Effects of a Participative Workplace Intervention on Work Strategies and Expectations of Availability Among Office-Based Employees With Flexible Work Arrangements

Occupational Applications

Work strategies changed following a participatory workplace intervention among office-based employees with flexible work arrangements (FWA). Also, the intervention likely led to clearer rules and routines for FWA within the work group. As FWA increases, especially after the COVID-19 pandemic, it will be important to revise work strategies for both the individual and the work group. The results of this study are relevant in the context of interventions that can support organizations and employees in adopting work strategies promoting good working conditions and health in FWA.

TECHNICAL ABSTRACT

Background

Flexible work arrangements (FWA) are common, but knowledge on how to organize flexible work to reduce negative consequences and preserve positive aspects is currently sparse, which hampers organizational initiatives.

Purpose

This study aimed to determine the extent to which work strategies, work-related use of information and communication technology (ICT) outside regular working hours (i.e., use of laptop, tablet, or smartphone, to handle information and facilitate communication), perceived productivity, expectations of availability, and clarity of expectations about availability, had changed among office-based employees with FWA 2 and 4 months after a participative two-step workplace intervention.

Methods

An intervention group (n = 97) was compared to a control group working as usual (n = 70). The intervention, initiated and approved by the top management of the organization, included individual education intended to change work strategies, and workshops developing common rules and routines for FWA within the work group.

Results

Participants were satisfied with the intervention and reported larger changes than the control group in work strategies. No statistically significant effects were found, however, on ICT use, perceived productivity, or expectations of availability.

Conclusions

This participative workplace intervention was successful in changing employees work strategies but may not be effective in influencing ICT use outside regular working hours, perceived productivity, expectations of availability, or clarity of expectations about availability. The results should be treated with caution due to a possible selection bias of the study population, both for technical reasons and due to the specific occupational context.

Keywords:

• Digitalization
• work control
• intervention
• participative
• work strategies

1. Introduction

Flexible work arrangements (FWA) appeared for the first time around 1950, when flextime (i.e., flexibility in working hours within specified time frames) was introduced as an initiative to reduce labor shortages (Rubin, 1979). A pronounced increase in FWA occurred during the Fourth Industrial Revolution at the beginning of the twenty first century, when new ways of working were introduced, relying on digitalization and use of information and communication technology (ICT), and enabling employees to connect to work anytime and anywhere (Bouzol-Broitman et al., 2016; Plantenga & Remery, 2010). FWA has become a common way to increase the autonomy of employees regarding when (e.g., how working hours are allocated), where (i.e., where work is done), and how to perform the work (e.g., which work tasks should be performed and in what way) (Eurostat, 2019; The European Commission’s Science & Knowledge Service, 2020).

Before the COVID-19 pandemic, 65% of employees had FWA in terms of time, and around 35% worked from home to some extent in Sweden (Eurostat, 2019; The European Commission’s Science & Knowledge Service, 2020). Furthermore, FWA, and especially work from home, have increased among office-based employees due to the COVID-19 pandemic (Contreras et al., 2020) and will likely persist in the future (De Lucas Ancillo et al., 2021). The good side of FWA is that it allows employees to combine work and personal life based on their own preferences, so that everyday life can be handled more easily (Johnson et al., 2020). This flexibility can lead to positive outcomes in terms of improved mental health (Johnson et al., 2020), better organizational commitment (Hashmi et al., 2022), reduced stress (Johnson et al., 2020) and less somatic symptoms (Shifrin & Michel, 2021).

However, FWA may also have downsides. The possibility of always being connected to work can create “a norm of availability” within a work group, with expectations of availability outside regular working hours, and expectations of more productivity and better performance (Johnson et al., 2020; Stadin et al., 2019). This can lead to an “always-on” culture within the work group, where in employees feel obliged to respond to work-related duties during free time (Schlachter et al., 2018), which in turn can lead to longer work days, more unpaid work, increased work-home interference, poor sleep quality, mental exhaustion (Johnson et al., 2020) and reduced psychological detachment from work (Mellner, 2015). Also, the possibility of connecting with colleagues at any time and in several different ways (e.g., by using email, smartphone, Skype, or Teams) can increase multitasking, switching between work tasks and work interruptions, which in turn reduces productivity (Rennecker & Godwin, 2005). Thus, FWA has influenced employees’ work strategies in how they handle work (Rennecker & Godwin, 2005).

A major challenge for employers may be to organize work so that the negative consequences of FWA are reduced and the positive aspects preserved. It is therefore important to identify interventions that effectively improve working conditions among employees with FWA (Fox et al., 2021). A review (Schlachter et al., 2018) concluded that organizations should support employees in finding strategies to work “smarter” and more efficiently with ICT, in ways that facilitate detachment from work during leisure. One way of achieving this can be to support employees in developing work strategies for how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions.

