Abstract
Past studies on service waits have primarily applied two theoretical perspectives—the expectancy–disconfirmation model and the psychological‑cost model—leading to conflicting explanations of how customers react to waits. To reconcile these perspectives, we propose a conceptual framework that integrates the dual perspectives of customer wait perceptions and identify four key moderators that shape which mechanism becomes more influential: wait stage, wait knowledge, wait regret, and wait importance. We validate our conceptual framework through a structural‑equation‑modeling meta‑analysis of 129 studies from 103 articles (672 effect sizes; N = 38,967). Our conceptual integration highlights a coherent system of interdependent mechanisms that shape how customers experience, evaluate, and emotionally respond to waiting. Its validity, however, depends on the efficacy of the moderators. During the wait, subjective time primarily drives psychological cost, heightening anxiety and anger—especially when regret is present. After the wait, subjective time instead shapes how acceptable the wait feels. Disconfirmation becomes more influential when customers have clearer information that strengthens their expectations. When the wait is important, wait acceptability becomes the dominant predictor of service satisfaction. These findings underscore the need to balance strategies that reduce psychological cost and manage expectations, particularly as technology transforms waiting experiences.
Keywords
Introduction
Service waits are almost inevitable for everyone: people wait at the Department of Motor Vehicles (DMV) to renew a driver’s license, queue at theme parks for rides, and sit in walk-in clinics for medical check-ups. To understand how customers react to these waits, prior research has largely relied on two theoretical frameworks: the expectancy–disconfirmation model (Oliver 1980, 2010) and the psychological-cost model (Osuna 1985).
The expectancy–disconfirmation model argues that satisfaction reflects customers’ expectations, the degree of disconfirmation, perceived performance, and the emotions tied to specific service attributes (Oliver 2010). In this view, disconfirmation and subjective time jointly shape wait acceptability, which then influences satisfaction directly and indirectly through anger (Kumar 2005; Oliver 2010; Pruyn and Smidts 1998; Taylor 1995). Waiting is therefore treated as a service attribute (service-attribute model; Figure 1a).
(a) Service-attribute model. (b) Service-cost model. (c) Integrated “Dual-Perspective” wait model.
By contrast, the psychological-cost model holds that satisfaction is driven by negative emotions arising from anxiety about the remaining wait and discomfort from perceived time loss (Osuna 1985). Subjective time increases stress and anger—both directly and through anxiety—which subsequently lowers satisfaction (Kumar et al. 1997; Miller et al. 2008; Voorhees et al. 2009). Here, waiting is viewed as an added cost customers must bear (service-cost model; Figure 1b).
These perspectives have shaped different wait-management strategies: the service-attribute model motivates providing wait information and improving the waiting environment (Baker and Cameron 1996; Grewal et al. 2003; Hui and Tse 1996; Hui et al. 1997; Kumar et al. 1997; Kumar and Dada 2021), whereas the service-cost model inspires distraction-based interventions and research on the social meaning of waits and queue length (Giebelhausen et al. 2011; Taylor 1994, 1995).
Although both models have empirical support, it remains unclear when and why one perspective becomes more salient in shaping customers’ reactions to waiting. Each theoretical viewpoint addresses service satisfaction in the context of waits, yet their explanations for how satisfaction forms reveal both notable parallels and important divergences—creating fertile ground for integrative research questions. This paper aims to (i) develop a conceptual framework that unifies the dual perspectives on customer wait perceptions (see Figure 1c) and (ii) resolve the theoretical divergences by identifying and empirically testing potential moderators through a meta-analytic study.
We identify three key theoretical divergences concerning the sources of wait (un)acceptability, anger, and service satisfaction across the customer wait journey. These divergences reveal important gaps and set the stage for three integrative research questions that guide our effort to conceptually unify the two models. We now turn to the three research questions that motivate our study.
Key Research Questions
Under What Condition Is a Specific Measure of Wait Time—Disconfirmation Versus Subjective Time—More Diagnostic in Shaping Customer Wait (Un)acceptability?
The expectancy–disconfirmation model proposes that wait acceptability is shaped by both subjective time and the discrepancy between perceived and expected wait duration (Oliver 1980, 2010), with disconfirmation often exerting the stronger influence (Caruelle et al. 2023; Maister 1985; Pruyn and Smidts 1998). Subjective time reflects the performance component, whereas disconfirmation captures the evaluative comparison against expectations (Oliver 1980, 2010; Szymanski and Henard 2001). In contrast, the psychological-cost model emphasizes affective responses, arguing that subjective time drives anxiety about uncertain time loss and anger from realized time loss (Hui et al. 1997; Kumar et al. 1997; Miller et al. 2008; Osuna 1985; Voorhees et al. 2009).
Although both models draw on the cognitive-timer framework—where limited attention shapes subjective time and reduced attention makes waits feel shorter (Hornik and Zakay 1996; Zakay 1989)—they differ fundamentally in how they conceptualize expectations. The service-attribute model treats expectations as point estimates of the “most likely” wait, which are updated through disconfirmation as new information arises during the wait (Bolton 1998; Zeithaml et al. 1993). In contrast, the service-cost model views expectations as anticipated losses represented by a probability distribution shaped by prior experience, with the passage of subjective time itself serving as the new information that drives emotional responses (Osuna 1985).
These theoretical differences yield opposing managerial strategies: distraction reduces subjective time by lowering time awareness (Taylor 1995), while wait-time information reduces disconfirmation by heightening it (Hui and Tse 1996; Kumar et al. 1997). This trade-off is evident in settings such as the DMV, where managers must choose between real-time queue updates that improve expectation accuracy and attention-absorbing content that shortens perceived wait duration. Reducing perceived wait duration and reducing expectation disconfirmation require fundamentally opposing psychological mechanisms.
Under What Condition Is the Elicitation of Customer Anger in the Wait More as a Result of Anxiety Versus Wait (Un)acceptability?
The expectancy–disconfirmation model argues that anger arises when customers judge a wait as unacceptable because it blocks need fulfillment and violates the implicit promise of prompt service (Taylor 1994). This transgression triggers anger—a high-arousal, attribution-dependent emotion (Roseman 1984)—consistent with Weiner’s (1985) attribution–affect model, in which individuals first evaluate the outcome and then experience anger when they attribute the negative outcome to others’ controllable actions. Thus, customers must assess the wait’s acceptability and assign blame to the provider before anger emerges.
