The Role of General and Daily Control Beliefs for Affective Stressor-Reactivity Across Adulthood and Old Age

Abstract

Background: General and situational control beliefs have been examined separately as buffers of the effects of daily stressors on affective well-being. However, general (trait) control beliefs reflect perceived ability to adapt, change, and influence overall life circumstances, whereas situational (daily) control beliefs reflect perceived ability to manage current circumstances and achieve desired outcomes. Method: Using 9 weeks of daily reports from 150 adults aged 18–89 years, we examined the extent that general and daily control beliefs buffer the between- and within-person associations involving stressors and negative and positive affect (i.e., daily stress processes) and whether/how the extent of buffering differs with age. Results: Aligning with prior findings, both greater average stressor exposure and experiencing a daily stressor compromised daily affective well-being and both higher general and daily control beliefs facilitated daily affective well-being. Specific to the motivating hypotheses, both general and daily control beliefs buffered daily stressor-reactivity. Age was associated with individuals’ daily stressor-buffering, such that stressor-reactivity was more effectively damped at older ages. Associations between general control beliefs and daily stress processes were age invariant. Discussion: Mixed evidence of age differences across general and daily control beliefs highlights how within-person processes may differentially contribute to well-being as individuals accommodate age-related strengths and vulnerabilities.

Daily stressors—“hassles” people experience in daily life that threaten psychosocial resources—and the psychological and physiological responses to daily stressors are associated with proximal and long-term well-being (Almeida, 2005; Charles, Piazza, Mogle, Sliwinski, & Almeida, 2013; Lazarus & Folkman, 1984; Mroczek et al., 2015). Theoretical models suggest that stressors have both direct and indirect impacts on emotional systems (Almeida, 2005; Lazarus, 1999; Lazarus & Folkman, 1984; Ong & Bergeman, 2004). Indeed, when looking at between-person associations, greater overall exposure to stressors is associated with poorer affective well-being (Birditt, Cichy, & Almeida, 2011; Bolger, DeLongis, Kessler, & Schilling, 1989; Bolger & Schilling, 1991; Zautra, Affleck, Tennen, Reich, & Davis, 2005). When looking at within-person associations, individuals are generally reactive to stressors, such that they have poorer affective well-being on days when a stressor occurred compared to stressor-free days (Almeida, 2005; Hay & Diehl, 2010). In identifying risk and resilience factors (Diehl & Hay, 2010; Ong & Bergeman, 2004), researchers find that a variety of sociodemographic (e.g., age, gender; socioeconomic status; Almeida & Horn, 2004; Diehl, Hay, & Chui, 2012; Grzywacz, Almeida, Neupert, & Ettner, 2004); psychosocial (e.g., personality, control beliefs; Mroczek & Almeida, 2004; Neupert, Almeida, & Charles, 2007; Ong, Bergeman, Bisconti, & Wallace, 2006); and mental and physical health factors (e.g., anxiety, depression, chronic health conditions; Diehl, Hay, & Chui, 2012; Ong, Bergeman, & Bisconti, 2005; Piazza, Charles, & Almeida, 2007) either exacerbate or buffer the effects of stressor experiences on affective well-being. In this article, we examine the role of both trait and state control beliefs as potential buffers of affective stressor-reactivity.

The Role of Control Beliefs for Daily Stress Processes

Individuals’ perceived control over life and circumstances reflects subjective beliefs about their ability to apply effective coping strategies (Lachman & Weaver, 1998a; Lazarus & Folkman, 1984; Levenson, 1981). High control beliefs serve to motivate the individual to use coping strategies in the face of stressors, thereby increasing the likelihood that coping strategies effectively buffer the impact of stressors on negative affect and overall well-being (Bandura, 1997; Heckhausen, Wrosch, & Schulz, 2010; Lachman, Neupert, & Agrigoroaei, 2011; Lazarus & Folkman, 1984). Control beliefs operate in two ways. General (trait) control beliefs reflect individuals’ perception that they have the resources and ability needed to face life’s demands and obtain desired outcomes (Bandura, 1997; Skinner, 1996). Adding proximal information to these general control beliefs, more situational (example.g., daily) control beliefs adjust individuals’ beliefs about whether they have the resources and ability to cope and obtain the desired outcome in specific situations. When situational (e.g., daily) control beliefs are high, individuals supplement their general control beliefs with more specific perceptions about whether they can successfully face a particular situation. Thus, it is possible for an individual to have high general control beliefs but low control in some situations or to have low general control beliefs and high control in some situations (e.g., when faced with a specific stressor). Daily control beliefs reflect an aggregate of the situational sense of control from a particular day. While both general and daily control beliefs should buffer the effects of daily stressors on well-being, the more proximal, day-to-day feelings of control may be more strongly associated with daily stress processes than trait-like general control beliefs (Ong et al., 2005).

Initial evidence from observational and experimental studies suggests that higher general and daily control beliefs are associated with ameliorated reactivity to daily stressors. For example, Diehl and Hay (2010) found that higher daily control beliefs buffered affective reactivity to daily stressors. Neupert et al. (2007) also found that general control beliefs (operationalized as high personal mastery and low environmental constraint), buffered affective reactivity to stressors in domains salient to specific ages. Using subjective stress as the outcome, Bollini, Walker, Hamann, & Kestler (2004) found that greater general control reduced biological reactivity to an experimentally induced (non-)controllable stressor. Ong et al. (2005) also examined the extent that daily buffering was moderated by general control. They found, for recently bereaved widows, higher daily control beliefs were associated with buffered stress–anxiety associations and this buffering effect was stronger when general control beliefs were higher. Evidence supports the notion that both general and daily control beliefs buffer individuals’ affective response to stressors. However, empirical studies have so far examined the role of general and daily control only separately rather than simultaneously, with the exception of Ong et al., (2005). It is possible that, for example, daily control beliefs may buffer daily stress processes differently for those with high as compared to low general control beliefs. Also unknown (as pointed out by Villarreal & Heckhausen, in press), is how the buffering effects provided by general and daily control beliefs change with age.

