Trait Boredom as a Lack of Agency: A Theoretical Model and a New Assessment Tool

Abstract

Trait boredom plays a significant role in well-being. However, this construct suffers from conceptual ambiguity and measurement problems. The aim of this study was to propose a comprehensive theory and a strong assessment tool to address these limitations. We defined trait boredom as the frequent experience of state boredom resulting from a chronic lack of agency. We developed a six-item self-report scale of the tendency to often experience boredom. Results confirmed a uni-dimensional scale with strong psychometric properties, including adequate internal consistency (ω = .84–.93), interindividual stability (69.04% of variance accounted by a trait factor), and acceptable model fit (CFI = .977–.998, TLI = .962–.997, RMSEA = .025–.090, SRMR = .014–.029). Results confirmed the validity of the scale by showing its associations with related measures. Our findings provide clarity on trait boredom and a strong, new measure to be used in future work.

Boredom is the aversive state of wanting, but being unable, to engage in satisfying activity and it is underpinned by two psychological components—a desire bind and an unoccupied mind (Eastwood & Gorelik, 2019). That is, boredom is characterized by wanting to do something, but not wanting to do anything that is available, which leaves the mind unoccupied and cognitive capacities underused. To be bored is to struggle to engage in agentic action (Eastwood & Gorelik, 2022). When bored, we are not authoring our lives, we are not engaged in activity that flows from, and gives expression to, our abilities and desires. Alternatively, it could be said that when bored, a person is not forming, acting on, and achieving their goals.

Boredom is not only caused by boring situations (e.g., monotony, constraint, mismatched challenge, devalued activities; Danckert & Eastwood, 2020) but also by internal psychological factors or individual differences. This notion is consistent with early theorizing on two types of boredom: one caused by external, environmental circumstances and the other caused by internal factors—boredom that comes from within (Bernstein, 1975; Neu, 1998; Todman, 2003). This distinction has been made by multiple theorists. For example, Neu (1998) proposed that there is a “reactive” (i.e., a reaction to external object or situation) and “endogenous” boredom (i.e., caused by some features of the person). Similarly, Bernstein (1975) believed that there is “responsive boredom”—a short-term, transient boredom evoked by the situation, and “chronic boredom,” and that the difference between them is the cause (i.e., whether boredom is caused by the situation/environment, or by internal characteristics of the person; Bernstein, 1975). Finally, Todman (2003) distinguished between “situation-dependent” and “situation-independent” boredom. Recent work suggests that psychological factors might even predispose some people to experience boredom regardless of the situation (Mercer-Lynn et al., 2014).

The internal psychological factors, that are thought to cause boredom, have been classified into five categories (Danckert & Eastwood, 2020). Cognitive factors include poor attention ability and executive dysfunction (e.g., Gerritsen et al., 2014; A. Hunter & Eastwood, 2018; Kass et al., 2001, 2003; Malkovsky et al., 2012). Motivational factors include extrinsic motivation (i.e., doing things for the sake of external rewards) and a strong need to minimize pain or maximize pleasure (e.g., Mercer-Lynn et al., 2014; Mercer-Lynn, Flora, et al., 2013; Mercer-Lynn, Hunter, & Eastwood, 2013). Volitional or self-regulatory factors include poor self-control (e.g., Isacescu & Danckert, 2018; Struk et al., 2016), a preference to “to the right thing” rather than getting on with action (Mugon et al., 2018), and state rather than action orientation (i.e., the tendency to focus on one’s thoughts and emotions about the present, past or future rather than take action; Blunt & Pychyl, 1998). Emotional factors include difficulty identifying emotions (e.g., Eastwood et al., 2007; Harris, 2000; Mercer-Lynn, Flora, et al., 2013; Mercer-Lynn, Hunter, & Eastwood, 2013), emotional unawareness (Bernstein, 1975), experiential avoidance (e.g., Mercer-Lynn, Flora, et al., 2013), and feelings of meaninglessness (e.g., Fahlman et al., 2009; McLeod & Vodanovich, 1991; Melton & Schulenberg, 2007; van Tilburg & Igou, 2012). And finally, physiological factors include nonoptimal arousal and low levels of alertness (e.g., Barmack, 1938; Hamilton, 1981; O’Hanlon, 1981; Zuckerman, 1979).

Thus, while boredom can be fleeting and situationally determined, some people experience boredom more often than others. This observation gave rise to the concept of “boredom proneness” (here referred to as trait boredom; Farmer & Sundberg, 1986). A substantial body of work suggests that trait boredom plays an important role in well-being. It has been linked to a range of clinical, psychological, and social issues (e.g., Danckert & Eastwood, 2020; Fahlman et al., 2013; Goldberg et al., 2011; Li et al., 2021; Mercer & Eastwood, 2010). Research has even demonstrated that trait boredom uniquely predicts some psychosocial problems—such as depression and anger—over and above other potentially confounding variables (Mercer-Lynn, Hunter, & Eastwood, 2013), suggesting the unique relevance of trait boredom and emphasizing the importance of understanding trait boredom.

Most research on trait boredom has used two self-report measures: the Boredom Susceptibility Scale (ZBS; Zuckerman et al., 1978) and the Boredom Proneness Scale (BPS), the latter being the most widely used tool (Farmer & Sundberg, 1986; Vodanovich & Watt, 2016). Of note, however, the ZBS and BPS are related to distinct psychosocial variables and thus are thought to be measure different constructs or types of trait boredom (Mercer-Lynn, Flora, et al., 2013).