In contrast, the psychological-cost model proposes that anger stems from realized time loss that customers are motivated to avoid (Osuna 1985). Anxiety reflects uncertainty about anticipated time loss, whereas anger is the response to actual time wasted (Menon and Dube 2004), aligning with cognitive view of emotions that distinguish threat-based emotions (anxiety) from loss-based emotions (anger; Beck 1976; Roseman 1984). As subjective time accumulates, anxiety builds, and anger intensifies in response to mounting wasted time.
Despite these differences, both models draw on Weiner’s (1985) appraisal–attribution framework and treat waiting as a negative event whose anger response strengthens when delays are attributed to the provider’s controllable actions, such as perceived incompetence or negligence (Beck 1976; Menon and Dube 2004). Where they diverge is in motivational assumptions: the service-attribute model adopts an appetitive view, emphasizing how waits obstruct need satisfaction, whereas the service-cost model adopts an aversive view, emphasizing anxiety about time loss that culminates in anger (Miller et al. 2008).
These distinctions yield different managerial implications. Priority-based systems can reduce average waits but may evoke perceptions of unfairness, whereas first-come-first-served systems enhance procedural fairness but can heighten anxiety as customers perceive accumulating time loss (Zhou and Soman 2008). This tension is especially salient in monetized priority systems—such as Disneyland’s Genie+ and Lightning Lane—which improve wait acceptability for paying guests but can increase anger and perceived time loss among non-paying guests who view the process as unfair.
Under What Condition Does Service Satisfaction Depend More on Wait (Un)acceptability Versus Anger Resulting From the Wait?
The expectancy–disconfirmation model argues that both cognitive evaluations and emotional reactions shape satisfaction, but because memory retrieval dominates information processing, wait acceptability becomes the primary driver (Durrande-Moreau and Usunier 1999; Kumar et al. 1997). This aligns with mood-congruency theory, which holds that people more readily access mood-consistent information when forming judgments (Forgas 1995; Gardner 1985). By contrast, the psychological-cost model positions emotional association—particularly anger—as the central mechanism influencing satisfaction. Here, anger from perceived time loss exerts a stronger negative effect than cognitive evaluation (Menon and Dube 2004; Miller et al. 2008), consistent with the affect-as-information view that individuals use their emotions as judgment cues (Forgas 1995).
Both models rely on mood-congruency principles: in the service-attribute model, mood shapes evaluations by making mood-consistent information more accessible (Forgas 1995; Gardner 1985), while in the service-cost model, anger similarly biases evaluations downward (Pruyn and Smidts 1998; Taylor 1994). Where they diverge is in the pathway to satisfaction. The service-attribute model emphasizes a memory-based process in which customers recall and apply judgments of wait acceptability, whereas the service-cost model highlights an online emotional process in which anger from time loss more immediately shapes satisfaction.
These differences lead to distinct managerial strategies. Managers may reduce the wait itself through operational improvements (Menon and Dube 2004) or enhance wait acceptability through environmental or informational interventions (Durrande-Moreau and Usunier 1999). This trade-off is evident in walk-in health clinics, where long waits can undermine satisfaction: managers must choose between shortening the wait through process redesign or improving the waiting experience through clearer communication and more comfortable environments.
Toward Conceptual Integration
The psychology of waiting reflects core tradeoffs in how people form expectations, what motivates them, and whether they rely on real‑time experience or memory. Time perception involves both prospective monitoring and retrospective reconstruction (Block and Zakay 1997; Zakay and Block 1997), paralleling expectation‑updating processes in which consumers compare elapsed time with initial estimates and anticipate what remains (Bolton 1998; Hogarth and Einhorn 1992). Service evaluations similarly draw on both predictive (“could be”) and normative (“should be”) standards (Tse and Wilton 1988; Zeithaml et al. 1993). Motivation research shows that people pursue desired outcomes while remaining sensitive to threats under uncertainty or delay (Berkowitz and Harmon‑Jones 2004; Miller et al. 2008), and affective research demonstrates that online emotions and memory‑based judgments jointly shape evaluations (Forgas 1995).
Because the service‑attribute and service‑cost models capture complementary mechanisms, an integrated framework is needed—one that compares their assumptions, highlights their points of convergence and divergence, and clarifies the boundary conditions under which each model applies (Mayer and Sparrowe 2013; Shepherd and Suddaby 2017). To reconcile the theoretical tension between the two models, we examine four potential moderators—wait stage, wait knowledge, wait regret, and wait importance—each aligned with the three research questions discussed earlier. Personal control theory (Averill 1973) provides the overarching logic for selecting these moderators. In stressful situations such as waiting, individuals seek to restore personal control through cognitive (information acquisition and reappraisal), behavioral, or decisional strategies to improve their physical and psychological responses.
These moderators, discussed later, illuminate the theoretical mechanisms that reconcile the service‑attribute and service‑cost models at their key points of divergence. First, they clarify how different roles and forms of expectation—retrospective versus prospective time orientations and “could be” versus “should be” standards—shape customer judgments. Second, they explain the origins of anger by distinguishing appetitive from aversive motivational processes. Third, they differentiate the foundations of service satisfaction by contrasting memory‑based evaluations with online processing strategies. Together, these mechanisms help integrate the two models. Table 1 summarizes the similarities and differences between the service‑attribute and service‑cost models.
Similarities and Differences between Service-Attribute Model and Service-Cost Model.
We validate our conceptual framework through a structural‑equation‑modeling meta‑analysis of 129 studies from 103 articles (672 effect sizes; N = 38,967). Most studies (64%) drew on the psychological‑cost model (Osuna 1985), while 22% relied on the expectancy–disconfirmation model (Oliver 1980). Only 14 percent combined elements of both frameworks. 1 However, these hybrid studies tended to focus on isolated aspects of waiting—such as duration information, queue structure, fairness, regulatory focus, perceived control, or distractors—without attempting to conceptually integrate the psychological‑cost and expectancy–disconfirmation models.
Our study is the first to integrate the expectancy–disconfirmation and psychological-cost perspectives into a unified conceptual framework and to reconcile their theoretical tension by identifying four moderators—wait stage, wait knowledge, wait regret, and wait importance. These meta-analytic findings clarify when subjective time versus disconfirmation drives customer reactions: subjective time shapes psychological cost during the wait but influences wait acceptability after the wait; disconfirmation becomes more diagnostic when wait information strengthens predictive anchors; anger arises primarily from anxiety, amplified by regret; and wait acceptability dominates satisfaction when the wait is important. Complementing the meta-analysis, our experiment shows that high-arousal anger distorts subjective time perception. Together, these insights specify when and why different cognitive and emotional mechanisms govern customers’ wait experiences.
