Managing expectations: How stress, social support, and aging attitudes affect awareness of age-related changes

Abstract

The degree to which social support (SS) moderates the effects of stress on self-perceptions of aging may depend on individual differences in general aging attitudes. We examined how stress, different types of SS, and general expectations regarding aging (ERA) affect awareness of age-related changes (AARCs). The sample included 137 adults (21–76 years; 56.2% women) who took an online survey on Amazon’s Mechanical Turk. Regression analyses showed differential moderation of stress effects due to ERA and the SS measure (perceived and received) and function (emotional and instrumental). Received emotional SS was only associated with AARC losses, whereas perceived support—both emotional and instrumental—was associated with AARC gains and losses. Findings may help guide future work aimed at promoting health and well-being in adulthood.

Individuals play an increasingly active part in monitoring their own development as they age (Brandtstädter & Rothermund, 2002). The kind and nature of normal changes they experience in multiple life domains, such as in physical, cognitive, and socioemotional functioning, may differentially influence evaluations of their personal aging. For instance, poorer performance on physical and cognitive tests can alert one to negative aspects of aging and elicit feelings of loss, whereas more positive emotional and interpersonal experiences may draw attention to positive aspects and bring about feelings of gain. These experiences accordingly contribute to an individual’s awareness of age-related changes (AARCs) or perception of “how his or her behavior, level of performance, or ways of experiencing life have changed [and continue to change] as a consequence of having grown older” (Diehl & Wahl, 2010, p. 340).

Subjective perceptions of aging have reliable and robust effects on health outcomes in adulthood, especially in older age. Studies typically show health benefits related to positive perceptions of aging, including higher cognitive ability (Robertson, King-Kallimanis, & Kenny, 2016), less decline in physical functioning (Wurm & Benyamini, 2014), reduced mortality (Levy, Slade, Kunkel, & Kasl, 2002), and less emotional reactivity to daily stressors (Bellingtier & Neupert, 2016). Some work, however, has alluded to potential vulnerabilities or limitations by finding poorer memory performance in older adults following exposure to extremely or unrealistically positive images of aging as opposed to more realistic ones (Fung et al., 2015). They have also shown that, in certain cases, adults with generally positive aging attitudes exhibit more depressive symptoms (Sharifian & O’Brien, 2018) and greater increases in negative affect (Neupert & Bellingtier, 2017). These findings consequently raise interesting questions about what experiences contribute to awareness of AARCs and under what conditions.

Stress

Much research has investigated the effects of aging attitudes on responses to stressors, which we broadly define as the physical, cognitive, and personal events that elicit active or passive coping processes. Adults who hold more negative aging attitudes show (a) greater increases in cortisol across 30 years (Levy, Moffat, Resnick, Slade, & Ferruccci, 2016), (b) elevated cardiovascular reactivity to mental challenges (e.g., Levy, Hausdorff, Hencke, & Wei, 2000), and (c) higher likelihoods of experiencing a hospitalization over 10 years (Levy, Slade, Chung, & Gill, 2015). Stressors that occur on a daily basis can also elicit stronger same-day negative affect among these individuals (Bellingtier & Neupert, 2016). Fewer studies have examined the reverse relationship with stress predicting attitudes. Those that do, however, suggest that experiencing a greater than usual number of stressors, physical health problems, and negative affect leads to feeling older than one’s chronological age (Kotter-Grühn, Neupert, & Stephan, 2015).

Thus, this body of work largely demonstrates—with two qualifications—that individuals who report less stress also tend to feel more positive about aging both in general and in relation to themselves. However, given that AARC has its roots in adults’ everyday age-related experiences (Miche et al., 2014), an investigation of its relation to everyday stressors—within a broader social setting—has practical implications for our interpretations. Specifically, it may help contextualize the previously mentioned qualifications whereby, counterintuitively, more stress predicts more negative outcomes in people with positive perceptions of aging (e.g., Neuptert & Bellingtier, 2017; Sharifian & O’Brien, 2018). These issues served as the basis for the general question in present study, what factors moderate the impact of stress on (self-)perceptions of aging? Based on the foregoing, two categories of factors, one related to dispositional characteristics and the other to characteristics of individuals’ social environments, could moderate the impacts of stress on perceptions of aging (i.e., AARC). In the current study, we attempted to integrate perspectives to understand the interrelationships among stress, social support (SS), and attitudes toward aging.

Social support

SS refers to the structures and functions of an individual’s social network (e.g., Uchino, 2004). Commonly examined functional measures include emotional support, which enhances a person’s self-esteem by highlighting his or her own value, and instrumental support, which enables a person to carry out a specific task through the provision of material or tangible resources (Cohen & Willis, 1985, p. 313). It also involves either instances of actual support whereby one person receives aid from another or simply the perception of access to resources should the need for them arise.