Age Differences in Control Beliefs as a Stress Buffer

Following Lazarus and Folkman’s (1984) cognitive approach to stress and control beliefs, actual or perceived age-related decline of psychosocial resources would suggest that older adults are faced with more and more difficulties when attempting to draw on daily control beliefs to buffer the effects of daily stressors. In accordance with this deficit model, the Strength and Vulnerability Integration model (SAVI; Charles, 2010) suggests that age-related vulnerability, specifically, operationalized as reduced ability to emotionally self-regulate, occurs when older adults are unable to employ avoidant coping strategies. In terms of stressors and affect, age-related increases in stressor-reactivity manifest when individuals lack control of the day’s situations or stressors. That is, older adults’ affective well-being is embedded in the more proximal, daily appraisals of control. As such, older adults should be less emotionally reactive to stressors when they can use avoidance or distraction techniques. However, in cases when the stressors are viewed as uncontrollable, older adults should be more emotionally reactive (more vulnerable). At the same time, developmental theories of control suggest that across the life span, higher general control beliefs protect affective well-being by promoting proactive coping strategies (Lachman, 2006; Wrosch, Heckhausen, & Lachman, 2006).

Life-span developmental trajectories of stressor exposure and general control beliefs must also be considered when examining age-related differences in affect. Mean levels of general control beliefs increase through early adulthood, peaking around midlife, and plateauing and/or declining in old age (Gerstorf, Ram, Lindenberger, & Smith, 2013; Mirowsky, 1995; Mirowsky & Ross, 2007; Pearlin, Nguyen, Schieman, & Milkie, 2007). Stressor exposure declines with older age due to selection out of stressor experiences (Baltes & Baltes, 1990; Baltes, Wahl, & Schmid-Furstoss, 1990; Brose, Scheibe, & Schmiedek, 2013). Differences in how age-related declines in general control beliefs are coupled with age-related declines in stressor exposure effects will alter the extent that general control beliefs buffer stressor exposure. When stressor exposure declines first, stability in general control may also manifest as increased buffering. When general control beliefs decline first, stability in exposure may manifest as decreased buffering. When the timing of changes also differ across persons (interindividual differences in timing of developmental sequences; see Ram & Grimm, 2015), the evidence of buffering gets even more complicated to parse.

Indeed, empirical evidence regarding age differences in how daily and general control beliefs buffer individuals’ stressor-reactivity is limited and inconsistent. With a sample of adults age 25–74 years, Neupert and colleagues (2007) found that general control beliefs buffered affective reactivity to stressors in domains that were salient to the specific age (e.g., interpersonal stressors, particularly relevant for younger adults who were aiming to form relationships). In a sample of adults age 18–89 years, Hay and Diehl (2010) found that higher daily control beliefs buffered stressor-reactivity to interpersonal, home, and health stressors but the effects were age invariant. For non–age-specific stressor domains, such as home overload stressors, stressor-reactivity was age invariant or remained high in older age and higher general control beliefs buffered stressor-reactivity. In a study of bereaved older adults (age 61–83 years), Ong and colleagues (2005) found that daily, not general (mastery), control beliefs, buffered the association between daily stress and anxiety. Although there is limited empirical evidence, theory suggests that older age will be associated with stronger daily stressor buffering due to increasing relevance of daily stressors and ability to apply avoidant coping strategies (Charles, 2010). Acknowledging that individuals are overlaying their daily control beliefs on their general control beliefs, there is a need to examine how the unique contribution of daily control beliefs differs across trait-level and/or age-related differences in general control beliefs.

Present Study

Both general and daily control beliefs are associated with buffered stressor-reactivity but still unknown are the extent that general and daily control beliefs uniquely buffer daily affective stressor-reactivity and whether/how the extent of buffering differs with age. Using an age heterogeneous sample (age range 18–89 years), the present study examines age differences in the associations among daily stressor-reactivity, stressor exposure, general and daily control beliefs, and daily negative and positive affect. By examining the unique contributions of general and daily control beliefs for affective well-being, the current study investigates important notions from life-span developmental theory suggesting that development is shaped by a complex interplay of gains and losses across timescales (e.g., Nesselroade, 1991; Ram & Gerstorf, 2009). The present study aims to corroborate and extend earlier findings by attempting to disentangle the affective implications of control beliefs and stressors at multiple time scales and across age. Specifically, the present study adds to the literature by expanding upon previous work that focuses on either daily or general control beliefs in the stress process (see Diehl & Hay, 2010) and is the only study to date that combines a multitimescale approach to control beliefs with a multidomain approach to affective well-being in an adult life-span sample of healthy adults.

Using a multilevel model that allows separation of (daily) within-person and (general) between-person associations, we examine (a) the extent that general and daily control beliefs buffer the (between- and within-person) associations between stressors and negative affect and (b) whether/how the extent of buffering differs with age. Following theoretical propositions (i.e., Charles, 2010; Lazarus & Folkman, 1984) and empirical findings, we expect that both general and daily control beliefs will serve as a stress buffer with negative affect and that the extent of buffering increases with age. Due to mixed results regarding positive affective stressor-reactivity (e.g., Schilling & Diehl, 2014; Stawski, Sliwinski, Almeida, & Smyth, 2008; Uchino, Berg, Smith, Pearce, & Skinner, 2006; Zautra et al., 2005), we also explore the extent that control beliefs support positive affect in the face of stressors across adulthood.