Conceptual Framework: An Integrated Dual-Perspective Model
We advance a conceptual framework that integrates the dual perspectives on customer waits (see Figure 2). Our discussion is organized around the three research questions discussed earlier. For each question, we offer theoretical reconciliation by examining the roles of key moderators that clarify when each model pathway is more influential within the integrated framework. Figure 3 presents the empirical model tested, highlighting these moderating relationships. Table 2 summarizes and defines the key constructs used in the conceptual framework.
Conceptual framework: an integrated “dual-perspective” model of customer wait.
Empirical model: an integrated “dual-perspective” model of customer wait.
Definition of Key Constructs.
Wait Acceptability—Subjective Time Versus Disconfirmation (Research Question 1)
The service‑attribute model conceptualizes wait expectation as a comparison standard used to evaluate the waiting experience. This standard takes the form of a single point estimate representing the “most likely” wait duration (Bolton 1998; Zeithaml et al. 1993). In contrast, the service‑cost model views expectation as an anticipated loss, represented not by a single value but by a probability distribution shaped by prior experiences (Osuna 1985). These differing assumptions about expectation formation lead to distinct time orientations.
Role of Expectation—Retrospective Versus Prospective Time Orientation
The service‑attribute model adopts a retrospective time orientation, assuming customers begin with a point‑estimate expectation and update it by comparing elapsed perceived time with their initial estimate. This memory‑based, backward‑looking process (Block and Zakay 1997; Soman 2003) involves forming and revising a series of discrete expectations, with disconfirmation providing new information as the wait unfolds (Bolton 1998; Hogarth and Einhorn 1992).
In contrast, the service‑cost model reflects a prospective time orientation in which individuals actively monitor the passage of subjective time, treating it as a source of new information, experiencing accumulating stress as waiting continues (Osuna 1985), and anticipating both the remaining duration and the perceived value of lost time (Festjens and Janiszewski 2015; Kumar and Krishnamurthy 2008; Lim et al. 2015; Soman 2003; Zakay and Block 1997).
Reconciling Moderator 1: Wait Stage (During Versus After the Wait)
Personal control theory (Averill 1973) suggests that individuals use cognitive reappraisal to regain control in stressful situations. In the service‑attribute model, expectations function as a point‑estimate comparison standard that guides retrospective reappraisal. In the service‑cost model, reappraisal supports emotion regulation and coping (Folkman 1984), enabling a prospective orientation in which customers interpret subjective time more positively. Thus, cognitive reappraisal can activate both time orientations.
We propose wait stage—whether evaluation occurs during or after the wait—as a key moderator that reconciles these orientations. Wait stage refers to the point in time at which evaluations are made—either during the wait itself or after the wait has ended. “During” captures real‑time feelings and evaluations as the wait unfolds, whereas “after” captures assessments made once service is delivered. This distinction determines whether customers rely on prospective, online processing during the wait or retrospective, memory‑based processing afterward, shaping the diagnostic value of subjective time and disconfirmation.
Subjective time plays different roles across models. In the service‑attribute model, it matters primarily after the wait, serving as an ex post indicator of performance and influencing wait acceptability only once customers retrospectively compare perceived duration with expectations. In the service‑cost model, subjective time is influential both during and after the wait: customers prospectively monitor time, which fuels anxiety about remaining duration, and this anxiety accumulates throughout the wait, amplifying subjective time’s effect after the wait (Osuna 1985).
Disconfirmation affects wait acceptability in the service‑attribute model during and after the wait because customers continually revise expectations, generating a sequence of retrospective contrast effects. These disconfirmations accumulate (Bolton 1998), similar to cumulative negative disconfirmations in service failure and recovery (De Matos et al. 2007; Smith and Bolton 1998), making their total impact stronger after the wait than during the wait. In sum, we posit:
Form of Expectation—Could Be Versus Should Be Expectation
The two models also differ in how they conceptualize wait expectations. The service‑attribute model treats expectations as predictive point estimates, whereas the service‑cost model frames them as normative probability distributions. These assumptions map onto two well‑established comparison standards in the customer expectation literature: “could be” and “should be” expectations (Caruelle et al. 2023; Miller et al. 2008; Tse and Wilton 1988; Zeithaml et al. 1993).
“Could be” expectations are predictive, expressed as point estimates of when the service is likely to occur based on external cues and situational factors. In contrast, “should be” expectations are normative, represented as probability distributions reflecting what customers believe ought to happen, shaped by internal factors such as desires, needs, and prior experiences (Caruelle et al. 2023). Both forms of expectations shape service evaluations by influencing evaluative congruity—the alignment between predictive and normative standards (Szymanski and Henard 2001; Tse and Wilton 1988).
Reconciling Moderator 2: Wait Knowledge (Known Versus Unknown Wait)
According to personal control theory (Averill 1973), individuals also regain control in stressful situations through information gain. In waiting contexts, such information includes expected wait times, queue status (Hui and Tse 1996), social‑environment cues (Grewal et al. 2003), and wait‑time guarantees (Kumar et al. 1997), all of which enhance situational predictability. The service‑attribute model, grounded in point‑estimate expectations, aligns with “could be” expectations, where disconfirmation provides the new information. The service‑cost model, which assumes expectations take the form of a probability distribution, aligns with “should be” expectations, where subjective time supplies the new information.
We propose wait knowledge—whether the wait is known or unknown—as a second moderator that determines which expectation type customers rely on. Waits are classified based on whether customers possess a confident “could be” expectation supplied by the service provider. Known waits provide clear, updated information that supports confident predictive expectations; unknown waits lack such information, prompting reliance on internal, normative standards. Using confidence in the “could be” expectation as the criterion reflects the core distinction between predictive and normative expectations. Wait knowledge, therefore, shapes the diagnostic value of subjective time and disconfirmation.
“Should be” expectations are relatively stable because they stem from enduring needs, desires, and prior experiences, whereas “could be” expectations are context‑dependent and shift with situational cues (Zeithaml et al. 1993). Because predictive wait estimates are more variable, confidence is most meaningfully captured in customers’ confidence in their “could be” expectation (Spreng and Page 2001). Low confidence heightens uncertainty, which in the service‑cost model intensifies anxiety about anticipated time loss. Thus, customers facing unknown waits experience greater ambiguity and stronger anxiety as time progresses.
