Smartphone withdrawal creates stress: A moderated mediation model of nomophobia, social threat, and phone withdrawal context (2018)

Tams, Stefan, Renaud Legoux, and Pierre-Majorique Léger.

Computers in Human Behavior 81 (2018): 1-9.

https://doi.org/10.1016/j.chb.2017.11.026

Highlights

Focus on Nomophobia, an important phenomenon that we need to better understand.
Explicating how and why Nomophobia influences stress (mediation).
Explaining under what conditions Nomophobia leads to stress (moderation).
Taking a theory-driven approach to study Nomophobia (demand-control-person model).

Abstract

A growing body of literature demonstrates that smartphone use can become problematic when individuals develop a technology dependency such that fear can result. This fear is often referred to as Nomophobia, denoting the fear of not being able to use one’s phone. While the literature (especially on technostress and problematic smartphone use) has shed ample light on the question of which factors contribute to the development of Nomophobia, it remains less clear how, why, and under what conditions Nomophobia, in turn, results in negative consequences, especially stress. Drawing on the demand-control-person model, this study develops a novel research model indicating that Nomophobia impacts stress through the perception of a social threat and that this indirect effect depends on the context of a phone withdrawal situation. Data collected from 270 smartphone users and analyzed using multi-group path analysis supported our model. The results showed that the proposed indirect effect is non-significant only when situational certainty and controllability come together, that is, when people know for how long they will not be able to use their phones and when they have control over the situation. Managers can help their nomophobic employees by instilling in them trust and perceptions of social presence while also giving them more control over their smartphone use during meetings.

1. Introduction

A growing trend in corporate environments is to require employees to leave their communication devices, especially smartphones, outside the meeting room (Forbes, 2014). This well-intended policy is often meant to create more productive and respectful work contexts in which employees are not constantly distracted by technological interruptions (e.g., checking and writing e-mails via smartphones). However, we argue in this article that such a policy may have unintended consequences for employees and organizations alike because smartphone withdrawal may create a new social phobia: Nomophobia or the fear of not being able to use one’s smartphone and the services it offers (Kang & Jung, 2014; King, Valença, & Nardi, 2010a, 2010b; King et al., 2013; Park, Kim, Shon, & Shim, 2013). Nomophobia is a modern phobia related to the loss of access to information, the loss of connectedness, and the loss of communication abilities (King et al., 2013, 2014; Yildirim & Correia, 2015). Nomophobia is situation-specific such that it is evoked by situations that engender the unavailability of one’s smartphone (Yildirim & Correia, 2015).

As a situation-specific phobia, Nomophobia has recently been suggested to lead to strong perceptions of anxiety and distress (Cheever, Rosen, Carrier, & Chavez, 2014; Choy, Fyer, & Lipsitz, 2007; Yildirim & Correia, 2015). In fact, some suggested that Nomophobia could be so stressful that it warrants to be considered a psychopathology (Bragazzi & Del Puente, 2014). Recent empirical research supported this idea, indicating that nomophobic individuals suffer from stress when their smartphones are out of reach (Samaha & Hawi, 2016). Stress, in turn, has various negative consequences for individuals and organizations, including reduced well-being, acute and chronic health problems, as well as diminished organizational productivity (Ayyagari, Grover, & Purvis, 2011; Lazarus & Folkman, 1984; Lazarus, 1999; Riedl, Kindermann, Auinger, & Javor, 2012; Tams, Hill, de Guinea, Thatcher, & Grover, 2014). Hence, stress is an important dependent variable to study in the context of Nomophobia.

Yet, while recent research offers clear and comprehensive explanations of how Nomophobia develops (Bragazzi & Del Puente, 2014; Hadlington, 2015; King, Valença, & Nardi, 2010a, 2010b; King et al., 2014; Sharma, Sharma, Sharma, & Wavare, 2015; Smetaniuk, 2014; Yildirim & Correia, 2015), it remains unclear how, why, and when (i.e., under what conditions) Nomophobia, in turn, leads to stress. Absent understanding of the mechanisms connecting Nomophobia to stress, research can offer only limited practical guidance to individuals as well as to health-care practitioners and managers on how to develop intervention strategies (MacKinnon & Luecken, 2008). To more fully understand the implications of Nomophobia for stress and to offer enhanced practical guidance, research must generate more detailed and specific explanations of intervening and contextual factors. First, research must generate more comprehensive explanations of the causal pathways involved in the process by which Nomophobia-related impacts unfold (i.e., mediation).¹ Second, it has to shed light on the contextual factors on which Nomophobia-related impacts depend (i.e., moderation). In other words, research needs to generate explanations of factors that carry the influence of Nomophobia on to stress (mediation) and of contextual factors on which this influence depends (moderation). Consequently, the present study begins to open the black box of the interdependencies between Nomophobia and other factors that explain in greater detail how and why Nomophobia can lead to stress (mediation) and when or under what conditions the stress-related effects of Nomophobia crystallize (moderation).