Method

Participants

Data are drawn from the Intraindividual Study of Affect, Health, and Interpersonal Behavior (iSAHIB; Ram et al., 2014), a multiple time-scale experience sampling study of N = 150 adults recruited from the community surrounding Pennsylvania State University. The sample was stratified by gender (51% women) and age (five 14-year age-bins) to provide coverage across the entire adult life span, age 18–89 years (M_Age = 47.64, SD_Age = 18.85). Participants had obtained between 2 and 24 years of formal education (M_Educ = 16.36, SD_Educ = 3.90), earned median annual household income of “$50,000 to $74,999” (range_Income: “under $20,000” to “$200,000 and over”; Mode_Income = “$20,000 – $49,999”), and self-identified as Caucasian (91%), African American (4%), Asian American (1%), and Mixed or other (4%) ethnicity.

Procedure

Participants completed three, 21-day measurement bursts spaced at approximately 4.5 months intervals (M = 5.25 months between Bursts 1 and 2; M = 4.25 months between Bursts 2 and 3). During each measurement burst, individuals provided (in addition to within-day reports about their social interactions) end-of-day reports about their feelings, thoughts and behaviors using a customized “iSAHIB Surveys” application on a study-provided smartphone. Prior to and after each measurement burst, participants visited the laboratory to receive training/debriefing and complete web-based questionnaires on their demographics, health, and personality. Participants received $500 compensation for completing the entire protocol. In total, participants provided between 13 and 76 days of data (M = 57.05 SD = 12.68), with 95% providing over 20 daily reports and 82% providing over 50 daily reports over the course of about one year (note that two participants provided data beyond the requested 63 days due to rescheduled postburst visits; all “extra” days of data were checked and deemed valid).

Measures

Daily affect

As part of each end-of-day questionnaire, participants indicated to what degree they felt various affective states that day answering “Today I felt [affect],” on a “touch-point continuum” (slider-type interface) scaled from 0 (Not at All) to 100 (Strongly; numbers not visible to participants). Participants’ daily negative affect (NA) was calculated as the average of responses to nine items: nervous, embarrassed, upset, tense, sluggish, sad, bored, disappointed, and depressed (Cronbach’s α = .87). Participants’ daily positive affect (PA) was calculated as the average of responses to 10 items on the same 0 to 100 touch-point continuum: enthusiastic, happy, alert, proud, excited, calm, peaceful, satisfied, relaxed, and content (Cronbach’s α = .93).

Daily stressors

Individuals’ level of stress each day was assessed as response to “Today I felt stressed” on the same 0 (Not at All) to 100 (Strongly) touch-point continuum scale. Immediately after that item, participants were prompted, “Based on the stress you just indicated, what were the sources of your stress?” and asked to check as many of the following nine boxes as appropriate: being evaluated, work/education, health/accident, events that happened to others, interpersonal tensions, finances, home, other, or none (adapted from the Daily Inventory of Stressful Events, Almeida, Wethington, & Kessler, 2002). Summarizing across the nine boxes, we created a binary daily stressor variable indicating whether (=1) or not (=0) any stressor had occurred on that day. Stressor exposure, a person-level variable, was quantified for each individual as the proportion of study days that a stressor occurred. On average, participants reported experiencing one or more stressors (M = 1.28, SD = 1.22) on 72.15% of study days.

Daily Control Beliefs

As part of each end-of-day questionnaire, participants indicated the degree of feeling that “I had control over the things that happened to me today” using a touch-point continuum scaled from 0 (Strongly Disagree) to 100 (Strongly Agree). Daily control beliefs were person-mean centered to reflect daily deviations from an individual’s average level of daily control.

General control beliefs

General control beliefs were measured at each lab visit (prior to and after each 21-day burst) using three survey questions (1 = No control at all to 7 = Very much control): “I often feel helpless in dealing with the problems of life,” “What happens in my life is often beyond my control,” and “I have little control over the things that happen to me.” (adapted from Lachman & Weaver, 1998b). Participant responses were averaged across all six visits to obtain a general control beliefs score.

Sociodemographic characteristics

Age, gender, and years of education were assessed at the baseline lab visit.

Data Analysis

The nested nature of the data (repeated days nested within persons) was accommodated using a multilevel model (Snijders & Bosker, 1999) where the repeated measures of daily affect for person i at occasion t, Negative Affect_it or Positive Affect_it, were modeled as

$$\begin{array}{l}Affec{t}_{it}={\beta }_{0i}+{\beta }_{1i}DailyStresso{r}_{it}+{\beta }_{2i}DailyContro{l}_{it}\\ +{\beta }_{3i}DailyStresso{r}_{it}*DailyContro{l}_{it}+e_{it}\end{array}$$ (1)

where β_0i is a person-specific baseline NA or PA (intercept) coefficient, β_1i is a person-specific daily stressor-reactivity coefficient that indicates the extent that an individual’s NA or PA changes in response to daily stressors, β_2i is a person-specific daily control-reactivity coefficient that indicates the extent that an individual’s NA or PA changes in response to day-to-day changes in perceived daily control, β_3i is a person-specific daily stressor-buffering coefficient that indicates the extent that (higher) daily perceived control buffers an individual’s stressor-reactivity, and e_it is residual error that is assumed normally distributed. The person-specific baseline (intercept), daily stressor-reactivity, daily control-reactivity, and daily stressor-buffering coefficients are, in turn, modeled as a function of between-person differences in stressor exposure, general control beliefs, and age (with gender and education included as covariates). Starting with a fully saturated model, non-significant higher order interactions that were peripheral to the hypotheses were pruned, to obtain a final between-person model that was

