personality traits predict changes in social relationship variables during adolescence and early adulthood (Asendorpf & van Aken, 2003; Asendorpf & Wilpers, 1998). Moreover, research has shown evidence for correlated changes between social relationships and the Big Five traits during this developmental period (e.g., Neyer & Lehnart, 2007), and into middle adulthood (Mund & Neyer, 2014). That said, it is important to note that these studies emphasize the potential for nuance in the longitudinal associations between the Big Five and relationship variables; findings from these studies run counter to any suggestions that adaptive changes in social relationships correspond with seemingly positive changes in manifold across the Big Five, or that these longitudinal associations will be consistent across different indices of social relationships. As an example, Mund and Neyer (2014) found that personality-relationship associations differed with respect to (a) investigating positive versus negative relationship aspects, (b) the Big Five trait of interest, as some (such as neuroticism) may exhibited more significant correlated changes with relationship variables than other traits (e.g., openness to experience), and (c) the social partner being considered.

Research has, however, been limited with respect to investigating changes in social relationships and personality change later in life, with existing work typically focusing on two fronts. First, research has investigated how specific social role losses may influence personality change, typically with respect to romantic partners. For instance, research demonstrates only modest and often inconsistent effects of divorce on personality trait trajectories in adulthood (Allemand, Hill, & Lehmann, 2015; Costa, Herbst, McCrae, & Siegler, 2000; Specht et al., 2011). However, other social role events, such as children leaving home, death of a parent, and widowhood, fail to significantly predict trait change (Specht et al., 2011). Research on this front is restricted though by the typically low sample sizes that have experienced a given event, as well as uncertainty regarding the subjective nature of the event. Accordingly, researchers also have considered how perceptions of social engagements may predict or coincide with trait change.

Second, researchers have examined the role of social support and social well-being on trait change, as these variables may provide greater insight into participants’ perceptions of their social environments. For instance, research suggests that older adults who perceive stronger social support are those more likely to gain on conscientiousness over time (Hill, Payne, Jackson, Stine-Morrow, & Roberts, 2014). In addition, research with middle-aged adults shows correlated changes between the Big Five traits and perceived social support, in the form of increased support being associated with seemingly adaptive personality trait changes (Allemand, Schaffhuser, & Martin, 2015). Similarly, research suggests that longitudinal changes on social well-being are tied to Big Five trait changes (Hill, Turiano, Mroczek, & Roberts, 2012). Specifically, middle adults who reported increases on social well-being between assessments almost a decade apart tended also to increase on extraversion, agreeableness, conscientiousness, and openness, but decrease on neuroticism. A primary limitation of these studies though is their inability to investigate specific social relationships (e.g., family versus friends), due to the broad nature of the social perceptions being measured. These distinctions are important given that research has noted that adults distinguish familial and friend relations with respect to perceived levels of closeness and reciprocity, with differences even evident across specific familial relations (Neyer, Wrzus, Wagner, & Lang, 2011), underscoring the need for research to move beyond broader social perception variables.

The comparison of support from different social partners is particularly important to consider during middle and older adulthood for at least three reasons. First, the benefits accrued from social networks appear to differ by social partner; for instance, research suggests that friend networks are more predictive of mortality risk among older adults than those with respect to family members (Giles, Glonek, Luszcz, & Andrews, 2005). Second, given the greater potential for older adults to experience losses in given relationships, an aggregate perception of social support may fail to appropriately capture the more acute changes occurring with respect to any individual social role. Third, previous research with adolescents and younger adults has suggested that the associations between the Big Five personality traits and perceived support may differ across social relationships (e.g., Branje, van Lieshout, & van Aken, 2004). As such, one would predict that not all forms of social support equally influence personality development in adulthood, though this question remains underexplored in older adult samples.

The current research addressed this limitation by considering how different social relationships play a role on personality trait development during middle and older adulthood. Specifically, we employed two-wave data from the Health and Retirement Study of American adults to investigate whether perceived positive or negative support from family and friends predicted changes on the Big Five, or if changes on these social variables coincide with personality trait changes. One would expect that having positive social relationships would motivate individuals toward dispositional characteristics that maintain or promote social relations, such as greater conscientiousness and agreeableness, along with reduced neuroticism. However, it is unclear whether these results would hold across different social partners. Though we investigated predictive effects (i.e., social support predicting trait change), correlated changes between social support and personality trait change are more likely to be evidenced given that some of the sample only had two longitudinal assessment occasions.

Method

Participants

A subsample of the Health and Retirement Study (HRS) was used for these analyses. Participants are eligible to be recruited into the study during a recruitment period, which occurs roughly once every 6 years, if they or their spouses are 55 years of age or older. While the HRS has collected data every 2 years starting in 1992, the personality measures used in the current study have been included since 2006. Participants receive this questionnaire every other wave, or every 4 years. We selected participants who completed the questionnaire during at least one point (in 2006, 2008, 2010, 2012, or 2014) for a final sample size of 20,422 (59% female). Age at the time of first assessment ranged from 30 to 97 years (M = 65.91, SD = 11.06). Participants had on average 12.65 years of formal education (SD = 3.14). The questionnaire assessed participants on a variety of psychosocial variables, and the measures of interest for the current work are described in what follows. It should be noted, however, that not all participants completed all the social support or personality items. For example, if a participant did not have children, they were not included in the child support models. Sample sizes for each univariate model are presented in Table 1. The sample sizes for the bivariate growth models range from 13,857 to 19,341.

