Ecological momentary assessment of the unique effects of trait worry on daily negative emotionality: does arousal matter?

ABSTRACT

Worry proneness is a transdiagnostic trait that predicts increased negative affect (NA), potentially in the service of preventing negative emotional contrasts. Although discrete types of NA vary along the dimension of arousal, the extent to which trait worry predicts high vs. low arousal forms of NA in daily life is unclear. This distinction has important implications for conceptualising how worry may perturb adaptive emotionality in various disorders. The present study (not pre-registered) aimed to isolate the effects of trait worry on high (N = 88) and low (N = 122) arousal NA in daily life using ecological momentary assessment while controlling for potential physical and psychological confounds. Participants were assessed for trait worry and depressive symptoms at baseline then reported their affect, heart rate, and exercise three times per day for one week. Multilevel models revealed that trait worry predicted both increased high and low arousal NA after controlling for momentary heart rate, daily exercise, and depression. In contrast, baseline depressive symptoms only predicted low arousal NA in daily life. Findings support the contrast avoidance model of worry and suggest that worry is linked to increased state NA in daily life, independent of arousal.

KEYWORDS:

• Worry
• contrast avoidance
• ecological momentary assessment
• negative affect
• physiology

Worry is a form of repetitive negative thinking about anticipated future threats (Davey & Wells, 2006). While worry is a normal cognitive process for coping with uncertain stressors, worry can become maladaptive when it is employed excessively, is experienced as uncontrollable, and/or involves catastrophizing or other forms of problem elaboration (Davey & Wells, 2006). Indeed, worry has been shown to engender distress, behavioural avoidance, and ineffective problem solving (Llera & Newman, 2020; Ruscio et al., 2011), and elevated trait worry (i.e. one’s overall proneness to engage in worry) has been robustly linked to diverse forms of psychopathology including anxiety, post-traumatic stress, and depression (Brosschot et al., 2006; Olatunji et al., 2010). Given that excessive worry confers risk for psychopathology broadly, further investigation into the effects of trait worry on intermediary mechanisms may yield important insights with transdiagnostic implications.

Several theories have been proposed to explain how worry may function to increase risk for psychopathology. The contrast avoidance model posits that individuals use worry to sustain a negative emotional state to avoid aversive, unexpected shifts towards negative emotions (Newman & Llera, 2011). This model implies that worry plays a causal role in maintaining chronic negative affect (NA), which is, in turn, a core risk factor for and feature of diverse forms of psychopathology (Stanton & Watson, 2014). A large literature supports the contrast avoidance model of worry, showing that worry predicts increased negative emotionality (Brosschot et al., 2006), more restricted reactivity to acute stressors (i.e. blunted emotional contrasts; Baez et al., 2023; Newman & Llera, 2011), and a decrease in negative emotional contrasts following negative events (Baik & Newman, 2023). However, negative emotional experiences vary across important dimensions and the effects of trait worry on different types of negative emotions in daily life remain unclear.

Dimensional accounts of emotion suggest that the spectra of valence (i.e. positive/pleasurable to negative/displeasurable) and arousal (i.e. degree of alertness, mobilisation, or energy) well characterise discrete emotions (Feldman Barrett & Russell, 1998). For instance, fear is a negatively valenced, high arousal emotion, whereas sadness is a negatively valenced, low arousal emotion. Importantly, different emotions are thought to have evolved to serve specific functions and thus differentially influence behaviour, with high arousal forms of NA generally precipitating fight, flight, or freeze responses and low arousal forms of NA typically motivating behavioural withdrawal or passivity (Lench et al., 2011). Further, high vs. low arousal forms of NA are differentially associated with distinct spectra of psychopathology (e.g. fear loads onto anxiety disorders vs. sadness loads onto depressive disorders; Stanton & Watson, 2014). Although worry has been shown to engender NA broadly (Newman & Llera, 2011), it is unclear if trait worry is differentially associated with high vs. low arousal negative emotionality. Given the distinct functions and consequences of high vs. low arousal NA, estimating the magnitude of worry’s effect on these different types of NA has important implications for conceptualising the role of worry in conferring risk for psychopathology via disturbance in emotionality. For instance, if trait worry is found to predict both high and low arousal NA in daily life, then that may imply a shared mechanistic role of worry in disorders characterised by high arousal NA (e.g. anxiety disorders) and low arousal NA (e.g. depressive disorders). Alternatively, if trait worry is only linked to one emotional profile, then that would suggest that worry’s NA-engendering function may be specific to certain types of emotions and thus worry may operate and be targeted differently in disorders characterised by high vs. low arousal emotional profiles.