$$\begin{array}{c}{\beta }_{0i}= {\gamma }_{00}+{\gamma }_{01}StressorExposur{e}_i+{\gamma }_{02}GeneralContro{l}_i\\ +{\gamma }_{03}Ag{e}_i +{\gamma }_{04}StressorExposur{e}_i*GeneralContro{l}_i\\ +{\gamma }_{05}StressorExposur{e}_i*Ag{e}_i+{\gamma }_{06}GeneralContro{l}_i\\ *Ag{e}_i+{\gamma }_{07}StressorExposur{e}_i*GeneralContro{l}_i\\ *Ag{e}_i+{\gamma }_{08}Gende{r}_i+ {\gamma }_{09}Educatio{n}_i+u_{0i}\end{array}$$ (2)

$$\begin{array}{l}{\beta }_{1i}= {\gamma }_{10}+ {\gamma }_{11}StressorExposur{e}_i + {\gamma }_{12}GeneralContro{l}_i\\ + {\gamma }_{13}Ag{e}_i +{\gamma }_{14}GeneralContro{l}_i*Ag{e}_i+ u_{1i}\end{array}$$ (3)

$$\begin{array}{l}{\beta }_{2i}= {\gamma }_{20}+{\gamma }_{21}StressorExposur{e}_i+{\gamma }_{22}GeneralContro{l}_i\\ + {\gamma }_{23}Ag{e}_i+{\gamma }_{24}StressorExposur{e}_i*Ag{e}_i\\ + {\gamma }_{25}GeneralContro{l}_i*Ag{e}_i\end{array}$$ (4)

$${\beta }_{3i}= {\gamma }_{30}+{\gamma }_{31}StressorExposur{e}_i+{\gamma }_{32}Ag{e}_i$$ (5)

where u_0i and u_1i are residual between-person differences that are assumed multivariate normally distributed with variances ${\sigma }_{u_0}^2$ , ${\sigma }_{u_1}^2$ and correlation $r_{u_0{u}_1}^{}$ . Specific parameters of interest for the general hypotheses that control serves as a buffer of stress, were whether, on average, daily control beliefs moderate stressor-reactivity (γ₃₀; “daily stressor-buffering”) and whether general control beliefs moderate the between-person effects of stressor exposure (γ₀₄; “general buffering of stressor exposure effects”) and within-person daily stressor-reactivity (γ₁₂; “general buffering of daily stressor-reactivity”). Specific parameters of interest with respect to age-related differences were those that captured age differences in “daily stressor-buffering” (γ₃₂) and in “general buffering” of stressor exposure effects (γ₀₇) and stressor-reactivity (γ₁₄).

The multilevel model was fit to the data using the nlme library in R (Pinheiro, Bates, DebRoy, Sarkar, & R Core Team, 2016) with REML estimation and incomplete data (<1%) treated as missing at random (Little & Rubin, 1987). Statistical significance was evaluated at p less than .05. All person-level predictors were grand-mean centered so the parameter estimates depict effects for the typical person in the study (as described in the participants section above) on a no stressor day.

Results

Descriptive statistics and intercorrelations for the study variables are shown in Table 1. Consistent with expectations and prior studies (Almeida, 2005; Almeida & Horn, 2004; Zautra et al., 2005), stressor exposure was lower for men (M_diff = −.19), positively correlated with average negative affect (r = .58) and inversely related to positive affect (r = −.43) and age (r = −.24). Note that the percentage of stressor days is larger than other studies that use the DISE, due to measurement differences such that the present study asks participants to indicate their “sources of stress,” rather than specify particular events. General and average daily control beliefs were highly related (r = .72) and correlated with other variables in similar ways; inversely related to individuals’ stressor exposure (general control beliefs: r = −.41; average daily control beliefs: r = −.43) and average negative affect (general control beliefs: r = −.54; average daily control beliefs: r = −.49) and positively correlated with positive affect (general control beliefs: r = .46; average daily control beliefs: r = .73). The intraclass correlation coefficient of daily control beliefs indicated that 58% of the total variance is accounted for by between-person variance. Descriptively, the data appear to conform to the theory that control beliefs are associated with fewer stressor experiences, lower negative affect, and higher positive affect. Importantly, the correlation between measures of average daily and general control beliefs reflect the concordant and complementary nature of the daily (state) and general (trait) assessments of individuals’ control beliefs.

Table 1.

Sample-Level Descriptive Statistics and Correlations

	M	SD	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Construct
(1) Average negative affect (0–100)	19.85	10.31	—	−.454	.577	−.537	−.486	−.061	−.084	−.092
(2) Average positive affect (0–100)	57.47	14.56		—	−.426	.464	.730	.104	.094	−.032
(3) Stressor exposure (0–1)	0.72	0.28			—	−.408	−.425	−.235	−.188	.011
(4) General control beliefs (0–100)	5.16	0.90				—	.720	−.047	.031	.042
(5) Average daily control beliefs (0–100)	67.54	16.38					—	−.049	.088	−.043
(6) Age (18–89)	47.10	18.76						—	.011	.031
(7) Gender (1 = men)	0.49	0.50							—	.092
(8) Education (0–18)	16.36	3.90								—

Note: N = 150. M = Mean; SD = Standard deviation.