Wait knowledge also determines which diagnostic cue customers rely on when evaluating waits. In unknown waits, low confidence in the “could be” expectation pushes customers to rely more heavily on their “should be” expectation, making subjective time the primary basis for evaluating wait acceptability. In known waits, confident predictive expectations make disconfirmation the key evaluative cue when actual wait time exceeds expectations. In sum, we posit:
Elicitation of Anger—Wait Acceptability Versus Anxiety (Research Question 2)
From the service‑attribute perspective, unacceptable waits signal a violation of the service provider’s implicit promise to deliver prompt service (Taylor 1994), and this perceived transgression evokes anger. From the service‑cost perspective, anger arises as a downstream reaction to anxiety: whereas anxiety reflects the stress of uncertainty about anticipated time loss, anger represents the emotional response to the actual loss incurred (Menon and Dube 2004). As noted earlier, both models draw on Weiner’s (1985) appraisal–attribution theory to explain how anger is triggered during waiting experiences.
Source of Anger—Appetitive Versus Aversive Motivation
The service‑attribute model explains anger as stemming from blocked goal attainment, reflecting an appetitive motivational system in which unmet needs and low wait acceptability trigger frustration. In contrast, the service‑cost model adopts an aversive perspective, emphasizing how anxiety about anticipated time loss intensifies into anger once that loss is realized (Miller et al. 2008). These appetitive and aversive systems structure behavior: the former drives approach toward desired goals, the latter avoidance of threat (Carver and Harmon‑Jones 2009; Miller et al. 2008). Accordingly, anger serves as an internal signal of either goal obstruction or emerging threat (Berkowitz and Harmon‑Jones 2004). Thus, wait (un)acceptability and anxiety represent distinct motivational orientations through which customers interpret and respond to waiting.
Reconciling Moderator 3: Wait Regret (With Versus Without Regret)
According to personal control theory (Averill 1973), individuals cope with stressful situations not only through cognitive control but also through behavioral control—efforts to prevent, shorten, or alter the stressor. In waiting contexts, behavioral control reflects customers’ ability (or missed opportunity) to avoid the wait, making wait regret a key moderator. Regret, as a comparison-based and self-blaming emotion, occurs when individuals recognize that a better outcome was possible had they chosen differently (Zeelenberg and Pieters 2007). Rather than being an outcome of waiting, wait regret functions as a moderator that arises only when individuals reflect on missed opportunities to avoid the wait. As an aversive emotion, it heightens sensitivity to negative outcomes and strengthens the anxiety‑to‑anger link within the service‑cost pathway.
Regret also threatens self‑image by calling past decisions into question, prompting customers to cope either by shifting blame to the provider or by downplaying the wait (Zeelenberg and Pieters 2007). In waiting contexts, customers often attribute time loss to the provider; under regret, they may exaggerate this transgression by inflating the perceived value of lost time. Because the opportunity cost of time is inherently ambiguous (Okada and Hoch 2004; Soman 2001), regret facilitates counterfactual inflation of time loss—for example, perceiving other lines as moving faster in multi‑line queues (Kumar and Dada 2021)—thereby amplifying the effect of subjective time on anger. Conversely, when the delay is seen as reasonable or unavoidable, customers may minimize the wait to protect self‑esteem, weakening the influence of wait acceptability on anger. In short, we posit:
Service Satisfaction—Wait Acceptability Versus Anger (Research Question 3)
From the service‑attribute perspective, both cognitive evaluations and affective reactions shape service satisfaction, consistent with the expectancy–disconfirmation framework (Oliver 1980, 2010). From the service‑cost perspective, negative emotions bias customers toward lower satisfaction, aligning with the mood‑congruence bias whereby affective states color subsequent judgments (Forgas 1995; Gardner 1985).
Source of Service Satisfaction—Memory-Based Versus Online Processing
The mood‑congruency model proposes two routes through which affect shapes evaluation: a direct route, where current mood influences judgments, and an indirect route, where mood‑consistent memories are more easily retrieved (Forgas 1995; Gardner 1985). Applied to service waits, these routes map onto distinct satisfaction antecedents. Anger during the wait reflects the direct, online emotional pathway, whereas perceptions of wait acceptability engage the indirect, memory‑based pathway. Accordingly, the service‑attribute model aligns with memory‑based processing, while the service‑cost model corresponds to online emotional processing.
Reconciling Moderator 4: Wait Importance (Important Versus Unimportant Wait)
According to personal control theory (Averill 1973), individuals also cope with stress through decisional control—the ability to choose among options during the stressful event. In waiting contexts, this reflects whether customers can remain in the queue, leave, or return later. Unlike regret, which concerns missed past opportunities, the opt‑out option concerns present‑moment control. The feasibility of opting out depends on wait importance: when the wait is critical, and alternatives are limited, decisional control is effectively constrained. We propose wait importance (important vs. unimportant) as a moderator that shapes the diagnostic value of anger versus wait acceptability.
Wait importance reflects the strength of customers’ motivation to obtain the service, rooted in the desirability of their goals (Giebelhausen et al. 2011; Sheth et al. 1991). High importance prompts deeper cognitive processing (Craik and Lockhart 1972), leading customers to rationalize the wait and form stronger memory traces—such as perceptions of wait acceptability—that later guide evaluations. Unimportant waits elicit shallower processing and weaker memories. Even when customers feel “trapped” due to limited alternatives (Averill 1973; Hui and Tse 1996), high importance sustains goal focus, enhances tolerance for delays, and therefore reduces anger’s spillover into satisfaction (Giebelhausen et al. 2011).
High importance also suppresses heuristic, emotion‑driven judgments. Whereas customers may normally rely on anger as a shortcut in evaluating service quality (Forgas 1995), important waits trigger more deliberate processing in which customers weigh the wait’s cost against the service’s value, weakening the anger–satisfaction link. Moreover, when the service is essential, customers discount alternative uses of time (Caruelle et al. 2023; Festjens and Janiszewski 2015), reframing the inability to opt out as a necessary trade‑off rather than a loss. This reduces anxiety about wasted time and further insulates satisfaction from anger. Thus, we posit:
Methodology
We conducted a citation‑based search to identify studies on customer service waits and service satisfaction, beginning with twenty-three seed articles from the literature that produced over 8,000 citation results. Keyword search was impractical, yielding more than 50,000 entries for “waiting service” in ABI/INFORM Collection. After removing duplicates and irrelevant entries, roughly 4,000 results remained. Topic screening reduced this set to about 900 articles, and full‑text review ultimately identified 129 recordable studies across 103 articles. For details on data collection, seed paper summary with references, and a full list of references for papers included in the meta-analysis, please refer to Web Appendix 1.