Isolating the unique effects of trait worry on high and low arousal NA will require ruling out some potential “confounds”. While there are several potential confounders that may account for the link between habitual worry and daily emotionality, two important constructs to consider and rule out are the physiological correlates of emotionality and baseline negative mood. First, emotions are thought to be comprised of both physiological and subjective components (Lench et al., 2011). Indeed, different types of emotions have been associated with different patterns of autonomic nervous system (ANS) responding that function to prepare the body for distinct adaptive behaviours. For instance, high arousal negative emotions (e.g. fear, anger) are linked to physiological activation including increased heart rate (HR), skin conductance level (SCL), and respiration that primes the body to respond to threats/aggressors, whereas low arousal negative emotions (e.g. sadness) are linked to ANS responses such as decreased HR and SCL that prime the body to withdraw (Kreibig, 2010). It is possible that the relation between trait worry and high arousal NA may be attributable to worry’s effects on physiological arousal, as worry has been shown to precipitate activating ANS responses including increased HR (Baez et al., 2023; Brosschot et al., 2006; Newman & Llera, 2011). However, we predict that trait worry may also be linked to increased NA at the cognitive level, above and beyond physiological activation, by exacerbating cognitive representations of stressors (Davey & Wells, 2006). Furthermore, it is important to consider one’s baseline mood when examining the relation between trait worry and daily emotionality. Depressive symptoms are made up of a broad scope of factors that underlie low mood (e.g. anhedonia, hopelessness, negative self-concept). Perhaps unsurprisingly, trait worry is significantly associated with symptoms of depression (Olatunji et al., 2010). Accordingly, the relation between worry and low arousal NA may then be explained by worry’s link to depressive symptoms, as depression is uniquely associated with reduced positive emotion, low mood, and under-arousal (Stanton & Watson, 2014). Alternatively, we predict that worry may predict increased NA over and above baseline depression, which would suggest that worry is an independent trait determinant of affective experience.

Much of the literature on the affective consequences of worry proneness has taken place in controlled laboratory settings, potentially limiting the generalizability of findings to real-world emotional experiences. Ecological momentary assessment (EMA) involves the repeated, real-time assessment of phenomena in individuals’ natural environments, thus addressing problems of retrospective bias and ecological validity that are common in laboratory research (Shiffman et al., 2008). Prior EMA research probing the effects of worry on daily emotionality have found that worry predicts increased concurrent and sustained NA, as well as blunted emotional contrasts following negative events (Baik & Newman, 2023; Newman et al., 2019). These findings suggest that worry is associated with increased negative emotionality in daily life that may function to prevent negative emotional contrasts (Newman & Llera, 2011). However, this experience sampling literature is in its nascence, and the specificity of worry’s effects on different types of NA remains unclear.

The present study aimed to examine trait worry as a unique predictor of high and low arousal NA in daily life. This study was not pre-registered, and the present study aims and data analytic plan were developed following the conclusion of data collection. Using an EMA design, we assessed high arousal NA, low arousal NA, and HR (as a biomarker of physiological arousal) three times per day for one week in a student sample. We used multilevel modeling (MLM) to probe the effects of trait worry on high and low arousal NA over time. Because of worry’s role as a transdiagnostic risk factor (Brosschot et al., 2006; Olatunji et al., 2010), we hypothesised that trait worry would predict higher levels of both high and low arousal NA throughout the week, even after accounting for concurrent HR and baseline differences in symptoms of depression.

Material and methods

Participants

The sample (full N = 122; n = 88 with complete stress data) included students at a private university in the southeastern United States. Inclusion criteria were being age 18 or older and owning and wearing a smart watch that tracks heart rate. There were no exclusion criteria. Participants’ mean age was 19.12 (SD = 1.38), ranging from 18 to 28 years. The sample predominantly identified as female (n = 93; 76.2%), and the racial/ethnic composition was as follows¹: White (n = 67; 54.9%), Asian/Pacific Islander (n = 35; 28.7%) African American/Black (n = 20; 16.4%), Hispanic/Latino (n = 10, 8.2%), American Indian/Alaska Native (n = 1; 0.8%), and other (n = 1; 0.8%).