The Role of Control Beliefs for Daily Stress Processes and Negative Affect

Results from the multilevel model used to assess the extent that general and daily control beliefs buffer stressor exposure and stressor-reactivity are shown in Table 2. As seen in Model 1, the prototypical iSAHIB participant’s baseline NA on a nonstressor day was estimated as γ₀₀ = 15.484 (p < .001) on the 0 to 100 scale. As expected, there was evidence of stressor-reactivity, with NA γ₁₀ = 4.329 (p < .001) points higher on stressor days. Also as expected, individuals with greater stressor exposure tended to have higher baseline NA, γ₀₁ = 11.645 (p < .001). However, level of stressor exposure was not related to the extent of stressor-reactivity, γ₁₁ = 0.092 (p = .96). Consistent with prior studies, individuals with higher general control beliefs tended to have lower baseline levels of NA (γ₀₂ = −1.996, p = .004), and there was evidence of control-reactivity, with each unit increase in daily control decreasing NA by γ₂₀ = −0.179 (p < .001) points. General control beliefs did not moderate daily control-reactivity (γ₂₂ = 0.014, p = .05). In line with hypotheses, the pattern of associations indicated that both greater stressor exposure and daily stressors compromise daily affective well-being and that both higher general control beliefs and higher daily control beliefs facilitate daily affective well-being.

Table 2.

Results from Multilevel Model Examining Age Differences in How General and Daily Control Beliefs Buffer Stressor Exposure and Stressor-Reactivity

	Model 1: Negative affect		Model 2: Positive affect
	Est.	(SE)	Est.	(SE)
Fixed Effects
Intercept (${\gamma }_{00})$	15.484*	(0.920)	59.724*	(1.675)
Age (${\gamma }_{03})$	−0.006	(0.038)	0.111	(0.069)
Gender (${\gamma }_{08})$	0.293	(1.233)	−0.027	(2.287)
Education (${\gamma }_{09})$	−0.105	(0.157)	−0.267	(0.291)
Daily stressor (${\gamma }_{10})$	4.329*	(0.424)	−2.216*	(0.498)
Stressor exposure (${\gamma }_{01})$	11.645*	(2.899)	−11.399*	(5.254)
General control beliefs (${\gamma }_{02})$	−1.996*	(0.688)	4.278*	(1.246)
Daily control beliefs (${\gamma }_{20})$	−0.179*	(0.023)	0.345*	(0.024)
Age × Daily stressor (${\gamma }_{13})$	0.009	(0.022)	−0.001	(0.026)
Age × Stressor exposure $({\gamma }_{05})$	0.104	(0.122)	0.184	(0.225)
Age × General control beliefs $({\gamma }_{06})$	−0.060	(0.039)	0.099	(0.069)
Age × Daily control beliefs $({\gamma }_{23})$	0.001	(0.001)	−0.004*	(0.001)
Stressor exposure × Daily stressor $({\gamma }_{12})$	0.092	(1.662)	0.022	(1.969)
General control beliefs × Daily stressor $({\gamma }_{13})$	−0.897*	(0.427)	−0.044	(0.505)
Daily stressor × Daily control beliefs $({\gamma }_{30})$	−0.086*	(0.025)	−0.008	(0.026)
General control beliefs × Stressor exposure (${\gamma }_{04})$	−4.369	(2.756)	1.723	(5.075)
Stressor exposure × Daily control beliefs $({\gamma }_{21})$	−0.120	(0.073)	0.315*	(0.075)
General control beliefs × Daily control beliefs $({\gamma }_{22})$	0.014	(0.007)	−0.022*	(0.008)
Stressor exposure × Daily stressor × Daily control beliefs $({\gamma }_{31})$	−0.194*	(0.025)	−0.053	(0.090)
Age × General control beliefs × Daily control beliefs $({\gamma }_{25})$	−0.001*	(0.001)	0.001	(<0.001)
Age × General control beliefs × Daily stressor $({\gamma }_{14})$	−0.002	(0.131)	0.045	(0.030)
Age × Daily stressor × Daily control beliefs $({\gamma }_{32})$	−0.003*	(0.002)	0.005*	(0.001)
Age × Stressor exposure × General control beliefs $({\gamma }_{07})$	−0.086	(0.087)	0.088	(0.241)
Age × Stressor exposure × Daily control beliefs $({\gamma }_{24})$	0.004	(<0.001)	−0.007*	(0.002)
Random effects	Est.	CI	Est.	CI
Variance of intercept (${\sigma }_{{\text{u}}_0}^2$ )	41.95	(5.42, 7.74)	118.12	(154.79, 202.85)
Correlation intercept, daily stressor (${\text{r}}_{{\text{u}}_0{\text{u}}_1}^{})$	−0.060	(−0.540, 0.450)	−.156	(−.015, .028)
Variance daily stressor (${\sigma }_{{\text{u}}_1}^2)$	7.231	(1.924, 3.758)	8.12	(13.24, 21.53)
Residual variance (${\sigma }_{\text{e}}^2$ )	80.533	(8.833, 9.116)	82.24	(84.87, 87.59)
Fit indices
AIC	58,375		59,011
−2LL	−29,160		−29,478

Note: N = 150, T≈63; AIC = Akaike information criterion; CI = 95% confidence interval; SE = Standard error; −2LL = −2(Log Likelihood).

*p < .05.