The ZBS was designed to measure the “aversion for repetitive experiences of any kind, routine work, or dull and boring people and extreme restlessness under conditions when escape from constancy is impossible” (Zuckerman, 1979, p. 103). It is a subscale of the Sensation Seeking Scale and considered to be part of the general construct of sensation seeking (Zuckerman, 1979). It has been associated with low levels of neuroticism and experiential avoidance, increased motor impulsivity (i.e., acting without forethought), increased sensitivity to reward, decreased sensitivity to punishment, and increased problematic gambling behavior and alcohol use (Mercer-Lynn, Flora, et al., 2013). The ZBS, however, has consistently been shown to suffer from low internal consistency reliability, with values ranging from .52 to .65 (see Vodanovich & Watt, 2016 for a review).

The BPS was designed to measure the tendency toward experiencing boredom, and more specifically captures “one’s connectedness with one’s environment on many situational dimensions, as well as the ability to access adaptive resources and realize competences” (Farmer & Sundberg, 1986, p. 10). Unlike the ZBS, the BPS has been associated with high levels of neuroticism and experiential avoidance, increased attentional and non-planning impulsivity, and decreased emotional awareness. It has also been associated with anxiety, depression, dysphoria, and emotional eating (Mercer-Lynn, Flora, et al., 2013). The BPS has had some ongoing psychometric issues since its development. It was originally a 28-item scale in a true-false format, although currently most researchers use a 7-point Likert-type scale, ranging from “highly disagree” to “highly agree” (Farmer & Sundberg, 1986; Vodanovich & Watt, 2016). The original BPS appeared to be a multifactorial scale, involving multiple factors or subscales, that have been inconsistent across different studies (Ahmed, 1990; Struk et al., 2017; Vodanovich & Kass, 1990). Specifically, previous studies suggested anywhere from two to five factors (see review in Vodanovich & Watt, 2016). Struk et al. (2017) suggested that previous inconsistencies in the BPS factor structure were primarily due to reverse scored items, and that modifying those would yield a single factor scale (Struk et al., 2017). In their study, the authors reduced the 28-item BPS to an eight-item scale comprised of a single factor (Struk et al., 2017). The eight-item, single factor scale appears to perform the best psychometrically and is currently the most often used version of the BPS, although the 28-item version continues to be used as well.

Despite its stronger factor structure, the eight-item BPS (Short BPS, or SBPS) was recently shown to lack validity (Gana et al., 2019). Gana et al. (2019) found that 64% of the variance in the SBPS scores was unexplained (error variance), 8% was due to state boredom, and only 28% was due to trait boredom. Since most of the variance in the SBPS is unaccounted for by trait components, this finding cast doubt on whether this scale truly assesses trait boredom (Gana et al., 2019).

In addition to problems with its measurement, there is increased recognition of the fact that trait boredom suffers from conceptual ambiguity (Gana et al., 2019; A. Hunter et al., 2016; Vodanovich & Watt, 2016), including a lack of agreed upon definition and understanding of what it means to have high levels of trait boredom. Increased efforts have been made to rectify these problems, and several characterizations of trait boredom have been proposed in the literature (A. Hunter et al., 2016; Mercer-Lynn et al., 2014; Tam et al., 2021). According to A. Hunter et al. (2016), for example, trait boredom could be defined as the frequent, intense, prolonged, distressing experience of state boredom, or as difficulty tolerating or reducing boredom. It could also be defined in terms of the tendency to experience boredom in specific situations/circumstances or more pervasively across situations. Alternatively, trait boredom could be conceptualized as some combination of the psychological causes of boredom (see also Bernstein, 1975; Neu, 1998; Todman, 2003).

Recently, Tam et al. (2021), examined three characterizations of trait boredom—frequency of state boredom, intensity of state boredom, and perceived life boredom (i.e., an evaluation of one’s life as boring overall)—to assess which most closely represents what is being measured by the BPS. They found that perceived life boredom was most strongly associated with the BPS, and reproduced the association between BPS and personality, life satisfaction, depression, anxiety, and stress. They concluded that trait boredom as measured by the BPS can be thought of as representing perceived life boredom (Tam et al., 2021). However, the authors point out that although their findings “help clarify the characterizations of boredom proneness, they do not address the psychometric limitations of the (BPS) scale” (Tam et al., 2021, p. 842). Moreover, they call for more theoretical work on understanding the construct of trait boredom, outside of its measurement (Tam et al., 2021).

It is important to note that, existing measures of trait boredom were developed prior to a theoretical understanding and validated measure of state boredom. As such, their items are not grounded in more recent knowledge of state boredom; specifically, boredom’s underlying mechanisms, causes, and correlates. This is a significant limitation of existing measures. As others (e.g., A. Hunter et al., 2016) recommended, conceptualizations and measures of trait boredom need to be grounded in emerging understandings of state boredom.

Present Research

In the present work we begin the process of developing a new self-report measure of trait boredom, which is grounded in theory and exhibits strong psychometric properties. First, we propose a definition of trait boredom. Second, we document the process of refining items to arrive at a reliable scale with acceptable model fit. Third, we present data to support the convergent and predictive validity of this new self-report measure of trait boredom, referred to as the Trait Boredom Scale (TBS).

A Model of Trait Boredom

Building on past proposals (Farmer & Sundberg, 1986; A. Hunter et al., 2016; Tam et al., 2021) our proposed definition of trait boredom consists of two parts; namely we suggest that to be highly trait-bored means that one often experiences state boredom, and that one possesses at least some of the psychological factors (individual differences) thought to cause state boredom. Thus, we are proposing to define trait boredom in terms of the frequency and cause of the state of boredom. The first component—frequently experiencing boredom—can be measured directly via self-report and the second component—possessing the psychological factors thought to cause state boredom—establishes a nomological network for demonstrating the validity of a self-report measure of trait boredom. By defining trait boredom in this manner, we can draw on existing empirical research and theory related to state boredom; in particular, building on our understanding of the psychological causes of state boredom. At this point we have chosen to prioritize frequency of state boredom, as opposed to other qualities of the state such as pervasiveness, duration, intensity, and tolerability. However, in our view, if subsequent work demonstrates these other qualities of the state can be measured distinctly from frequency, then they should be considered for possible inclusion in future definitions and measurements of trait boredom.