To understand the effect of Nomophobia on stress in greater detail, we draw on the demand-control-person model developed by Bakker and Leiter (2008) as well as Rubino, Perry, Milam, Spitzmueller, and Zapf (2012). This theoretical framework is an extension of Karasek (1979) demand-control model, one of the most important theories of stress (Siegrist, 1996). The demand-control-person model can provide a theoretical explanation for the negative impacts of Nomophobia on stress in a context where phobic traits of the individual (Nomophobia) are exacerbated by stressful demands, particularly uncertainty, and by a lack of management interventions in terms of providing control. The model further suggests that stressors, such as a nomophobic personality facing a phone withdrawal situation, lead to stress by threatening other valued resources (e.g., social esteem, social acceptance, or social respect). Using this model, we examine whether the impact of Nomophobia on stress is mediated by social threat and whether this indirect effect varies under different conditions of uncertainty and control, which are important work conditions in contemporary organizational arrangements (Galluch, Grover, & Thatcher, 2015).

By investigating interdependencies between Nomophobia, social threat, uncertainty, and control in the prediction of stress, this study makes important contributions. Perhaps most importantly, the study helps research on Nomophobia progress toward more detailed and specific explanations of the process by which Nomophobia results in stress (we find that Nomophobia leads to stress by generating a perceived social threat). Furthermore, the study establishes certain work conditions (uncertainty and control) as contextual factors on which the negative impacts of Nomophobia depend. Overall, this study yields an enriched explanation and prediction of how, why, and when Nomophobia leads to stress.

The paper proceeds as follows. The next section provides a background on the study context as a means to frame an integrative research model of Nomophobia, stress, as well as relevant mediating and moderating factors. This integrative model hypothesizes that Nomophobia leads to stress via a perceived social threat and that this indirect effect is strengthened by uncertainty about the phone withdrawal situation and weakened by control over the situation. The section thereafter reports on the method employed to test our integrative model and on the results obtained. Finally, we discuss implications for research and practice.

2. Background and hypotheses

Our approach focuses on integrating the concepts of Nomophobia, stress, and social threat as well as work conditions (i.e., uncertainty and control), which have mostly been studied in isolation before (see Fig. 1). Only a few studies have looked at the intersection of two such areas (e.g., Samaha and Hawi (2016) examined whether Nomophobia can generate stress), and no research to date has examined empirically the point at which all three areas intersect. It is precisely this intersection that holds strong potential to explain the stress-related impacts of Nomophobia in greater detail; according to recently-advanced conceptual ideas, social threat could be relevant to both Nomophobia and stress, and work conditions such as uncertainty and lack of control could be relevant factors in exacerbating phobic traits such as Nomophobia (Cooper, Dewe, & O’Driscoll, 2001; Dickerson, Gruenewald, & Kemeny, 2004; Dickerson & Kemeny, 2004; King et al., 2014; Rubino et al., 2012; Yildirim & Correia, 2015).

Fig. 1. Illustrative Studies in the Contexts of Nomophobia, Stress, and Social threat as well as Work conditions.

To integrate the concepts of Nomophobia, stress, and social threat as well as work conditions, we draw on the demand-control-person model (Bakker & Leiter, 2008; Rubino et al., 2012), an extension of Karasek (1979) demand-control model. The latter indicates that environmental demands interact with the control people have over their environment in generating stress, that is, it is the interaction between demands and control that determines the amount of stress people experience. As regards demands, these are generally perceived as stressful; therefore, stress increases with high demands. An important demand in the context of our study is uncertainty (Best, Stapleton, & Downey, 2005). Uncertainty is an ambiguity-type stressor that refers to the lack of information people perceive in relation to their environment (Beehr, Glaser, Canali, & Wallwey, 2001; Wright & Cordery, 1999). For example, the lack of information on the duration of a meeting can be perceived as stressful. According to the literature on organizational stress, this lack of information, or uncertainty, can generate different types of stress, such as dissatisfaction, burnout, and general perceived stress (Rubino et al., 2012).

As regards the control dimension of Karasek (1979) model, it refers to decision latitude, that is, control refers to peoples’ freedom, independence, and discretion in terms of determining how to respond to a stressor. As such, control enables people to better manage environmental demands. In doing so, control serves as a buffer against stress, as a shield protecting people from the adverse consequences of stressors in their lives. In line with this notion, research has consistently shown that people who control their environment are less stressed (Van der Doef & Maes, 1999).