More specific to our research questions, results also revealed that control beliefs moderated associations between stressors and NA. Specifically, daily control beliefs buffered daily stressor-reactivity (γ₃₀ = −0.086, p < .001). On days when daily control beliefs were higher, individuals tended to have lower daily stressor-reactivity. General control beliefs also buffered daily stressor-reactivity (γ₁₂ = −0.903, p = .04). Individuals with higher general control beliefs tended to have lower daily stressor-reactivity. There was no evidence that general control beliefs buffered the stressor exposure effect (γ₀₄ = −4.369, p = .12). In sum, and supporting the hypotheses, both general control beliefs and daily control beliefs buffer daily stressor-reactivity. However, it is important to note that individual differences in stressor exposure were related to extent of stressor-buffering (γ₃₁ = −0.194, p = .03). Displayed in Figure 1, at low levels (−1SD) of stressor exposure, daily control beliefs did not buffer daily stressor-reactivity; while, at high levels (+1SD) of stressor exposure, daily control beliefs did buffer daily stressor-reactivity. In other words, the within-person buffering provided by daily control beliefs is only evident among those with sufficiently high exposure.

Figure 1.

Stressor exposure moderates daily stressor-buffering of negative affect. At low levels (−1SD = −0.28) of stressor exposure, the negative affect stressor-reactivity slopes did not differ between low (solid line) and high (dashed line) daily control beliefs, indicating no daily buffering effect. At high levels (+1SD = +0.28) of stressor exposure, days with high daily control (the dashed line) have a shallower slope (lower negative affect stressor-reactivity) than days with low daily control beliefs (solid line), indicating a daily buffering effect.

Age Differences in the Extent Control Beliefs Buffer Stressor-Negative Affect Associations

We then examined if and how the above associations differed by age. Results indicated that age was not directly associated with baseline NA (γ₀₃ = 0.006, p = .87), daily stressor-reactivity (γ₁₃ = 0.009, p = .68), or the link between stressor exposure and NA (γ₀₅ =0.104, p = .40). Similarly, there were no age differences in the link between daily control beliefs and NA (γ₂₃ = −0.110, p = .18), or the link between general control beliefs and NA (γ₀₆ = −0.060, p = .12). However, these (non-)associations existed in the context of several three-way interactions. There were age differences in how general control beliefs moderated daily control-reactivity (γ₂₄ = −0.001, p = .001). At younger ages, lower general control beliefs were associated with stronger daily control-reactivity, while at older ages, general control beliefs did not moderate the link between daily control beliefs and NA. In sum, the associations between stressors and affective well-being do not appear to differ across age, but the associations among NA, daily control beliefs, and general control beliefs do differ across age.

Our main interest, however, was in whether there were age differences in how general and daily control beliefs buffered the effects of stressor exposure and daily stressors on NA. There was indeed evidence of age differences, in that age was associated with individuals’ daily stressor-buffering (γ₃₂ = −0.003, p = .03). Looking at the left panel of Figure 2a, we see that at younger age (−1SD = age 29 years), the extent of stressor-reactivity (slope of the lines) did not differ on days with low (dashed line) and high (solid line) levels of daily control (i.e., the lines are parallel). In contrast, when looking at the right panel, we see that at older age (+1SD = age 66 years), the lines are not parallel. Older adults have lower stressor-reactivity on days with high daily control compared to days with low daily control beliefs (dashed line has shallower slope than solid line). Age did not moderate the extent to which general control beliefs moderated stressor-reactivity (γ₁₄ = −0.002, p = .93). That is, the slopes of the four lines showing stressor-reactivity in Figure 2b are not significantly different from one another. There was also no evidence that general control beliefs buffered the effects of stressor exposure on daily NA (γ₀₇ = −0.086, p = .51). In sum, while buffering provided by general control beliefs appears to be age invariant, the buffering of stressor-reactivity provided by daily control appears to be larger at older ages.

Figure 2.

Age differences in extent that control beliefs buffer negative affect stressor-reactivity. Panel a (left): At younger age (−1SD = age 29 years), the slopes (negative affect stressor-reactivity) do not differ on days with low (dashed line) and high (solid line) levels of daily control. Panel a (right): At older age (+1SD = age 66 years), days with high daily control (the dashed line) have a shallower slope (lower stressor-reactivity) than days with low daily control beliefs (solid line). Panel b: The slopes of the four lines showing negative affect stressor-reactivity at low and high age and low and high general control beliefs are not significantly different from one another, indicating no age differences in the extent to which general control beliefs moderated negative affect stressor-reactivity.

The Role of Control Beliefs for Daily Stress Processes and Positive Affect

As seen in Model 2, the prototypical iSAHIB participant’s baseline PA on a nonstressor day is γ₀₀ = 59.724 (p < .001) on the 0 to 100 scale. PA was γ₁₀ = 2.216 (p < .001) points lower on stressor days and individuals with greater stressor exposure tended to have lower baseline PA, γ₀₁ = −11.399 (p = .03). Again, level of stressor exposure was not related to extent of stressor-reactivity, γ₁₁ = 0.022 (p = .99). Consistent with the NA results, individuals with higher general control beliefs tended to have higher baseline levels of PA (γ₀₂ = 4.278, p = .001) and there was evidence of daily control-reactivity, such that each unit increase in daily control was associated with γ₂₀ = 0.345 higher daily PA (p < .001). General control beliefs moderated daily control-reactivity (γ₂₂ = −0.022, p = .004), such that those with low general control beliefs had daily control-reactivity, while those with high general control beliefs did not. Replicating findings from NA, both greater stressor exposure and daily stressors compromise daily PA, and both higher general control beliefs and higher daily control beliefs facilitate higher daily PA.Contrary to the NA results, control beliefs mostly did not moderate associations between stressors and PA. Neither general nor daily control beliefs buffered daily stressor-reactivity (γ₃₀ = −0.044, p = .93; γ₁₂ = −0.008, p = .75). Additionally, there was no evidence that general control beliefs buffered the stressor exposure effect (γ₀₄ = 1.723, p = .73). However, Figure 3 demonstrates that for those with high levels of stressor exposure, days with low control beliefs were related to much lower PA (γ₂₁ = 0.315, p < .001), particularly for younger adults (γ₂₄ = −0.007, p < .001). In sum, neither general control beliefs nor daily control beliefs buffer daily PA stressor-reactivity, though stressor exposure moderates control-reactivity.