Given our definition of trait boredom includes a commitment to the presence of presumed psychological causes, we sought to integrate existing research on the various causes of state boredom into a singular, coherent, and overarching articulation of the psychological cause of boredom. In sum, we propose that the various psychological causes reviewed above—cognitive, motivational, volitional/self-regulatory, emotional, and physiological—can be effectively represented with a singular overarching construct: agency (Danckert & Eastwood, 2020; Eastwood & Gorelik, 2019, 2022). For example, in terms of cognitive factors, individuals who are not able to maintain their attention are likely to have difficulty staying cognitively engaged and following through with tasks, leaving their minds unoccupied. In terms of motivational factors, individuals who possess a strong need to minimize pain are likely to have difficulties with making choices and following through with goals as they are likely to perceive many things as potentially painful/fearsome, and thus tend to engage in avoidance behaviors. Individuals who, however, have a strong need to maximize pleasure may find themselves unable to follow through with tasks that are not sufficiently rewarding, and thus, are likely to have difficulty executing goals. In terms of volitional factors, individuals who prefer to “do the right thing” rather than get on with action and “just do it” may struggle to make choices—they are likely to be more preoccupied with whether there is a better course of action to take and whether their choice is the “right” one, resulting in a failure to act altogether. Similarly, individuals who are not able to identify their emotions, or who tend to avoid them altogether, would have difficulty deciding what matters to them and therefore struggle to form goals/intentions. In a similar vein, feeling that life is meaningless drains the world of significance and possible targets for action. Finally, individuals with chronically low levels of alertness are likely to struggle to stay cognitively engaged with activities at hand and will need strongly stimulating environments to following through with goals. Our contention is that the concept of “agency” provides a useful framework for organizing and understanding the various psychological factors that have been shown to cause state boredom.

We suggest that individuals high in trait boredom suffer from a chronic lack of agency. Agency refers to people’s capacity to influence the course of their lives—their ability to form, act on, and follow through with goals. A chronic lack of agency makes it difficult to realize intentions, to execute, and persist with activities to achieve desired goals, and thus results in the frequent experience of boredom. Therefore, trait boredom can be said to be caused by a chronic lack of agency.

To be clear, we are not proposing that a lack of agency is a new cause of boredom. Rather, we see it as an overarching framework for organizing existing literature on the causes of boredom. The idea that trait boredom is associated with a chronic lack of agency is not new. Past work supports the idea that trait boredom is associated with diminished agency. For example, Bargdill (2000) conducted a qualitative study of participants experiencing chronic boredom. They reported having difficulties realizing their goals and consciously felt this was primarily due to external forces/beyond their control (Bargdill, 2000). Furthermore, trait boredom has been negatively correlated with constructs associated with agency including locus of control (J. P. Hunter & Csikszentmihalyi, 2003), assertiveness (Tolor, 1989), self-actualization (McLeod & Vodanovich, 1991), and self-control or self-regulation (Mugon et al., 2018).

Bandura (1997) argued that people strive to control events that affect their lives. He thought that agency reflects the capacity to act based on choices to shape one’s life circumstances. As such, he proposed four core properties of agency—intentionality, forethought, self-reactiveness, and self-reflectiveness—and argued that to be an agent, one needs to have these four properties. Intentionality refers to making choices or deciding what one wants to do. Forethought refers to setting goals and motivating oneself by thinking of or envisioning the future. By self-reactiveness, Bandura (1997) means an ability to self-regulate; in other words, to motivate and regulate the execution of goal directed actions. Finally, self-reflectiveness refers to a capacity to reflect on one’s thoughts or actions and the meaning of one’s pursuits, and to adjust as necessary—to be self-aware (Bandura, 1997). Though these properties are distinct, in some respects they are also interrelated with fuzzy boundaries. For example, part of regulating the execution of one’s actions (self-reactiveness), likely involves thinking of and visualizing future events (forethought).

We propose that the trait-bored person has chronic difficulties with all four domains proposed by Bandura, which leads to frequent state boredom. As such, we anchor our use of the term “agency” in Albert Bandura’s (2006) model, which provides a framework for understanding the core difficulties of the trait-bored person. Indeed, the psychological factors (individual differences) found in previous work to cause boredom can be roughly mapped onto Bandura’s four components of agency. For instance, attention problems have been discussed as a cause of state boredom and have been correlated with trait boredom (e.g., Gerritsen et al., 2014; A. Hunter & Eastwood, 2018; Kass et al., 2001, 2003; Malkovsky et al., 2012). Using Bandura’s agency framework, attention problems can be characterized as difficulties with self-reactiveness—regulating the execution of goals or plans. Similarly, a strong need to maximize pleasure can be characterized as difficulties with intentionality—individuals who have a strong need to seek rewards may find few sufficiently rewarding options to motivate action (Mercer-Lynn et al., 2014; Mercer-Lynn, Flora, et al., 2013; Mercer-Lynn, Hunter, & Eastwood, 2013). As another example, alexithymia (which has been associated with boredom; Eastwood et al., 2007; Harris, 2000; Mercer-Lynn, Flora, et al., 2013; Mercer-Lynn, Hunter, & Eastwood, 2013) can be thought of as a difficulty in the domain of self-awareness (self-reflectiveness). A sense of meaninglessness (e.g., van Tilburg & Igou, 2012) could be characterized as difficulties with forethought and intentionality, whereby it is not possible to envision a future in which personal choices even matter. Finally, avolition (e.g., Gerritsen et al., 2015), could be characterized as difficulties with the execution of goal directed pursuit, namely self-reactiveness.