Table 1.

Means and variances of univariate growth models.

Model	Intercept Mean	Intercept variance	Slope Mean	Slope variance	N
Personality traits
Extraversion	3.191* [3.181, 3.201]	0.193* [.180, .206]	−0.031* [−.038, −.023]	0.014* [.007, .022]	20,309
Agreeableness	3.519* [3.510, 3.528]	0.138* [.127, .150]	−0.024* [−.031, −.017]	0.004 [−.002, .011]	20,287
Conscientiousness	3.361* [3.353, 3.370]	0.129* [.118, .139]	−0.030* [−.036, −.023]	0.007* [.001, .013]	20,293
Neuroticism	2.062* [2.051, 2.073]	0.222* [.206, .239]	−0.046* [−.054, −.038]	0.017* [.007, .026]	20,255
Openness	2.940* [2.930, 2.950]	0.193* [.180, .206]	−0.047* [−.054, −.040]	0.009* [.001, .016]	20,268
Social support
Spouse positive	3.470* [3.456, 3.483]	0.261* [.237, .285]	−0.016* [−.026, −.005]	0.022* [.008, .036]	13,938
Spouse negative	3.007 [2.992, 3.023]	0.299* [.271, .327]	0.027* [.015, .039]	0.033* [.018, .049]	13,924
Child positive	3.232* [3.218, 3.246]	0.349* [.323, .374]	0.023* [.012, .033]	0.025* [.011, .039]	18,336
Child negative	3.262* [3.249, 3.275]	0.249* [.228, .269]	0.044* [.034, .054]	0.020* [.009, .032]	18,332
Family positive	2.901* [2.885, 2.917]	0.407* [.370, .444]	−0.007 [−.020, .006]	0.027* [.004, .050]	19,412
Family negative	3.376* [3.364, 3.388]	0.203* [.183, .224]	0.059* [.049, .068]	0.015* [.002 .028]	19,396
Friends positive	3.065* [3.051, 3.079]	0.305* [.276, .334]	−0.020* [−.031, −.008]	0.020* [.003, .038]	18,997
Friends negative	3.544* [3.535, 3.554]	0.125* [.111, .139]	0.042* [.034, .050]	0.011* [.002, .020]	18,980

Note. 99% confidence intervals are presented in brackets. *indicates p < .01. The intercepts of the latent variables could have ranged from 1 to 4, the end points of the individual items in the scale. The slopes ranged from −0.47 to 0.45.

Personality

The MIDI (Midlife Development Inventory) personality scale was assessed at each time point (Lachman & Weaver, 1997). The 26 items were adjectives that participants rated on how well the adjectives described them on a scale from 1 (Not at all like me) to 4 (A lot like me). For extraversion, the items are outgoing, friendly, lively, active, and talkative (α_T1 = .74, α_T2 = .75, α_T3 = .77). For agreeableness, the items were helpful, warm, caring, soft-hearted, and sympathetic (α_T1 = .77, α_T2 = .78, α_T3 = .80). For conscientiousness, the items were organized, responsible, hardworking, careless (reverse-coded), and thorough (α_T1 = .65, α_T2 = .67, α_T3 = .67). For neuroticism, the items were moody, worrying, nervous, and calm (reverse-coded; α_T1 = .71, α_T2 = .71, α_T3 = .72). For openness, the items were creative, imaginative, intelligent, curious, broad-minded, sophisticated, and adventurous (_T1 = .79, α_T2 = .79, α_T3 = .80).

Social support

Using items from previous work on support and depressed mood (Schuster, Kessler, & Aseltine, 1990), social support was assessed separately for four different types of social relationships: spouse/partner, child, family members, and friends. For each relationship partner, participants reported on three positive social support items (assessed on a scale from 1 = Not at all to 4 = A lot) including “How much do they [people in this domain] understand the way you feel about things”; and four negative social support items (assessed on the scale of 1 to 4) including “How much do they criticize you?” Negative support was scored so that higher levels indicated less strain from a social partner, in order for higher scores on both support variables to be viewed as adaptive. Positive and negative social support were not strongly correlated within a relationship; thus, we estimated positive and negative social support separately for each relationship. This resulted in scales with relatively good reliabilities, with α ranging from .71 to .84.

Analytic plan

Attrition analyses were conducted to examine whether participants with a single assessment differed from those with longitudinal data. Full attrition analyses are presented in Supplemental Table 1.4. With respect to demographics, participants who failed to complete multiple assessments tended to be male, older, less educated. On personality variables, these participants were lower on extraversion, agreeableness, conscientiousness, and openness, but higher on neuroticism. For support variables, these participants were lower on positive friend support, but reported more negative support from spouses, family, and friends, as well as more positive support from children. Most effect sizes were small in magnitude, with the largest differences with respect to age, conscientiousness, and education.