Figure 3.

Stressor exposure moderates daily positive affect control-reactivity. For individuals with high levels of stressor exposure (+1 SD = 0.28; dashed lines), days with low control beliefs have a steeper slope (greater positive affect control-reactivity) than for individuals with low levels of stressor exposure (–1SD = –0.28; solid lines). At younger ages, the slope differences is greater, such that younger adults experience even stronger positive affect control-reactivity in the presence of high levels of stressor exposure.

Age Differences in the Extent Control Beliefs Buffer Stressor-Positive Affect Associations

Similar to NA, age was not directly associated with baseline PA (γ₀₃ = 0.111, p = .11), daily stressor-reactivity (γ₁₃ = −0.001, p = .97), or the link between stressor exposure and PA (γ₀₅ = 0.184, p = .42). Older adults experience less control-reactivity (γ₂₃ = −0.004, p < .001) than younger adults, who experience particularly strong control-reactivity in the presence of high stressor exposure (γ₂₁ = 0.315, p < .001). There were no age differences in the link between general control beliefs and PA (γ₀₆ = 0.099, p = .16). As displayed in Figure 4a, age was associated with individuals’ daily stressor-buffering (γ₃₂ = 0.005, p < .001) such that younger adults’ (−1SD = age 29 years) stressor-reactivity, displayed on the left panel, did not differ on days with low and high levels of daily control. In contrast, older adults (+1SD = age 66 years), displayed on the right panel, have lower stressor-reactivity on days with high daily control compared to days with low daily control beliefs. Also displayed in Figure 4b, age did not moderate the extent to which general control beliefs moderated stressor-reactivity (γ₁₄ = 0.045, p = .13). There was again no evidence that general control beliefs buffered the effects of stressor exposure on daily NA (γ₀₇ = 0.088, p = .72). Replicating results from analysis of NA, the buffering provided by general control beliefs is age invariant, while the buffering of stressor-reactivity provided by daily control is larger at older ages.

Figure 4.

Age differences in extent that control beliefs buffer positive affect stressor-reactivity. Panel a (left): At younger age (–1SD = age 29 years), the slopes (positive affect stressor-reactivity) do not differ on days with low (dashed line) and high (solid line) levels of daily control. Panel a (right): At older age (+1SD = age 66 years), days with high daily control (the dashed line) have a shallower slope (lower stressor-reactivity) than days with low daily control beliefs (solid line). Panel b: The slopes of the four lines showing positive affect stressor-reactivity at low and high age and low and high general control beliefs are not significantly different from one another, indicating no age differences in the extent to which general control beliefs moderated positive affect stressor-reactivity.

Discussion

Using 9 weeks of daily reports obtained from 150 individuals aged 18–89 years, this study examined how both general and daily control beliefs buffer within-person affective reactivity to stressors and between-person effects of stressor exposure and how that buffering differed with age. As expected, we found evidence of both within- and between-person buffering of daily stress processes. Within-persons, for both positive and negative affect, higher daily control beliefs buffered stressor-reactivity. Between-persons, higher general control beliefs buffered negative affect stressor-reactivity. General buffering of stressor-reactivity was age invariant but the buffering of stressor-reactivity provided by daily control increased with age.

The Role of Control Beliefs for Daily Stress Processes

Results from the model of NA support the hypothesis that both general and daily control beliefs buffer the effects stressors have on affective well-being. Higher daily control beliefs were associated with ameliorated NA stressor-reactivity. NA stressor-reactivity was further ameliorated for individuals exposed to more stressors. Individuals with greater likelihood of experiencing a stressor (i.e., high stressor exposure) benefitted more from daily control beliefs’ stressor-buffering, supporting research and theory that those with more psychosocial resources may be better able to cope with higher stressor exposure (Hobfoll, 2001; Lazarus & Folkman, 1984).

Higher general control beliefs also buffered NA stressor-reactivity, confirming prior evidence that, at the trait level, control beliefs buffer the effects of stressors (Hay & Diehl, 2010). Surprisingly, general control beliefs were not related to stressor exposure effects. This may suggest that general control beliefs and stressor exposure influence affect through a similar underlying mechanism. Indeed, models of general control beliefs suggest they are a function of accumulated life experiences (Lachman, 2006); thus greater exposure and lower general control beliefs may form a feedback loop rather than a moderation effect. The cognitive appraisal of daily and general control beliefs related to stressor-reactivity, supporting a framework describing stress as a dynamic interaction between environment and evaluation of personal resources (Lazarus & Folkman, 1984).

Results from the model of PA similarly suggested higher daily control beliefs are associated with ameliorated stressor-reactivity, specifically for older adults. However, unlike NA, stressor exposure did not moderate daily control beliefs’ buffering of PA stressor-reactivity; rather, stressor exposure moderated daily control-reactivity. General control beliefs’ also did not moderate stressor-reactivity but rather moderated control-reactivity. Lack of general control beliefs or stressor exposure moderation of stressor-reactivity could again be due to the two constructs’ similar roles in perceived general circumstance. Although conflicting evidence for the role of stressors on PA abounds (e.g., Schilling & Diehl, 2014; Stawski et al., 2008; Uchino et al., 2006; Zautra et al., 2005), the present study provided evidence that daily stressors are linked to decreases in daily PA, and that this association may be buffered by daily control beliefs.

