Intra-individual variability and psychological flexibility: Affect and health in a National US sample
Introduction
Short-term changes that are more or less reversible and that may differ across individuals comprise within-person variability or intra-individual variability (IIV) (Nesselroade, 2001). However, IIV is only one way of characterizing within-person variability. There is a growing interest in psychological flexibility (PF), a construct that appears to hold promise for understanding resilience and vulnerability to psychological distress (Kashdan & Rottenberg, 2010). Operationally defining PF can be difficult, as can distinguishing between PF and IIV (Kashdan & Rottenberg, 2010). Empirical studies appear to disagree about the adaptiveness of within-person variability: IIV has been associated with poorer psychological and physical outcomes, whereas PF may be related to better psychological and physical health outcomes (Kashdan and Rottenberg, 2010, Röcke and Brose, 2013).
There is some evidence that both IIV and PF are stable over time and may represent persistent individual differences that could affect long-term health outcomes. IIV (total variability) in affect had test-retest correlations of 0.46–0.90 over periods of up to 2 months (Eid and Diener, 1999, Penner et al., 1994). One study of flexibility in affect regulation reported a test-retest correlation of 0.45 over 3 years (Westphal, Seivert, & Bonanno, 2010). However, further assessment of the reliability and validity of IIV and PF would advance our understanding of how these constructs relate to each other and to psychological and physical health and whether they constitute stable individual differences.
IIV has been operationalized in a number of ways, most commonly as the individual’s standard deviation (iSD) (see Ram and Gerstorf, 2009, Röcke and Brose, 2013 for excellent reviews of IIV methodology). IIV operationalized in this way is distinct from the mean level of the response; it specifically refers to variability around the means. It is important to note that assessment of IIV does not require characterization of the situation or pairing of response and situation. Thus, IIV represents the range or frequency of a response, uncharacterized by situational change. IIV in affect is therefore the range of emotional experience over time, typically operationalized as the iSD of the individual’s affect (Eid & Diener, 1999).
PF is more complex than IIV in both its definition and operationalization. One of the more common ways of defining PF is the ability to vary one’s responses in a contextually dependent manner in order to appropriately meet situational demands (Bonanno et al., 2004, Cheng, 2001, Fujimura and Okanoya, 2012, Tracey, 2005, Westphal et al., 2010). PF differs from IIV in that PF refers to within-person variability where the response is dependent on the situation and is patterned and predictable. PF in affect therefore refers to the ability to match one’s emotional experience appropriately to situational cues (Fujimura and Okanoya, 2012, Waugh et al., 2011).
Operationalizing PF in affect therefore requires a theory that defines what the “appropriate” emotional experience is and calculation of the difference between an individual’s affect in different situations and the “appropriate” affect as defined by the model. Mathematically, when the response being measured can be plotted on Cartesian coordinates, the Euclidean distance between “appropriate” and actual responses can be calculated; the smaller the mean distance across assessments, the better the fit (e.g., Cheng, 2001, Tracey, 2005). PF with regard to affect regulation has been defined as the ability to modulate affective expression in accordance with situational demands or to enhance or suppress affective expression when prompted. In one study, PF in affect was operationalized as the difference in emotional expression during positive and negative stimuli, where those with more differentiation between positive and negative expressions were considered to have greater flexibility (Waugh et al., 2011). In another, PF was calculated as the difference between emotional expression in the control condition and emotional expression when instructed to enhance or suppress expression, summed to get an overall index of flexibility (Bonanno et al., 2004).
In sum, operationalizing PF requires a theory from which one can derive an index of fit by comparing observed responses with theoretically adaptive responses across different situations. Therefore, PF concerns the covariation of response and situation and, importantly, compares this covariation with a theoretical standard.
There are a number of theories that provide a possible standard for positive affect (PA) and negative affect (NA) in situational context. Most of them imply relationships between NA and PA in the situation. First, NA and PA may be inversely correlated, such that increases in NA imply decreases in PA and vice versa (Feldman Barrett and Russell, 1998, Russell and Carroll, 1999). Second, NA and PA may be orthogonal, such that changes in one do not imply changes in the other (Cacioppo and Berntson, 1994, Watson and Clark, 1997; Watson & Tellegen, 1985). Third, the relationship between NA and PA may itself vary, switching between a bipolar relationship and an orthogonal relationship, depending on the situation (Davis, Zautra, & Smith, 2004).