First, we constructed univariate second-order latent growth models for each of the personality traits and each of the social support domains (Jackson & Allemand, 2014). In these models, we created latent constructs for Time 1, Time 2, and Time 3 variables where indicators were set to be equal across the three time points, and the item residuals were allowed to correlate across time. A latent intercept factor as created by constraining the loadings from Time 1, Time 2, and Time 3 to 1, whereas the slope factor was constrained to 0 for Time 1, constrained to 1 for Time 2, and constrained to 2 for Time 3. Intercepts for the latent time variables were constrained to 0. Intercepts for items across time were constrained to be equivalent (i.e., the item “outgoing” was constrained to have the same intercept at each of the time points). Finally, we applied effects coding to the model by constraining both the sum of the loading estimates and the sum of the item intercepts to equal the number of items at each time point (Little, Cards, Slegers, & Ledford, 2007). The full code is included in the supplementary material. We used the lavaan package (version 0.5–20; Rosseel, 2012) in R (version 3.3.0; R Core Team, 2016) to estimate these models.

Second, we fit bivariate latent growth models with all parameter estimates (i.e., the loadings, variances, covariances and intercept) for the personality and social support latent variables to equal those found in the univariate growth models and allowed the paths between latent intercepts and slopes to be estimated freely. Separate models were fit for each pairing of personality trait and social variables. In these bivariate models, the latent slopes were regressed onto the latent intercept of the opposite construct (e.g., the slope of spouse support was regressed onto the intercept of extraversion, and the slope of extraversion was regressed onto the intercept of spouse support). Each of the latent intercept and slope variables were also regressed onto mean-centered age, to test for age-related effects. For all analyses below, we employed an alpha significance level of .01 to safeguard against Type I errors given the number of tests run and the exploratory nature of many of the associations.

Results

Univariate models

As an initial step, we examined measurement invariance of personality and social support variables. Given our large sample size, we used changes in CFI, with change in CFI ≤ .005 as the cutoff for invariance (Chen, 2007). All univariate growth models passed the test of weak invariance and the conscientiousness model was the only one to fail the test of strong invariance (▵CFI = .015; see Supplemental Tables 2.1), though the overall fit was still acceptable. Thus, some caution should be taken in the conscientiousness models. Next, we fit univariate latent growth curves were fit for each of the five personality variables and for positive and negative support across the four social partners (see Table 1). All models fit the data adequately (CFIs > .92; RMSEAs < .08.

Small mean level changes were found for all the Big Five personality traits, such that the sample decreased in each of the traits over time estimates range from −.02 to −.05). In addition, there were significant individual differences in change for each of these traits, as indicated by the significant slope variance components, and thus, while on average, participants decreased on extraversion, agreeableness, conscientiousness, neuroticism, and openness across the period, some people bucked this trend and either increased, decreased more dramatically, or demonstrated greater stability on these traits.

Similar to the personality variables, small normative mean level changes were evidenced for positive and negative support. Specifically, positive social support declined across the years from one’s spouse and friends (std. est. = −.02) but slightly increased for child (std. est. = .02). Negative social support also increased from spouses (std. est. = .03), children (std. est. = .04), family (std. est. = .06) and friends (std. est. = .04). As with personality traits, however, there were significant individual differences in change for all types of social support.

Prior to examining how these changes in personality and social support coincide, we also examined whether age predicted these trajectories, testing whether younger or older adults differed in their trajectories of change (see Supplementary Tables 3.1 for full details). With respect to personality traits, age negatively predicted change in extraversion (est. = −.002), agreeableness (est. = −.002), conscientiousness (est. = −.002), and openness (est. = .002), but positively predicted change in neuroticism (est. = .002). For support, age predicted more negative support from children (est. = −.003), family (est. = −.001), and friends (est. = −.001). No other age differences were evident with respect to trajectories of change.

Within-domain (personality or social support) bivariate latent growth models

Next, we sought to investigate the joint development within the personality traits and social support domains. First, we examined the correlated intercepts and changes among the personality variables. Table 2 describes the initial correlations between personality traits as well as the association between changes in one personality trait with changes in another. Overall, there was large overlap between personality traits both for initial levels and correlated change. This overlap may indicate that the sources responsible for these changes are, in part, shared across traits or there are difficulties in distinguishing traits at the assessment level.

Table 2.

Correlated mean and change in latent personality and social support variables.