In sum, we have proposed a theoretical model of trait boredom that consists of two key components: frequently experiencing state boredom and chronic difficulties with agency. In the following section we develop a scale that measures the frequency of state boredom, is correlated with various measures of agency and predicts future episodes of state boredom.

Psychometric Development, Convergent and Predictive Validity

Scale items for the TBS were initially derived by creating trait versions of items from the Multidimensional State Boredom Scale (MSBS; Fahlman et al., 2013), which is a reliable and valid tool for measuring state boredom (Alda et al., 2015; Baratta & Spence, 2015; Craparo et al., 2017; Goldberg et al., 2011; J. A. Hunter et al., 2016; Liu et al., 2013; Mercer-Lynn et al., 2014; Ng et al., 2015; Oxtoby et al., 2018).

A preliminary 10-item version of the TBS has been explored in two previous studies (Britton, 2018; Gerritsen et al., 2014). These studies showed the TBS items had promising reliability and validity (correlated with BPS and ZBS and self-reported measures of inattention, Gerritsen et al., 2014; as well as low approach motivation, Britton, 2018).

We began by adding 5 items to the preliminary 10-item TBS introduced by Gerritsen et al., (2014). We also revised 3 existing items judged to have poor wording (Willis & Lessler, 1999). Next, using several data sets ¹ (see Table 1) we reduced the 15-item TBS to arrive at a brief unidimensional scale that measures the tendency to often experience boredom. We also sought to evaluate the stability of scores on the TBS using a Trait-State Occasion (TSO) analysis, demonstrate its correlation with agency and evaluate its predictive validity. The final 6-item TBS is presented in the Appendix.

Table 1.

Data Sets Used in Part 2.

Data Set number	Source
Data Set 1	Gorelik (2019)—study 1, dataset 2
Data Set 2	Bambrah (2020)
Data Set 3	Hunter & Eastwood (2021)
Data Set 4	New data collected for present TBS project
Data Set 5	Bambrah et al. (2022)

Data Set 1

Data Set 1 examined the 15-item TBS (N= 476, Age range: 18-88 years, M_age = 48.62, SD _age =16.18). Four items were removed –two items with high skew (.73 and .74), and two items with low corrected item-total correlations (.60 and .66). A confirmatory factor analysis was conducted for a one-factor model of the 11-item TBS. Fit indices suggested that the model fit was relatively poor for the 11-item TBS, Robust CFI = .943, Robust TLI = .928, Robust RMSEA = .104, Robust SRMR = .038. Reliability assessed using Coefficient Omega was .95. We examined modification indices to better understand the relatively poor fit. The highest index (MI= 100.493) was for items 2b and 9, suggesting the fit of the model would improve if the errors of items 2b and 9 were correlated. These two items ask two questions each (i.e., feeling “forced” or “stuck” and having “no value” or “meaningless”), making them potentially confusing. Thus, items 2b and 9 were removed from the scale.

Data Set 2

Dataset 2 examined the 9-item TBS (N= 572, Age range: 16-57, M_age = 20.47, SD = 4.13). One item was removed from the scale due to having high skew (–.75). A confirmatory factor analysis was conducted for a one-factor model of the 8-item TBS. Fit indices suggested that the model fit was relatively poor for the 8-item TBS, Robust CFI = .925, Robust TLI = .896, Robust RMSEA = .112, Robust SRMR = .047. Reliability assessed using Coefficient Omega was .89.

Data Set 3

In this analysis we re-examined the 8-item TBS in a new dataset to identify weaker items that could be removed to improve the model fit (N= 491, Age range: 17-38, M_age = 19.63, SD = 3.06). One item was removed due to having high skew (–.67). A confirmatory factor analysis was then conducted for a one-factor model of the 7-item TBS. Fit indices suggested that the model fit was good for the 7-item TBS, Robust CFI = .97, Robust TLI = .95, Robust RMSEA = .072, Robust SRMR = .035. Factor loadings were strong ranging from .63 to .73. Reliability assessed using Coefficient Omega was .85.

Data Set 4

In this analysis, we re-examined the 7-item TBS in a new dataset. Also, given previous work has shown that the BPS has little trait variance but large error variance (Gana et al., 2019), we sought to partition the variance in TBS scores into state, trait, and error components to assess the stability (i.e., rank-order consistency) ² and thus validity of the scale using a longitudinal design and a Trait-State Occasion (TSO) model analysis (Cole et al., 2005).

Next, we aimed to validate the TBS by exploring its association with direct measures of agency. We also examined the correlations between the TBS and a measure of social desirability to establish that the TBS responses are not overly biased by social desirability.

The Subjective Personal Agency Scale, which measures one’s perception of their ability to make purposeful choices (Yamaguchi et al., 2020), and the Sense of Agency Scale, which measures “individuals’ beliefs about being agents in the sense of generally experiencing control over one’s body, thought and immediate environment” (Tapal et al., 2017, p. 7), were both expected to have a negative association with trait boredom, such that high levels of trait boredom would be associated with lower levels of agency on these scales. The Avolition Scale (Gerritsen et al., 2015) assesses the inability to initiate and persist in goal-directed activities, and we expected trait boredom to be associated with higher levels of avolition, reflecting lower levels of intentionality in Bandura’s (1997) model.