The demand-control model (Karasek, 1979) has been very successful in the study of stress (Siegrist, 1996). However, the model has important limitations, especially regarding construct dimensionality; the model has been criticized for not being sufficiently comprehensive (Van der Doef & Maes, 1999). Therefore, recent research suggests extending the model by incorporating peoples’ individual differences (Bakker & Leiter, 2008). Individual differences determine how people perceive their environment and react to it. In doing so, they determine peoples’ predispositions to being stressed. Based on these ideas, Rubino et al. (2012) developed the demand-control-person model. This model is an extension of the demand-control model that includes individual differences. Thus, the demand-control-person model specifies three factors that determine the level of stress: environmental demands such as uncertainty, control over one’s environment, and individual differences. While Rubino et al. (2012) examined emotional stability as an individual difference, these authors concluded that other individual differences (e.g., social phobias such as Nomophobia) could also influence peoples’ experiences of stress as well as the impacts of environmental demands and control on their stress levels.

The demand-control-person model is a general and comprehensive theoretical framework for examining stress formation in individuals. Therefore, the model can be applied to various stressful environments and situations (Bakker & Leiter, 2008; Rubino et al., 2012). With its emphasis on individual differences, such as social phobias, the model is germane to our study context. Hence, we draw on this model to examine the impact of Nomophobia on stress.

According to the demand-control-person model, and consistent with Karasek (1979) demand-control model as described earlier, uncertainty in the context of smartphone use can be stressful (for example, the lack of information about the duration of a meeting during which employees cannot use their smartphones can be experienced as taxing by nomophobic individuals). By contrast, control can help reduce stress (for example, some decision latitude as to whether a smartphone can be used during a meeting can buffer against the otherwise stressful impacts of Nomophobia). Finally, Nomophobia can cause stress, and this effect of Nomophobia can be exacerbated by uncertainty and lack of control. The question remains of how, and why, Nomophobia causes stress. According to the demand-control-person model, stressors such as social phobias cause stress by threatening other valued resources (e.g., social esteem, social acceptance, or social respect; (Rubino et al., 2012)). This notion implies that social phobias, such as Nomophobia, lead to stress by generating feelings of being socially-threatened; that is, according to the demand-control-person model, Nomophobia and stress are connected through a perceived social threat. This idea is consistent with research on attentional biases.

Recent research indicates that clinical anxiety is associated with attentional biases that favor the processing of threat-related information specific to particular anxiety syndromes (Amir, Elias, Klumpp, & Przeworski, 2003; Asmundson & Stein, 1994; Hope, Rapee, Heimberg, & Dombeck, 1990). For example, people with a social phobia are more likely than others to perceive a social threat in their environment (Amir et al., 2003; Asmundson & Stein, 1994). The mechanism involved is selective attention, which is responsible for the efficient allocation of mental resources (i.e., information processing resources). Selective attention refers to the ability to selectively attend to some information sources while ignoring others (Strayer & Drews, 2007). In the case of individuals with anxiety disorders, such as those suffering from a social phobia, selective attention targets negative stimuli; that is, individuals with anxiety disorders selectively attend to threatening information that is specifically related to their particular disorder (Asmundson & Stein, 1994).

This attentional bias has been demonstrated using several cognitive psychology paradigms. For example, an early study into attentional biases associated with social phobia used a dot-probe paradigm to show that when attention was allocated in the spatial location of a stimulus cue, individuals with social phobia responded faster to probes that followed social threat cues than to probes following either neutral cues or physical threat cues, an effect that was not observed among control subjects (Asmundson & Stein, 1994). These findings demonstrated that individuals with social phobia selectively process threat cues that are social-evaluative in nature; that is, they seek out information that makes them feel socially-threatened. Another study into attentional biases associated with social phobia used a paradigm with valid and invalid cues that were presented at different locations on the computer screen (Amir et al., 2003). In this study, people with social phobia demonstrated significantly longer response latencies when detecting invalidly cued targets than did the controls, but only when the probe followed a social threat word. These results further confirmed the notion that people with social phobia have difficulty disengaging their attention from socially threatening information, implying that people with social phobia are more likely to feel socially-threatened than people without social phobia. Social threat, in turn, has been established as a major stressor. For example, the Trier Social Stress Test with its focus on social threats is one of the most prominent stress paradigms (Granger, Kivlighan, El-Sheikh, Gordis, & Stroud, 2007).

Since Nomophobia is a social phobia to which the demand-control-person model and the attentional bias literature apply (Bragazzi & Del Puente, 2014; King et al., 2013), one can argue that social threat carries the influence of Nomophobia on to stress. We expect social threat in the context of Nomophobia to manifest in feelings of not meeting others’ expectations regarding constant availability and immediate responsiveness to such technologies as emails, instant messages, Voice over IP, tweets, and Facebook posts (King et al., 2014). Thus, social threat can explain in more detail the link between Nomophobia and stress. Furthermore, the indirect effect of Nomophobia on stress via social threat should be exacerbated by uncertainty as well as lack of control as argued above (based on the demand-control-person model). Overall, on the basis of the demand-control-person model and the literature on attentional biases we advance the following hypotheses (please also see Fig. 2):

Social threat mediates the positive relationship between Nomophobia and Stress.