Personality traits	1	2	3	4	5
1) Extraversion	–	.61	.44	−.22	.55
2) Agreeableness	.52	–	.49	−.06	.39
3) Conscientiousness	.40	.44	–	−.18	.49
4) Neuroticism	−.13	−.06	−.11	–	−.19
5) Openness	.51	.44	.46	−.10	–
Social support domains	1	2	3	4	5	6	7	8
1) Spouse positive	–	.61	.24	.18	.15	.14	.09	.08
2) Spouse negative	.32	–	.22	.41	.11	.33	.05	.29
3) Child positive	.14	.06	–	.47	.44	.24	.30	.18
4) Child negative	.10	.19	.28	–	.20	.55	.06	.45
5) Family positive	.09	.04	.16	.09	–	.28	.33	.09
6) Family negative	.01	.15	.07	.28	.09	–	.01	.52
7) Friends positive	.07	.02	.09	.01	.15	.02	–	.09
8) Friends negative	.03	.09	.06	.19	−.02	.22	.06	–

Note. The upper triangles show the correlations between the latent intercepts, while the lower triangles show the correlations between the latent slopes. Correlations larger in magnitude than r = .05 are significant at p < .01. Sample sizes for these correlations ranged from 2,079 to 3,952.

Second, we considered the correlated intercepts and changes across social support types (also shown in Table 2). It is possible that declines in perceived support in one domain may be offset by increases in another domain. Such a pattern would result in negative change correlations among the types of support. Alternatively, positive correlated change would again indicate that the sources responsible for the change in one type of support also impact other types of social support. Results suggest that changes among social support indices are largely unrelated to one another (ranging from r = −.02 to .31). Overall, these estimates suggest that participants did distinguish types of social support from one another, bolstering the need to look at the different types separately.

Between-domain bivariate latent growth models

To examine the joint development of personality traits and social support, we fit multivariate growth models, which fit the data moderately well (CFIs > .90; RMSEAs < .05). First, we examined the initial correlations between personality traits and social support indices (see Table 3). All five personality traits were correlated with each type of social support at the initial wave of assessment. Neuroticism was negatively correlated with social support whereas the remaining traits were positively correlated with social support.

Table 3.

Relationships between latent intercepts and slopes of personality and social support growth models.

Initial covariances
	Spouse		Child		Family		Friends
	Positive	Negative	Positive	Negative	Positive	Negative	Positive	Negative
Extraversion	0.059* [.052, .067]	0.033* [.025, .042]	0.086* [.078, .094]	0.025* [.018, .032]	0.095* [.086, .104]	0.017* [.011, .023]	0.106* [.098, .114]	0.014* [.009, .019]
Agreeableness	0.037* [.031, .044]	0.024* [.017, .031]	0.078* [.072, .085]	0.022* [.016, .028]	0.078* [.070, .086]	0.014* [.008, .020]	0.099* [.092, .106]	0.024* [.019, .028]
Conscientiousness	0.047* [.040, .054]	0.036* [.028, .043]	0.053* [.046, .060]	0.031* [.025, .037]	0.046* [.038, .053]	0.027* [.021, .032]	0.060* [.053, .067]	0.031* [.027, .036]
Neuroticism	−0.068* [−.076, −.059]	−0.103* [−.112, −.093]	−0.051* [−.059, −.042]	−0.098* [−.105, −.090]	−0.039* [−.049, −.030]	−0.087* [−.094, −.080]	−0.025* [−.033, −.016]	−0.065* [−.070, −.059]
Openness	0.048* [.040, .056]	0.016* [.008, .024]	0.042* [.035, .050]	0.010* [.004, .017]	0.040* [.031, .049]	0.000 [−.006, .006]	0.067* [.060, .075]	0.006* [.000, .011]
Initial standing on personality predicting change in social support
	Spouse		Child		Family		Friends
	Positive	Negative	Positive	Negative	Positive	Negative	Positive	Negative
Extraversion	−0.031 [−.063, .000]	0.008 [−.027, .042]	−0.056* [−.086, −.025]	0.002 [−.025, .030]	−0.033 [−.072, .005]	0.002 [−.027, .030]	−0.042* [−.076, −.008]	0.000 [−.024, .025]
Agreeableness	−0.020 [−.055, .017	−0.015 [−.056, .026]	−0.053* [−.090, −.017]	−0.006 [−.039, .029]	−0.020 [−.065, .026]	−0.019 [−.053, .016]	−0.025 [−.065. .016]	−0.018 [−.047. .011]
Conscientiousness	−0.013 [−.055, .028]	−0.020 [−.066, .026]	−0.041* [−.081, −.001]	0.013 [−.023, .049]	−0.039 [−.089, .012]	−0.017 [−.054, .021]	−0.033 [−.078, .012]	−0.028 [−.060, .004]
Neuroticism	0.002 [−.030, .034]	0.018 [−.016, .053]	0.015 [−.016, .046]	0.051* [.024, .078]	0.012 [−.027, .051]	0.056* [.027, .084]	0.034 [−.001, .069]	0.038* [.014, .061]
Openness	−0.034* [−.065, −.003]	−0.002 [−.036, .033]	−0.056* [−.086, −.026]	0.001 [−.025, .028]	−0.056* [−.094, −.019]	0.007 [−.021, .035]	−0.045* [−.079, −.012]	0.002 [−.021, .026]
Initial standing on support predicting change in personality
	Spouse		Child		Family		Friends
	Positive	Negative	Positive	Negative	Positive	Negative	Positive	Negative
Extraversion	−0.014 [−.036, .008]	−0.005 [−.029, .018]	0.005 [−.013, .022]	0.017 [−.006 .040]	−0.023* [−.039, −.008]	0.001 [−.026, .029]	−0.029* [−.047, −.010]	0.023 [−.016, .062]
Agreeableness	−0.015 [−.036, .005]	0.00 [−.021, .021]	0.002 [−.014, .019]	0.008 [−.013, .029]	−0.023* [−.038, −.009]	0.014 [−.011, .039]	−0.022* [−.040, −.004]	0.035 [−.001, .070]
Conscientiousness	−0.020 [−.040, .001]	0.004 [−.017, .024]	−0.002 [−.018, .014]	0.011 [−.009, .032]	−0.017* [−.031, −.003]	0.009 [−.015, .034]	−0.017 [−.034, .001]	0.022 [−.012, .056]
Neuroticism	0.042* [.017, .067]	0.054* [.029, .080]	0.015 [−.006, .035]	0.055* [.030, .080]	0.009 [−.009, .027]	0.052* [.022, .082]	0.008 [−.014, .029]	0.060* [.018, .102]
Openness	−0.021* [−.043, .000]	0.00 [−.022, .022]	0.005 [−.012, .022]	0.012 [−.010, .034]	−0.019* [−.034, −.004]	0.009 [−.017, .035]	−0.015 [−.033, .004]	0.020 [−.017, .057]
Correlated changes (covariances)
	Spouse		Child		Family		Friends
	Positive	Negative	Positive	Negative	Positive	Negative	Positive	Negative
Extraversion	0.009* [.006, .012]	0.004* [.001, .007]	0.006* [.003, .009]	0.002 [−.001, .005]	0.010* [.006, .013]	0.002 [−.001, .004]	0.012* [.009, .016]	0.001 [−.002, .003]
Agreeableness	0.005* [.002, .008]	0.003* [.000, .006]	0.005* [.002, .008]	0.003* [.000, .005]	0.009* [.005, .012]	0.002 [−.002, .005]	0.007* [.004, .010]	0.002 [.000, .004]
Conscientiousness	0.006* [.004, .009]	0.003 [.000, .006]	0.005* [.002, .007]	0.002 [−.001, .004]	0.006* [.002, .009]	0.002 [−.001, .004]	0.007* [.004, .010]	0.003* [.001, .005]
Neuroticism	−0.009* [−.012, −.006]	−0.016* [−.019, −.012]	−0.007* [−.010, −.004]	−0.014* [−.017, −.011]	−0.004 [−.008, .000]	−0.011* [−.014, −.008]	−0.005* [−.009, −.001]	−0.008* [−.010, −.005]
Openness	.009* [.006, .012]	0.003 [.000, .006]	0.006* [.003, .009]	0.002 [−.001, .004]	0.010* [.006, .014]	0.000 [−.003, .003]	0.009* [.006, .013]	0.001 [−.001, .003]