Study coding involved (1) extracting reported data from each study and (2) classifying studies according to proposed moderators, using definitions and guidelines provided in Web Appendix 2. Correlation coefficients served as the primary effect size metric; other effect size metrics (e.g., d, t-value, and covariance matrix) were converted when possible. When multiple effect sizes existed for the same relationship link, we applied sample‑size‑weighted averaging of Fisher’s Z‑transformed values. 2 We also contacted authors for supplemental data when relevant studies lacked effect‑size information. In total, 672 effect sizes were recorded from 129 studies (N = 38,967).
Two independent coders classified all studies following detailed instructions. Ambiguous cases were judged contextually or left unclassified when necessary. Intercoder reliability was high (95.4% agreement; Cohen’s kappa = .905). Full coding details including moderator coding appear in Web Appendix 2.
Meta-Analysis and Results
Bivariate Association
We conducted univariate analyses of all pairwise correlations among the constructs using a random-effects approach. For each construct pair, we report the sample‑weighted average correlation (
To assess potential publication bias, we report File Drawer N (Rosenthal 1979)—the estimated number of null effect studies required to reduce the observed correlation below .05 level—as well as Begg’s test z value. We also provide Q-statistics of homogeneity. Table 3 summarizes these results. Web Appendix 3 (Table 1 and Figure 1) reports additional tests of publication bias (e.g., Egger’s, Begg’s, trim-and-fill analyses) including funnel plots. Overall, the results indicate that publication bias is not a concern and that the estimated effects are stable and robust.
Pairwise Analysis.
The univariate analysis shows that wait‑time constructs strongly influence wait outcomes and that wait‑process constructs play meaningful mediating roles, with correlations ranging from .137 to .708. Most of these effects appear robust with regard to the number of null studies needed to render the observed effects insignificant (mean file-drawer N is 176). Begg’s test revealed no significant publication bias, with all z values falling between −1.78 and 1.78. All Q‑statistics of homogeneity were significant, suggesting the presence of potential moderators across these relationships.
Two-Stage SEM
Using two‑stage structural equation modeling (TSSEM; Cheung and Chan 2005), we validated the paths in the integrated model based on summarized data. TSSEM is a multivariate meta‑analytic SEM technique consisting of two stages: (1) deriving a pooled correlation matrix via multi‑group SEM, and (2) fitting this pooled matrix to the structural model using weighted least squares. This approach yields unbiased parameter estimates and has been shown to outperform alternatives such as generalized least squares (Cheung and Chan 2005).
We conducted the TSSEM in R using a random‑effects model to address missing data and study heterogeneity variance across studies. Weighted mean correlation matrices were used to compute the asymptotic covariance matrix, and substantial heterogeneity across correlations justified the random‑effects approach. Table 4 presents the pooled correlation matrix.
Pooled Correlation Matrix.
Lower Triangle is the estimated correlations. Upper Triangle is the heterogeneity index (I2).
Model Fit Results
We first evaluated the service‑attribute and service‑cost models. Model fit indices were acceptable for both: for the service‑attribute model, χ²/df = 8.15, RMSEA = .0177, CFI = .9656, NFI = .9622; for the service‑cost model, χ²/df = 9.13, RMSEA = .0243, CFI = .9351, NFI = .9320. In the service‑attribute model, subjective time and disconfirmation were strongly correlated (β = .625, p < .001), and both significantly reduced wait acceptability (β = –.242 and –.379, p < .001, respectively). Wait acceptability decreased anger (β = –.544, p < .001) and increased service satisfaction (β = .405, p < .001), while anger negatively affected satisfaction (β = –.143, p < .001). In the service‑cost model, subjective time increased anxiety (β = .235, p < .001) and anger (β = .317, p < .001). Anxiety also increased anger (β = .448, p < .001), and anger reduced service satisfaction (β = –.388, p < .001).
We then tested the integrated dual‑perspective model, which showed strong fit (χ²/df = 6.292, RMSEA = .0135, CFI = .9744, NFI = .9700). 3 Subjective time significantly increased anxiety (β = .293, p < .001) and reduced wait acceptability (β = –.190, p < .01), but did not directly affect anger (β = .082, p > .05). Disconfirmation again reduced wait acceptability (β = –.439, p < .001). Both anxiety and wait acceptability significantly influenced anger (β = .458 and –.319, p < .001), and both wait acceptability and anger significantly affected service satisfaction (β = .359, p < .001; β = –.248, p < .01).
All hypothesized paths were supported except the direct link from subjective time to anger (β = .082, CI [–0.03, 0.18]), indicating full mediation by anxiety and wait acceptability (Zhao et al. 2010). Bootstrapped indirect effects confirmed significant mediation through both anxiety (indirect = .134, CI [0.094, 0.184]) and wait acceptability (indirect = .084, CI [0.039, 0.123]). These results show that the two perspectives are complementary in that anger arises from both perceived time‑loss threat (anxiety) and judgments of unacceptable waiting (wait acceptability). Finally, service satisfaction is jointly shaped by wait acceptability and anger (β = .359, CI [0.289, 0.426]; β = –.248, CI [–0.323, –0.168]).
Moderator Analysis
To assess the effects of the proposed moderators, we conducted subgroup analyses following Jak and Cheung (2018). This approach applies TSSEM separately to two moderator‑split subgroups and then tests whether model parameters differ significantly between them. Heterogeneity is evaluated using chi‑square difference test comparing an unconstrained model with a constrained model.
We examined moderator effects at three levels: (a) the overall model, (b) each research question, and (c) each focal construct. The model‑level constrained model assumes all parameters are equal across subgroups. At the research‑question level, each of the three constrained models (for three research questions) restricts all parameters associated with one specific research question to equality. At the construct level, six constrained models each impose equality constraints on all parameters related to a specific construct. This hierarchical approach allows us to determine whether a moderator influences the model globally, affects specific research questions, or operates at the construct level. Results are summarized in Table 5, and the pooled correlation matrices for all subgroups appear in Web Appendix 4 (Tables 1–8). 4
Moderator Testing Using Subgroup Analysis in the Integrated Dual-Perspective Model.
p < .001. **p < .01. *p < .05.