Uncertainty regarding the duration of a phone withdrawal situation moderates the indirect effect of Nomophobia on Stress (via Social threat) such that this indirect effect will be stronger for greater levels of Uncertainty.

Control over a phone withdrawal situation moderates the indirect effect of Nomophobia on Stress (via Social threat) such that this indirect effect will be weaker for greater levels of Control.

Fig. 2. Research Model.

3. Method and results

An experiment was conducted to test our hypotheses. The experimental design involved two factors to manipulate uncertainty and control, yielding four experimental groups. 270 young business professionals were recruited via a university research panel and, subsequently, divided into these four groups by random allocation. Participation was voluntary and the study was approved by the institutional review board. The experiment employed a questionnaire as a method of data collection. The questionnaire was developed on the basis of prior research.

3.1. Protocol: details on the questionnaire used as the method of data collection

The participants were randomly assigned to one of four conditions: 1) low uncertainty, low control, 2) low uncertainty, high control, 3) high uncertainty, low control, and 4) high uncertainty, high control. Dependent on their respective conditions, the participants were, then, presented with a scenario. They were given clear instructions to imagine themselves in a fictitious business meeting during which they could not use their smartphones. In the low uncertainty condition, the scenario indicated the duration of the meeting (i.e., a 1-h meeting), whereas in the high uncertainty condition the length of the meeting was left unspecified. In the high control condition, the scenario indicated that the participants could exit the meeting at any time to use their smartphones. By contrast, in the low control condition it was clearly indicated that stepping out of the meeting to use one’s phone was not possible. The four scenarios are presented in Table 1:

Table 1. Scenarios.

Low Uncertainty, High Control	Low Uncertainty, Low Control
The meeting will last 1 h. Even if you cannot use your smartphone during the meeting, you may leave the meeting to use it for incoming calls or messages, or to obtain important information from the internet. Note: You have no possibility of accessing a laptop computer.	The meeting will last 1 h. During the meeting, you CANNOT exit the room, which means you CANNOT leave the meeting to use your smart phone for incoming calls or messages, NOR to obtain important information from the internet. Note: You have no possibility of accessing a laptop computer.
High Uncertainty, High Control	High Uncertainty, Low Control
You do NOT know the length of the meeting. Even if you cannot use your smartphone during the meeting, you may leave the meeting to use it for incoming calls or messages, or to obtain important information from the internet. Note: You have no possibility of accessing a laptop computer.	You do NOT know the length of the meeting. During the meeting, you CANNOT exit the room, which means you CANNOT leave the meeting to use your smart phone for incoming calls or messages, NOR to obtain important information from the internet. Note: You have no possibility of accessing a laptop computer.

A French version of the NMP-Q questionnaire developed by (Yildirim & Correia, 2015) was used to measure nomophobia. A double translation was performed to ensure the validity of the French questionnaire (Grisay, 2003). Perception of stress was measured with a likert scale developed by Tams et al. (2014) on the basis of Moore (2000, pp. 141–168) measure. Social threat was measured using a likert scale adapted from (Heatherton & Polivy, 1991). The list of measurement items that were used is presented in Appendix 1.

3.2. Measurement assessment

The psychometric quality of our measures was assessed by estimating reliability as well as convergent and discriminant validity. The internal consistency reliability, as evaluated by Cronbach’s coefficient alpha, was satisfactory for all measures. As shown in Table 2, all alphas exceeded the 0.70 threshold (Nunnally, 1978).

Table 2. Quality criteria and descriptives of construct measures.

Construct	N. of items	AVE	Alpha	Mean	SD	Range
Nomophobia	20	0.51	0.95	2.95	1.26	6
Social threat	6	0.67	0.90	2.13	1.19	6
Stress	8	0.64	0.92	3.11	1.32	6

AVE = Average Variance Extracted.

Convergent validity is increasingly being assessed on the basis of a construct’s average variance extracted (AVE). The AVE represents the amount of variance a construct measure captures from its associated items relative to the amount that is due to measurement error. An AVE of at least 0.50 indicates sufficient convergent validity, demonstrating that the construct accounts for the majority of the variance in its items (Fornell & Larcker, 1981). The discriminant validity of a construct is commonly regarded as adequate when the square root of the construct’s AVE is higher than the inter-construct correlations in the model (Chin, 1998). All AVE values were above 0.50 (see Table 2) and the square root of the AVE for each construct (0.71, 0.82, and 0.80 for Nomophobia, social threat, and stress, respectively) was higher than the correlations between that construct and all other constructs in the model (ρ_Nomo-Threat = 0.44, ρ_Nomo-Stress = 0.53 and ρ_{Threat-Stress} = 0.61), indicating sufficient convergent and discriminant validity.