Measures

Mini International Neuropsychiatric Interview (MINI; Sheehan et al., 1998). The MINI is a brief structured diagnostic interview that assesses 17 DSM-5 disorders. Bachelor- and master-level research assistants who were trained and supervised by a licensed clinical psychologist administered the MINI to assess for diagnostic criteria for major psychiatric disorders.

Penn State Worry Questionnaire (PSWQ; Meyer et al., 1990). The PSWQ is a 16-item self-report measure of trait worry. Items (e.g. “Many situations make me worry.”) are rated on a Likert scale from 1 (not at all typical of me) to 5 (very typical of me), yielding total scores from 16 to 80 with higher scores indicating higher tendencies to worry. Internal consistency of the PSWQ in the present sample was excellent (Cronbach’s alpha = .95).

Depression, Anxiety, and Stress Scales 21 – Depression Subscale (DASS-D; Lovibond & Lovibond, 1995). The DASS-D is a seven-item self-report measure of past-week symptoms of depression (e.g. “I felt that I had nothing to look forward to”.). Items are rated on a Likert scale from 0 (Did not apply to me at all) to 4 (Applied to me very much, or most of the time) and multiplied by two, yielding total scores ranging from 0 to 56. The internal consistency of the DASS-D in the present sample was excellent (Cronbach’s alpha = .91).

Ecological Momentary Assessment Items. At each EMA timepoint (i.e. three times per day), participants provided (1) five momentary affect ratings in response to the prompt “Please read each item then indicate how you feel each emotion right now” for anxiety, depression, sadness, anger, and stress² on a visual analogue scale from 1 (very slightly or not at all) to 5 (extremely), (2) their current heart rate as measured by their smart watch, and (3) an indication of whether they had exercised that day (i.e. “Did you exercise today?”). To measure high arousal NA at each time point, we computed a sum of anxiety, anger, and stress ratings. To measure low arousal NA at each time point, we computed a sum of depression and sadness ratings.

Procedure

The university’s Institutional Review Board approved all study procedures (IRB approval # 210287). All study procedures were performed in compliance with relevant laws and institutional guidelines. Data were collected between October 2021 and April 2023. Participants were recruited from psychology classes and were compensated with course credit. Data collection took place over the course of eight consecutive days. On day one, participants completed a one-hour study visit via Zoom videoconferencing that included informed consent, administration of the MINI, and baseline self-report measures (i.e. demographics, PSWQ, DASS-D). During days two through eight, participants received the EMA survey three times per day via email at 8:00 am, 2:00 pm, and 8:00 pm (totaling 21 total EMA timepoints). Participants were instructed to complete EMA surveys as soon as possible after receipt. A reminder email was sent to participants if they had not completed the survey after one hour, and EMA surveys only remained open for two hours. Participants were also instructed to wear their smart watches continuously throughout the week for heart rate monitoring, and to report their current heart rate as measured by their smart watch in each EMA survey.

Data analytic overview

Data may be shared upon request. Our hypotheses were tested using three-level MLMs in the SAS Mixed procedure (script is available at osf.io/zj3e8/). This approach accounted for the nested structure of the data, whereby surveys constituted level one, days constituted level two, and participants constituted level three. We ran two separate models to examine trait worry as a predictor of daily high arousal NA and low arousal NA, respectively. To isolate the unique effects of trait worry on subjective negative emotionality, covariates included momentary heart rate, daily exercise (to control for the effects of metabolic activity on heart rate; Brouwer et al., 2018), past-week symptoms of depression, and time of day (given prior evidence for diurnal shifts in emotions; English & Carstensen, 2014). In each model, the level one predictors were time of day (linear variable, coded such that morning = 0, afternoon = 1, and evening = 2), momentary heart rate, and daily exercise (dummy coded, such that no exercise = 0, exercise = 1). Level three predictors included trait worry, past-week depression, and mean heart rate and exercise. We did not include any level two predictors, but day-level clustering was employed to allow for the potential clustering of assessments within each day. To parse within- and between-person effects, level 1 predictors (excluding time of day) were person-mean centered, level 3 predictors were grand-mean centered, and the person means of heart rate and exercise were entered as level 3 covariates in the model. Supplementary models tested the unadjusted effects of trait worry on high and low arousal NA (see Supplementary Table 1). All models included random intercepts to account for differences in the mean level of the dependent variable across participants, and random effects were limited to the intercepts to preserve power and simplicity in models. We used unstructured covariance for the random and repeated effects. Restricted maximum likelihood estimation was used to handle missing data.