Note. n = 3952. *indicates that the relationship is significant at p < .01. 99% confidence intervals are presented in brackets. In cases where latent relationships were estimated at greater than 1, we censored those values to be equivalent to 1.00.

Second, we investigated whether initial standing on personality was associated with changes in social support. Initial levels of openness, agreeableness, conscientiousness, and extraversion were associated with declines in positive child social support, and initial levels of extraversion were associated with less positive social support from friends. Initial levels of neuroticism were associated with increases in negative child social support and negative family social support. Overall, it seems that higher initial levels of neuroticism predict increases in negative social support, while the other traits predict decreases in positive social support.

We next considered the associations between initial levels of social support and changes in personality traits. Initial levels of positive family support were associated with decreases in extraversion, agreeableness, conscientiousness, and openness, whereas initial levels of positive friend support coincided with decreases in extraversion and agreeableness. Also, initial positive spouse support was associated with decreases in openness. Finally, initial levels on almost all support variables (except positive family and positive friend support) were associated with increases in neuroticism. Overall, it seems that greater positive support predicts decreases in all traits except neuroticism; increases in neuroticism are predicted by greater initial levels of both positive and negative support.

Third, we evaluated whether correlated changes occurred between personality traits and social support controlling for initial levels. Mirroring the initial correlations, consistent evidence was found to suggest that personality traits and social relationship variables change in tandem. For all Big Five traits, evidence for correlated changes were found with respect to the different forms of social support. Correlated changes were evidenced across all positive support variables (e.g., increases in extraversion coinciding with increases in positive support from spouse, child, family, and friends). Moreover, increases in neuroticism coincides with increases in negative support across all four social partners. Increases on negative support from spouse coincided with decreases in extraversion and agreeableness, while increases in negative child support were associated with declines in agreeableness, and increased negative friend support was correlated with declines in conscientiousness.

General discussion

The current study investigated the co-development of personality traits and perceived social support with respect to children, friends, family members, and spouses in middle to later adulthood. At the initial assessment, greater support across social partners was related to higher levels of extraversion, agreeableness, conscientiousness, and openness to experience, but lower levels of neuroticism. Evidence was found for small associations between intercepts on traits and changes in social support and vice versa. However, these effects were small and few consistent patterns were found. We did find consistent evidence that traits and support variables co-develop during middle and older adulthood, as all Big Five traits exhibited correlated changes with multiple indices of social support. We focus our discussion below on the cross-lagged and correlated change results, and how the latter support the calls for continued research on how social relationships may influence personality development in adulthood (Neyer et al., 2014).