The measurement of nomophobia through the NMP-Q questionnaire developed by (Yildirim & Correia, 2015) originally comprises four dimensions. In the context of this study, we treated the construct as unidimensional. First, theoretical development and our hypotheses were laid out at the overall construct level and not by individual dimensions. Second, the scree plot from a factor analysis, through the examination of the point of separation or the “elbow”, suggests that a unidimensional operationalization is adequate. The eigenvalue associated with the first dimension was 10.12. It dropped to 1.89, 1.22, and 0.98 for the subsequent dimensions. The first extracted factor explained 50.6% of the total variance. The absolute factor loadings were all greater than 0.40, suggesting a good indicator-factor correspondence (Thompson, 2004). Third, when assessing construct validity of the NMP-Q, Yildirim and Correia (2015) also used a unidimensional approach to the measurement of the concept.

Following Podsakoff et al. (2003), procedural as well as statistical remedies were used to control for common method bias. In terms of procedure, we guaranteed response anonymity and separated the measurement of the predictor and criterion variables. Statistically, the single factor test revealed that a single factor explains only 40.32% of the variance. Additionally, the marker-variable technique was applied to the analyses (Malhotra, Kim, & Patil, 2006). Gender was chosen as the marker variable since there is no theoretical link between this variable and nomophobia, a necessary condition for the marker-variable technique. The average correlation with other constructs was less than 0.10 in the four groups. Adjusting the correlation matrices to fit the path analyses yielded analogous results to the ones from the main analyses (presented below). Thus, common method bias did not appear to be an issue in this research (Podsakoff et al., 2003).

3.3. Model specification

A multi-group path analysis approach was used to test our conditional indirect effect hypotheses. This approach allowed for a straightforward and simultaneous way of assessing the effects of two potential moderators (i.e., uncertainty and control). Multi-group path analysis was particularly appropriate in that we could consider each experimental condition as a different group in which we, then, conducted a path analysis. The regression weights, the covariances, and residuals could be estimated separately and compared in such a multi-group setting. This approach was, thus, more flexible in estimating moderated mediation effects than prepackaged macros, such as (Preacher, Rucker, & Hayes, 2007) macro. The AMOS statistical software was used to estimate the model (Arbuckle, 2006). The Maximum likelihood method was used.

In order to assess invariance between experimental conditions, four successive parametrizations were fitted. Model 1 constrained residuals, covariances and regression weights to be equal between experimental conditions; Model 2 allowed for unconstrained residuals but constrained covariances and regression weights; Model 3 for constrained regression weights; and Model 4 for a fully unconstrained specification.

As shown in Table 3, unconstraining covariances and residuals does not add significantly to the fit of the model; p > 0.10. Yet, regression weights appear to vary between experimental conditions; Δ χ² = 26.38, Δdf = 9, p < 0.01. Thus, the remainder of this analysis will report model specifications where residuals and covariances are invariant between experimental conditions.

Table 3. Model comparison.

Model	Model comparison	Δdf	Δ χ²
Model 1: Constrained residuals + C + R	2 vs. 1	6	3,65
Model 2: Constrained covariances (C) + R	3 vs. 2	3	2,88
Model 3: Constrained regression weights (R)	4 vs. 3	9	26,38	^∗∗

^∗∗p < 0.01.

4. Results

Table 4 presents the unconstrained regression weights for the model with constrained covariances and residuals. Fit indices show a good fit to the data; GFI = 0.961 and NFI = 0.931. The chi-square statistic is close to its expected value; CMIN = 14.394, df = 16. In other words CMIN/df is close to 1. This measure of fit, on which other indices are derived, causes the RMSEA to be exceptionally low (<0.001) and the CFI to be high (>0.999). The relationship between Social Threat and Stress (Path B in Table 4) was significant and positive for all groups; all Betas >. 45 with all p-values < 0.001. Path A – Nomophobia to Social Threat – and C – Nomophobia to Stress – was not significant for the high control, low uncertainty condition; β_A = 0.091, Critical Ratio (C.R.) = 0.82, p > 0.10 and β_B = 0.118, C.R. = 1.15, p > 0.10. These two paths were significant for all the other experimental conditions; all Betas > 0.25 with all p-values < 0.05.

Table 4. Regression weights for the path analysis.

Control	Uncertainty	Regression weights
Control	Uncertainty	Nomophobia – > Social threat (Path A)	Social threat – > Stress (Path B)	Nomophobia – > Stress(Path C)
Low	Low	0.490 (0.108)^∗∗∗	0.457 (0.120)^∗∗∗	0.512 (0.115)^∗∗∗
Low	High	0.483 (0.104)^∗∗∗	0.468 (0.115)^∗∗∗	0.597 (0.110)^∗∗∗
High	Low	0.091 (0.112)	0.582 (0.124)^∗∗∗	0.118 (0.103)
High	High	0.577 (0.109)^∗∗∗	0.461 (0.121)^∗∗∗	0.263 (0.122)^∗

^∗∗∗p < 0.001, ^∗∗p < 0.01, ^∗p < 0.05.