Results

Descriptive statistics and correlations among study variables

Participants completed between 9 and 21 EMA timepoints (M = 18.93, SD = 2.14), with only 9.8% of observations considered missing. Missingness was not significantly associated with any study variable. 25% (n = 31) of the sample met diagnostic criteria for a current DSM-5 disorder. Specifically, 18% (n = 22) of the sample met clinical criteria for a current anxiety-related disorder (including generalised anxiety disorder, panic disorder, agoraphobia, social anxiety disorder, post-traumatic stress disorder, and obsessive-compulsive disorder) and 4.1% (n = 5) met clinical criteria for a current mood disorder (including major depressive disorder and bipolar disorder) as determined by the MINI. The most common presenting diagnoses were panic disorder (8%), generalised anxiety disorder (7%), and social anxiety disorder (5%). Most participants used an Apple Watch (n = 101; 82.8%) for heart rate monitoring throughout the study period.³

Descriptive statistics and correlations between study variables are presented in Table 1. PSWQ scores were normally distributed and ranged from 19 to 79, with a mean score of 49.17 (SD = 14.97). Thus the sample ranged from low to clinically significant (i.e. PSWQ > 62; Behar et al., 2003) levels of trait worry. There were small, significant associations between trait worry and average high arousal NA (r = .38, p < .001), as well as between trait worry and average low arousal NA (r = .33, p < .001). Symptoms of depression also demonstrated small, significant correlations with high arousal NA (r = .26, p < .05) and low arousal NA (r = .36, p < .001). Trait worry and symptoms of depression were significantly positively correlated, r = .36, p < .001.

Table 1. Descriptive statistics and correlations among study variables.

Variable	1.	2.	3.	4.	5.
1. PSWQ	–
2. DASS-D	.36**	–
3. Average High Arousal NA	.38**	.26*	–
4. Average Low Arousal NA	.33**	.36**	.61**	–
5. Average HR	−.02	.05	.03	−.02	–
M	49.17	5.38	5.22	2.59	84.06
SD	14.97	7.90	1.47	0.74	8.68
Range	19–79	0–40	3.16–9.52	2.00–6.14	64.00–119.47

Note. PSWQ = Penn State Worry Questionnaire; DASS-D = Depression, Anxiety, and Stress Scales 21 – Depression Subscale; Average High Arousal NA = average of High Arousal Negative Affect (anxiety, anger, stress) ratings at each timepoint; Average Low Arousal NA = average of Low Arousal Negative Affect (sadness, depression) ratings at each timepoint; Average HR = average of heart rates reported at each timepoint. *p < .05, **p < .001

Effects of trait worry on high arousal negative affect

We first examined high arousal NA as a function of trait worry, while controlling for momentary HR, daily exercise, time of day, and depression (see Table 2). The main effect of trait worry was significant and positive, with higher worry independently predicting higher levels of subjective high arousal NA throughout the day after controlling for covariates (b = .02, p < .01). There was also a significant positive effect of the time-varying covariate HR, suggesting that increases in momentary HR were associated with concurrent increases in high arousal NA at the same time point (b = .01, p < .01). Exercise, depression, and time of day all had non-significant relations with high arousal NA (ps > .25).

Table 2. Multilevel model results predicting momentary (1) High Arousal NA and (2) Low Arousal NA as a Function of Trait Worry.

	Model 1 (High Arousal NA)					Model 2 (Low Arousal NA)
Predictor	Coeff.	SE	df	t	p	Coeff.	SE	df	t	p
Intercept	7.708	0.964	84.6	7.99	<.001	2.427	0.632	119	3.84	<.001
Time of Day	−0.062	0.054	1280	−1.14	0.253	0.097	0.026	1714	3.73	<.001
PSWQ	0.024	0.007	82.4	3.40	0.001	0.012	0.005	117	2.69	0.008
DASS-D	0.005	0.014	82.3	0.37	0.709	0.026	0.009	117	3.00	0.003
HR	0.010	0.003	1552	3.02	0.003	0.002	0.002	2107	1.32	0.187
Exercise	0.035	0.124	1435	0.28	0.776	−0.128	0.063	2161	−2.02	0.044
HR Mean	−0.032	0.011	82.7	−2.89	0.005	−0.001	0.007	117	−0.16	0.876
Ex Mean	−0.265	0.522	82.1	−0.51	0.614	0.292	0.307	117	0.95	0.344

Note. NA = Negative Affect; PSWQ = Penn State Worry Questionnaire; DASS-D = Depression, Anxiety, and Stress Scales 21 – Depression Subscale; HR = momentary heart rate, person-mean-centered; Exercise = whether or not participant exercised that day (0 = no, 1 = yes), person-mean-centered; HR Mean = person mean of momentary heart rate; Ex Mean = person mean of daily exercise (0 = no, 1 = yes).