Furthermore, using the Work Preference Inventory (Robinson et al., 2014), we predicted that trait boredom would be positively associated with extrinsic motivation and negatively associated with intrinsic motivation. Self-determination theory (Ryan & Deci, 2000) proposes a more differentiated continuum of motivation. Intrinsic motivation is highest on the self-determination continuum and is the most agentic, followed by extrinsic motivation, which is further divided into integrated, identified, introjected, and external regulation—from highest to lowest levels of self-determination, and amotivation which is the least agentic. These various types of motivation are assessed by the Global Motivation Scale (Sharp et al., 2003). We broadly expected the TBS to be negatively associated with intrinsic motivation, and the high end of extrinsic motivation continuum. We expected the TBS to be positively associated with the low end of the extrinsic motivation continuum, as well as positively associated with amotivation.

Finally, with respect to social desirability, we expected trait boredom to be negatively associated with Self-Deceptive Enhancement, consistent with findings for the MSBS (Fahlman et al., 2013), and consistent with the fact that items on this subscale are associated with psychological adjustment. We expected no significant association with Impression Management (IM).

Methods

Participants

We determined sample size based on recommendations for Trait-State Occasion (TSO) models (e.g., Cole et al., 2005), while also considering attrition rates. As such, we aimed to recruit a sample of about 1,400 participants in wave 1 of this study, such that with 30% attrition at each part we would obtain a minimum of 480 participants in wave 4. We recruited 1406 participants through Prolific Academic. Participants who failed an attention check at waves 1, 2, and 3 were not invited to take part in subsequent waves and their data was removed. In wave 1, 21 participants failed an attention check. Furthermore, data from 72 participants were removed because they completed the study too quickly (i.e., in less than 2 minutes ³ ). The final sample in wave 1 was N = 1313 (age range: 18-76, M_age = 26.96, SD = 9.6). The total sample contained 785 individuals who identified their gender as male (59.8%); 517 individuals who identified their gender as female (39.4%); 10 individuals who identified their gender as other (.8%); and 1 individual who preferred not to answer (.08%). Participants identified with the following ethnicities: 84.3% White/Caucasian, 1.7% Black, 7.5% Latin American, 1.4% South-Asian, 1.1% East-Asian, .76% South-East Asian, .53% Arab/West-Asian, 1.1% Other, and 1.6% as multiracial. In wave 2, 1290 of the participants from wave 1 completed the study. A total of 21 participants failed an attention check, and data from another 109 participants were removed because they completed the study too quickly (i.e., in less than 2 minutes). The final sample in wave 2 was N = 1160. In wave 3, 1242 participants completed the study, but 23 failed an attention check and another 99 were removed for completing the study too quickly. The final sample in wave 3 was N = 1120. In wave 4, 1199 participants completed the study, but 13 failed an attention check and 194 participants were removed for completing the study too quickly. The final sample in wave 4 was N = 992.

Design and Procedure

This study consisted of four waves completed entirely online, and it took approximately five minutes for a participant to complete the measures in each wave. The waves were separated by 2-week intervals. Participants had a maximum of three days to complete each wave before it was no longer available to them. Compensation was increased exponentially for participation in each wave of the study. Participants were paid £.53 for completing wave 1, £.58 for wave 2, £.68 for wave 3, and £.93 for wave 4.

In wave 1, participants answered demographic questions about their age, gender, and ethnicity. They then completed the TBS followed by a self-report measure of socially desirable responding. In waves 2 to 4, participants again completed the TBS, followed by self-report measures that directly assessed agency including general belief in having agency (wave 2), avolition and work motivation (wave 3), and regulatory orientations (wave 4). In each wave, the measures of agency were presented in counterbalanced order.

Measures

Trait Boredom Scale (TBS; Completed in Waves 1, 2, and 3)

The 7-item TBS (from the analysis of data set 3) was used for the current study.

Balanced Inventory of Desirable Responding Short Form (BIDR-16; Completed in Wave 1)

This questionnaire measures socially desirable responding (Hart et al., 2015). It includes two subscales: Self-Deceptive Enhancement and Impression Management. Participants were asked to respond to each item on a Likert-type scale from 1 (=not true) to 7 (=very true) to items such as “I never cover up my mistakes.” Cronbach’s alpha reliability for the current study was .74 for the Self-Deceptive Enhancement subscale, and .70 for the Impression Management subscale.

Subjective Personal Agency Scale (SPA-5; Completed in Wave 2)

This five-item questionnaire measures personal agency (Yamaguchi et al., 2020). Participants were asked to respond to each item on a Likert-type scale from 1 (=strongly disagree) to 5 (=strongly agree) to items such as “I have an idea of what I want to do and/or how I want to be.” Cronbach’s alpha reliability for the current study was .74.

Sense of Agency Scale (SAS; Completed in Wave 2)

This 13-item questionnaire measures “chronic” or general belief in having “core” agency (Tapal et al., 2017). Participants were asked to respond to each item on a Likert-type scale from 1 (=strongly disagree) to 7 (=strongly agree) to items such as “My actions just happen without my intention.” Cronbach’s alpha reliability for the current study was .83.

Avolition Scale (AS; Completed in Wave 3)

This 13-item questionnaire measures the inability to initiate and persist in goal-directed activities (Gerritsen et al., 2015). Participants were asked to respond to each item on a Likert-type scale from 1 (=strongly disagree) to 7 (=strongly agree) to items such as “I have trouble getting started on tasks.” Cronbach’s alpha reliability for the current study was .86.

Work Preference Inventory (WPI; Completed in Wave 3)

This ten-item questionnaire measures work motivation. It includes four subscales: Extrinsic outward, extrinsic compensation, intrinsic challenge, and intrinsic enjoy (Robinson et al., 2014). Participants were asked to respond to each item on a Likert-type scale from 0 (=Never or almost never true of me) to 3 (=Always or almost always true of me) to items such as “I enjoy trying to solve complex problems.” Cronbach’s alpha reliability for the current study was .68 for the extrinsic outward subscale, .78 for extrinsic compensation, .84 for intrinsic challenge, and .77 for intrinsic enjoy.

Global Motivation Scale (GMS; Completed in Wave 4)

This 18-item questionnaire measures different forms of individuals’ enduring regulatory orientations: intrinsic motivation, extrinsic motivation - integrated, identified, introjected and external regulation, and amotivation (Sharp et al., 2003). Participants were asked to respond to each item on a Likert-type scale from 1 (=Not agree at all) to 7 (=Completely agree) to items such as “In general, I do things. . .because I like making interesting discoveries.” Cronbach’s alpha reliability for the current study was .69 for the intrinsic motivation subscale, .86 for extrinsic motivation integrated, .75 for identified, .83 for introjected, .78 for external, and .76 for amotivation.

Results and Discussion

Factor Analyses

The psychometric properties of the 7-item TBS were examined first (see Table 2). The scale was further refined by removing one final item which had a low corrected item-total correlation (.60) and was thought to be redundant with existing items. This resulted in the final 6 item TBS.

Table 2.

Descriptive Statistics for Final Set of 6 Items in Data Set 4.

Item	M	SD	Skew	Kurtosis	Corrected item-total correlation	Standardized factor loading
10	4.39	1.67	−.39	−0.95	.72	.799
13b	4.59	1.68	−.50	−0.80	.76	.837
17b	4.20	1.76	−.23	−1.05	.70	.782
22	5.25	1.61	−.88	−0.08	.66	.692
28	4.32	1.74	−.35	−0.93	.70	.736
31	3.83	1.72	.14	−1.01	.70	.678

A confirmatory factor analysis was then conducted for a one-factor model of the 6-item TBS using wave 1 data. The overall model was significant, χ²(14) = 41.967, p < .001. Although the chi-square result was significant, this measure is particularly influenced by large sample sizes and its usefulness has thus been questioned (Bentler & Bonett, 1980; Brown, 2006; Struk et al., 2017). Fit indices suggested that the model fit was good for the 6-item TBS, Robust CFI = .981, Robust TLI = .968, Robust RMSEA = .079, Robust SRMR = .026. Factor loadings were strong ranging from .68 to .84 (see Table 2). Reliability assessed using Coefficient Omega was .89.

Trait-State Occasion (TSO) Model

Next, a trait state occasion model analysis was conducted. First, because the TBS was shown to be unidimensional, the six items were randomly assigned into parcels—three parcels of two. This was done so that the parcels could serve as observed indicators of the latent state factor for each of the four waves (see Figure 1). Furthermore, as recommended by Cole and colleagues (2005), several constraints were imposed on the model. Specifically, all factor loadings were fixed to 1 except those of manifest indicators (i.e., TBS parcels). The model also allowed for correlated measurement residuals among indicators for each subscale, which can help to account for any common method variance that may contribute to stability of scores over time (Newsom, 2015).

Figure 1.

Trait-State Occasion (TSO) Model

The fit of the trait-state-occasion model was good, χ²(38) = 77.728, p < .001, CFI = .997, TLI = .995, RMSEA = .030, SRMR = .026. To assess the assumption of the homogeneity of the autoregressive paths (i.e., their equality), the fit of the models with and without stationarity constraints was compared. The fit of the trait-state-occasion model with stationarity constraints was χ²(42) = 105.899, p < .001, CFI = .995, TLI = .992, Robust RMSEA = .036, and SRMR = .033. Comparing the two models, results showed that the assumption of homogeneity did significantly worsen the model fit, (Δχ² = 24.955, Δdf = 4, p < .001). However, the fit indices and parameter estimates of the models with and without constraints were similar. Overall, we selected and proceeded to interpret the TSO model without the stationarity constraints.

To decompose the total variance in the model into trait, state, and error components, the unstandardized estimates of each variance component were used (see Table 3). Each estimate was then divided by the sum of all three components (e.g., trait variance was calculated as follows: trait variance/ [trait variance + occasion variance + average indicator variance]). Results showed the trait factor accounted for about 69.04% of the total variance in this model, the state factor accounted for 7.94% of variance and finally, based on the average measurement residual variance, error comprised the remaining 23.02% of the variance in this model.

Table 3.

Variance Components in Trait-State-Occasion (TSO) Model.

Variances	Estimate	SE	Standardized estimate
Occasion 2	0.255	0.186	0.802
Occasion 3	1.182	0.115	0.994
Occasion 4	1.080	0.114	0.685
State 1	0		0
State 2	0		0
State 3	0		0
State 4	0		0
TBS parcel 1, time 1	2.054	0.146	0.211
TBS parcel 2, time 1	2.039	0.122	0.235
TBS parcel 3, time 1	2.538	0.141	0.286
TBS parcel 1, time 2	1.976	0.133	0.215
TBS parcel 2, time 2	1.818	0.123	0.226
TBS parcel 3, time 2	2.353	0.133	0.282
TBS parcel 1, time 3	1.886	0.138	0.187
TBS parcel 2, time 3	1.630	0.124	0.187
TBS parcel 3, time 3	2.476	0.168	0.267
TBS parcel 1, time 4	1.595	0.151	0.156
TBS parcel 2, time 4	1.700	0.129	0.186
TBS parcel 3, time 4	2.559	0.173	0.264
Occasion 1	0.708	0.141	1
Trait	6.155	0.233	1

Note. TBS = Trait Boredom Scale.

Convergent Validity

Table 4 summarizes the results. As expected, results showed that the TBS was negatively associated with direct measures of agency assessed in Data Set 4, including the Sense of Agency Scale (SAS; Tapal et al., 2017), r = –.38, p < .001, and the Subjective Personal Agency Scale (Yamaguchi et al., 2020), r = –.40, p < .001. The TBS was positively related to the Avolition scale (Gerritsen et al., 2015), r = .69, p < .001, reflecting difficulties in initiating and persisting in goal-directed activities. On the Global Motivation Scale (GMS; Sharp et al., 2003) the TBS was positively related to amotivation r = .28, p < .001, external (motivated by external awards and limitations), r = .13, p < .001, and introjected (motivated by internal desire to please others, for example) subscales, r = .29, p < .001, all of which are on the lower end of the self-determination continuum (and similarly lower on agency). It was negatively related to identified (r = –.15, p < .001), integrated (r = –.18, p < .001), and intrinsic motivation (r = –.22, p < .001), which are on the higher end of the self-determination continuum. Similarly, on the Work Preference Inventory (WPI; Robinson et al., 2014), the TBS was positively related to Extrinsic—outward motivation (e.g., motivation coming through recognition from others), r = .11, p < .001, and negatively related to Extrinsic—compensation (i.e., awareness of income and/or promotion goals for oneself), r = –.14, p < .001, Intrinsic—challenge (i.e., motivation to engage in challenging or complex problems or tasks), r = –.23, p < .001, and intrinsic—enjoy motivation (i.e., motivation to engage in tasks because they are enjoyable), r = –.07, p < .05. Contrary to our predictions, the negative correlation with the extrinsic—compensation subscale is, in hindsight, possibly explained by the fact that the extrinsic-compensation subscale may tap into difficulties forming goals (i.e., intentionality and forethought) more so than being externally motivated per se.

Table 4.

Summary of All Correlations With the Final 6-Item TBS.

Measure	Data Set 4	Data Set 5
Direct measures of agency
Subjective Personal Agency Scale (SPA-5)	−.40***
Sense of Agency Scale	−.38***
Avolition Scale	.69***
Work Preference Inventory
Extrinsic: outward	.11***
Extrinsic: compensation	−.14***
Intrinsic: challenge	−.23***
Intrinsic: enjoy	−.07*
Global Motivation Scale
Intrinsic	−.22***
Integrated	−.18***
Identified	−.15***
Introjected	.29***
External	.13***
Amotivation	.28***
Desirable responding
BDIR
SDE	−.54***
IM	−.31***
Predictive validity
Short MSBS after 1 week		.63***
Short MSBS after 2 weeks		.62***

Note. Data Set 4: N = 992–1,313 (each wave had a different N). Data Set 5: N = 345. TBS = Trait Boredom Scale; BDIR = Balanced Inventory of Desirable Responding; SDE = Self-Deceptive Enhancement; IM = Impression Management; MSBS = Multidimensional State Boredom Scale.

^*p < .05. **p < .01. ***p <.001.

Finally, we examined the relation between the TBS and socially desirable responding. As predicted, the TBS was negatively associated with Self-Deceptive Enhancement (SDE), r = –.54, p < .001, consistent with previous state boredom (MSBS) findings (Fahlman et al., 2013). The SDE subscale measures a type of social desirability characterized by unintentionally providing positively biased self-reports (Paulhus, 1988). The negative correlation between the SDE and the TBS indicates that items on the TBS are “undesirable” in nature. Furthermore, Paulhus found that high scores on SDE are associated with an “adjusted personality”—including high self-esteem, low neuroticism, depression, empathic stress, and anxiety (Fahlman et al., 2013; Paulhus, 1991; Paulhus & Reid, 1991), all of which have been shown to be associated with boredom.

There was also a significant negative correlation between trait boredom and Impression Management (IM), r = –.31, p < .001. This was unexpected, as this was not the case for state boredom as measured by the MSBS (Fahlman et al., 2013). Recent findings may help explain this unexpected finding. Namely, evidence suggests that the IM subscale may not be an unconfounded measure of socially desirable responding per se (Uziel, 2010, 2014; see also de Vries Zettler & Hilbig, 2014; Zettler et al., 2015), and that it may in fact assess stable levels of self-control or regulation. That is, individuals may not score high on IM scales because they act to impress others, or because they depend on social approval, but rather because they have a high self-regulatory capacity (Uziel, 2010). As highlighted in the present work, individuals with high levels of trait boredom have difficulties with self-control, as part of their chronic lack of agency. As such, the significant negative correlation between IM and trait boredom may reflect these difficulties rather than an attempt to engage in impression management.

Data Set 5

In this analysis, we aimed to validate the TBS by exploring its ability to predict state boredom. This was examined using data previously collected in our lab (Bambrah et al., 2022). We expected trait boredom to predict state boredom (MSBS) in a longitudinal manner 1 and 2 weeks later.

Methods

Participants and Procedure

A full description of the methods, including the procedure and all measures used, is described in (Bambrah et al., 2022). Participants were recruited from a Qualtrics Panel. The study consisted of three waves completed entirely online. Participants who completed questionnaires too quickly (i.e., in less than 400 seconds in Wave 1, and less than 300 seconds in Waves 2 and 3), or took too long to complete them (i.e., completed waves 1, 2, or 3 in over 36,000 seconds) were excluded from all three waves. Participants who demonstrated insufficient effort in responding were also excluded from all three waves. Specifically, an intra-individual response variability (IRV) was calculated for each participant on two questionnaires that included both positively and negatively worded items. Participants with an IRV value greater than two standard deviations below the mean IRV on either of the two questionnaires at any wave were excluded. The final sample at Wave 3 was N = 345 (age range: 18–88, M_age = 49.26, SD _age = 16.68). The total sample contained 234 individuals who identified their gender as female, 110 individuals who identified as male, and one who did not disclose their gender. Most participants resided in the United States (i.e., over 99%).

Measures

Trait Boredom Scale (6-Item Final Version; Completed in Wave 1)

A confirmatory factor analysis was conducted for a one-factor model of the six-item TBS using Wave 1 data (N = 2743). The fit of the TBS was good, Robust CFI = .98, Robust TLI = .97, Robust RMSEA = .09, Robust SRMR = .02. Reliability assessed using Coefficient Omega was .91.

Multidimensional State Boredom Scale—Short Form (Completed in Waves 1, 2 and 3)

This questionnaire measured participants’ boredom over the past week at each wave (J. A. Hunter et al., 2016). Participants were asked to respond to each item on a Likert-type sale from 1 (=strongly disagree) to 7 (=strongly agree), to items such as “I felt like I was wasting time that would be better spent on something else.” Cronbach’s alpha reliability for the current study was .91 at Wave 1.

Results

Scores on the six-item TBS predicted scores on the Short Form MSBS 1 week later, r = .63, p < .001, and 2 weeks later, r = .62, p < .001, suggesting it possesses good predictive validity.

General Discussion and Conclusion

A substantial body of work has shown that trait boredom plays a pivotal role in well-being. However, trait boredom has suffered from conceptual ambiguity—including a lack of agreed upon definition and understanding of this construct, and its measures have been shown to lack reliability and validity. This work aimed to address these problems by proposing a comprehensive model of trait boredom and developing a psychometrically strong assessment tool informed by this model.

First, we proposed a definition of trait boredom, which consisted of two key components: frequently experiencing state boredom and possessing at least some of the psychological factors thought to cause state boredom (i.e., cognitive, motivational, volitional/self-regulatory, emotional, and physiological factors). Critically, we integrated existing literature and proposed that the core difficulties of the trait bored person can be captured by the concept of chronically lacking agency. We suggested that a chronic lack of agency makes it difficult to realize intentions, to execute, and persist with activities to achieve desired goals, and thus results in the frequent experience of boredom. Specifically, we situated our model in Bandura’s (1997) four properties of agency (i.e., intentionality, forethought, self-reactiveness, and self-reflectiveness). This model of agency provides a lens through which to better understand the internal psychological factors thought to cause boredom. ⁴

Our model can be integrated with other characterizations of trait boredom that have been recently proposed in the literature. For example, Tam and colleagues (2021) proposed that trait boredom as measured by the BPS most closely represents perceived life boredom, as well as frequency and intensity of boredom. While the model presented in the current study focused on the frequency of boredom, we suggested that other qualities of state boredom (e.g., intensity, pervasiveness, distress, etc.) could be integrated as well. For example, perceived life boredom may reflect a pervasive experience of boredom (i.e., experiencing boredom in many areas of one’s life).

Next, we sought to address problems with the measurement of trait boredom (e.g., Gana et al., 2019; Tam et al., 2021) by developing a new, psychometrically strong, and theoretically grounded scale. We aimed to develop a scale that measures the frequency of state boredom and that is correlated with the psychological causes of state boredom in the manner that would be expected based on our agency-based model. Our analyses across five independent samples supported the unidimensional structure of the TBS. We showed that the TBS possesses good internal consistency, and importantly, that it captures stable interindividual differences. These findings point to the strengths of this measurement tool and importantly, show that this scale addresses key limitations in existing measures. For example, as recommended in past work (e.g., A. Hunter et al., 2016), the TBS is grounded in recent knowledge and theoretical understanding of state boredom. In contrast, existing measures were developed before the emergence of these theories. The TBS is also based on a clear definition or model of trait boredom, while existing measures lack such theoretical grounding. Finally, the TBS demonstrated good psychometric properties including strong reliability and validity, and unlike the BPS it appears to mostly capture trait variance. The TBS is a six-item unidimensional measure that offers a succinct and efficient way of examining trait boredom in future work.

The TBS was associated with lower levels of personal agency on all direct measures of agency assessed, consistent with our conceptualization of trait boredom. The TBS also predicted state boredom (on the MSBS) longitudinally, consistent with our definition of trait boredom. These findings provide evidence for the validity of our proposed model of trait boredom and for the TBS.

Limitations and Future Directions

There are several limitations to the present studies. One limitation was that the TBS did not include reverse-scored items. Although there is some controversy on this matter (e.g., van Sonderen et al., 2013; Zhang et al., 2016), the lack of reverse scored items may mean that the trait variance of the TBS is confounded by response style variance, or the tendency to give stronger or weaker endorsement of self-report items. Given that our goal in this work was to create a brief, unidimensional scale, adding reverse scored items would likely introduce a second factor and thereby complicate this goal. Future work should address this limitation.

Future research will need to further validate the TBS. For example, by examining additional qualities of state boredom to possibly be included in the definition and assessment tool of trait boredom. These additional qualities of the state of boredom may include pervasiveness, duration, intensity, and tolerability. Our agency model also gives rise to new predictions of other constructs that may be associated with trait boredom. For example, given our model proposes that individuals with high levels of trait boredom would have difficulties with forethought, they may evidence weak prospective memory and difficulties with future oriented thinking. These kinds of questions can be explored in the future to further validate our agency-based understanding of trait boredom. In addition, our definition of trait boredom can contribute to a more nuanced understanding of its relationship with psychological well-being. For example, future work may examine whether a chronic lack of agency alone can explain the observed association between trait boredom and well-being or if the frequent experience of state boredom is also critical. Our definition can also aid in the development of interventions to help individuals who are struggling with high levels of boredom, including interventions that focus on facilitating increased agency.

Appendix

Instructions: Please respond to each question indicating how you generally feel about yourself and your life, even if it is different from how you feel right now.

10. I often feel bored

13b. I often do not know what I want to do

17b. I often feel like there is nothing fun to do

22. I often feel like I am wasting time that would be better spent on something else

28. I often feel like I’m sitting around waiting for something to happen

31. It is difficult for me to stay interested in what I’m doing