To test further this pattern of results, we did a chi-square difference test between an unconstrained regression weight model with a model where the A and C paths were allowed to vary only for the high control, low uncertainty condition; Δ χ² = 6.805, ΔDF = 8, p > 0.10. Thus, constraining the low control, low uncertainty, the low control, high uncertainty, and the high control, high uncertainty conditions to have the same regression weights for path A and C as well as having all B paths to be equal among all conditions did not reduce significantly the fit. The aggregated paths for the three conditions were all positive and significant: β_A = 0.521, C.R. = 8.45, p < 0.001, β_B = 0.480, C.R. = 7.92, p < 0.001, and β_C = 0.431, C.R. = 6.58, p < 0.001. Paths A and C remained non-significant for the high control, low uncertainty condition: β_A = 0.091, C.R. = 0.82, p > 0.10, and β_C = 0.128, C.R. = 1.22, p > 0.10.

The indirect effect of Nomophobia on Stress for the high control, low uncertainty condition was 0.053. The bootstrapping procedure developed by Preacher and Hayes (2008) showed that this mediation effect was non-significant (LL = −0.048, UL = 0.156, p > 0.05). For the three other conditions, the indirect effects of Nomophobia on Stress were 0.224, 0.226, and 0.226. The bootstrapping procedure showed that these three indirect effects were all significant, with 0 outside of the 95% confidence intervals (LL = 0.097, UL = 0.397; LL = 0.113, UL = 0.457; and LL = 0.096, UL = 0.481, respectively). Thus, Hypothesis 1 was partially supported in that the mediated relationship between nomophobia and stress through social threat was present only when uncertainty was high or control low.

These results suggest that a high level of control and a low level of uncertainty are necessary for the nomophobia – > social threat – > stress link to be avoided. Nomophobic people show less inclination for experiencing feelings of social threat (Path A) that lead to stress in situations of high control and low uncertainty. This pattern of results confirms Hypotheses 2 and 3 in that uncertainty and control moderate the indirect effect of nomophobia on stress. Also, the direct relationship between nomophobia and stress is dampened only for situations of high control and low uncertainty (Path C). In other words, if control is low or uncertainty high, nomophobia will lead to stress but also to social threat that will, in turn, lead to stress.

5. Discussion

Past research focusing on whether Nomophobia has downstream negative consequences showed that stress is an important problem associated with Nomophobia (direct effect), but it has not offered theoretical explanations for how and why Nomophobia leads to stress (indirect effect). To advance knowledge in this area and offer more specific guidance to individuals, healthcare practitioners, and managers, this study examined the process by which Nomophobia’s effect on stress unfolds. In doing so, the study helps research on Nomophobia progress from offering general explanations of the relationship between Nomophobia and stress toward more detailed and specific explanations of the causal pathway involved. This research has shown that Nomophobia leads to stress by generating feelings of being socially-threatened; in other words, Nomophobia exerts its influence on stress through social threat.

Additionally, this study extends past work by yielding a more nuanced understanding of the moderating factors that bound the applicability of Nomophobia’s effects. We found that Nomophobia leads to stress via social threat when uncertainty or lack of control are present. Only under the condition of low uncertainty and high control does Nomophobia not lead to stress. Thus, as a second contribution, our results help research on Nomophobia progress from investigating the general association between Nomophobia and its negative consequences, such as stress, toward more detailed and specific explanations of when, or under what conditions, Nomophobia leads to stress. In other words, the results shed light on the boundary conditions, or contextual factors, on which the stress-related effects of Nomophobia depend, a critical contribution to theory development and testing (Bacharach, 1989; Cohen, Cohen, West, & Aiken, 2013). The stress-related consequences of Nomophobia are reduced only when two positive conditions come together. This finding can help healthcare professionals and managers design interventions aimed at relieving stress in nomophobic individuals. Besides, the finding suggests that Nomophobia leads to stress in most situations and is, thus, a quite powerful stressor.

Overall, this study makes three important contributions to our understanding of the Nomophobia phenomenon. First, this research reveals that social threat is a causal pathway through which Nomophobia leads to negative consequences, especially stress. Before this study, Nomophobia was shown to correlate with stress; that is, prior research has advanced our understanding of whether Nomophobia has negative consequences such as stress. However, there was a lack of understanding of the causal pathways involved in the relationship between Nomophobia and stress. In other words, the direct effect of Nomophobia on stress was established, but it remained unclear what factors are responsible for carrying the influence of Nomophobia on to stress. This study shows how and why Nomophobia impacts stress (by generating the perception of a social threat). In doing so, this study yields an enriched theoretical understanding of the relationship between Nomophobia and stress, revealing social threat as a pertinent mediating mechanism. From a practical standpoint, managers must be aware that Nomophobia can generate feelings of being socially-threatened, ultimately leading to stress (Bragazzi & Del Puente, 2014; Samaha & Hawi, 2016; Yildirim & Correia, 2015).