Effects of trait worry on low arousal negative affect

We next examined low arousal NA as a function of trait worry, while controlling for momentary HR, daily exercise, time of day, and depression (see Table 2). Worry was a significant, positive predictor of subjective low arousal NA above and beyond the effects of covariates (b = .01, p < .01). Further, symptoms of depression significantly predicted higher levels of low arousal NA (b = .03, p < .01), and daily exercise significantly predicted decreased levels of low arousal NA (b = -.13, p < .05). Time of day had a positive relation with momentary low arousal NA, such that levels of low arousal NA increased from morning to evening (b = .10, p < .001). Heart rate had a non-significant relation with low arousal NA (p > .18).

Discussion

The present study examined the unique effects of trait worry on high and low arousal forms of NA over the course of one week. To isolate worry’s independent effects on subjective emotional experiences, models controlled for momentary HR, daily exercise, time of day, and symptoms of depression. Importantly, this study employed an EMA approach to maximise ecological validity and leveraged a sample spanning the full spectrum of worry proneness, from low, nonclinical levels of worry to those meeting clinical criteria for disorders characterised by pathological worry (i.e. generalized anxiety disorder). The findings showed that trait worry predicted increased levels of both high and low arousal NA throughout the week-long EMA period in both the full and unadjusted models. This finding is noteworthy given the diverse functions of and conceptual distinctions between high and low arousal forms of NA (Feldman Barrett & Russell, 1998), and suggests that worry may be a nonspecific predictor of negative emotionality broadly.

The demonstrated effect of trait worry on high and low arousal forms of NA may have important implications for conceptualising the role of worry in conferring risk for psychopathology via disturbance in emotionality. High trait worry may predispose individuals towards a more negative affective “home base” around which specific emotional states fluctuate, independent of arousal (Kuppens et al., 2007). These results are consistent with the contrast avoidance model of worry (Newman & Llera, 2011) and suggest that worry may function to maintain chronically elevated NA of diverse types. Considered alongside additional EMA research showing that worry predicts reduced negative emotional contrasts following negative events in daily life (Baik & Newman, 2023), together these findings suggest that the increased NA associated with trait worry may function to prevent various types of negative emotional contrasts. Given the present findings, future research will be needed to determine the context(s) that may determine the extent to which worry results in daily fluctuation in high vs. low arousal NA. For example, the link between worry and high arousal NA may be mediated by maladaptive threat appraisals whereas the link between worry and low arousal NA may be explained by maladaptive appraisals about loss.

The present study did find that momentary HR significantly predicted increased concurrent high arousal, but not low arousal, NA. This offers further validation of the distinction made between high and low arousal NA in the present study. Of note, by controlling for daily exercise, we aimed to approximate the effects of non-metabolic HR, which is thought to be a more sensitive correlate of emotionality (Brouwer et al., 2018). This finding extends laboratory research on the physiological correlates of emotion into daily life and replicates the finding that high arousal forms of NA (e.g. anxiety, anger, fear) are typically characterised by activating ANS responses including increases in HR (Kreibig, 2010). Importantly, the present study found that trait worry predicted increased NA even after accounting for physiological arousal (i.e. HR), suggesting that worry may play a role in shaping the cognitive level of emotional experience. This result is noteworthy given the large body of evidence demonstrating the effects of worry on physiological activation (see Brosschot et al., 2006 for a review). Indeed, emotional responses are thought to be the product of the interaction between bodily sensations and higher-order cognitive processes, including appraisals, thoughts, and goals (Lench et al., 2011). Dispositions to worry may act on this cognitive level of emotional experience by predisposing individuals towards prolonged or amplified cognitive representations of stressors (Brosschot et al., 2006). Consistent with this notion, elevated worry has been linked to more catastrophic interpretations of potential threats, including exaggerated notions of their likelihood and severity (Provencher et al., 2000).