The Emotional Congruency Model (Congruent) theory predicts that it is most adaptive to experience emotions congruent with the situation. That is, “appropriate” responses comprise increased PA and decreased NA in positive situations; decreased PA and increased NA in negative situations; and the ability to switch between these profiles. In one study (Waugh et al., 2011) affect was rated using a dial, with negative at 0 degrees and positive at 180 degrees. This measurement strategy implies an inverse or bipolar relationship between NA and PA. Situation was defined as exposure to a positive or negative image. People whose affect rating more consistently matched the situation (affect closer to the positive pole when viewing a positive image, and affect closer to the negative pole when viewing a negative image) had higher trait resilience scores. The authors defined affective flexibility in this study as the ability to “switch responses when the emotional valence of the event change, and maintain responses with the emotional valence when the events do not change.” In another study (Fujimura & Okanoya, 2012) affect was measured using a grid with valence and arousal dimensions. Participants indicated their current affect by checking one area on the grid. This measurement strategy also implies inverse or bipolar NA and PA, with arousal being a separate dimension. Situation was defined as exposure to positive, negative, or neutral images, with the arousal properties of the images consistent in all of the tasks. People whose affect ratings more closely matched the image valence also had higher HRV, which has been linked to successful self-regulation (Segerstrom, Hardy, Evans, & Winters, 2011). Higher affective flexibility, as match between affect and situation, was therefore suggestive of better self-regulatory ability. These studies reflect the Emotional Congruency Model, where NA and PA are inversely correlated, and affect and the valence of the situation match.
The Maintenance of Emotional Complexity Model (Complex) predicts that the ability to experience positive affect during a stressor may buffer against development of depressive symptoms (Fredrickson, Tugade, Waugh, & Larkin, 2003), as well as shortening the cardiovascular recovery time following negative events (Tugade & Fredrickson, 2004). In the first study (Fredrickson et al., 2003), PA and NA were rated using a modified Differential Emotions Scale (DES), and separate subscales were created for PA and NA. This operationalization of PA and NA as discrete scales reflects a model of PA and NA as separate dimensions (Watson & Tellegen, 1985). Those with higher PA during a stressful event also had higher ratings of resilience and lower incidence of depressive symptoms. In the second study (Tugade & Fredrickson, 2004), a modified version of the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) was used to construct separate PA and NA subscales. Again, those with higher PA during a stressful task also had higher resilience and a faster physiological recovery from a stressful task. In these studies, emotional flexibility was the ability to maintain positive affect in the face of negative events. Therefore, in contrast to Congruent, the Complex theory implies that although NA may increase in negative situations, PA should be maintained. That is, levels of PA should be resilient to the influence of negative events and should not be inversely correlated with NA, which may be reactive to negative events. In this model, NA should depend on the situation, whereas the level of PA should remain stable across situations, resulting in dissociation between NA and PA.
The Dynamic Model of Affect (Dynamic) theory specifies that the relationship between PA and NA depends on the presence of negative events (Zautra, Smith, Affleck, & Tennen, 2001). Under non-stressful conditions, it is thought that people receive the most benefit from independence between PA and NA. Independence yields the maximum amount of information about situations, because emotional responses on one affective dimension are not limited by experience or lack of experience on the other dimension (i.e., there is higher emotional complexity). However, the added information of greater emotional complexity results in higher cognitive demand. The Dynamic theory states that emotional complexity is therefore adaptive in low-stress situations but maladaptive in high-stress situations, in which cognitive resources are scarce. In stressful situations, adopting a simpler representation of one’s affective experience reduces cognitive load, freeing up resources for managing the situation. In addition, as affect becomes more unidimensional, more PA during a stressful situation would be related to less NA. The Dynamic theory suggests that positive experiences also have the potential to decrease NA under stress. In this study, NA and PA were measured using the PANAS, and separate subscales for PA and NA were calculated. Hierarchical Linear Modeling was used to examine the relationship between NA, PA, and increased stress due to physical pain. The relationship between NA and PA changed under stress: NA and PA became more inversely correlated as pain increased. In this model, therefore, the inverse relationship between PA and NA is stronger during stressful situations and weaker during non-stressful situations, and the ability to switch between modes (NA and PA are bipolar during high stress; NA and PA are orthogonal during low stress) adaptively uses cognitive resources and provides an additional means of decreasing NA under stress, to wit, PA.