SS can have positive impacts on emotional and physical health outcomes in the absence and presence of stressors (Cohen & Wills, 1985). For example, people who perceive greater responsiveness from their intimate partner during a conversation about a stressful topic also report more positive mood (Collins & Feeney, 2000). One possible explanation for the stress-buffering capabilities of SS links stress closely to ego threat (e.g., feelings of helplessness and loss of self-esteem) and assumes that SS has a compensatory function (Cohen & Wills, p. 312). In support of this idea, emotional SS reduces the extent to which stressors associated with salient social roles (e.g., death of a child) diminish role-specific feelings of control (Krause, 1987; Krause & Borawski-Clark, 1994). This suggests that SS may restore feelings of personal control and self-worth potentially lost during the coping process (e.g., Pearlin, Menaghan, Lieberman, & Mullan, 1981).

That said, SS may only buffer stress effects in some situations (e.g., Krause,1986) and for some people, such as for the oldest-old but not young-old (Krause, 2005). This underscores a different but complementary explanation whereby the effectiveness of SS as a stress buffer depends upon its alignment with individual needs, considering the nature and appraisals of the stressors themselves. Whereas emotional SS can optimize coping responses and outcomes in uncontrollable situations (e.g., loss of social roles due to transitions or death of a loved one) by fulfilling a need for attachment, instrumental SS may optimize responses primarily in controllable situations (e.g., loss of assets) by providing material resources (Cutrona & Russell, 1990). The absence of a buffering effect associated with SS might consequently reflect a mismatch between the coping requirements mobilized by the stressor and the support provided.

The presence of an amplification or proliferation effect (e.g., Bolger & Amarel, 2007; Bolger, Zuckerman, & Kessler, 2000; Nguyen et al., 2017; Paukert et al., 2010; Reinhardt, Boerner, & Horowitz, 2006; Sharifian & O’Brien, 2018) may also derive from a stress–support mismatch but emphasizes the self-related implications of potential or received support. Someone who feels generally positive about the self will likely attribute support received to internal causes (e.g., personal inadequacies) and respond negatively, especially in salient or highly valued life domains. In contrast, someone who feels more negatively about the self may respond more positively and experience more benefits (for a review, see Fisher, Nadler, & Whitcher-Alagna, 1982). Individuals’ appraisals may therefore dictate how SS moderates the impact of stressors on health and well-being.

Laboratory-based findings provide indirect evidence for the idea that SS induces ego threat. In one study, individuals who received help from an unfamiliar person on a task supposedly indicative of ability (i.e., ego-relevant) reported worse self-perceptions (e.g., intelligence, successfulness) than those who received help on a task indicative of chance factors (i.e., non-ego-relevant; Nadler, Fisher, & Itzhak, 1983). Studies on the perceptions of support in both healthy people and cancer patients also find that those in the latter group often perceive supportive behaviors from those in the former as inappropriate or unhelpful. For example, expressions of pity and forced cheerfulness elicit feelings of isolation and being “less normal” (Peters-Golden, 1982). More direct evidence suggests that SS can result in actual ego-related losses, with the receipt of emotional, instrumental, and informational support from familiar or close relationships resulting in decreased self-esteem and increased negative affect (Lepore, Glaser, & Roberts, 2008). These findings taken together suggest that SS can often operate in a counterproductive manner, magnifying rather than offsetting the noxious impacts of stress. In other words, SS itself seems to become an identity-related stressor that potentiates the very issues it is mobilized to cure by damaging rather than repairing important aspects of the self (similar to the discussion by Throits, 1991).

One additional factor is worth noting in discussing inconsistencies in SS effects. First, perceived SS and the actual receipt of SS may have differential impacts (e.g., Cohen, Towbes, & Flocco, 1988; Cummins, 1998; Cutrona & Russell, 1990). For instance, in older adults, perceived emotional support (PES) and received emotional support (RES) predict significantly fewer depressive symptoms, whereas perceived instrumental support (PIS) has no effect and received instrumental support predicts significantly more symptoms (Reinhardt et al., 2006). These results suggest that perceived support is associated with more beneficial outcomes and received support with no benefits or worse outcomes (Wethington & Kessler, 1986). One possible explanation for this difference, as noted by Reinhardt and colleagues, bears again on the implications for self-perceptions. People may feel comforted in simply knowing that they can turn to others in times of need but feel inadequate and dependent in receiving support even if it helps in a tangible way.