The model level comparisons indicate that all four proposed moderators have significant impacts on the integrated model (Δχ2 = 36.79, 50.39, 36.78, and 29.54, all Δdf = 9, and all p < .01). Specifically, the research question level examination suggests that the diagnostic measure (Research Question 1) varies between studies conducted “during” and “after” the wait and between “known” and “unknown” waits (Δχ2 = 7.78 and 36.05, both Δdf = 3 and p < .05); that the source of anger (Research Question 2) is impacted by “regret in wait” (Δχ2 = 14.31, Δdf = 3, and p < .01); and that the source of service satisfaction (Research Question 3) is impacted by whether the service is “important” or “unimportant” (Δχ2 = 7.25, Δdf = 2, and p < .05).
More specifically, construct-level contrasts reveal the impacts of the proposed moderators at the hypothesis level.
First, the impact of subjective time on anxiety (β = .200 and .343) is significant both “during” and “after” the wait, respectively, in the service-cost component of the integrated model. Also, the impact “after” the wait is significantly larger than the impact “during” the wait (Δχ2 = 7.72, Δdf = 1, and p < .01), validating H1a. However, wait acceptability is influenced differently “during” versus “after” the wait. In “during” the wait, disconfirmation is the only source to influence wait acceptability (β = −.384, p < .05) but not subjective time (β = −.217, p > .05). Conversely, both subjective time and disconfirmation have significant influences on wait acceptability “after” the wait (β = −.195 and −.418, both p < .001). These results are aligned with the hypothesis about the roles of subjective time (H1b supported): its non-significant impact “during” the wait implies the influence of subjective time on wait acceptability is ex post facto. Finally, the results partially supported our hypothesis about disconfirmation (H1c): disconfirmation has a significant impact in both “during” and “after” the wait studies, but its impact is not significantly larger in “after” the wait studies.
Second, customers in “known” and “unknown” waits have different approaches in the formation of wait acceptability (Δχ2 = 20.67, Δdf = 2, and p < .01). Customers in “unknown” waits are easier to get anxious about the passage of wait time, compared to those in “known” waits (β = .500 vs. .276, Δχ2 = 12.88, Δdf = 1, and p < .001), supporting H2a. Also, in “unknown” waits, customers use both subjective time and disconfirmation to determine wait acceptability (β = −.490 and −.238, p < .001 and .01). As hypothesized, here the primary driver is subjective time.5 This contrast confirms our hypotheses about the use of “should be” expectation. In known waits, the effect of subjective time is insignificant (β = .010, p > .05). This supports H2b. Further, in “known” waits, customers rely on disconfirmation solely to determine the wait acceptability (β = −.575, p < .001) and this effect is magnified when compared to unknown waits (β = −.238, p < .01). This result confirms our hypothesis about the use of “could be” expectation, supporting H2c.
Third, customers “with regret” and those “without regret” weigh time loss differently (Δχ2 = 14.31, Δdf = 3, and p < .01). Customers “with regret” get more angry from the time loss in the wait than those “without regret” (Δχ2 = 10.03, Δdf = 2, and p < .01). Particularly, such a stronger impact is achieved primarily via the anxiety about wait time (β = .571 vs. .336, Δχ2 = 8.59, Δdf = 1, and p < .01), supporting H3a. However, in either condition, subjective time has no significant impact on anger, rejecting H3b. Finally, customers “with” and “without” regret use wait acceptability similarly to influence anger (β = −.316 vs. −.215, Δχ2 = .919, Δdf = 1, and p > .05), rejecting H3c. These results partially confirm that regret exerts extra weight on the aversive motivational system.
Fourth, the “importance” of the service leads to different levels of use of memory retrieval in the service assessment (Δχ2 = 4.26, Δdf = 1, and p < .01): wait acceptability (i.e., memory retrieval) has a more significant influence on service satisfaction when service is important (β = .447 vs. .301). Also, customers tend to use less emotion association in their service satisfaction evaluation when service is important (Δχ2 = 7.12, Δdf = 1, and p < .01), as indicated by the difference in the anger-satisfaction link (β = −.109ns vs. −.322). These results are compatible with our hypotheses about divergent utilization of emotion association and memory retrieval (H4a and H4b supported) 6 . The findings are summarized in Table 6.
Summary of Findings, Insights, and Implications.
Experimental Study
Our meta‑analytic test of the dual‑perspective wait model treats subjective time as exogenous—shaping anxiety and anger but not shaped by them. Yet high‑arousal negative emotions are known to lengthen perceived time (Gil and Droit-Volet, 2012), and internal‑clock models attribute this distortion to additive attentional and multiplicative arousal mechanisms (Gibbon et al. 1984). This study, therefore, tests whether high‑arousal anger inflates subjective wait time in a service‑delay setting.
Method
We conducted a 2 (Attention Toward Wait: Low vs. High) × 2 (Anger: Low vs. High) between‑subjects experiment. We recruited 600 participants via Prolific (53.5% female; M age = 43.8) who read a scenario in which they imagined waiting at the DMV to renew their driver’s license. All participants received the same introductory common scenario description before being randomly assigned to one of four conditions with condition-specific scenarios.
Attention toward the wait was manipulated by varying the quality of phone-based distraction. In the low‑attention condition, participants viewed engaging social media content; in the high‑attention condition, the content was uninteresting and quickly became boring. Anger was manipulated by varying the perceived cause of slow progress in the renewal line. In the low‑anger condition, delays were attributed to legitimate procedural demands and staff working on other tasks; in the high‑anger condition, delays stemmed from an incompetent officer and other staff visibly neglecting customers.
Subjective wait time was assessed with three slider items (1–100; e.g., short–long, minutes–hours, flies by–drags on). Anger was measured with four items: (angry/irritated/frustrated/annoyed). Participants also rated the DMV’s responsibility and blame for the wait. All manipulation checks were confirmed. Web Appendix 6 provides the scenarios, measures, manipulation, and detailed results including figure plots.
Results
A 2 (Attention Toward Wait: Low vs. High) × 2 (Anger: Low vs. High) ANOVA on subjective time revealed a significant main effect of attention, F(1, 596) = 17.35, p < .001, ηₚ² = .028 such that participants reported longer subjective waits when their attention was directed toward the queue rather than distracted. The main effect of anger was not significant, F(1, 596) = .12, p = .734, ηₚ² < .001 nor was the interaction, F(1, 596) = 2.13, p = .145, ηₚ² = .004.
Moderated Regression: Treating Anger as Continuous
We estimated a moderated regression treating anger as a continuous moderator and retaining the two‑level attention manipulation. Subjective time served as the dependent variable. The model was significant, F(3, 596) = 56.41, p < .001. Anger (β = .32, p < .001), attention (β = –.36, p < .001), and their interaction (β = .20, p < .01) were all significant.
Consistent with the ANOVA, greater attention to the wait (vs. phone) increased subjective time. The interaction indicates that anger amplifies this effect: at low anger (1–2 on the scale), attention condition strongly differentiates subjective time, whereas at higher anger levels the gap between attention conditions narrows. At very high anger (6–7), anger itself exerts the strongest influence on subjective time.
In sum, although the Attention × Anger interaction was not significant in the ANOVA, treating anger as a continuous variable revealed a clearer pattern. The moderated regression showed that, in addition to the additive attention effect, anger exerted a multiplicative influence on subjective time, consistent with internal clock predictions.
General Discussion
Conceptual Integration
Our contingency-based dual-perspective framework integrates expectancy–disconfirmation and psychological-cost theories, showing that their influence varies with four moderators—wait stage, wait knowledge, wait regret, and wait importance. By specifying when each pathway becomes more diagnostic, the framework resolves prior conceptual tensions and, consistent with MacInnis’s (2011) call for boundary-conditioned theorizing, offers a unified lens for understanding how cognitive and emotional mechanisms jointly shape the waiting experience.
Our meta‑analytic results show that the service‑attribute and service‑cost models function as complementary rather than competing explanations of customers’ reactions to waits. This dual‑process perspective is consistent with research demonstrating that people rely on both prospective and retrospective time‑perception mechanisms (Block and Zakay 1997; Zakay and Block 1997) and update expectations by comparing elapsed time with initial estimates while projecting what remains (Bolton 1998; Hogarth and Einhorn 1992). Consumers also draw simultaneously on predictive (“could‑be”) and normative (“should‑be”) expectations when evaluating service performance. Both types of standards independently shape satisfaction judgments, as shown in expectation‑formation models and empirical evidence on dual benchmarks (Caruelle et al. 2023; Miller et al. 2008; Tse and Wilton 1988; Zeithaml et al. 1993).
Motivational processes likewise operate in parallel. Appetitive and aversive systems jointly regulate responses to waits: the aversive system signals loss or wasted time, generating anger, while the appetitive system remains engaged because the service goal is still active. Research on goal conflict confirms that approach‑ and avoidance‑oriented motivations frequently co‑occur under uncertainty or delay (Berkowitz and Harmon‑Jones 2004; Miller et al. 2008). Finally, affective processing involves both online emotions and memory‑based judgments. Current mood shapes real‑time evaluations and biases the retrieval of expectations and norms (Forgas 1995), meaning that anger and acceptability assessments jointly inform satisfaction.
The conceptual integration also mirrors the broader distinction in the time-versus-money literature: the service-attribute model reflects the experiential accumulation of time use (Mogilner and Aaker 2009), whereas the service-cost model reflects the evaluative ambiguity of time as a resource that is lost and cannot be stored or replenished (Okada and Hoch 2004; Soman 2001). Service waits uniquely require customers to engage both experiential and evaluative modes of processing, making them an ideal context for bridging these theoretical domains. Together, these literatures reinforce our central contribution: customers’ reactions to waiting emerge from co‑active cognitive, motivational, and affective mechanisms rather than single‑process explanations.
We further advance wait research by situating the service‑attribute and service‑cost models within a unified framework grounded in personal control theory (Averill 1973). Cognitive, behavioral, and decisional control organize how customers interpret and respond to waits, linking mechanisms identified across prior work. Cognitive control clarifies how time orientations shape expectations and emotion; information gain enables point-estimate expectation and disconfirmation; behavioral and decisional control explain how regret and opt‑out feasibility shape reactions to delays. By showing that personal control integrates time orientation, expectation type, motivational direction, and information‑processing style, we offer an overarching foundation that reconciles the two models and specifies when each mechanism is most influential.
In sum, our conceptual integration highlights a coherent system of interdependent mechanisms that shape how customers experience, evaluate, and emotionally respond to waiting. Its validity, however, depends on the efficacy of the moderators.
Theoretical Insights: Moderators
Our framework shows that each theoretical lens becomes diagnostic only under specific conditions. For the first research question—whether disconfirmation or subjective time drives wait (un)acceptability—the critical moderator is the wait stage. Subjective time functions as a flow variable during the wait, influencing psychological cost but not acceptability (Lim et al. 2015; Soman 2003), and becomes a stock variable after the wait, strongly shaping wait acceptability alongside disconfirmation (Zhou and Soman 2003). Disconfirmation, however, affects acceptability in both stages, especially when customers reduce dissonance by adjusting expectations rather than subjective time (Tse and Wilton 1988).
These effects also depend on wait knowledge. Distinguishing predictive (“could be”) from normative (“should be”) expectations, we show that customers in unknown waits—lacking predictive anchors—rely more on subjective time. In known waits, predictive expectations guide judgments, making disconfirmation more diagnostic. Providing wait information reduces uncertainty and weakens subjective time’s influence (Grewal et al. 2003; Hui and Tse 1996) but simultaneously strengthens predictive anchors, increasing the impact of disconfirmation and mitigating one dissatisfaction pathway while amplifying another.
Our second research question—whether anger arises more from anxiety or from wait (un)acceptability—depends partly on wait regret. Drawing on motivational direction theory (Carver and Harmon-Jones 2009), we show that regret, reflecting self-attribution for poor prior decisions threatening self-esteem (Averill 1973), intensifies anger during waits. Regret acts as an emotional amplifier even though subjective time does not directly predict anger (H3b), likely because service-attribute and service-cost pathways offset its direct effect. 7 The absence of support for H3c may reflect customers’ tendency to blame the provider when waits become unacceptable. We extend motivational direction theory by demonstrating how regret shapes the translation of anxiety into anger.
Our third research question—whether satisfaction is driven more by wait (un)acceptability or by anger—depends on wait importance. Integrating service-attribute and service-cost models, we show that when waits are important and decisional control is low (e.g., healthcare), customers engage in deeper, goal-directed processing (Craik and Lockhart 1972), rationalizing delays and weighting wait acceptability more heavily (Sheth et al. 1991). In these contexts, anger becomes less predictive of satisfaction because customers suppress transient frustration to align with higher-order goals (Festjens and Janiszewski 2015). In less important waits, shallow processing dominates, and anger directly shapes satisfaction. This boundary condition clarifies when cognitive versus affective mechanisms drive service satisfaction.
The experiment showed that directing attention to the wait reliably lengthened subjective time, consistent with resource-allocation accounts (Zakay 1989) and internal-clock models in which attention produces an additive expansion of perceived duration (Gibbon et al. 1984). Moderated regression further revealed that anger imposed a multiplicative distortion: as anger increased, differences between attention conditions narrowed, and at high arousal, anger itself dominated time perception. Thus, anger can reverse-causally shape subjective time, but its influence is intensity-dependent—attention drives distortion under low arousal, whereas anger overrides attention under high arousal.
Osuna’s (1985) psychological-cost model assumes that subjective waiting time is an exogenous input to stress, and later mathematical extensions retain this one-way structure by modeling stress solely from waiting-time distributions rather than emotional arousal (Denuit and Genest 2001; Suck and Holling 1997). This implicitly presumes low-arousal negative affect, since only mild emotions can be represented as smooth, time-driven accumulations without feeding back into time perception (Gzyl and Osuna 2013). However, internal-clock theories (Gibbon et al. 1984; Zakay and Block 1997) and emotion–time research (Gil and Droit-Volet 2012) show that high-arousal negative emotions lengthen perceived time, contradicting the exogeneity assumption and revealing an unacknowledged low-arousal boundary condition in the psychological-cost model—and thus in our conceptual framework.
Managerial Implications
First, customers view waits through two shifting lenses—the time they lose and the experience of waiting—so managers must consider how their strategies shape that focus. Virtual queues (e.g., Waitwhile, Queue-it, QLess, etc.) reduce the salience of waiting by letting customers join remotely, but notification needs to be used carefully. Occasional updates are helpful, whereas frequent notifications can draw attention back to the wait and weaken the benefits of virtual queuing.
Second, wait‑knowledge conditions call for different managerial approaches. In known waits, customers depend on disconfirmation, so early, accurate estimates are crucial, while frequent revisions heighten retrospective evaluation and undermine satisfaction. When accurate predictions are difficult, as in settings like the DMV, distraction tactics become more effective. In unknown waits, customers rely on subjective time, making environmental design and engaging activities essential for easing anxiety, shaping emotions, and increasing willingness to wait.
Third, motivational direction and regret strongly shape customers’ emotional reactions. Aversive motivation heightens anger more than appetitive motivation, and regret from perceived poor choices amplifies this anger further. Technologies like virtual queues or paid priority access can unintentionally trigger regret among those excluded. More equitable or transparent designs—such as clearer explanations of capacity limits and real‑time indicators that standby lines are moving—can help reduce these emotional costs. For instance, Disneyland could explain how Genie+ and Lightning Lane capacity is capped to protect standby waits and provide real‑time signals that standby lines are progressing predictably.
Fourth, in high‑importance waits, customers rely less on emotion and more on how the wait is managed. In healthcare, for example, patients judge care quality less by frustration and more by transparency, empathy, and support. Proactive updates, sincere apologies, and helpful distractions increase perceived control and make waits more acceptable. Although reducing wait times should always remain the priority, resource limits often prevent major improvements despite advanced optimization methods. When that happens, managing the wait experience becomes the most effective way to maintain satisfaction.
Future Research
We acknowledge that our meta‑analytic study relied on correlational rather than causal evidence, limiting our ability to capture the dynamic effects of heightened anger on time distortion. Osuna’s (1985) psychological‑cost model—focused on waiting‑time distributions rather than emotional arousal—should be complemented by internal‑clock theories (Gibbon et al. 1984; Zakay and Block 1997) and emotion‑time research (Gil and Droit-Volet, 2012). Building on our experimental findings, future work could examine how subjective time expands as anger intensifies, using a non‑recursive feedback‑loop framework to better represent its dynamic nature.
Second, research should investigate how primacy and recency effects shape subjective time and disconfirmation across prospective versus retrospective time orientations (Lim et al. 2015; Soman 2003). Because our dual-perspective model shows that post-wait acceptability integrates both orientations, understanding how speed patterns (e.g., fast–slow–fast) and sequential expectation updates differentially weight early versus late segments of the wait could inform optimal service design. This includes testing whether later disconfirmations exert disproportionate influence in multi-stage waits.
Third, the interplay between temporal versus nontemporal information and time orientation warrants deeper examination. Memory-based models predict longer duration estimates with greater informational complexity, whereas attention-based models predict the opposite (Zakay 1989). Future research should clarify how these mechanisms operate under prospective versus retrospective estimation modes, and how information design can strategically shape perceived duration.
Fourth, the dual-perspective model should be extended to online waiting contexts, where attentional resources and temporal motives differ from physical queues. Evidence suggests that visual content can either lengthen or shorten perceived waits depending on wait duration (Hong et al. 2013). Understanding when individuals shift from retrospective to prospective time processing online—and how digital interface design influences this shift—can reconcile prior inconsistencies in the literature.
Fifth, subjective time perception may also be shaped by goal conflict. Stress and anxiety arising from competing goals (Etkin, Evangelidis, and Aaker 2015) can compress or expand perceived time even when goals do not directly compete for temporal resources. Future research should examine how goal conflict during waits interacts with emotional states and time orientation to influence subjective duration and wait acceptability.
These directions could extend our dual‑perspective framework encouraging further theory building and boundary-spanning research on the psychology of waiting.
Supplemental Material
sj-DOC-1-jsr-10.1177_10946705261453760 – Supplemental material for Dual Perspectives of Customer Wait Time Perception: A Conceptual Framework and Meta-Analysis
Supplemental material, sj-DOC-1-jsr-10.1177_10946705261453760 for Dual Perspectives of Customer Wait Time Perception: A Conceptual Framework and Meta-Analysis by Yizhe Lin and James Agarwal in Journal of Service Research
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