Although the present findings show that trait worry and depressive symptoms were significantly correlated, the pattern of effects for worry contrasts with those of depressive symptoms, which were found to predict increased low arousal, but not high arousal, NA. Symptoms of depression may therefore demonstrate more specificity than trait worry in terms of their effects on negative emotionality, increasing the likelihood that individuals will experience negative emotions in their daily life that are characterised by passivity, withdrawal, lethargy and low ANS arousal (Kreibig, 2010). Further, we found that exercise had an inverse relationship with momentary low arousal, but not high arousal, NA. This extends prior research showing that physical activity has beneficial effects on mood states and suggests that exercise may specifically function to buffer against low arousal forms of NA, perhaps by inducing adaptive arousal states (Audiffren, 2009). Low arousal NA was also found to increase from morning to evening, which is consistent with prior evidence for a diurnal increase in negative emotionality (English & Carstensen, 2014). This diurnal shift in NA may be specific to emotions that are characterised by lethargy and low physiological activation. Conversely, the finding that trait worry was a nonspecific predictor of negative emotionality is consistent with prior research suggesting that worry is transdiagnostic (Brosschot et al., 2006; Olatunji et al., 2010), and points to broad affective dysfunction as a potential mechanism through which worry may confer risk for diverse forms of psychopathology.

Given the growing literature suggesting that trait worry is a transdiagnostic mechanism that predicts broad affective dysfunction (Brosschot et al., 2006; Olatunji et al., 2010), trait worry may be a potent treatment target for individuals across diagnostic categories with high levels of negative emotionality. Through a contrast avoidance framework, it has been recommended that the treatment of excessive worry involves exposures to negative emotional contrasts paired with applied relaxation techniques to build distress tolerance and emotional acceptance (Newman et al., 2014). The present study suggests that such interventions will likely need to target both high and low arousal forms of negative emotions. However, directly targeting the daily high and low arousal NA that is characteristic of chronic worriers may require addressing other mechanisms, such as excessive attention towards emotionally distressing cognitions (Wells & Papageorgiou, 1995).

Although the present study suggests that trait worry has unique effects on emotionality above and beyond those of physiology and depression, the findings must be interpreted within the context of several study limitations. The sample had restricted demographic diversity and was not selected for any psychological traits, but rather on the basis of owning a smart watch. This may have introduced a selection bias into the sample and limited the generalizability of findings. Thus, replication of the present study in a more diverse and clinical sample is warranted. We did not assess momentary worry during the EMA period; thus the present study is not able to disentangle the effects of state vs. trait worry. While our method of measuring momentary HR showed promise (see Supplement for further information), future research may also further evaluate the validity of measuring HR via participants’ self-report from their personal smart watches by comparing such self-reported HR to HR measured via a standardised and well-validated ambulatory monitor. Relatedly, replication using existing measures of high and low arousal NA with demonstrated psychometric properties is also recommended. Additionally, daily exercise was included as a covariate to control for metabolic determinants of HR, but additional factors such as sleep, diet, and other forms of physical activity affect HR and accordingly may be measured and covaried for in future research, along with other variables (e.g. intolerance of uncertainty) that may influence the relation between trait worry and emotionality. Similarly, it is important to note that numerous bodily systems are activated in concert to create the distinct feeling states associated with different emotions beyond solely HR (Kreibig, 2010). To increase measurement feasibility, HR was selected as the sole biomarker of NA in the present study, but future research is needed to examine the unique contribution of dispositional worry to affective experience after accounting for additional indicators of physiological reactivity (e.g. muscle tension, SCL, respiration, etc.). Also to increase measurement consistency, EMA surveys were sent at the same time every day via email. This non-random prompting may have distorted the data if participants waited for the email and thus altered their natural behaviour in reaction to the assessment, and/or delayed their responding until they checked their email. Future research should send EMA assessments at pseudo-random times throughout the day using push notifications to address this limitation. Finally, the lack of experimental manipulation prevents conclusions about directionality and causality. Future research addressing these study limitations will further clarify how trait worry contributes to negative emotionality and associated dysfunction in daily life across various disorders.

Acknowledgements

We thank all study participants for their time and effort, without whom this research would not have been possible.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Participants could select more than one race/ethnicity.

2 Stress ratings were added to the EMA survey mid-way through the study, so only n = 88 participants have complete stress data. Accordingly, only n = 88 participants have complete “High Arousal NA” data.

3 Other smart watches used included Fitbits, Garmins, and others. Adding smartwatch type as a covariate to the primary analyses did not change the results.