Aging attitudes

People can conceive of their own aging in various ways, for example, in terms of the age group to which they most identify (i.e., age identity), the age they feel in relation to their chronological age (i.e., felt age), and their anticipated or realized experiences (i.e., expectations). We concentrate on general attitudes since they are based in specific and concrete life experiences and have received much attention in psychological research (Diehl et al., 2014). Additionally, few studies have empirically investigated the interrelationships among subjective aging constructs but those that have do demonstrate the expected associations. People across age groups who report generally positive attitudes toward their own aging also report being more aware of age-related gains and less aware of age-related losses (Brothers, Miche, Wahl, & Diehl, 2017). Aging attitudes also relate to other self-beliefs, such as personal control (e.g., O’Brien et al., 2017) and general self-efficacy (e.g., Dutt & Wahl, 2019), which prior work also implicates as potential moderators or mediators that link stress to SS.

Emerging theory (e.g., Diehl & Wahl, 2010; Levy, 2009) suggests that attitudes operate as part of a (cognitive) lens through which individuals derive meaning from their personal experiences. Expectancies either generated or triggered by these attitudes can prime awareness of either the positive or negative consequences of aging (i.e., gains or losses), with generally positive self-perceptions of aging predicting greater AARC gains and generally negative perceptions predicting greater AARC losses (Brothers et al., 2017). Moreover, scholars have largely characterized positive perceptions as a protective psychological resource (e.g., Brothers, Diehl, Gabrian, & Wahl, 2018; Levy, 2003; Levy et al., 2000, 2015; cf., Sindi et al., 2012). Only recently have they alluded to such perceptions as a potential risk factor. In one study, more age-related losses resulted in more negative affect in adults who tended to view their own aging in a positive light compared to those who viewed their aging more negatively (Neupert & Bellingtier, 2017). Thus, we examined aging attitudes as an additional moderator of the effect of stress on AARC.

Combined effects of stress, support, and attitudes

Much research suggests that SS and aging attitudes should moderate the effect of stress, adopting one of two complementary interpretations based either on resources or on context. On the one hand, a resource approach argues that high SS and positive attitudes should mitigate stress effects by compensating for a lack of coping resources. On the other hand, a context-based approach also accounts for an individual’s coping needs within a particular situation. It suggests that the SS and aging attitudes could either mitigate stress effects if aligned with needs or magnify them if mismatched. Each interpretation notably assumes that people with positive attitudes tend to have more “psychological” resources than those with negative attitudes. In this view, high SS provided to a person with positive attitudes (rich in coping resources) would constitute a mismatch. This would elicit more negative response by respectively decreasing and increasing the saliency of age-related losses and gains.

The present study

We integrated the perspectives and findings from the foregoing literature in order to examine the impact of stress on AARCs (gains and losses) and the moderating roles of SS and aging attitudes. We limit our focus to emotional and instrumental SS to remain consistent with prior work that has also included aging attitudes (Sharifian & O’Brien, 2018); however, we expand our investigation by exploring the effects of different types of SS, given the evidence suggestive of more positive effects for perceived versus received and emotional versus instrumental support. We also had separate hypotheses for AARC losses and gains, with the main difference being the direction of effects associated with each predictor. Thus, we hypothesized higher stress, lower perceived but higher received SS, lower emotional but higher instrumental SS, and more negative aging attitudes to predict more losses (H1_Loss, H2_{Loss: a–d}, and H3_Loss) and fewer gains (H1_Gain, H2_{Gain: a–d}, and H3_Gain). We further hypothesized that SS and aging attitudes would moderate the impact of stress. In line with a contextual approach and some research findings (Neupert & Bellingtier, 2017; Sharifian & O’Brien, 2018), higher received SS was hypothesized to result in even greater losses (H4_Loss) and fewer gains (H4_Gain) only in persons with more positive attitudes. (Note that for H4 we made no prediction about potential differences between types of SS to reduce complexity.)

Method

Participants

We conducted a priori power analysis using G*Power version 3.1 (Faul, Erdfelder, Buchner, & Lang, 2009) to determine our sample size. This analysis indicated that a total sample size of 160 would provide sufficient power at 1 – β (Type II error rate) = .80 to detect a medium effect size (f² = .15) with 21 predictors (including controls) and α at .05. Participants were recruited through Amazon’s Mechanical Turk (MTurk). We restricted participation to individuals who resided within the U.S. and to those who had a 95% approval rating based on other completed Human Intelligence Tasks. We initially recruited 164 participants from MTurk. To ensure the validity of responses, chronological age and date of birth were collected on two separate pages of the survey. We then calculated each participant’s age using the specified birth date and compared it to the reported chronological age. Participants (n = 23) whose chronological age and birth date did not match were excluded. Additionally, three participants completed the survey in under 400 s (average completion time: 17.06 min), and one participant completed the survey in 3 hr. This suggests that these participants may have been engaged in other activities during the survey, so we excluded them from analyses as well. The final sample included 137 participants who ranged in age from 21 years to 78 years (M = 50.58, SD = 17.08) and who were majority female (56.2%), non-Hispanic White (82.50%), and college educated (74.60% with some college or more).

Measures

Awareness of age-related change.

AARC losses and AARC gains were measured using a 50-item scale (Brothers, Gabrian, Wahl, & Diehl, 2016). Each item was preceded by the stem “With my increasing age, I realize that...” and included either the description of a gain (e.g., “...I have a better sense of what is important to me.”) or a loss (e.g., ““...my ability to move around has gotten worse.”). Items reflected positive or negative experiences in five behavioral domains (e.g., physical functioning). Participants rated their agreement with 25 gain items and 25 loss items on a 5-point scale. Scores ranged from 1 (not at all) to 5 (very much) and were averaged by valence. Higher scores reflect greater awareness of either gains or losses. Internal consistencies were adequate (losses: α = .94; gains: α = .93).

Stress.

We assessed participants’ current levels of stress using the Weekly Inventory of Stressful Events (WISE; adaptation of the Daily Inventory of Stressful Events; Almeida, Wethington, & Kessler, 2002). Participants reported whether they experienced any of the seven different types of stressors (e.g., interpersonal conflicts, work-related stress, discrimination). For each stressor, they indicated Yes (1) if a stressor in-question occurred or No (0) if it did not, which means they could have experienced none, any, or all of the stressors. If participants indicated “Yes,” they rated how stressed they felt when they experienced the stressor on a 7-point scale that ranged from 1 (not at all) to 7 (extremely). We averaged severity ratings across stressors for analyses.¹

Social support

RES.

In cases where the participant responded “Yes” to a stressor, subsequent questions followed concerning whether he or she: (a) “discussed personal feelings and/or private concerns about the event,” (b) “received encouragement and/or moral support about the event,” (c) “received advice about the event,” and (d) “received any other type,” with specification requested in the case of (d). The total sum of instances of emotional support received across stressors, which could range from 0 (no SS) to 21 (reported all types of SS), was then calculated and used for analysis.

PES and PIS.

Perceived SS was assessed using the 12-item Interpersonal Support Evaluation List (ISEL; Cohen, Mermelstein, Kamarck, & Hoberman, 1985), which includes measures of emotional and instrumental support.² Items such as “I feel that there is no one I can share my most private worries or fears with” (PES) and “If I had to go out of town for a few weeks, it would be difficult to find someone who would look after my house or apartment(i.e., plants, pets, garden, etc.)”(PIS) were rated by participants on a 4-point scale from 1 (definitely false) to 4 (definitely true). Ratings were then reverse-scored if necessary and then averaged within each type of support. Higher scores reflect more support. Internal consistencies were adequate (emotional support: α = .90; instrumental support: α = .81).

Aging attitudes.

Aging attitudes were assessed with the 12-item Expectations Regarding Aging (ERA) Questionnaire (Sarkisian, Steers, Hays, & Mangione, 2005). Items such as “When people get older, they need to lower their expectations of how healthy they can be.” and “Forgetfulness is a natural occurrence just from growing older.” were rated on a 4-point scale ranging from 1 (definitely false) to 4 (definitely true). Items were averaged, with higher scores indicating more negative ERA and lower scores indicating more positive ERA. Internal consistency was adequate (α = .91).

Covariates.

We controlled for self-reported age, gender, race, health, marital status, and education in all analyses. Sex, race, and marital status were treated as binary variables, with males, non-Hispanic Whites, and non-married people serving as reference groups. Finally, we assessed physical health by averaging 2 items (e.g., “Overall, how would you rate your physical health?”) and education using each participant’s report of the highest level of education he or she obtained. Age, health, and education were standardized prior to including them in analyses. Tables 1 and 2 list the descriptive statistics and correlations among the main variables, respectively.

Table 1.

Means (SDs) of sample characteristics and variables of interest.

	M	SD
Female (%)	42.10
Married (%)	48.80
Age	49.88	17.15
Health	3.64	0.85
Education	2.43	1.09
Stress	2.73	2.26
RES	1.48	2.37
PES	3.17	0.83
PIS	3.04	0.76
ERA	2.41	0.64
AARC gains	86.98	17.72
AARC losses	62.73	20.77

Note. RES = received emotional support; PES = perceived emotional support; PIS = perceived instrumental support; ERA = expectations regarding aging; AARC = awareness of age-related change.

Table 2.

Bivariate correlations between variables of interest.

	1	2	3	4	5	6	7
1. Stress
2. ERA	.14
3. Gains	.13	−.25***
4. Losses	.25***	.60***	−.01
5. RES	.49***	−.02	.21**	.15***
6. PES	−.05	−.30***	.43***	−.38***	.17
7. PIS	−.01	−.40***	.30***	−.48***	.14	.77****

Note. ERA = expectations of aging; RES = received emotional support; PES = perceived emotional support; PIS = perceived instrumental support.

^*p < .05;

^**p < .01;

^***p < .001.

Procedure

Participants completed a consent form followed by a background questionnaire. They then completed a battery of questionnaires that assessed their current mood, recent stress, SS, AARC, other measures not related to the current study, and expectations regarding their own aging. Upon completion of the survey, participants were debriefed and offered the opportunity to view five positive images to alleviate any negative emotions potentially elicited by the recall of stressful life events. This study was approved by the institutional review board at NC State University.

Results

We conducted separate multiple linear regressions on AARC losses and gains. These models treated stress, SS, and ERA as predictors and age,³ health, education, gender, race, and marital status as controls.⁴ Simple slopes analyses were examined when significant interactions emerged, with corrections for multiple comparisons within each of these tests at p = .025.

AARC losses

Received support.

We first examined whether RES⁵ and ERA moderated the relationship between stress and AARC losses. As presented in Table 3, significant main effects of stress and ERA were obtained which indicated that more stress and negative ERA resulted in greater losses. A significant Stress RES interaction was also obtained. Figure 1 depicts this interaction, showing the effect of stress at high and low levels of RES (±1 SD).⁶ A nonsignificant and relatively flat slope associated with low RES (t = −1.41, p = .160) suggests no impact of stress at this level. In contrast, a significant and positive slope at high levels of RES (t = 2.86, p = .005) indicated that higher stress resulted in more losses for those with higher levels of RES. The difference in losses when stress was low was not significant (t = 2.26, p = .03).

Table 3.

AARC losses regressed on stress, SS, and ERA.

	Received support			Perceived support
	B	SE	β	B	SE	β
Step 1
Age	2.63	1.39	.04	2.45	1.46	.12
Health	−4.42	1.26	−.22***	−3.93	1.27	−.20**
Sex	−6.43	2.59	−.16*	−4.67	2.54	−.11
Race	0.52	3.44	.01	−0.65	3.37	−.01
Education	−1.45	1.37	−.02	−2.12	1.32	−.10
Marital status	2.14	2.59	.02	6.83	2.73	.17*
Step 2
Stress	4.58	2.05	.07*	1.39	1.65	.07
ERA	10.96	1.88	.17***	12.80	2.15	.64***
ES	−4.67	3.06	−.07	−1.53	2.38	−.08
IS				−5.55	2.21	−.29*
Step 3
Stress × ES	7.27	2.49	.14**	1.69	2.41	.09
Stress × ERA	0.22	1.94	.00	−0.09	2.12	−.00
ES × ERA	3.27	3.07	.04	−9.43	2.90	−.63**
Stress × IS				−0.66	2.20	−.03
IS × ERA				3.92	2.45	.25
ES × IS				−1.50	1.25	−.11
Step 4
Stress × ES × ERA	3.34	2.88	.05	6.02	2.94	.37*
Stress × IS × ERA				−5.31	2.35	−.35*
Stress × ES × IS				0.62	1.44	.05
ES × IS × ERA				−3.30	1.15	−.45**
Step 5
Stress × ERA × ES × IS				−1.28	1.20	−.16
R 2			.58			.64

Note. SE = standard error; AARC = awareness of age-related change; SS = social support; ERA = expectations regarding aging; ES = emotional support; IS = instrumental support. Sex: female = 0, male = 1; race: non-Hispanic White = 0, persons of color = 1; marital status: unmarried = 0, married = 1; R² for model at last step.

^*p < .05;

^**p < .01;

^***p < .001.

Figure 1.

AARC losses as a function of stress and RES. AARC = awareness of age-related change; RES = received emotional support.

Perceived support.

We ran the same analysis but with both PES and PIS included in the model. Significant main effects of PIS and ERA were obtained which indicated that more PIS predicted fewer losses and more negative ERA predicted greater losses. A significant Stress × PES × ERA interaction, which qualified a lower order ERA × PES interaction, and a significant Stress × PIS × ERA were also obtained. Figure 2 depicts these three-way interactions, showing the effect of stress at high and low levels of PES or PIS separately for individuals with negative and positive ERA (±1 SD). Follow-up tests suggested that PES had a stronger effect on losses in the negative compared to the positive ERA group. High compared to low PES was associated with significantly fewer losses in individuals with negative ERA (t = −2.84, p = .005) but not in those with positive ERA (t = 2.03, p = .045; Figure 2(a)). In contrast, PIS had a stronger effect on losses in the positive compared to the negative ERA group, with high levels associated with fewer losses (t = −2.81, p = .006; Figure 2(b)). Moreover, we obtained a significant PES × PIS × ERA interaction (see Figure 3) primarily due to higher PES predicting fewer losses at high PIS in the negative ERA group (t = −3.39, p =.001).

Figure 2.

AARC losses as a function of stress, (a) PES or (b) PIS, and ERA. AARC = awareness of age-related change; PES = perceived emotional support; PIS = perceived instrumental support; ERA = expectations regarding aging.

Figure 3.

AARC losses as a function of PES, PIS, and ERA. AARC = awareness of age-related change; PES = perceived emotional support; PIS = perceived instrumental support; ERA = expectations regarding aging.

AARC gains

Received support.

The model assessing the effects of stress, RES, and ERA on AARC gains did not yield any significant effect (ps > .11; Table 4).

Table 4.

AARC gains regressed on stress, SS, and ERA.

	Received support			Perceived support
	B	SE	β	B	SE	β
Step 1
Age	2.06	1.48	0.12	0.68	1.51	0.04
Health	6.44	1.39	0.38***	5.50	1.39	0.33***
Sex	−3.87	2.87	−0.11	−2.38	2.82	−0.07
Race	10.94	3.73	0.24**	10.22	3.62	0.23**
Education	−0.85	1.46	−0.05	0.00	1.40	0.00
Marital status	−0.54	2.86	−0.02	0.32	2.99	0.01
Step 2
Stress	0.59	2.22	0.03	2.70	1.79	0.15
ERA	−3.33	2.07	−0.19	1.05	2.34	0.06
ES	2.68	3.31	0.15	10.25	2.58	0.61***
IS				−4.28	2.40	−0.25
Step 3
Stress ES	1.26	2.71	0.08	−1.09	2.62	−0.07
Stress × ERA	3.39	2.13	0.20	−1.06	2.32	−0.06
ES × ERA	−4.64	3.35	−0.22	−7.12	3.16	−0.55*
Stress × IS				1.50	2.44	0.09
IS × ERA				5.95	2.66	0.44*
ES × IS				1.27	1.28	0.11
Step 4
Stress × ES × ERA	4.47	3.17	0.23	5.66	3.19	0.41
Stress × IS × ERA				−4.13	2.53	−0.32
Stress × ES × IS				−0.37	1.43	−0.04
ES × IS × ERA				−1.40	1.23	−0.22
Step 5
Stress × ERA × ES × IS				0.19	1.30	0.03
R 2			0.32			0.42

Note. SE = standard error; AARC = awareness of age-related change; SS = social support; ERA = expectations regarding aging; ES = emotional support; IS = instrumental support. Sex: female = 0, male = 1; race: non-Hispanic White = 0, persons of color = 1; marital status: unmarried = 0, married = 1; R² for model at last step.

^*p < .05;

^**p < .01;

^***p < .001.

Perceived support.

As in the case of AARC losses, we ran an analysis with both PES and PIS as predictors. A significant main effect of PES as well as a PES × ERA interaction was obtained. Higher PES predicted more AARC gains, especially in individuals with positive ERA (t = 3.88, p < 0.001) but not negative ERA (t = 0.86, p = 0.39; Figure 4(a)). We also obtained a significant PIS × ERA interaction, which was again driven by the positive ERA group. In contrast to the pattern associated with PES, however, higher PIS predicted fewer gains in these individuals (t = −2.50, p = 0.014) but had no effect those with negative ERA (t = 0.56, p = 0.58; Figure 4(b)).

Figure 4.

AARC gains as a function of (a) PES or (b) PIS, and ERA. AARC = awareness of age-related change; PES = perceived emotional support; PIS = perceived instrumental support; ERA = expectations regarding aging.

Discussion

The current study investigated the synergistic effects of stress, SS, and aging attitudes on awareness of age-related gains and losses. We expected stress would be negatively related to gains and positively related to losses, with SS attenuating this effect for people with negative aging attitudes and strengthening it for those with positive attitudes. We also explored whether perceived and emotional SS would have more positive effects than received and instrumental SS.

Results revealed several effects associated with stress, SS, and aging attitudes that were in the hypothesized directions. More stress and negative attitudes predicted greater losses (H1_Loss and H3_Loss) and more perceived SS predicted greater gains (H2_{Gain: a}). We also observed interactions that were partially consistent with our expectations. Received emotional SS moderated stress effects on losses, with higher levels of received SS associated with even more losses (H2_{Loss: b}). Additionally, perceived emotional and instrumental SS as well as aging attitudes moderated the impact of stress on losses. Whereas greater losses were associated with lower levels of perceived emotional SS in people with negative aging attitudes, they were associated with lower levels of PIS in those with positive attitudes (H2_{Loss: a, c, d}; H4_Loss). Analyses similarly revealed significant interactions between perceived SS and aging attitudes in the context of gains. There were no effects among people with negative attitudes, but higher levels of perceived emotional and instrumental SS were respectively associated with greater and fewer gains among those with positive attitudes (H2_{Gain: a, c}). Thus, although stressful events themselves can have a negative effect on AARCs (i.e., greater losses), their impact can generally be exacerbated by lower levels of SS. In the subsequent sections, we tease apart these findings further and interpret them in the light of extant research as well as the strengths and limitations of the current study.

Domain-specificity of SS effects

We found a somewhat consistent pattern of results associated with the measure (i.e., perceived and received) and the function (i.e., emotional and instrumental) of SS. Related to measures, received emotional SS only affected AARC losses, whereas perceived emotional SS affected both AARC gains and losses. The difference in the effect of emotional SS appears to concern the correspondence between the SS measure and the valence of AARC. SS types associated with negative effects (i.e., received) potentially have the strongest impacts in negatively valenced domains (i.e., losses). The fact that we observed exaggerated losses at higher levels of received emotional SS but reduced losses at higher levels of perceived emotional SS further underscores this idea. This interpretation aligns with our predictions and prior findings of more negative impacts of received support compared to perceived support (e.g., Reinhardt et al., 2006; Wethington & Kessler, 1986) but expands upon previous work by revealing their domain-specific characteristics (which we did not anticipate). The same rationale may apply to domain differences between measures of instrumental SS but remains speculative since we did not have both measures of this function in the current study.

Related to functions, we also observed differences between AARC domains in the nature of the effects associated with emotional and instrumental SS. On the one hand, higher levels of emotional SS corresponded to greater gains on average in people with positive aging attitudes. This suggests compounding benefits associated with a combination perceived and emotional SS (e.g., Cohen, 2004; Cohen & Willis, 1985; Krause, 1986; Lee & Goldstein, 2016; Rafaeli et al., 2013; Reis, Clark, & Holmes, 2004). The fact that we observed this only for those with positive attitudes also suggests that having strong socioemotional resources forms a basis for feeling positively about one’s aging in general. On the other hand, higher levels of perceived instrumental SS resulted in fewer losses as well as fewer gains in these same individuals. The finding related to losses supports work showing increases in same-day vigor and decreases in next-day anger associated with instrumental SS (Shrout et al., 2010) and suggests net benefits associated with perceived instrumental SS, with the positive aspects of the measure outweighing the negative aspects of its function. The finding related to gains, however, highlights the importance of considering the implications of SS on self-appraisals.

Thus, accounting for domain differences in the effects of SS involves consideration of both the measure and function of SS. Differences in the strength of effects may partly revolve around a match between the valence of the outcome domain (i.e., positive [gain] or negative [loss]) and that associated with the measure (i.e., positive or negative). Evidence based on developmental theories provides indirect support for this argument. For example, implicit beliefs about aging have their strongest effects on behavioral outcomes when the domain associated with the outcome in-question matches the domain associated with the content of such beliefs (e.g., cognitive vs. physical; Levy & Leifheit-Limson, 2009). Differences in the direction of effects may concern a match between the function and an individual’s coping needs.

Limitations of SS and positive aging attitudes as psychological resources

Two findings from the current study allude to potential boundaries within which SS and positive attitudes generally promote adaptive responses to stress but that beyond which undermine coping efforts and worsen stress effects (e.g., for SS; see Cutrona & Russell, 1990). First, opposite to our expectations, the level of perceived SS—emotional and instrumental—tended to result in the largest differences in AARC losses when stress was low. This suggests that SS and attitudes have limited efficacy in helping people cope with high amounts of stress, irrespective of the resources afforded by them. Second, high compared to low perceived SS was associated with fewer average losses regardless of ERA, forwarding the idea that SS acts predominately as a stress buffer. Yet high perceived instrumental SS resulted in fewer average gains in people with positive attitudes. Interestingly, in this group, high SS may conflict with a positive sense-of-self (e.g., high self-efficacy and self-reliance), whereas low SS may reaffirm it. Moreover, these same individuals may interpret this type of support as an unexpected loss, which could explain some of the just-described domain differences.

In line with this view, people who hold positive attitudes about their own aging report feeling emotionally worse on days when they perceived more age-related losses than individuals who hold more negative attitudes (Neupert & Bellingtier, 2017). Other findings have also shown that SS received during times of stress can lead to heightened reactivity or worse psychological adjustment and well-being (e.g., Bolger et al., 2000), potentially by undermining feelings of self-esteem (Lepore et al., 2008). Together, our findings advance a match perspective whereby SS compensates for depleted psychological resources (i.e., pessimism about one’s aging or ego loss) in the presence of stress, but may not do so in the absence of it.

Limitations and future directions

We take care to interpret the results of the current findings in the light of a few limitations. First, we utilized an online crowdsouring platform, MTurk, for the recruitment of participants and the administration of the study. Although these types of samples tend to represent more demographically diverse populations compared to standard internet and typical college samples (Buhrmester, Kwang, & Gosling, 2011), they may or may not respond similarly on psychological assessments compared to traditional laboratory-based samples (Behrend, Sharek, & Meade, 2011; Buhrmester et al., 2011). To this end, older adults in our study may differ in meaningful ways from those who participate in other online- or laboratory-based studies. Future studies should replicate these results in different assessment settings. Relatedly, we restricted participation to people within the U.S. Given differences in belief systems (e.g., filial piety; individualistic vs. collectivistic) between different cultures, studies conducted in non-U.S. samples may cast a different light on the current effects.

Second, the cross-sectional design of the study precludes us from making strong causal conclusions and leaves open the question as to whether SS makes people more aware of age-related changes or an increased awareness of such changes makes them more reactive to SS. We also make the strong assumption that reductions in self-esteem, presumably a psychological resource, and negative perceptions of SS in the group with positive aging attitudes resulted in stronger stress effects but could not assess these pathways directly in the current work. New studies that deploy measures across multiple time points and include other self-appraisals and cognitive representations of SS could help validate these claims.

Third, future work may also benefit from considering the impacts of specific relationship characteristics (e.g., nature and quality of relationship between the provider and the receiver). For instance, support received from social ties high in both negativity and positivity predicts greater stress reactivity compared to support received from a positive-only tie (Holt-Lunstad, Uchino, Smith, & Hicks, 2007). Similarly, certain characteristics of the provider (e.g., age, relation, affect) may influence SS effects. We also treated age as a control variable, but new studies that examine age differences could offer meaningful insights on the efficacy of these results at different points in the life span. People make important transitions throughout adulthood in terms of their social roles and relationships, personal goals, and life’s circumstances. Such changes partly reflect concrete experiences that play a key part in our ability to monitor our own development. Fourth, the current study did not include a measure of received instrumental SS. If perceived and emotional SS compounds its stress-buffering impacts, then received and instrumental SS might exacerbate its stress-amplifying effects. Future research should incorporate more comprehensive measures of received and perceived SS to aid in comparisons. Finally, we can only speculate why effects depended upon domain of perceived age-related change. Future research could take a more domain-specific approach by examining the associations between individual types of SS and the behavioral domains assessed in the AARC measure (e.g., interpersonal, cognitive, physical).

Conclusions

Findings from the current study point to a potential pathway by which SS can moderate the effect of stress on psychological outcomes. They suggest that the match or mismatch between internal factors (e.g., coping resources associated with generalized attitudes) and external forces (e.g., availability of SS) can influence self-perceptions of aging. Whereas individuals may feel comforted knowing that they can turn to members of their social networks when necessary and view SS offered as a normal occurrence, the realization of its receipt may elicit negative feelings about the self, including one’s age, and enhance the negative impact of stress. Moreover, in a society that tends to idealize youthfulness, the tendency to internalize the need or receipt of support in a negative fashion may be especially pronounced in people with positive (and possibly unrealistic) attitudes about their own aging.

Our findings may help guide health promotion efforts in adulthood. Some work shows that individuals who participated in an educational intervention to reduce negative aging attitudes and promote physical activity also reported increases in exercise (i.e., type, minutes, intensity), awareness of age-related gains, and control beliefs (Brothers & Diehl, 2016). One study has also illustrated synergistic effects of SS and self-efficacy on similar outcomes, with people low in self-efficacy having a lower likelihood of activity than those high in self-efficacy even with SS (Warner, Ziegelmann, Schüz, Wurm, & Schwarzer, 2011). Could an intervention that promotes healthy aging attitudes and appraisals of SS as well as encourages caregivers to offer support that facilitates self-reliance and self-efficacy be beneficial? Our findings emphasize the need to increase our understanding of mechanisms and expand upon already promising intervention efforts.