Though few cross-lagged effects were evidenced, all came in a direction counter to the initial relationships between the constructs of interest. Given that there is little theoretical rationale for these directions (such as that having positive support from a spouse makes one more neurotic), it is more likely that they should be interpreted as being similar to ceiling effects. It is interesting to note that the only coherent pattern across these effects came with respect to positive family support, where higher levels on this construct predicted declines on extraversion, agreeableness, conscientiousness, and openness. Positive family support was unique in the univariate models as well, in that it had lowest initial intercept and was the only support variable that failed to show mean-level change. Given these distinctions, future research should pinpoint which specific family relationships were being appraised by participants, particularly given that adults do view their kin relations as differing in closeness and reciprocity (Neyer et al., 2011), and that family was assessed uniquely from children and spouses (often the closest kin relatives for most individuals). Moreover, cross-lagged effects are better identified using more than two or three waves of data collection, and future research is needed to test whether the cross-lagged effects observed in the current study are indeed artefactual, or if they may represent “true” conflicts with theoretical expectations (c.f. Bleidorn, 2012). That said, contradictory cross-lagged effects have been evidenced in previous two-wave research using similar constructs (Bleidorn, 2011; Hill et al., 2012).

Given the current number of measurement waves, the correlated change effects may present better evidence that social relationships play a role in personality development during middle and older adulthood. Across all relationships, greater social support was associated with more adaptive changes on the Big Five traits, and the effects were consistent with respect to having more positive support relative to less negative support. These findings point to a potentially integral link between social relationships and personality development that likely operates in a bidirectional nature. On one hand, certain personality traits (e.g., being agreeable, extraverted, conscientious) are preferred in a relationship partner (Botwin, Buss, & Shackelford, 1997), and these traits have been linked to greater relationship success (Malouff, Thorsteinsson, Schutte, Bhullar, & Rooke, 2010). Evidence of these benefits, in the form of having positive relations with friends and family, may encourage one to maintain or deepen these seemingly adaptive personality characteristics (Hill & Jackson, in press; Roberts, Wood, & Caspi, 2008). Research though is needed to understand the extent to which adults explicitly associate relationship benefits with personality trait dimensions.

Though one may expect such findings for more obviously “social” traits such as extraversion or agreeableness, it is interesting to consider how these correlated change findings held across neuroticism, conscientiousness, and openness to experience as well. With respect to neuroticism, these findings would appear to support claims that social relationships play an important role in emotional regulation during middle and older adulthood (Carstensen, 1993; Carstensen, Isaacowitz, & Charles, 1999), insofar that having more positive social support coincided with lower levels of neuroticism. Moreover, it is worth noting that neuroticism was relatively unique with respect to generally exhibiting stronger associations with negative support than with positive support, both concurrently and longitudinally. Accordingly, when considering changes in neuroticism, it appears at least as important for individuals to regulate negative relationship features (e.g., conflict, stress) as to promote positive relationship aspects, if not more. Conscientiousness has been increasingly implicated as playing an important role in romantic relationships (Hill, Nickel, & Roberts, 2014; Solomon & Jackson, 2014), in part because conscientious partners may be more prone to sustain pro-relationship behaviors (Kammrath & Peetz, 2011). It remains a question for future research whether the health benefits associated with conscientiousness or social support may be explained in part because changes on these constructs coincide with one another, as recent work has posited that conscientiousness adults experience better health in part because of their more positive romantic relationships (e.g., Hill, Nickel, & Roberts, 2014; Hill & Jackson, in press).

Although the correlated changes with respect to openness replicate past work with social well-being (Hill et al., 2012), the mechanisms linking social support to openness remain somewhat uncertain. One possibility is that social partners lead one to consider new experiences and try novel activities, which in turn lead to feeling more connected and supported to those social partners. Alternatively, when focusing on the intellectual component of openness, it is worth noting that social engagements appear adaptive for cognitive functioning later in life (e.g., Fratiglioni, Paillard-Borg, & Winblad, 2004; Seeman, Lusignolo, Albert, & Berkman, 2001). In part, these effects may result because engaging in social relationships provides “exercise” of one’s cognitive faculties, which may lead one to report higher levels of intellectual engagement or feel more intelligent. Future research though is needed to understand whether the longitudinal relationships between social relationships and openness to experience are better explained by virtue of enabling new experiences, or promoting positive perceptions of cognitive ability. More broadly, additional measurement occasions would not only allow a better understanding of the potential for cross-lagged effects, but also of the mechanisms implicated in explaining the correlated change findings across all Big Five traits (Jackson & Allemand, 2014).