Age Invariance and Differences in Control Beliefs as a Stress Buffer

Considering the theoretical meaning of traits and states, and age-related changes in how each operates, control beliefs have a complex relation with affective well-being (Lachman, 2006; Folkman, 1984; Lachman, Rosnick, & Röcke, 2009). Our study provides additional support that higher control beliefs, both general (trait) and daily (state) are adaptive, regardless of age. Additionally, we find that associations between daily control beliefs and stressor-reactivity are particularly strong at older ages. Age-related decline in availability and effectiveness of psychosocial resources may lead to greater relevance of situational and contextual characteristics when reacting to stressors (Baltes & Baltes, 1990; Gerstorf, Ram, Goebel, Schupp, Lindenberger, & Wagner, 2010). The age invariance of general control beliefs’ buffering may reflect (but fail to show the nuance of) older adults’ shifting control strategies. Older adults’ coping techniques often rely on avoidance (secondary control) strategies as opposed to more active (primary control) strategies commonly used by younger adults (Charles, 2010; Heckhausen, Wrosch, & Schulz, 2010; Lachman, 2006). Thus, while the specific type of control strategies used in specific situations may change with age, older adults may maintain the same buffering effects of higher general control beliefs as younger adults.

Our study provides further evidence for the adaptive associations between “trait” and “state” control beliefs and daily affect, and cautions that control beliefs, particularly at a daily level, may be more important with age. Such associations may have strong implications for interventions that teach behaviors and work to improve control beliefs (e.g., Lachman et al., 1997). Although parsing the moderating effects of stressor exposure and general control beliefs needs further investigation, our findings suggest interventions focused on improving both daily and general control beliefs may be more successful than focusing on general control beliefs alone. Indeed, interventions that improve daily sense of control may be particularly useful for buffering stressor effects for older adults.

Limitations and Future Directions

Of course the findings here need to be considered with respect to several limitations embedded in the study design. The age-heterogeneous sample and 63-day design provided the persons and density of repeated measures needed for examination of age differences in within-person moderation. However, we are fully relying on daily retrospective self-reports about occurrence of stressors, and using multilevel regressions to infer individuals’ daily stressor-reactivity. More objective and more momentary measures of stressor-reactivity and stressor exposure would be additionally useful. Indeed, we believe that the daily measurement scheme does not provide the granularity needed to parse the directionality of effects, as it is likely the processes associating control beliefs, stressors, and affect occur at faster timescales. Physiological markers of stress (e.g., diurnal cortisol) could provide further insight into when, how, and on what time-scale control beliefs buffer stressor-reactivity. Further, the selectivity of the sample, particularly that the older participants are more active, healthier, and educated than the general population, temper the generalizability of our findings. General and daily control beliefs may function differently for more vulnerable older adults (e.g., control beliefs become especially important for individuals living in lower socioeconomic contexts).

Additionally, we recognize several analytical limitations. Conceptually, constructs that exacerbate or buffer stressor-reactivity form a complicated web of interactions. Focusing solely on control beliefs as buffers of stress is extremely simplifying and it is important to understand how many factors jointly associate with well-being. Although the importance of theoretically and empirically understanding how general and daily control beliefs function in relation to the daily stress process merited this simplification, future work would benefit from also examining how the buffering provided by general and daily control beliefs operate alongside risk and resilience factors, such as sociodemographics, mental and physical health, and psychosocial resources (Hay & Diehl, 2010) to influence other components of well-being (e.g., general or domain-specific life satisfaction). Analytically, the multilevel models do not provide for causal inferences about association between daily affect, the stress process, and control beliefs. Our main focus was to describe how within- and between-person differences in affect are associated with stressors and control beliefs. We note that further studies, and experimental/intervention designs, and more intensive experience sampling designs are needed to investigate process-oriented mechanisms more closely and to delineate pathways of age-related changes in daily affect (see Ram & Gerstorf, 2009). Relatedly, while we examined the role of both daily and general control beliefs on daily stress processes over nine weeks, we look forward to long-term multiple time-scale longitudinal studies with the data needed to link individuals’ day-to-day stressor-buffering and long-term developmental trajectories (for discussion, see Gerstorf, Hoppmann, & Ram, 2014).

Synopsis

This study distinguished the roles of daily and general control beliefs for daily stress processes. Regardless of age, there is evidence that daily and general control beliefs buffer effects of daily stressors on daily affective well-being. At the same time, our results highlight the importance of age differences, such that daily control beliefs were more strongly associated with stressor-reactivity for older adults than younger adults. We emphasize the theoretical need to include daily and general control beliefs, particularly for understanding older adults’ affective reactivity to stressors, and we look forward to future studies considering such measures across multiple-time scales.

Funding

We gratefully acknowledge the support provided by the National Institutes of Health (R01-HD076994, R24-HD041025, UL1-TR000127), Big Data Social Science IGERT (National Science Foundation: Award Abstract 1144860), the Pathways T32 (National Institutes of Health; T32 AG049676), the German Research Foundation (DFG, GE 1896/6-1, GE 1896/7-1), and the Penn State Social Science Research Institute. The Intraindividual Study of Affect, Health, and Interpersonal Behavior was supported by RC1-AG035645. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.

Conflict of Interests

The authors have no conflicts of interest to disclose.