Carefully distinguishing among IIV and the Congruent, Complex, and Dynamic theories of PF in affect allows one to determine which theory is the best predictor of psychological and physical health. If IIV and PF are distinct, then they ought to predict different outcomes. Higher IIV in affect has been related to higher distress, depression, and neuroticism and lower agreeableness and extraversion (Eid and Diener, 1999, Gruber et al., 2013, Kuppens et al., 2007, Timmermans et al., 2010). This relationship appeared whether the response was measured multiple times within an experiment, multiple times a day, or daily over a period of days or years. However, when affective IIV is confined to a specific, limited time period, such as immediately after a stressful event, greater affective IIV may signal adaptive coping with the stressor (Röcke & Brose, 2013). Therefore, situational context may be crucial to understanding the effects of IIV on outcomes.
There are fewer studies examining health effects of affective PF. However, those studies indicate that higher PF is related to better psychological and physical health: higher resilience, higher heart rate variability (HRV), less systemic inflammation, and less distress (Bonanno et al., 2004, Fujimura and Okanoya, 2012, Sin et al., 2015, Waugh et al., 2011, Westphal et al., 2010). Therefore, the extant literature suggests that IIV is associated with worse health, whereas PF is associated with better health. However, no study has compared them directly, using the same outcomes in the same sample.
The current study addressed this need by assessing IIV and PF (as defined by the Congruent, Complex, and Dynamic theories) as well as health, longitudinally, in a large sample of US adults. The study had the following 2 aims:
Aim 1: Provide evidence for reliability and stability of IIV and the three operationalizations of PF in affect, as well as the relationships within and between constructs.
Aim 2: Test the hypothesis that less IIV and more PF in affect predicts better psychological and physical health.
Section snippets
Participants
Data for this study came from the National Study of Daily Experiences (NSDE) Waves 1 and 2, an 8-day daily diary portion of the Midlife Development in the United States (MIDUS I) and MIDUS II surveys. Participants (N = 793) completed both waves of MIDUS and the NDSE. NSDE Wave 2 was completed approximately 10 years after Wave 1.
The final study sample had mean age 46.7 years (SD = 12.5; range 24–74); was 55.9% women; and was 92.6% Caucasian, 3.5% African American, and 3.3% other. Education level was
Between- and within-wave stability of IIV and PF
Means, standard deviations, ranges, stability, and test-retest correlations for each IIV and PF variable can be found in Table 1. Bootstrapping was used to obtain estimates of within-wave stability. A randomly selected subset of 4 days of data was used to calculate IIV and PF variables and was correlated with scores calculated on the other 4 days of data. Random sampling with replacement was used to obtain 1000 resamples of the data and provide within-wave stability estimates and confidence
Discussion
As evidenced by the present issue of Journal of Research in Personality, within-person variability is an area of growing attention both in theoretical models and in empirical investigations. The present investigation focused on four constructs of within-person variability in affect, including IIV, in which all variability is of importance; and three versions of PF, in which only theoretically adaptive variability is of importance: Congruent PF (ideally, NA and PA should change in concert,
Conclusion
Within-person variability has multiple manifestations with different implications for psychological and physical health. In general, more total variability in affect (IIV) appears to capture emotional lability and is associated with poorer psychological and physiological outcomes. However, PF – within-person variability comprising a patterned, predictable response – represents increased emotional control and is related to better psychological and physical outcomes. Therefore, whether affective
Acknowledgements
This research was supported by the National Institute on Aging – United States Grants K02-AG033629, P01-AG020166, P01-AG0210166-02, and R01-AG19239 and the John D. and Catherine T. MacArthur Foundation Research Network. The authors thank Nilam Ram, Charles Carlson, Thomas Widiger, and Deborah Danner for their helpful suggestions on a previous version of this work.
References (53)
- et al.
Psychological flexibility as a fundamental aspect of health
Clinical Psychology Review
(2010) - et al.
Self-regulatory deficits in fibromyalgia and temporomandibular disorders
Pain
(2010) Interpersonal rigidity and complementarity
Journal of Research in Personality
(2005)- et al.
The daily inventory of stressful events an interview-based approach for measuring daily stressors
Assessment
(2002) - et al.
Regulatory flexibility: An individual differences perspective on coping and emotion regulation
Perspectives on Psychological Science
(2013) - et al.
The importance of being flexible: The ability to both enhance and suppress emotional expression predicts long-term adjustment
Psychological Science
(2004) - et al.
Relationship between attitudes and evaluative space: A critical review, with emphasis on the separability of positive and negative substrates
Psychological Bulletin
(1994) - et al.
Control processes and self-organization as complementary principles underlying behavior
Personality and Social Psychology Review
(2002) Assessing coping flexibility in real-life and laboratory settings: A multimethod approach
Journal of Personality and Social Psychology
(2001)- et al.
Chronic pain, stress, and the dynamics of affective differentiation
Journal of Personality
(2004)