The current findings are limited in ways that provide directions for future research. First, and foremost, these results should be replicated and extended in data sets with additional measurement occasions, in order to better test both cross-lagged effects and potential mediational mechanisms. Previous work has investigated whether correlated changes better reflect mechanisms responsible for the initial correlations in the first place or novel experiences that promote development in directions different than the original associations (Klimstra, Bleidorn, Asendorpf, van Aken, & Denissen, 2013). Our results indicate that many of the processes driving the effects are likely shared across traits and types of social support and potential reflect associations that were in place prior to the initial measurement wave. While this is not surprising for some associations between support variables, our findings must nonetheless be interpreted cautiously in terms of what is driving the joint development of personality and social support.

Second, ideally observer-reports of personality and social support should be integrated into these designs, in order to test whether reporter biases play a role in the covariation between social relationships and personality development. It also would be of interest to know whether both members of a social relationship perceive the levels of support similarly, and in turn whether differences in reporting mitigate some of the high levels of similarity across types of support. Third, lengthier measures of the Big Five would allow for a more fine-grained analysis of which specific facets develop in tandem with social relationships, which has been investigated in previous work on social role loss and personality development (e.g., Allemand, Hill, et al., 2015; Costa et al., 2000), in addition to better differentiating the broad traits. Moreover, the current work employed a broad Big Five trait measure that may be construed as occurring at a higher level of aggregation compared to the more specific social support variables, potentially influencing the relative stability of traits versus social support constructs. As such, having more specific trait measures would help to alleviate any concerns regarding differing levels of aggregation.

These caveats aside, the current research provides further evidence not only that personality trait fluctuations are evidenced later in the lifespan, but also that these developments may be related in part to adults’ changing perceptions of their social environments. Moreover, they provide further evidence that perceived social support provides a valuable variable for explaining personality trait development in older adulthood (see also Hill, Payne, et al., 2014). In this respect, our findings coincide with the previous suggestion that objective indices of social roles (e.g., having a spouse, number of children, etc.) may be less important for understanding personality trait changes than the adults’ subjective perceptions of these relationships (e.g., Lodi-Smith & Roberts, 2007). The next step is to better understand the mechanisms linking perceived social support to personality in middle and older adulthood (including changes in environments, health, or social responsibilities), in order to consider potential avenues for changing trait levels even during these periods largely defined by stability.

Footnotes

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental material

Supplementary material for this article is available online.

References

Allemand

Hill

P. L.

Lehmann

(2015). Divorce and personality development across middle adulthood. Personal Relationships, 22, 122–137.

Allemand

Schaffhuser

Martin

(2015). Long-term correlated change between personality traits and perceived social support in middle adulthood. Personality and Social Psychology Bulletin, 41, 420–432.

Asendorpf

J. B.

Van Aken

M. A.

(2003). Personality -relationship transaction in adolescence: Core versus surface personality characteristics. Journal of Personality, 71, 629–666.

Asendorpf

J. B.

Wilpers

(1998). Personality Effects on Social Relationships. Journal of Personality and Social Psychology, 74, 1531–1544.

Bleidorn

(2012). Hitting the road to adulthood: Short-term personality development during a major life transition. Personality and Social Psychology Bulletin, 38, 1594–1608.

Bleidorn

Klimstra

T. A.

Denissen

J. J.

Rentfrow

P. J.

Potter

Gosling

S. D.

(2013). Personality maturation around the world: A cross-cultural examination of social-investment theory. Psychological Science, 24, 2530–2540.

Botwin

M. D.

Buss

D. M.

Shackelford

T. K.

(1997). Personality and mate preferences: Five factors in mate selection and marital satisfaction. Journal of Personality, 65, 107–136.

Branje

S. J.

van Lieshout

C. F.

van Aken

M. A.

(2004). Relations Between Big Five Personality Characteristics and Perceived Support in Adolescents’ Families. Journal of Personality and Social Psychology, 86(4), 615–628.

Carstensen

L. L.

(1993, 1). Motivation for social contact across the life span: A theory of socioemotional selectivity. In Jacobs

J. E.

(Ed.), Nebraska symposium on motivation (Vol. 40, pp. 209–254). Lincoln: University of Nebraska Press.

10.

Carstensen

L. L.

Isaacowitz

D. M.

Charles

S. T.

(1999). Taking time seriously: A theory of socioemotional selectivity. American Psychologist, 54, 165–181.

11.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling, 14, 464–504.

12.

Costa

P. T.

Herbst

J. H.

McCrae

R. R.

Siegler

I. C.

(2000). Personality at midlife: Stability, intrinsic maturation, and response to life events. Assessment, 7, 365–378.

13.

Fratiglioni

Paillard-Borg

Winblad

(2004). An active and socially integrated lifestyle in late life might protect against dementia. The Lancet Neurology, 3, 343–353.

14.

Giles

L. C.

Glonek

G. F.

Luszcz

M. A.

Andrews

G. R.

(2005). Effect of social networks on 10 year survival in very old Australians: the Australian longitudinal study of aging. Journal of Epidemiology and Community Health, 59, 574–579.

15.

Hill

P. L.

Jackson

J. J.

(2016). The invest-and-accrue model of conscientiousness. Review of General Psychology, 20, 141–154.

16.

