Personality Profiles that Put Users at Risk of Perceiving Technostress

An individual’s personality consists of personality traits, which are relatively enduring ways in which the individual tends to think, feel, and behave. These tendencies, i.e. personality traits, characterize an individual, distinguish individuals from one another, and predispose the individual to have various thoughts, feelings, and behaviors (Roberts et al. 2008).

Great efforts have been made to identify the specific personality traits that constitute individuals’ personality. The Big Five personality traits (Costa and McCrae 1997) are typically conceptualized as core characteristics constituting the essential basis of individual differences (McCrae and Costa 2008). They thus provide a valuable reflection and holistic picture of an individual’s personality. The Big Five personality traits include openness to experience, which refers to the tendency to prefer new experiences over routines. Neuroticism is the tendency to experience unpleasant emotions such as anxiety easily. Agreeableness describes the tendency to cooperate with others and conscientiousness the tendency to act in a planned and duty-oriented manner. Finally, extraversion refers to the tendency to seek the stimulation of others. These five traits are commonly seen as context-free and stable, and are useful to understand thoughts, feelings, and behaviors across different situations (Kandler et al. 2014). This means that the Big Five personality traits change little over time and influence various aspects of an individual’s perceptions and behaviors. As a group, the Big Five personality traits form a personality profile, a set of coexisting Big Five personality traits. In a personality profile, the degree of each personality trait can vary widely, resulting in a virtually infinite number of possible personality profiles. However, only some personality profiles lead to the thoughts, feelings, or behaviors of interest in this research (Vollrath and Torgersen 2000).

The focus on a personality profile instead of individual personality traits enables the investigation of multiple coexisting personality traits to reveal their multifaceted influence and interactions. Personality literature in the fields of management or psychology inform us that it is not sufficient to focus on the effect of a single personality trait to describe and understand the effects of personality traits. Rather, the personality traits work together in leading to perceptions (Grant and Langan-Fox 2006). This means that whether a personality trait results in a perception depends on whether the other personality traits of a profile are high or low. We see these interacting effects of personality traits in related streams such as organizational research. Consider, for example, how an employee’s job performance is influenced by the personality traits extraversion and conscientiousness. Research indicates that high extraversion leads to an increase in job performance among employees with a high level of conscientiousness, but to a decrease in job performance among employees with a low level of conscientiousness (Witt 2002). Thus, the simultaneous investigation of extraversion and conscientiousness reveals more information about an employee’s behavior, i.e. job performance, than the separate investigation of each personality trait. Studying extraversion alone might have led to an insignificant relationship between the personality trait and job performance, which might have led to insufficient conclusions and precluded the multifaceted influence of extraversion on job performance. Similar interactive effects have been shown for further personality traits, e.g., conscientiousness and agreeableness (Witt et al. 2002), as well as other variables of interest, e.g., stressful events or counterproductive work behavior (Bardi and Ryff 2007). When investigating the influence of a personality profile on variables of interest, the interactive effects of multiple coexisting personality traits are critical. Focusing on personality profiles allows to comprehensively understand and predict an individual’s perceptions and behavior based on personality traits. We find empirical support for these considerations in personality literature, where a personality profile called the Type D personality was revealed. The Type D personality is a combination of negative affectivity, which is strongly related to the Big Five personality trait neuroticism, and social inhibition, which is strongly negatively related to extraversion (Horwood et al. 2015). Research shows that this personality profile, i.e. the Type D personality, is effective in understanding and predicting how personality elicits negative consequences such as stress perception (Reich and Schatzberg 2010).

Turning to information systems (IS) literature, the Big Five personality traits were used in different contexts to understand user behavior, particularly in the context of ICT use, information technology professionals or privacy. For an overview of personality in IS literature see Maier (2012) or Online Appendix A (available online via http://​link.​spinger.​com). Among others, it has been shown that the Big Five personality traits influence ICT acceptance, e.g., by affecting whether ICTs are perceived as useful (Devaraj et al. 2008) and whether individuals intend to use or actually use the ICT (McElroy et al. 2007). These insights are helpful to understand the relevance of the Big Five personality traits for various IS domains. However, IS literature that deals with personality has disregarded the interactive nature of coexisting personality traits. The empirical studies have investigated how separate personality traits lead to behaviors such as ICT use, but have not accounted for interactive effects or investigated personality profiles. Accounting for interactive effects in terms of personality profiles is important in order to reveal multifaceted influences of personality traits on IS relevant perceptions and behaviors. This allows to investigate under which combinations low and high levels of the Big Five personality traits lead to negative effects.

In the next section, we provide relevant background knowledge on technostress before highlighting why the investigation of personality profiles is pivotal to understand how technostress emerges.

Technostress is a process that starts with a stimulation condition, called techno-stressor, which, when perceived, can lead to a reaction called techno-strain. Techno-stressors are stressful demands, situations or stimuli caused by ICTs that can result in techno-strain. Techno-strain refers to the consequences of perceived techno-stressors (Tarafdar et al. 2019). For a summary of relevant technostress literature, see Fischer and Riedl (2017), Tarafdar and colleagues (2019) or Online Appendix B, which includes the respective results.

Users of ICTs are confronted with a variety of techno-stressors, including interruptions, unreliability, and overload (Galluch et al. 2015; Gimpel et al. 2018; Riedl et al. 2012; Tams et al. 2014). Drawing from this variety, previous research has predominantly discussed five techno-stressors (Ragu-Nathan et al. 2008). Techno-overload describes situations in which users experience increased work volume and speed due to ICTs. Techno-invasion refers to situations where users feel the need to be permanently connected to work and where the line between work and personal life becomes blurred due to ICTs. Techno-complexity describes situations where ICT-related complexity leads users to feel they lack the skills and they need to spend time and effort to understand the different aspects of ICTs. Techno-insecurity refers to situations where users fear losing their job to other employees with better ICT skills or due to the replacement by an ICT. Finally, techno-uncertainty describes situations where users feel uncertainty because of ongoing changes in ICTs and where they are constantly forced to adapt, learn, and educate themselves about new ICTs (Ragu-Nathan et al. 2008).

The perception of techno-stressors can lead to techno-strain, which is a non-beneficial reaction to perceived techno-stressors (Tarafdar et al. 2019). Techno-strain includes adverse behavioral reactions like low job performance or lack of innovation (Tarafdar et al. 2015). Additionally, it incorporates adverse psychological reactions such as exhaustion or job burnout (Maier et al. 2015a). Finally, perceived techno-stressors can lead to adverse physiological reactions, including increased stress hormone production (Riedl et al. 2012).

Beyond studying the reactions to perceived techno-stressors, some initial studies explain why an individual perceives techno-stressors and why the reactions to perceived techno-stressors arise. This ranges from technological characteristics (Ayyagari et al. 2011) to individual differences (Srivastava et al. 2015). Some technostress research has also investigated the role of a user’s Big Five personality traits in perceiving technostress (see Fig. 1), drawing three primary conclusions. First, the Big Five personality traits influence users’ reactions to perceived techno-stressors (Maier et al. 2015c). For example, users high on neuroticism or low on extraversion are more exhausted from using ICTs. Second, the Big Five personality traits influence how perceived techno-stressors influence job burnout and job engagement (Srivastava et al. 2015). For example, whether a user is high or low on agreeableness influences the extent to which a given level of perceived techno-stressors produces job burnout. Third, the Big Five personality trait neuroticism leads users to perceive a high level of techno-stressors (Maier et al. 2019).

In summary, IS research illustrates the detrimental effects of technostress and shows that individual personality traits influence IS-related experiences and behavior, including technostress. Building on these findings, the next section illustrates why and how personality profiles are important for comprehending why people perceive techno-stressors, and introduces the differential exposure-reactivity model (Bolger and Zuckerman 1995) for linking personality profiles and techno-stressors and understanding their relationship.

Our sampling strategy was to survey a broad spectrum of participants who work with ICTs regularly as part of their work. We prepared an online survey and used Amazon Mechanical Turk (mTurk), which is a well-established approach in IS research, considered equal to traditional data collection approaches (Maier et al. 2019). The characteristics of the final sample are shown in Table 1.

We followed recommendations for self-reported data and tested for common method bias (CMB) and late response bias. We can conclude that CMB and late response bias are not a cause for concern in our data set (see Online Appendix C).

To measure the Big Five personality traits, we used existing items, which have already been used in techno-stress research (Srivastava et al. 2015). Openness to experience, neuroticism, agreeableness, conscientiousness, and extraversion were measured with three items each. To measure the five techno-stressors, we used techno-overload, techno-invasion, techno-complexity, techno-insecurity, and techno-uncertainty (Ragu-Nathan et al. 2008). Techno-stressors were calculated as the sum of the five techno-stressors. All items were measured on a seven-point Likert-type agreement scale, ranging from 1 (completely disagree) to 7 (completely agree) (see Online Appendix D).

To ensure content validity, we used items that have been used and validated in previous research. We calculated the loading of each item and only used items with a loading above 0.707 (Carmines and Zeller 2008) (see Online Appendix D), thus indicator reliability is given (Carmines and Zeller 2008; Hair et al. 2019). We can assure construct reliability because the average variance extracted (AVE) of each construct is higher than 0.50 and the composite reliability (CR) is higher than 0.70 (see Online Appendix E). Furthermore, discriminant validity is given, as we calculated the heterotrait-monotrait (HTMT) ratio, which is 0.68 at most (see Online Appendix F) and consequently below the threshold of 0.85 (Hair et al. 2019). The square root of the AVE is higher than the corresponding correlations of the constructs (Fornell and Larcker 1981) (see Online Appendix E). In sum, the evaluation suggests that our data is reliable and valid.

As we aim to analyze which personality profiles predispose a user to perceive techno-stressors, we take a configurational approach. This is based on the fact that this approach allows us to study the Big Five personality traits in profiles instead of individually, and because the approach accounts for complex relationships between the distinct Big Five personality traits. We use fsQCA (Ragin 2014), which enables us to study perceived techno-stressors as the result of a sufficient configuration of personality traits. In a nutshell, fsQCA is grounded in set-theory and uses Boolean algebra to examine what configurations of conditions are sufficient configurations leading to an outcome (Ragin 2014). In this study, a sufficient configuration refers to a personality profile which predisposes to perceive a high level of techno-stressors and where each personality trait is expressed through a fuzzy set. Using fuzzy sets allows us to express the degree of membership to which a measure of a personality trait is high or low on this personality trait. This means that with fsQCA we can examine to what degree each of the five personality traits is high or low within a specific personality profile. To conduct the fsQCA analysis, we first need to calibrate the survey data into fuzzy sets.

As recommended in QCA literature (Mattke et al. 2020; Ortiz de Guinea and Raymond 2020), we applied the direct calibration to calibrate survey data, which was measured on a seven-point Likert scale, into fuzzy sets, which range from the value 0 to the value 1. We followed recommendations of QCA literature (Mattke et al. 2020) and used the value 7 as the full membership anchor, which, for instance, indicates that a personality trait has a high level. We used the value 1 as the full nonmembership anchor, which, for instance, indicates that a personality trait has a low level. Finally, we used the value 4 as the crossover point. It thus indicates that a personality trait is neither high nor low (more information in Online Appendix G). We calibrated based on the Likert scale values and not based on the data distribution (e.g., taking minimum/maximum values or percentiles of the data set as anchor) for the following reasons: First, the meaning of anchors based on the data distributions is obscure and, second, bound to the specific data sample (Wagemann et al. 2016). This makes the calibration idiosyncratic to the data sample, resulting in fuzzy sets which can only be interpreted for a specific data set. This departs from the true meaning of the data from the Likert-scale which can be compared across data samples. Thus, through a calibration based on the maximum, minimum and median value of the Likert scale, we can differentiate high and low values of a personality trait, which in turn are comparable to other data sets. Based on the resulting fuzzy sets, we incorporated the calibrated data into the truth table (see Online Appendix H) which lists all possible configurations of the personality traits, and calculated the consistency of each configuration. The consistency of a configuration (Ragin 2014) is a measure that captures the extent to which a configuration leads to the studied outcome.

To reduce the truth table to sufficient configurations, we applied a frequency threshold of three (Mattke and Maier 2020), meaning that all configurations with less than three observations were dropped from further analysis. In line with QCA literature (Ragin 2009), we used a consistency threshold of 0.90 which is higher than the minimum level of 0.85 and therefore leads to better robustness of the configurations. By applying those two thresholds, we revealed sufficient configurations that predispose to perceive techno-stressors. The reduced truth table is reported in Online Appendix H. As outlined above, sufficient configuration means that individuals who show this sufficient configuration – thus a personality profile – are predisposed to perceive high techno-stressors. Finally, we applied logical minimization with the Quine–McCluskey algorithm to minimize the sufficient configurations (see Online Appendix H). The Quine–McCluskey algorithm simplifies set-theoretic statements, so that logical redundant conditions can get omitted and multiple sufficient configurations can be merged into one (Ragin 2014). By applying the McCluskey algorithm we reveal more parsimonious sufficient configurations. To gain further insights about the sufficient configurations, we tested whether any of the five personality traits is necessary for perceiving high techno-stressors. In the context of this study, a necessary personality trait needs to exist in all configurations predisposing to perceive high techno-stressors, thus it is common that those personality traits need to exceed a consistency threshold of 0.90 and a coverage threshold of 0.60 (Mattke et al. 2020). When testing for necessity, consistency captures the extent to which a personality trait is consistently a necessary personality trait and coverage assesses the empirical relevance of the personality trait (Ragin 2006).

In this section, we outline the findings of the analysis of configurations sufficient for perceiving high techno-stressors. The results reveal six alternative configurations, thus six personality profiles, sufficient for predisposing to perceive high techno-stressors. We draw on the graphical illustration of the configurations for readability reasons (Fig. 3).

Configuration 1 depicts a personality profile where openness to experience plays a subordinate role and can be either high or low. Moreover, the configuration encompasses high neuroticism and agreeableness as well as low conscientiousness and extraversion. Thus, we have a user who tends to experience negative emotions easily, is more likely to want to please others, and has difficulties controlling his or her impulses, but does not seek social attention.

In Configuration 2, openness to experience can either be high or low. Moreover, it presents a configuration with high neuroticism and conscientiousness as well as low agreeableness and extraversion. Thus, we have a user who tends to experience negative emotions easily, is less likely to want to please others, is able to control his or her impulses, and does not seek social attention.

Configuration 3 encompasses a low level of openness to experience and extraversion, a don’t care situation for neuroticism, as well as high levels of agreeableness and conscientiousness. Thus, we have a user who does not prefer experiences over routines nor seeks social attention, but wants to please others and is able to control his or her impulses.

Configuration 4 presents a configuration with low openness to experience as well as high neuroticism, agreeableness, and conscientiousness. Extraversion plays a subordinate role that can be either high or low. Thus, we have a user who does not prefer experiences over routines and is likely to experience negative emotions easily. Moreover, just as in Configuration 3, this user wants to please others and is able to control his or her impulses.

Configuration 5 encompasses high levels of openness to experience, agreeableness, and extraversion as well as low levels of neuroticism and conscientiousness. Thus, we have a user who prefers experiences over routines, wants to please others, and seeks social attention and other rewarding experiences. Moreover, this user does not tend to experience negative emotions easily, but has difficulties controlling his or her impulses.

Finally, Configuration 6 presents a configuration with high levels of openness to experience and neuroticism, and low levels of agreeableness and extraversion. Conscientiousness can be either high or low. Here, we have a user who prefers new experiences over routines and tends to experience negative emotions easily, but in contrast to Configuration 5 does not want to please others nor seeks social attention.

The solution coverage of 0.60 indicates the degree of how much of the outcome, i.e. high techno-stressors, is covered by the six configurations. Thus, the six configurations account for 60% of the membership in the outcome, which illustrates a high explanatory power. The solution consistency of 0.81 is well above the minimum required consistency of 0.75, which attests the robustness of the results. The raw coverage of the six configurations expresses the extent to which the outcome is explained by each configuration (for detailed explanation see Ragin 2006) and ranges from 0.36 to 0.38. The comparable raw coverage scores indicate that all configurations are of equal empirical relevance for perceiving high techno-stressors. The six unique coverage values range from 0.01 to 0.03, expressing the unique contribution of each configuration when excluding the contribution of other configurations (Ragin 2006). A comparison of the six unique coverage scores shows that each configuration has its own unique contribution to explain the perception of high techno-stressors, and the unique contribution is nearly equivalent for each configuration.

The analysis for necessary personality traits does not reveal any personality trait to be necessary. Thus, there is no single personality trait which exists in all personality profiles predisposing to perceive high techno-stressors.

In summary, we see that there are six different personality profiles which predispose to perceive high techno-stressors. However, this does not necessarily inform us about what personality profiles predispose to perceive low techno-stressors. QCA literature informs us about causal asymmetry (Ragin 2014), meaning that inversing the profiles predisposing to perceive high techno-stressors (e.g., inversing a high level of extraversion to a low level of extraversion) might not lead to the profiles which predispose to perceive low techno-stressors. Therefore, we next perform a post-hoc analysis and test for sufficient configurations predisposing to perceive low techno-stressors.

For this, we use the same calibrated data as above and follow the equivalent steps and thresholds. The analysis for sufficient configurations reveals two sufficient configurations of personality traits predisposing to the perception of low techno-stressors (see Fig. 3).

Configuration 7 presents a configuration with low levels of openness to experience, agreeableness, and extraversion, and with high levels of neuroticism and conscientiousness.

Configuration 8 shows a configuration with high levels of openness to experience, neuroticism, and agreeableness as well as low levels of conscientiousness and extraversion.

Comparing the results from the analysis for high and low techno-stressors, we see that Configuration 7 is the inverted version of Configuration 5, whereas Configuration 8 is not the inverted version of any of the configurations predisposing to high techno-stressors. Thus, we see that there exists causal asymmetry, meaning that inversing the profiles predisposing to perceive high techno-stressors (e.g., inversing a high level of extraversion to a low level of extraversion) does only partly lead to the profiles which predispose to perceive a low level of techno-stressors. The analysis for necessary personality traits does not reveal any personality trait to be necessary.

We draw on the differential exposure-reactivity model (Bolger and Zuckerman 1995) to show which personality profiles predispose a user to perceive techno-stressors. Thereby, we extend technostress literature which has so far implied that single personality traits influence users’ perceptions of techno-stressors (Maier et al. 2019).

Personality profiles, consisting of the Big Five personality traits, predispose users to perceive techno-stressors. Organizational research has identified interactions between the Big Five personality traits (Witt 2002), meaning that the influence of a personality trait on an outcome depends on the level of another personality trait, e.g., because the effect of one personality trait can be buffered by another one. In line with the organizational research literature, our results show that the specific influence of a personality trait depends on the other personality traits, as we see that no personality trait is always high or always low in the configurations. Also, we reveal that all Big Five personality traits are relevant for techno-stressor perception as no personality trait plays a subordinate role in all configurations (indicated by a blank space). The post hoc analysis supports the notion that we need to take all Big Five personality traits into account because two personality traits alone do not determine whether the user perceives high or low techno-stressors and are therefore relevant for predicting health-related outcomes such as the perception of techno-stressors. We show that we can account for the interplay of all Big Five personality traits when we consider personality traits in profiles, with each personality profile consisting of a certain level of the Big Five personality traits. We contribute by revealing that investigating personality traits in profiles enables a more nuanced understanding of the interplay of personality traits predisposing to perceive high techno-stressors, which is grounded in a differential exposure (Bolger and Zuckerman 1995) to perceiving techno-stressors.

High and low levels of personality traits, depending on the level of the other personality traits of the personality profile, predispose users to perceive techno-stressors. Prior technostress literature has focused on a linear relationship between personality traits and perceived techno-stressors, i.e. a high level of neuroticism leads to a high level of perceived techno-stressors (Maier et al. 2019). Our results reveal that a high level of a personality trait can be a risk factor in one configuration but - in a different one - it can also be a low level, depending on the combination of the interdependent personality traits that constitute the personality profile. Let us consider neuroticism. Neuroticism can act in two ways, because despite the negative effect of high neuroticism (Maier et al. 2019), low neuroticism can also have a negative effect. In the case of a low level of neuroticism, when combined with low conscientiousness such as in C5, it makes individuals not take care of their issues that need attention (Sadahiro et al. 2015), which can provoke techno-stressor perception. We contribute that personality traits need to be interpreted relatively to the other personality traits, because drawing conclusions about a personality trait in isolation can hardly account for the finely granulated relationships.

There is not one but multiple personality profiles - including different sub-profiles of the Type D personality - that predispose to techno-stressor perception. Personality literature informs us about the Type D personality that has been linked to an increased stress perception (Reich and Schatzberg 2010). In line with the findings on the Type D personality, we find the combination of high neuroticism and low extraversion in the extracted personality profiles that predispose to the perception of techno-stressors. However, we discover that there is not only one Type D personality, but different sub-profiles, e.g., one with high openness to experience (C1) and one with high agreeableness (C2). Considering also the results of the post-hoc analysis, we contribute that it is relevant to distinguish between these different Type D personality sub-profiles, because there is also a Type D sub-profile that predisposes to low techno-stressor perception. This finding suggests that Type D is not necessarily “bad”, but can also lead to low techno-stressor perception.

Moreover, we extend the findings on the Type D personality by showing that there is another personality profile predisposing to techno-stressor perception that does not reflect the Type D personality (C5), but features a low level of conscientiousness. Recent personality literature deduces that the lack of considering conscientiousness in the Type D personality is a shortcoming, because conscientiousness is an important predictor for health-related outcomes (Horwood and Anglim 2017). Individuals who find it difficult to control their impulses, i.e. low level of conscientiousness, show worse health-related outcomes (Horwood and Anglim 2017). Thus, our findings support the considerations that a low level of conscientiousness is important for the prediction of health-related outcomes, i.e. the perception of techno-stressors. In addition, we extend these considerations and show that the other two Big Five personality traits, openness to experience and agreeableness, are also important for the prediction of health-related outcomes such as the perception of techno-stressors. In sum, we contribute that users with one of the identified personality profiles are prone to the modern disease of technostress, which can be a Type D sub-profile, but also a different one.

The insights into how highly perceived techno-stressors come about cannot simply be transferred to how low perceived techno-stressors come about. Considering the results of the post-hoc analysis, we can take these considerations one step further as we see that there is an asymmetric relationship between personality profiles predisposing to high and low techno-stressor perception. In line with a configurational approach, high and low levels of perceptions and behavior can be caused by diverging configurations (Ragin 2014), because high and low levels of the perception might be grounded in different mechanisms. We show that one profile predisposing to the perception of a low level of techno-stressors (C8) is different, and not just the inversion of the profiles predisposing to a high level of techno-stressors. This fact contributes to the insight that transferring the findings from a high level of techno-stressors is not sufficient to reveal how a low level of techno-stressors comes about. Instead, a separate investigation is necessary.

This research also has implications for practice. An important step to take action to mitigate and prevent perceived techno-stressors is that employees using ICTs could assess their personality profile to evaluate their own risk level. This evaluation would enhance the employees' self-awareness concerning whether they are predisposed to perceive techno-stressors. This enhanced self-awareness is the first and an important step to prevent employees from facing techno-stressors. An assessment of the personality profile is increasingly used in organizations (Harrell 2017), e.g., for personnel development processes. Thus, information on personality profiles is becoming more and more accessible. Given that employees might share this information with their executives, our findings help executives to better understand the risk and complexity of techno-stressor perception. At this point, a scattergun approach is not helpful nor effective. Technostress prevention should especially target employees whose personality profile makes them especially prone to perceive techno-stressors, and the techno-stressor prevention strategies should be specific for the respective personality profile. In that way, steps for preventing techno-stressor perception would be target-oriented and focus on the risk group. Interventions that aim at mitigating that users are likely to be stressed by ICTs could target the behavioral and cognitive aspects of the personality traits (Atherton et al. 2014). Let us consider C5: This user has difficulties controlling his or her impulses, i.e. low conscientiousness. Thus, an adjustment of notifications by ICTs could decrease external impulses and thereby the difficulties. At the same time, the strengths of the personality traits of a personality profile can also be used for preventing techno-stressor perception. For instance, a conscientious user such as in C2 or C3 might very likely be able to organize and prioritize his or her ICT-related tasks which would help to reduce the perceived techno-stressors. In line with the discussion on personality profiles, we derive that applying a single measure for a personality trait, e.g., the adjustment of notifications, without considering the whole personality profile is not effective. To offer some initial starting points for techno-stressor prevention for specific personality profiles, we gathered data via an online questionnaire from mTurk workers. We assessed the personality profile, and the final 109 included participants match to one of the personality profiles extracted in our study. Following an explorative approach, we used open questions to ask them which technological, managerial or self-directed strategies help them to reduce their techno-stressor perception. Two researchers of the research team independently carried out the descriptive and interpretive coding (Myers 2013). We summarize the most prevalent strategies in Table 2, respecting that these initial results should be elaborated in a more comprehensive and confirmatory way in future research. In summary, we see that the exploratory approach confirms that the identified personality profiles show different strategies to prevent techno-stressor perception.

Limitations. This study is limited in some ways, but the results offer interesting room for future research. We focused on techno-stressors, which are stressful demands, situations or stimuli caused by ICTs, thus conditions stimulating technostress (Tarafdar et al. 2019). In our research, we identify personality profiles that predispose to perceive technostress stimulating conditions. However, we did not include the adverse technostress reactions, i.e. techno-strain, in our research model. The perception of techno-stressors can lead to techno-strain, but may not always do so (Pirkkalainen et al. 2019). Thus, we cannot deduce from the findings of the current study that the extracted personality profiles also lead to techno-strain, e.g., job burnout or low job performance.

Moreover, our study is limited to profiles of the Big Five personality traits (Costa and McCrae 1997), which are the most general and stable personality traits. We chose the Big Five personality traits because they provide a holistic picture of an individual’s personality (McCrae and Costa 2008) and are found to influence the perception of techno-stressors (Maier et al. 2019). Technostress research has also found further factors that influence the perception of techno-stressors such as demographic (e.g., country of employment) (Krishnan 2017), organizational (Tarafdar et al. 2015), task-related (Galluch et al. 2015), and technological factors (e.g., presenteeism) (Ayyagari et al. 2011), which we did not include in our research but which could affect how personality profiles predispose to techno-stressor perception.

We measured personality with a short scale, because the brevity allows us to reduce missing values and thereby increase the reliability, and shorter measures have also proven reliable and valid (Herzberg and Brähler 2006). Thus, our results are limited to the short scale of measuring personality, which is a proxy for multi-item scales. While we can assure the reliability and validity of our data and short scales for measuring personality have been proven an acceptable proxy of multi-item scales (Herzberg and Brähler 2006), our empirical results do not allow quality measures that compare the short scale with a multi-item scale.

Comparing our data with the data of a large representative sample of Big Five personality profiles (Gerlach et al. 2018) and a data set of Big Five personality profiles applying fsQCA as well (Maier et al. 2020) suggests that our results on personality profiles are representative and transferable. Nevertheless, there are minor differences in the frequency of personality profiles, which make it possible that there exist sparse further personality profiles predisposing to high techno-stressor perception that require a separate investigation of personality profiles in different research contexts.

Finally, we focused on perceived techno-stressors as an aggregated construct that consists of five different dimensions, i.e. techno-stressors. This approach is in line with prior technostress research (Tarafdar et al. 2015) that reached an agreement about these five techno-stressors and treated them as an overall construct. However, technostress studies have theorized and investigated specific techno-stressors (Ayyagari et al. 2011; Galluch et al. 2015; Maier et al. 2015b; Riedl et al. 2012). Our results are limited to the overall construct as we did not investigate the predisposition of personality profiles for each techno-stressor separately. Thus, we do not provide insights into the five specific techno-stressors as our results do not offer a precise understanding of how each specific techno-stressor emerges, although the personality profiles predisposing to a specific techno-stressor might differ.

Avenues for future research. Turning to avenues for future investigations, research could study which personality profiles predispose specifically to perceive each of the five techno-stressors and investigate the role of personality profiles for each techno-stressor separately. This might also include an examination of additional techno-stressors, as there are findings for additional techno-stressors, such as unreliability (Fischer et al. 2019; Riedl et al. 2012).

Moreover, our study focuses on techno-stressors that lead to negative reactions in terms of techno-strain. Recent technostress literature theorizes that techno-stressors can be divided into threat techno-stressors and challenge techno-stressors (Tarafdar et al. 2019). Threat techno-stressors initiate negative effects in terms of techno-strain. Challenge techno-stressors have positive effects as they lead to positive outcomes and do not initiate techno-strain. Our study focuses and is limited to the perception of threat techno-stressors. As challenge techno-stressors are different from threat techno-stressors, challenge techno-stressors might be elicited by different personality profiles than the ones extracted in our study, and future research could investigate which Big Five personality profiles lead to positive effects, i.e. predispose to the perception of challenge techno-stressors.

Finally, with fsQCA we shed light on interactions between personality traits. While fsQCA takes a configurational approach to analyze those interactions that has also proven useful in related IS research contexts (Maier et al. 2020) to study the role of personality profiles, other methods, such as decision trees or polynomial regression modeling, could be used in future research to provide further insights into the interaction and non-linear effects of personality traits.