Second, this study established work conditions (uncertainty and control) as pertinent moderators in the Nomophobia phenomenon. Prior research has focused on drivers and consequences of Nomophobia to the exclusion of contextual factors on which Nomophobia-related impacts depend. Hence, there was a lack of understanding of the prominent role that work conditions can play in the Nomophobia phenomenon, by helping people cope with Nomophobia (i.e., moderators of the Nomophobia-stress link). From a practice point of view, managers must be aware of the central role of worker control and certainty in nomophobic individuals and of their potential to offset the damaging effects of Nomophobia (Bakker & Leiter, 2008; Bragazzi & Del Puente, 2014; Karasek, 1979; Riedl, 2013; Rubino et al., 2012; Samaha & Hawi, 2016).

Third, our use of the demand-control-person model increases the diversity of theoretical perspectives that are being brought to bear in the study of Nomophobia. This greater diversity enriches our theoretical understanding of Nomophobia along with our understanding of the phenomenon’s nomological network. Before this study, the literature on Nomophobia and Technostress were largely the only ones applied to understanding the stress-related consequences of Nomophobia. Although Technostress research and prior research on Nomophobia are very useful to understanding these stress-related consequences, they are not longstanding, precise stress theories. Hence, adding an extension of the Demand-Control model to the mix improves the prediction of Nomophobia’s consequences. In a word, our approach adds theoretical diversity to the study of Nomophobia, enriching how we study the Nomophobia phenomenon and what we can predict (Bakker & Leiter, 2008; Bragazzi & Del Puente, 2014; Rubino et al., 2012; Samaha & Hawi, 2016; Yildirim & Correia, 2015). For managers, they can gain a more refined understanding of the Nomophobia-stress process and of how to combat Nomophobia; they are no longer limited solely to the ideas put forth by research on technostress.

Additionally, this study demonstrates that Nomophobia is a strong stressor; Nomophobia leads to stress under all conditions studied here, except under the combination of (a) low uncertainty about the duration of a phone withdrawal situation and (b) high control over the situation.

To combat the stress arising from withdrawal situations, managers can, first and foremost, instill trust in their employees, making them believe that the withdrawal situation will not take any longer than absolutely necessary (i.e., trust that the duration of the withdrawal situation is strictly limited). Trust is a classic mechanism to reduce feelings of uncertainty (e.g., Carter, Tams, & Grover, 2017; McKnight, Carter, Thatcher, & Clay, 2011; Pavlou, Liang, & Xue, 2007; Riedl, Mohr, Kenning, Davis, & Heekeren, 2014; Tams, 2012). It builds perceptions of security and safety that are directly opposed to uncertainty (Kelly & Noonan, 2008). In doing so, trust can extinguish the negative emotions associated with uncertainty and other job demands (McKnight et al., 2011; Tams, Thatcher, & Craig, 2017). Future research can empirically examine this initial idea.

Another mechanism to help nomophobic employees deal better with uncertainty could be social presence. Social presence reduces problems related to uncertainty by creating the perception that important social encounters occur during the meeting. Managers could communicate to their employees the message that a given meeting is important and that it warrants everyone’s attention. To this end, the manager might also employ attention-grabbing formats of information presentation during the meeting. The resulting perception of social presence might reduce employees’ needs to use the phone (Pavlou et al., 2007). This idea could also be empirically verified in future research.

As with any research, there are certain limitations to our study that should be considered when interpreting our results. This study was conducted with young business professional. While this choice may limit the study’s external validity, it was appropriate for the study given the respondents’ familiarity with the focal technology and its relevance to their lives. Further, this approach was associated with high internal validity due to the homogeneity inherent in this sample population. Moreover, given that our target technology was the smartphone, which is widely used in all aspects of peoples’ lives (Samaha & Hawi, 2016), our findings may generalize to a variety of settings, including organizations. Additionally, our research is based on a psychometric monomethod approach that captures the perception of stress in a hypothetical situation. Future research should aim at replicating these results in an ecologically more valid situation, potentially using objective measures of stress, such as cortisol.

Furthermore, future research could examine other pathways through which nomophobia elicits stress responses in individuals. We focused on social threat as a mediator due to its particular relevance for nomophobic individuals. However, other variables might constitute additional, relevant mediators. For example, social overload could be of additional relevance in the context of our study. Research in the area of social network addiction, which is related to our study context, has found that social overload mediates the relationship between personality characteristics and addiction (Maier, Laumer, Eckhardt, & Weitzel, 2015). A study was conducted in the context of Facebook usage, showing that social support mediates the link between, for instance, number of friends on Facebook and exhaustion due to the extended use of Facebook (Maier et al., 2015). Social overload was defined as the negative perception of social network usage when users receive too many social support requests and feel they are giving too much social support to other people embedded in their social network. Given that the context of nomophobia also includes elements of addiction, social overload might well be an additional, relevant mediator in the context of our study, connecting nomophobia to stress.

Consistent with MacKinnon and Luecken (2008; p. S99), our findings, taken together, yield a “more sophisticated” understanding of how, why, and when (or under what conditions) Nomophobia has downstream negative consequences. This improved understanding facilitates the development of intervention strategies aimed at reducing the stress-related consequences of Nomophobia.

6. Conclusion

Past research has established stress as an important consequence of Nomophobia but has not examined the causal pathways or contextual factors involved in this important relationship, resulting in the need to further knowledge in this area. Based on the Demand-Control-Person model and its predictions about phobic traits, uncertainty, control, and social threat, this paper has produced a more refined understanding of the process by which Nomophobia leads to stress, as well as pertinent contextual factors on which this process depends. Accordingly, this study helps research on Nomophobia progress toward more detailed and specific explanations of how, why, and when Nomophobia results in stress. These explanations imply that research on Nomophobia is not yet saturated but that clearer guidance can, and should, be provided to individuals, healthcare practitioners, and managers in our increasingly smartphone-driven world.

Appendix 1. List of measurement items

	Mean scores	Standard deviation
Nomophobia
1. I would feel uncomfortable without constant access to information through my smartphone	2.52	1.81
2. I would be annoyed if I could not look information up on my smartphone when I wanted to do so	3.53	1.74
3. Being unable to get the news (e.g., happenings, weather, etc.) on my smartphone would make me nervous	1.89	1.65
4. I would be annoyed if I could not use my smartphone and/or its capabilities when I wanted to do so	3.45	1.87
5. Running out of battery in my smartphone would scare me	2.91	1.91
6. If I were to run out of credits or hit my monthly data limit, I would panic	2.45	1.91
7. If I did not have a data signal or could not connect to Wi-Fi, then I would constantly check to see if I had a signal or could find a Wi-Fi network	2.37	1.95
8. If I could not use my smartphone, I would be afraid of getting stranded somewhere	2.15	1.85
9. If I could not check my smartphone for a while, I would feel a desire to check it If I did not have my smartphone with me	2.81	1.95
10. I would feel anxious because I could not instantly communicate with my family and/or friends	3.67	1.75
11. I would be worried because my family and/or friends could not reach me	4.01	1.77
12. I would feel nervous because I would not be able to receive text messages and calls	3.92	1.77
13. I would be anxious because I could not keep in touch with my family and/or friends	3.45	1.71
14. I would be nervous because I could not know if someone had tried to get a hold of me	3.90	1.82
15. I would feel anxious because my constant connection to my family and friends would be broken	3.08	1.64
16. I would be nervous because I would be disconnected from my online identity	2.49	1.58
17. I would be uncomfortable because I could not stay up-to-date with social media and online networks	2.21	1.50
18. I would feel awkward because I could not check my notifications for updates from my connections and online networks	2.31	1.59
19. I would feel anxious because I could not check my email messages	3.43	1.94
20. I would feel weird because I would not know what to do	2.65	1.83
Stress
1. You would feel frustrated.	3.26	1.73
2. You would feel anxious.	3.31	1.66
3. You would feel strain.	3.52	1.70
4. You would feel stressed.	3.60	1.78
5. You would feel emotionally drained.	2.72	1.56
6. You would feel used up.	2.67	1.57
7. You would feel fatigue.	3.04	1.62
8. You would feel burned out.	2.82	1.56
Social threat
1. I would be worried about whether I am regarded as a success or failure.	1.89	1.28
2. I would feel self-conscious.	2.44	1.71
3. I would feel displeased with myself.	2.38	1.36
4. I would feel inferior to others at this moment.	1.69	1.16
5. I would feel concerned about the impression I am making.	2.43	1.73
6. I would be worried about looking foolish.	1.98	1.47

References

Amir et al., 2003

N. Amir, J. Elias, H. Klumpp, A. PrzeworskiAttentional bias to threat in social phobia: Facilitated processing of threat or difficulty disengaging attention from threat?

Behaviour Research and Therapy, 41 (11) (2003), pp. 1325-1335

Article PDF (121KB)View Record in Scopus

Arbuckle, 2006

J.L. ArbuckleAmos (version 7.0)[computer program]

SPSS, Chicago (2006)

Asmundson and Stein, 1994

G.J. Asmundson, M.B. SteinSelective processing of social threat in patients with generalized social phobia: Evaluation using a dot-probe paradigm

Journal of Anxiety Disorders, 8 (2) (1994), pp. 107-117

Article PDF (808KB)View Record in Scopus

Ayyagari et al., 2011

R. Ayyagari, V. Grover, R. PurvisTechnostress: Technological antecedents and implications

MIS Quarterly, 35 (4) (2011), pp. 831-858