Hill

P. L.

Nickel

L. B.

Roberts

B. W.

(2014). Are you in a healthy relationship? Linking conscientiousness to health via implementing and immunizing behaviors. Journal of Personality, 82, 485–492.

17.

Hill

P. L.

Payne

B. R.

Jackson

J. J.

Stine-Morrow

E. A. L.

Roberts

B. W.

(2014). Perceived social support predicts increased conscientiousness during older adulthood. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69, 543–547.

18.

Hill

P. L.

Turiano

N. A.

Mroczek

D. K.

Roberts

B. W.

(2012). Examining concurrent and longitudinal relations between personality traits and social well-being in adulthood. Social Psychological and Personality Science, 3, 698–705.

19.

Jackson

J. J.

Allemand

(2014). Moving personality development research forward: Applications using structural equation models. European Journal of Personality, 28, 300–310.

20.

Kammrath

L. K.

Peetz

(2011). The limits of love: Predicting immediate versus sustained caring behaviors in close relationships. Journal of Experimental Social Psychology, 47, 411–417.

21.

Klimstra

T. A.

Bleidorn

Asendorpf

J. B.

van Aken

M. A. G.

Denissen

J. J. A.

(2013). Correlated change of Big Five personality traits across the lifespan: A search for determinants. Journal of Research in Personality, 47, 768–777.

22.

Lachman

M. E.

Weaver

S. L.

(1997). The Midlife Development Inventory (MIDI) personality scales: Scale construction and scoring. Waltham, MA: Brandeis University.

23.

Little

T. D.

Card

N. A.

Slegers

N. W.

Ledford

E. C.

(2007). Representing contextual effects in multiple-group MACS models. In Little

T. D.

Bovaird

J. A.

Card

N. A.

(Eds.), Modeling contextual effects in longitudinal studies (pp. 121–147). New York, NY: Psychology Press.

24.

Lodi-Smith

Roberts

B. W.

(2007). Social investment and personality: A meta-analysis of the relationship of personality traits to investment in work, family, religion, and volunteerism. Personality and Social Psychology Review, 11, 68–86.

25.

Lucas

R. E.

Donnellan

M. B.

(2011). Personality development across the life span: Longitudinal analyses with a national sample from Germany. Journal of Personality and Social Psychology, 101, 847–861.

26.

Malouff

J. M.

Thorsteinsson

E. B.

Schutte

N. S.

Bhullar

Rooke

S. E.

(2010). The five-factor model of personality and relationship satisfaction of intimate partners: A meta-analysis. Journal of Research in Personality, 44, 124–127.

27.

Mund

Neyer

F. J.

(2014). Treating personality-relationship transactions with respect: narrow facets, advanced models, and extended time frames. Journal of Personality and Social Psychology, 107, 352–368.

28.

Neyer

F. J.

Lehnart

(2007). Relationships matter in personality development: Evidence from an 8-year longitudinal study across young adulthood. Journal of Personality, 75, 535–568.

29.

Neyer

F. J.

Wrzus

Wagner

Lang

F. R.

(2011). Principles of relationship differentiation. European Psychologist, 16, 267–277.

30.

Neyer

F. J.

Mund

Zimmermann

Wrzus

(2014). Personality-relationship transactions revisited. Journal of Personality, 82, 539–550.

31.

R Core Team. (2016). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org

32.

Roberts

B. W.

DelVecchio

W. F.

(2000). The rank-order consistency of personality traits from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126, 3–25.

33.

Roberts

B. W.

Walton

K. E.

Viechtbauer

(2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132, 1–25.

34.

Roberts

B. W.

Wood

Caspi

(2008). The development of personality traits in adulthood. In John

O. P.

Robins

R. W.

Pervin

L. A.

(Eds.), Handbook of personality: Theory and research (3rd ed., pp. 375–398). New York, NY: Guilford.

35.

Rosseel

(2012). Lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48, 1–36. Retrieved from http://ww.jstatsoft.org/v48/i02

36.

Schuster

T. L.

Kessler

R. C.

Aseltine

R. H.

Jr (1990). Supportive interactions, negative interactions, and depressed mood. American Journal of Community Psychology, 18, 423–438.

37.

Seeman

T. E.

Lusignolo

T. M.

Albert

Berkman

(2001). Social relationships, social support, and patterns of cognitive aging in healthy, high-functioning older adults: MacArthur studies of successful aging. Health Psychology, 20, 243–255.

38.

Solomon

B. C.

Jackson

J. J.

(2014). Why do personality traits predict divorce? Multiple pathways through satisfaction. Journal of Personality and Social Psychology, 106, 978–996.

39.

Specht

Egloff

Schmukle

S. C.

(2011). Stability and change of personality across the life course: The impact of age and major life events on mean-level and rank-order stability of the Big Five. Journal of Personality and Social Psychology, 101, 862–882.

40.

van Scheppingen

M. A.

Jackson

J. J.

Specht

Hutteman

Denissen

J. J. A.

Bleidorn

(2016). Personality trait development during the transition to parenthood: A test of social investment theory. Social Psychological and Personality Science, 7, 452–462.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB