I'd rather know what to expect … Work unpredictability as contemporary work stressor with detrimental implications for employees’ daily wellbeing

ABSTRACT

Particularly in knowledge-intensive jobs, employees are increasingly challenged by complex and dynamically changing work tasks. These developments make it difficult for employees to anticipate a day's upcoming work tasks and associated activities including methods, time requirements, and potential problems arising in the work process. We present three arguments why this work unpredictability represents a contemporary occupational stressor causing that affects employees until beyond working hours and is thus associated with lower daily wellbeing in the evening: Work unpredictability can be perceived as a lack of control at work, as a lack of mastery expectancies, and it might add high-effort planning and self-regulation demands to employees’ daily psychological workload. In a diary study with 105 employees, we collected 666 observations at three daily measurement occasions over two weeks. The results supported our hypotheses and demonstrated that work unpredictability relates negatively to evening serenity via employees’ elevated strain levels after work. These relationships were also found when controlling for time pressure as a representative of an established daily work stressor. We conclude that work unpredictability is a so far neglected work stressor that should receive more research attention in the future.

KEYWORDS:

• Work anticipation
• cognitive activation theory of stress
• stress reaction
• multi-level path model

Introduction

Over the last decades, occupations characterised by the performance of knowledge-intensive work with a heavy reliance on information and communication technology experienced a shift from regularly recurring line operations to situation-sensitive project work (Schoper et al., 2018). These developments brought two major challenges for occupational groups in the constantly growing service sector, including managers, professionals (e.g. science, engineering), service and sales workers, and clerical support workers (derived from the classification of occupations of the United Nations; International Labour Organization, 2016). The first challenge is that, due to the flexible prioritisation of projects within organisations dependent on the market situation, employees’ work tasks can dynamically change at any time (Mohr & Wolfram, 2010). The second challenge is the increase in complex work tasks, as project work involves irregularly changing activities and thus has a lower proportion of routine tasks (Schoper et al., 2018).

These profound changes in the world of work have been accompanied by a considerable fluctuation in the ability of employees to predict their daily work tasks and processes. “Predictability involves a relatively high stability of work and a lack of unexpected changes, which characterised the earlier industrial era which had stable production systems” (Väänänen et al., 2008, p. 2264). However, nowadays, the possibility that the prioritisation of projects due to organisational strategy can change any time makes it difficult for employees to predict which work tasks they will have to accomplish in a day (Väänänen et al., 2008). Moreover, with increasing task complexity, difficulties in anticipating the task completion process raised as well, including the methods and time needed to perform the work activities as well as any problems that may come up in the work process (Mohr & Wolfram, 2010).

The first purpose of this study is to provide theoretical reasoning why work unpredictability represents a contemporary, daily varying work stressor. Based on cognitive activation theory of stress (CATS; Ursin & Eriksen, 2004), we define a work stressor as employees’ perception of a mismatch between what they think is required from them in a certain situation and what they would have desired themselves. Drawing on stress and motivation literature, we present three arguments why this applies to work unpredictability: First, work unpredictability is likely to be perceived as a lack of control which hinders employees from feeling autonomous. Second, work unpredictability hinders employees from taking anticipatory actions and can thus cause low expectancies to master one’s work successfully (Meurs & Perrewé, 2011) and increase associated feelings of incompetence. Feeling autonomous and competent have both been described as innate human needs whose satisfaction and frustration, respectively, influence work-related motivation and wellbeing (Deci et al., 2017). Third, work unpredictability is associated with less low-effort routine work activities and more high-effort planning and regulation demands that add to employees’ psychological workload (Mohr & Wolfram, 2010). High levels of workload are known to be a daily varying stressful work experience (Ilies et al., 2010).

The second purpose of this study is to provide empirical evidence for a relationship between work unpredictability and employees’ daily wellbeing. Drawing again on CATS (Ursin & Eriksen, 2004), we hypothesise that employees’ work unpredictability is associated with a stress reaction that sustains beyond their working hour of a day and, in turn, links higher levels of day-specific work unpredictability to lower levels of their serenity in the evening on the same day. To test our hypotheses, we conducted a daily diary study among employees in knowledge-intensive jobs. This approach enables us to consider daily variations in the perception and interpretation of work situations and individuals’ psychological responses to them (see also Ohly et al., 2010). We chose serenity as an indicator of wellbeing because it is a low activated positive emotional state (Cropanzano et al., 2003) that indicates a reduction of stressor-induced elevated strain, enabling mental recovery outside of working hours (Sonnentag & Fritz, 2007). Daily serenity and thus calming down psychologically after the encounter of a work stressor is critical because, in the long-term, sustained stress reactions are pathogenic pathways (Brosschot et al., 2005) leading to chronic physical and psychological impairments such as cardiovascular disease, burnout, and reduced life satisfaction (Sonnentag & Frese, 2012).

Our study represents a notable and necessary addition to the work stress literature. Research so far acknowledged the role of predictability as a work resource that relates to higher wellbeing (Flovik et al., 2019) and buffers the detrimental impact of stressful work situations (Mohr & Wolfram, 2010; Rosen et al., 2020), but failed to see the relevance of work unpredictability for wellbeing. However, related empirical findings on the unpredictability of work schedules (Scholarios et al., 2017) and of major organisational changes (Nielsen et al., 2006) provided evidence for a detrimental influence on employees’ wellbeing and thus indicate that work unpredictability is more than just the absence of work predictability. More specifically, as it seems to be a work characteristic with raising salience in knowledge-intensive jobs, it is crucial to investigate the unpredictability of work tasks and activities (further referred to simply as work unpredictability) and its potential to harm employees’ wellbeing on a daily level.

Work unpredictability as daily work stressor

Based on Mohr and Wolfram's (2010) definition of work predictability, we define work unpredictability as employees’ inability to anticipate which work-related tasks and activities they will have to perform on a day, including methods, time requirements, and potential problems arising in the process. This new form of work unpredictability is likely to be prone to daily variations (e.g. on certain days employees might face more difficulties to predict what lies ahead than on other days). It further differs from the unpredictability of longer-term events such as major organisational changes (see Nielsen et al., 2006) and career developments such as job insecurity (Vander Elst et al., 2012) that are nevertheless thought of to be detrimental for employees’ wellbeing.

CATS (Ursin & Eriksen, 2004) elaborates that individuals tend to evaluate an environmental stimulus as a stressor if they perceive a discrepancy between their actual experiences and what they would have desired. If this discrepancy is accompanied by negative outcome expectancies, a stress reaction will be triggered (Ursin & Eriksen, 2004). Negative outcome expectancies refer to the subjective belief that a situation will not end well for oneself, whereas positive outcome expectancies refer to the subjective belief of being able to master the current situation which is associated with only a mild stress reaction that fades quickly (Meurs & Perrewé, 2011).

We present three reasons why work unpredictability is likely to be perceived as a discrepancy between what is and what would have been desired followed by negative consequences. First, work unpredictability can be experienced as a lack of control at work. The perception of control over one’s life is “a major focus of interest in understanding the stress process and in explicating the possible links among stressors, psychological wellbeing, health, and functioning” (Price et al., 2002, p. 304). More precisely, as individuals have an innate need to experience ownership of their actions (e.g. to plan and to make decisions for themselves) and to act with a sense of volition (referred to as the need for autonomy; Deci et al., 2017), a perceived lack of control at work will cause a discrepancy with their desire for control over their work activities. As work unpredictability may impede employees’ planning and decision-making, it seems likely that employees’ experience an inability to predict their work tasks and activities as a lack of control at work. In other words, we argue that work unpredictability can make employees aware of their lack of control at work, thus emphasising a discrepancy associated with negative personal consequences.

Second, work unpredictability can be associated with decreased expectancies to master one’s workday successfully. While work predictability permits to take specific proactive actions in order to adapt to upcoming situations (Lazarus, 1991), the inability to foresee whether or not particular challenging situations might occur at work hinders individuals from “psychologically defend against or distort the stressor” (Meurs & Perrewé, 2011, p. 1050). They may also struggle to take specific anticipatory actions, for example seeking additional resources in a timely manner because they were confronted with a last-minute task change. Consequently, by actually decreasing their scope of action, work unpredictability may lower employees’ prospects to master the upcoming work tasks and activities successfully. A lack of mastery expectancies, however, may cause a discrepancy as individuals have an innate desire to feel capable of mastering their environment (i.e. need for competence; Van den Broeck et al., 2008). The frustration of this need poses a personal threat and reduces employees’ work-related motivation and wellbeing (Deci et al., 2017).

Third, work unpredictability makes work more intellectually demanding due to high-effort planning and self-regulation demands. Whereas work predictability enables the implementation of routines that allow action regulation on an automatic, lower-demanding level, employees that are not able to foresee their work tasks and activities have to constantly develop action plans and to evaluate their actions, a situation associated with the exertion of high-level intellectual processes and cognitive overload (Mohr & Wolfram, 2010). Acts of intellectual regulation require psychological efforts and thus add to employees’ psychological workload (i.e. all currently perceived work demands and tasks) on a given day (Ilies et al., 2010). As high levels of workload up to cognitive overload are negatively associated with wellbeing (Bowling et al., 2015), we argue that employees may evaluate the unpredictability of work as a stressor due to the associated high workload, which is a discrepancy with a more desirable moderate level of workload. A negative relationship between psychological workload and performance (Fan & Smith, 2017) further suggests that employees might expect negative outcomes from this perceived discrepancy.

Our reasoning may not be conclusive for work unpredictability alone, it also applies to other, well-known work stressors such as time pressure and interruptions. Similar to our arguments why work unpredictability is a daily work stressor, Sonnentag et al. (2018) stated that work-related time pressure and interruptions may be associated with low goal achievement and high workload. However, time pressure conceptionally differs from the unpredictability of work tasks and activities insofar as work unpredictability does imply that one lacks information to foresee what is coming up next, but not that one has to rush in order to achieve the work goals. Moreover, while interruptions are inherently unpredictable, they are typically distractions from a primary task that delay goal achievement (Baethge & Rigotti, 2013). This is not the case for work unpredictability.

Sustained stress reaction and daily wellbeing

Stressors cause a stress reaction (Sonnentag & Frese, 2012) that remains persistent, even after the working hours have ended and the initial stressor is no longer present (Meurs & Perrewé, 2011). Individuals’ psychological reaction to a stressor is mostly conceptualised as high activated negative feelings of strain, irritation or anxiety (Sonnentag & Frese, 2012). The stress reaction is supposed to put the individual in an uncomfortable position in order to prompt them to take action and remove the cause of stress (Ursin & Eriksen, 2004). Actions to deal with a stressor are manifold and include “worry and rumination about the stressor, attempts to deal with the work-related problems, or engagement in additional work tasks during nonwork time” (Sonnentag & Fritz, 2015, p. 91). The stress reaction (i.e. feelings of strain) will sustain until the discrepancy between what is and what is desired could be dissolved (Ursin & Eriksen, 2004).

In the context of work unpredictability, feelings of strain may sustain beyond working hours because employees may continuously take actions that aim to cope with the aftermath of their lack of control, lack of mastery expectancies, or high psychological workload, or that aim to prevent these circumstances in the future. Employees might, for example, use their time after work to create a plan on how to handle the next situation in which the supervisor changes the prioritisation of tasks or projects, or to do research outside working hours in order to get familiar with a new project so that they can anticipate the next work steps and workdays. Anyway, constantly thinking about work outside working hours confronts them with the work stressor and thus re-create the stress reaction, even beyond the working hours of one day (Sonnentag & Fritz, 2015). Previous empirical findings accordingly linked low levels of perceived control (Vander Elst et al., 2012) and a frustration of one’s need for competence at work (Bartholomew et al., 2011) to increased levels of employees’ negative emotions on the same day. Moreover, Ilies et al. (2010) showed that high workload relates to higher feelings of strain after work on the same day. Consequently, we hypothesise that work unpredictability is associated with employees’ higher levels of strain reported after work on the particular day.

Hypothesis 1: The higher employees’ perceived work unpredictability on a particular day, the higher their level of strain after work on the same day.

The stress reaction and its duration is key in explaining the connection between stressors and more distant outcomes such as health and wellbeing (Brosschot & Thayer, 1998), yet it is often neglected by empirical stress research (Meurs & Perrewé, 2011). People feel subjectively well when they feel many pleasant and few unpleasant emotions (Cummins, 2010; Diener, 2000), and individuals’ affective states appear to be particularly mutable following stressful experiences (DeLongis et al., 1988). In this study, we assess serenity as a measure of positive feelings such as being calm and relaxed (see Abele-Brehm & Brehm, 1986) that are thought to be beneficial for employees as they put them in a low activated psychological state in which exhausted energy reserves can be restored (Sonnentag & Fritz, 2007). However, before a calm state of serenity can be achieved after work, employees’ physiological as well as psychological activation associated with the stress reaction must decrease (Geurts & Sonnentag, 2006).

A sustained stress reaction marked by feelings of strain will hinder employees from engaging in relaxing activities that would be needed to psychologically calm down after work (Sonnentag & Fritz, 2007) and to reach a comfortable state of serenity. Especially a person’s positive affective state can significantly fluctuate within the course of a day (Saksvik-Lehouillier et al., 2021). Positive affect further seems be influenced by negative feelings as empirical findings showed that the higher the strain employees felt following work stressors such as time pressure (Sonnentag et al., 2018), technical difficulties and ambiguous work tasks (Ohly & Schmitt, 2015), the less positive emotions they felt on the same day. Consequently, we hypothesise an indirect relationship between work unpredictability and lower levels of serenity in the evening via employees’ strain after work as mediator.

Hypothesis 2: Perceived work unpredictability will indirectly relate to lower serenity in the evening of the same day via higher levels of strain after work.

Method

As work unpredictability, strain, and serenity are each reflecting individual experiences and psychological states that are subject to intrapersonal variation from day to day, we conducted a diary study to observe the hypothesised path model on a daily level. To reduce common method bias (Podsakoff et al., 2012) and thus to increase the informative value of our study, we separated the daily measurement of all focal variables. Moreover, we included two control variables into our study: First, to demonstrate that work unpredictability causes a stress reaction independently of the influence of other common daily varying work stressors on participants’ after-work strain, we included daily time pressure in our study. Second, to underline the specific relevance of the sustained stress reaction for employees to develop high levels of serenity during their off-work hours until the evening, we controlled for after-work serenity when regressing evening serenity on after-work strain on the day-level.

Sample and procedure

Participants were recruited by psychology master’s degree students in the course of a class; students’ help with recruitment was an optional class task and thus voluntary. The students were instructed to recruit employees who work at least 20 hours in knowledge-intensive jobs because as we assumed that work unpredictability is especially salient for this occupational group (Beerheide & Katenkamp, 2011; Flovik et al., 2019). In the course of recruitment, a total of 126 individuals that met our inclusion criteria announced their interest to take part in the study and filled out an initial online questionnaire that collected information on demographics and on the job.

After receiving their completed initial questionnaire, the first author invited the participants to respond to three questionnaires per day over the course of two consecutive work weeks. From Monday to Friday, participants received the link via SMS or e-mail to the first daily questionnaire at 5 pm (which they should fill in at the end of the workday), to the second at 7 pm (which they should fill in about one to two hours after work), and to the third at 9 pm (which they should fill in before going to bed). We collected a total of 846 datasets from the at-work questionnaire (capturing work unpredictability and time pressure), 812 from the after-work questionnaire (capturing strain and serenity), and 750 from the evening questionnaire (capturing serenity). We matched day-level datasets from the same individual and for the same day if they were provided in the correct chronological order. Data matching was anonymously based on a code each participant entered on each questionnaire and resulted in 666 day-level observations nested within 105 individuals.

As 21 participants failed to provide day-level data and thus were excluded from the final sample, we tested for systematic attrition. Therefore, we conducted a two-sample t-test and compared the means of the dropped-out participants (N = 21) and those from the final study sample (N = 105). We examined differences in sociodemographic and occupational characteristics. Dropped out participants did not differ significantly from those in the final sample with regard to gender, age, education, job tenure, occupation, working hours per week, leadership position, and job satisfaction. These results indicated that the attrition of our study participants was random and not systematically biased.

In the final sample, participants’ mean age was 34.8 years (SD = 11.5), 56.2% were female. They worked on average 39.6 hours per week (including overtime, SD = 9.3) and reported a mean position tenure of 6.6 years (SD = 8.7). Over a third of the sample (35.2%) reported having leadership responsibilities. Categorisation based on the classification of occupations of the United Nations (International Labour Organization, 2016) indicated that our sample consisted of professionals (41.2%), clerical support workers (20.6%), service and sales workers (18.6%), and managers (13.7%). Among the represented occupations, service and sales workers (M = 2.6, SD = 0.9), professionals (M = 2.6, SD = 1.1) and clerical support workers (M = 2.6, SD = 2.0) reported the highest mean perceptions of work unpredictability, followed by managers (M = 2.1, SD = 0.8).

Measures

Questionnaires were provided in German. Participants’ were asked to rate the questionnaires items – which were all phrased as personal statements – on a 7-point Likert scale ranging from 1 (fully disagree) to 7 (fully agree). Reliability estimates (Cronbach’s alpha) of all scales are displayed in Table 1.

Table 1. Means, standard deviations, reliabilities, intraclass correlations and zero-order correlations on the day-level and the person-level.

Variables	Person-level			Day-level		ICC	1	2	3	4	5
	M	SD	α	SD	α
1. Work unpredictability (at work)	2.50	1.07	.99	0.87	.76	.60		.25***	.26***	−.20***	−.08
2. Time pressure (at work)	2.79	0.95	.91	1.14	.75	.41	.54***		.15***	−.17***	−.05
3. Strain (after work)	2.27	0.70	.94	1.02	.84	.32	.49***	.46***		−.71***	−.46***
4. Serenity (after work)	4.98	0.86	.97	1.02	.86	.42	−.36***	−.30**	−.82***		.53***
5. Serenity (in the evening)	5.27	0.80	.98	0.90	.82	.44	−.48***	−.37***	−.83***	.97***

Notes: ICC, intraclass correlation coefficient; α, Cronbach’s alpha reliability estimates. Correlation coefficients above the diagonal are day-level (N = 666), below the diagonal are person-level correlations (N = 105). *** p < .001, ** p < .01 (two-tailed).

Work unpredictability

We assessed the level of work unpredictability participants perceived during a workday in the first diary questionnaire of a day. We adapted four items from Schyns and Paul (2004) that had already been used to measure predictability in previous research (Mohr & Wolfram, 2010). We made sure that the adapted items explicitly focussed on employees’ desire for more work predictability on a given workday in order to capture their perception of a discrepancy between their actual work situation today and what they would have desired (which is what makes an environmental stimulus a stressor in the first place; Ursin & Eriksen, 2004). The items measured the perceived unpredictability of work tasks (i.e. “When I think back to my day at work, I would have liked to know earlier … what work tasks I have to accomplish today”), associated work activities (i.e. “ … what's facing me at work today”), their time requirements (i.e. “ … how much time it would take to complete today's work tasks”), and related problems (i.e. “ … what problems come with today's work tasks”).

Strain

We measured participants’ sustained stress reaction after work by their high activated negative feelings with the second daily questionnaire. We applied three items from the German mood survey (Abele-Brehm & Brehm, 1986) and asked participants to rate whether they felt “strained”, “irritated”, and “grumpy” at that moment.

Serenity

We assessed participants’ serenity in the evening as indicators for their day-specific wellbeing in the third questionnaire of a day. We used three items from the German mood survey (Abele-Brehm & Brehm, 1986) capturing low activated positive feelings and asked participants to indicate whether they felt “serene”, “relaxed”, and “calm”.

Control variables

First, to demonstrate that work unpredictability causes a stress reaction that goes beyond the influence of time pressure (as a representative of another common work stressor), we included participants’ daily time pressure in our study. Time pressure is defined as the extent to which workers perceive that they do not have enough time to complete their work tasks or need to work faster than usual (Semmer et al., 1999). We assessed time pressure in the first daily questionnaire immediately after work unpredictability with three items from the instrument for stress-oriented job analysis (Semmer et al., 1999) adapted to the diary study design. A sample item reads “Today, at work, I was under time pressure”.

We argue that time pressure is likely to cause a discrepancy between what is and what is desired due to similar reasons as work unpredictability: As time pressure stems from external inputs such as constraints or deadline or a failed planning of the employee, it may be associated with a lack of control at work. Moreover, it holds the risk of not being able to accomplish all work task of the day (i.e. a lack of mastery expectancy) and needs a high effort investment (see Sonnentag et al., 2018) that adds to employees’ daily psychological workload. However, in contrast to work unpredictability time pressure is an acknowledged work stressor with daily variations and a well-documented association with strain that can outlast working hours (Baethge & Rigotti, 2013). Furthermore, the stress reaction following time pressure was related to lower wellbeing (e.g. Höge, 2009).

Second, we controlled for after-work serenity (measured in the second daily questionnaire with the same items as in the evening) when regressing evening serenity on after-work strain. This approach should demonstrate that higher levels of strain after work following work unpredictability predict lower levels of employees’ evening serenity beyond the predictive power of their prior state of serenity as reported after work. However, we controlled for after-work serenity only on the focal day-level (and not on the person-level) after centering the variable around the group mean (see Paccagnella, 2006). We did so because of the difference between a person’s mood, which refers to milder affective states of longer duration and often is present outside a person’s awareness, and emotional state determined by their presently salient feelings (Cropanzano et al., 2003) that can “ebb and flow over the course of a day” (Russell & Carroll, 1999, p. 5). Accordingly, as their aggregated estimates are not sensitive to varying feelings within a few hours, emotional states tend to have high stabilities on the person-level (see Sonnentag & Lischetzke, 2018).

Multi-level confirmatory factor analysis

To determine the distinctiveness of our measures, we conducted a multi-level confirmatory factor analyses with Mplus 8 (Muthén & Muthén, 2017), using a robust maximum likelihood estimation (MLR). The hypothesised five-factor model (i.e. work unpredictability, time pressure, strain, after-work serenity, and evening serenity as one factor each) showed a good model-fit, χ²(188) = 291.14, comparative fit index (CFI) = .98, Tucker-Lewis index (TLI) = 0.97, root mean square error of approximation (RMSEA) = .03, and Akaike Information Criterion (AIC) = 30,092.97. The five-factor model fitted our data significantly better than the best-fitting four-factor model (strain and after-work serenity as one common factor), χ²(196) = 525.10, CFI = .93, TLI = 0.91, RMSEA = .05, AIC = 30,317.75, Sattora-Bentler scaled Δχ² = 228.53 (considering a scaling correction factor of 1.05), Δdf = 8, p < .001. We found a low but significant correlation between work unpredictability and time pressure. The five factor-model further had a better fit than the best-fitting three-factor model (work unpredictability and time pressure as well as strain and after-work serenity as one common factor), χ²(202) = 1,027.42, CFI = .80, TLI = 0.76, RMSEA = .08, AIC = 30,934.74, Sattora-Bentler scaled Δχ² = 953.20 (correction factor of 0.77), Δdf = 14, p < .001. The hypothesised five factor-model had a better fit than the best-fitting two-factor model (work unpredictability and time pressure as well as strain, after-work serenity, and evening serenity as one common factor), χ²(206) = 1,319.83, CFI = .76, TLI = 0.72, RMSEA = .09, AIC = 31,137.14, Sattora-Bentler scaled Δχ² = 874.65 (correction factor of 1.24), Δdf = 18, p < .001, and finally also a better fit than the one-factor model, χ²(154) = 2,097.35, CFI = .60, TLI = 0.54, RMSEA = .12, AIC = 31,988.42, Sattora-Bentler scaled Δχ² = 1,355.19 (correction factor of 1.43), Δdf = 20, p < .001. These results indicate that the study variables can be distinguished from one another.

Method of data analysis

As day-level data is nested within persons, we tested all hypotheses by means of a multi-level path model using Mplus 8 (Muthén & Muthén, 2017) which separated the variance of all study variables into their between person (i.e. person-level) and the focal within person (i.e. day-level) components (Zhang et al., 2009). We examined the hypothesised indirect associations between work unpredictability and serenity in the evening via after-work strain (Hypothesis 2) using the Monte Carlo method (Preacher & Selig, 2010) with 1,000,000 repetitions and a 95% confidence interval (CI).

Results

Preliminary analyses

The means, standard deviations, and zero-order correlations among the variables on the day-level and the person-level are shown in Table 1. Before testing the hypotheses, we examined between-person (i.e. person-level) and within-person (i.e. day-level) reliabilities, within-person consistencies, and the empirical distinctiveness of work unpredictability and the control variable time pressure. First, Cronbach’s alphas were estimated using a multi-level approach in Mplus (see Geldhof et al., 2014). Alpha estimates ranged, on the person-level, from .91 to .99 and, on the day-level, from .75 to .86 (see Table 1). This indicated that our data were derived from reliable scales.

Second, we calculated the intra-class correlation coefficients (ICCs) which reflect the percentage of within-person consistencies (as the opposite of variation) in participants’ responses to the items defining a variable at the day-level, calculated with the following formula (see Asparouhov, 2006): variance between person / (variance between person + variance within person). The ICCs (also shown in Table 1) ranged from 32% (strain after work) to 60% (work unpredictability), indicating that participants’ responses varied throughout the days of their study participation and justifying a day-level data analysis.

Third, we investigated the empirical distinctiveness of work unpredictability and time pressure as representative of another work stressor with similar implications for employees’ daily wellbeing. Data showed a low but significant correlation between the two variables that could also be distinguished by the confirmatory factor analysis: The five-factor model with all items loading on their respective factors – that thus distinguished between work unpredictability and time pressure – on the day and on the person-level fitted our data significantly better than all alternative models with a fewer number of factors. These preliminary results indicate that both work stressors can be statistically distinguished and are thus in favour of work unpredictability’s discriminant validity.

Hypotheses testing

Table 2 shows the significant direct regression estimates (direct effects) between the variables on the day-level derived from our multi-level path model including after-work serenity and time pressure at work as control variables. We report standardised estimates using the Mplus function StdYX for linear regression that “uses the variances of the continuous latent variables as well as the variances of the background and outcome variables for standardisation” (Muthén & Muthén, 2017, p. 799).

Table 2. Direct effects from the multilevel path model on the day-level and the person-level.

	Strain (after work)	Serenity (in the evening)
Day-level estimates
Work unpredictability (at work)	.23*** (.05)	.05 (.08)
Time pressure (at work)	.09* (.05)	.04 (.04)
Strain (after work)		−.21*** (.06)
Serenity (after work)		.40*** (.06)
Person-level estimates
Work unpredictability (at work)	.33*** (.13)	−.09 (.12)
Time pressure (at work)	.28*** (.13)	.10 (.12)
Strain (after work)		−.88*** (.07)

Notes: N = 105 participants, 666 daily observations, standardised regression estimates, and standard errors (in parentheses) are reported. *** p < .001, * p < .05.

In support of Hypothesis 1, our findings show that the higher participants’ perceived work unpredictability on a day, the higher their strain levels after work on the same day. This finding is in favour of Hypothesis 1 proposing a positive association between work unpredictability and employees’ stress reaction that sustains beyond working hours and beyond the impact of time pressure at work.¹

When considering sustained strain after work as a mediator, higher levels of strain after work linked the unpredictability of a workday to lower levels of serenity in the evening (γ = −.05, SE = .02, 95% CI [−.0459, −.0003]). These results are in favour of Hypotheses 2 proposing an indirect negative relationship between work unpredictability and employees’ wellbeing in the evening. The significant indirect effect further shows that after-work strain adds to the variance of lower levels of evening serenity beyond the influence of their prior state of serenity reported after work.²

Discussion

Given the rise of non-routine and dynamically changing project work in knowledge-intensive jobs (see Schoper et al., 2018), the goal of our study was to contribute to the understanding of work unpredictability as contemporary work stressor within this particular occupational group. We argued that employees may associate the unpredictability of their daily work tasks and activities as a lack of control at work, a lack of mastery expectancies, or an increase of high-effort intellectual work demands adding to their psychological workload. Either way, there might be days when employees experience a discrepancy between what is and what they would have desired along with the risk not being able to master their unpredictable work tasks and activities – a situation causing a stress reaction (i.e. feelings of strain) that sustains until the perceived discrepancy could be dissolved (Ursin & Eriksen, 2004).

Following CATS (Ursin & Eriksen, 2004), we proposed that high levels of work unpredictability cause a stress reaction that sustains beyond the working hours of a day and, in turn, relate to lower levels of employees’ wellbeing in the evening before going to bed. Results of our diary study supported our hypotheses: On days when employees were less able to foresee their work tasks and activities in terms of work methods, required time, and associated problems, they felt more strained, irritated, and grumpy after work as compared to days when they perceived less work unpredictability. And this relationship was significant even after considering the effects of time pressure as another daily work stressor with well documented negative implications for employees’ affective states during and after work (e.g. Baethge & Rigotti, 2013; Sonnentag et al., 2018). Furthermore, on days when employees reported to feel more strained, irritated, and grumpy after work, they also felt less serene, relaxed, and calm in the evening before going to bed than on workdays they felt less strain after work.

To our knowledge, our study is the first to conceptualise the unpredictability of work tasks and activities as a distinct work characteristic that is more than just the absence of work predictability (which has been describes as job resource; Mohr & Wolfram, 2010; Rosen et al., 2020), but a work stressor itself with negative implications for employees’ wellbeing. Our study further contributes to the stress literature by emphasising the relevance of the stress reaction and its duration in explaining the connection between stressors and more distant health and wellbeing outcomes (Brosschot & Thayer, 1998). As this link is often neglected by empirical stress research (Meurs & Perrewé, 2011), it is important that two of our findings provided support for it: First, work unpredictability and evening serenity were not directly associated but mediated via after-work strain. Second, the negative relationship between after-work strain and evening serenity was significant even after controlling for after-work serenity. This demonstrates that higher levels of strain after work following a work stressor statistically account for lower levels of evening serenity beyond the share of variance explained by their state of serenity a couple of hours before.

Limitations

Despite the new insights this study has facilitated, it has also some limitations. First, we applied a rather uncommon measure of work unpredictability which can be considered a limitation, especially when it comes to its comparability with other work stressors. We adapted the items of Schyns and Paul (2004) according to CATS (Ursin & Eriksen, 2004), stating that only a perceived discrepancy between one’s experiences and desires accounts for the transformation of a stimulus into a stressor. To measure the difference between the actual experiences (as subtrahend) and the desires (as minuend), we formulated the items in a way in which the participants indicated their wish for more predictability on the particular workday (i.e. “I would have liked to know earlier … ”). In contrast, we captured time pressure traditionally by asking participants directly about the experienced time pressure (items derived from Semmer et al., 1999) rather than the perceived discrepancy. We argue that our measurement of work unpredictability more validly depicts the actual experience of the specific stressor, yet it can be considered a limitation when it comes to its comparability with other work stressors such as time pressure.

Second, we did not distinguish work unpredictability from other work stressors besides time pressure. Although we conceptually separated work unpredictability from time pressure and interruptions (as representative of other daily varying work stressors), we only included items for time pressure into our questionnaire in order not to jeopardise the success of our rather elaborate study with three daily measurements. We could however show that work unpredictability is a statistically unique factor and relates to after-work strain beyond the impact of time pressure, but we did not demonstrate its distinctiveness from interruptions.

Third, as we conducted our study among employees with knowledge-intensive jobs, our findings do not provide information about the experience of other workers. We did so because we argued that this occupational group is especially challenged by the unpredictability of their work tasks and activities. Yet, of course, employees in other jobs may also experience work unpredictability. Future research should aim to further explore the distribution of this work characteristic. For example, collecting data from a broader sample would enable a comparison between the experiences of various occupational groups.

Avenues for future research

In order to maintain a healthy and thriving workforce, it is relevant to recognise and to further investigate work unpredictability which we introduced as contemporary work stressor. We specifically invite future research to examine other possible aspects of work unpredictability, its differences across jobs, and its relationship to other work experiences, such as motivation and procrastination – as elaborated in the following. The first avenue for future research could be the examination of other possible aspects of work unpredictability. We focused on employees’ inability to foresee their work tasks and activities of a workday in terms of methods, time requirements, and potential problems. However, over the last decades, work has, for example, also changed in terms of the implementation of flexible working arrangements (Korunka & Kubicek, 2017) that potentially add unpredictability of daily working hours and locations to the working life of many employees.

A second avenue for future research could be the investigation of differences across jobs. Although we focused on the day-level, we observed the hypothesised relationships between work unpredictability, strain, and serenity also at the person-level (see Table 2). These results not only reveal daily variations in work unpredictability between an employee's workdays, but also differences in the average levels of perceived work unpredictability among multiple employees of varying occupations. Consequently, different knowledge-intensive occupations might face different levels of unpredictability (e.g. office clerks might face more work unpredictability than managers who are more involved in the organisational strategies that determine task changes) which should be addressed by future studies.

A third avenue for future research is the investigation of the relationship between work unpredictability and other work experiences, such as motivation and procrastination. It seems likely that, over a certain period of time, certain critical levels of work unpredictability will have negative implications for workers’ intrinsic work motivation. We argued that employees might associate their inability to foresee their workday with a lack of control or low expectancies to master their workdays successfully. Both the need for autonomy and the need for competence are crucial antecedents to develop intrinsic motivation which refers to employees’ personal interest in and enjoyment of their work activities (Deci et al., 2017).

Moreover, an inability to foresee work activities might foster procrastination. Work procrastination refers to the delaying to begin or proceed an intended course of action which is usually seen as unfavourable work behaviour (Steel, 2007). However, procrastination has also been touted out as functional strategy to avoid rushing into things and wait until new information comes along (Chu et al., 2005). Consequently, employees experiencing uncomfortable high levels of work unpredictability might procrastinate in an attempt to decrease their negative outcome expectancies or to increase their positive outcome expectancies, respectively. And this delay of action will further decrease their intellectual effort and self-regulation (see Prem et al., 2018) and thus reduce work unpredictability’s affect as a stressor due to increased psychological workload.

Practical implications for the management of work

Although a certain degree of work unpredictability is probably unavoidable given the increasing complexity and dynamics of tasks associated with project work, we can think of some measures that counteract the detrimental implications of work unpredictability for employees’ wellbeing. Following the stress process described by CATS (Ursin & Eriksen, 2004), these measures may either aim to reduce individuals’ perception of work unpredictability per se (and thus the discrepancy between what is and what would have been desired) or their negative outcome expectancies following a perceived work unpredictability.

In order to reduce work unpredictability, organisations and supervisors should foster the transparency of their strategies as well as fast and accessible communication methods including early warnings when projects and their prioritisation are about to change. This might be achieved either directly by implementing strict communication guidelines or indirectly by supporting a climate of mutual exchange and transparency among the staff of various working teams and across hierarchies. This way, employees will receive more information on potential project changes enabling them to anticipate their upcoming tasks.

In order to reduce negative outcome expectancies, organisations and supervisors should foster employees’ learning from stressful situations. “If employees are allowed to make mistakes without being subjected to punishment or abuse and they are encouraged to learn from their mistakes, they will be more likely to anticipate positive outcomes and interpret their environment as challenging as opposed to threatening” (Meurs & Perrewé, 2011, p. 1057). In such a safe environment, regular feedback processes will further increase employees’ chances to learn from their prior work experiences (Chiaburu & Harrison, 2008) and thus to develop skills and self-efficacy beliefs (Holman & Wall, 2002). High skills and self-efficacy beliefs foster employees’ attitude to evaluate stressors as personal challenge associated with positive outcome expectancies (Meurs & Perrewé, 2011). However, employees can also take the chance and learn from their co-workers’ experiences which can be fostered by mentoring or the sharing of work experiences between team members.

Conclusion

Our diary study contributes to the literature by showing that perceived unpredictability of a workday’s tasks and activities has detrimental implications for employees’ evening wellbeing due to a stress reaction that sustains beyond working hours. With these findings, we underpin our theoretical argument that work unpredictability is a stressful, daily varying work experience that can be associated with a lack of control, a lack of competence expectancies, and high workload. In order to maintain a healthy and thriving workforce, it is relevant to recognise and to further investigate this contemporary work stressor salient in the context of knowledge-intensive work. Beyond that, we invite future research to examine other possible aspects of work unpredictability, its differences across jobs, and its relationship to other work experiences, such as motivation and procrastination.

Acknowledgements

This research was supported by the Austrian Academy of Sciences.

Disclosure statement

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

Notes

1 The association between work unpredictability and after-work strain was significant even when time pressure was not included in the analysis (γ = .26, SE = .05).

2 The association between after-work strain and evening serenity was significant even when after-work serenity was not included in the analysis (γ = −.13, SE = .03, 95% CI [−.1814, −.0746]).