However, even though an individual may have effective strategies to detach from work, high expectations of availability may challenge these strategies. Therefore, an important measure to maintain positive individual work strategies may be to also establish common rules and routines within the work group regarding availability (Johnson et al., 2020). Clear expectations from colleagues and management can make it easier for employees to detach from work without feelings of guilt and reduce the pressure to respond to emails during leisure (Rennecker & Godwin, 2005). Office-based employees with FWA have previously expressed a wish for changes in these areas, for example by an education in personal efficiency and by clarifying expectations about availability within the work group (Bjärntoft et al., 2021).

Previous intervention studies have mainly focused on the implementation of FWA or changes in policies in order to increase employees’ control over when, where, and how they perform their work (Fox et al., 2021). Little is known, though, about interventions aiming to improve work strategies among employees who already have FWA. A workplace intervention should preferably be implemented using a participative approach and involve employees throughout the intervention process, including problem identification, suggestions for improvements, and implementation of the intervention (Johnson et al., 2020; Van Eerd et al., 2010). Interventions using a participative approach have previously been effective at reducing physical and psychosocial risk factors at work (Van Eerd et al., 2010) and at improving mental health outcomes (Johnson et al., 2020).

Participation should be implemented at several levels within the organization (Dellve & Eriksson, 2017). As mentioned above, it may be difficult for the individual to change her work strategies if high expectations about availability outside regular working hours persist within the work group. Therefore, changes may be needed at the level of individuals, groups, and the organization. A systematic literature review (Montano et al., 2014) called for comprehensive interventions using a participative approach, with implementation of initiatives at different levels within the organization. The aim of the present study was to determine the extent to which work strategies (i.e., how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions), ICT use outside regular working hours (i.e., use of laptop, tablet, or smartphone to handle information and allow communication), perceived productivity, expectations of availability, and clarity of expectations about availability, had changed among office-based employees with flexible work arrangements (FWA) 2 and 4 months after a participative two-step workplace intervention, approved by the top management of the organization.

2. Methods

2.1. The intervention

This workplace intervention study was conducted in a large governmental organization in Sweden (i.e., the Swedish Transport Administration). In contacts with the research group, the Human Resource department and the top management of the organization requested an intervention aimed at developing employees’ ability to work “smarter” and more efficiently with ICT in FWA, and to develop common rules and routines for FWA within the work group. This request was based on results from two previous steps within the research project. First, problems in existing working conditions were identified in a survey on work environment and health in 2016 addressed to all employees in the organization with FWA (n = 3259; Bjärntoft et al., 2020; Edvinsson et al., 2023). Examples of problems identified in the questionnaire were high work demands, over commitment, and expectations of availability from colleagues.

Second, in 2017, suggestions for improvements that could promote a good work environment and health in FWA were collected in focus group interviews with 45 participants, divided into eight groups (Bjärntoft et al., 2021). Examples of suggestions for improvement were education in personal efficiency and creating common rules about availability within the work group. Based on the results from the survey and the focus group interviews (Bjärntoft et al., 2021), the research group, in collaboration with the Human Resource department and the top management, developed the intervention. Thus, the development of the intervention followed a participative approach (Johnson et al., 2020; Van Eerd et al., 2010).

The intervention consisted of two steps: (1) an individual education to change the employee’s work strategies, and (2) workshops to develop common rules and routines for FWA within the work group. The organization’s goal with the intervention was to “in the long term reduce stress and sickness absence, and promote recovery and work-life balance,” as stated in the management document outlining the expected result of the intervention. However, a prerequisite for this to happen is that the intervention is effective in changing the work strategies of those participating. Thus, the present study investigated work strategies after the intervention, while distal outcomes such as stress and sickness absence will be examined elsewhere.

The design of the intervention was quasi-experimental and comprised baseline measurement and two follow-ups, both including an intervention and a control group. The control group was aware of the intervention, but did not take part in it, and was intended to work as usual. The intervention took place between March and June 2019 (step one) and between August and October 2019 (step two), after negotiations with and eventual approval by the top management of the organization. Data were collected among both groups using a questionnaire on three occasions (Figure 1): (1) prior to the intervention (baseline); (2) about 2 months after step one (follow-up one); and (3) about 4 months after step two (follow-up two). The first, educational step of the intervention started approximately 1 month after the baseline questionnaire had been completed. Measurements were performed at these occasions for practical reasons. At all three data collections, the employees received an email from the research group containing information about the study and a link to a web-based questionnaire. The participants gave their informed consent to participate in the study by responding to the questionnaires, and the study was approved by the Ethical Review Board in Uppsala (Dnr 2017/528).

Figure 1. Intervention procedure (step one and step two), and the three measurements (baseline, follow-up one and follow-up two), which were completed in both the intervention and control group.

2.2. Participants

The selection of the participants was done in collaboration with the Human Resource department and the top management of the organization. Since the organization’s goal for the intervention was to reduce work-related stress and sickness absence, the organization wished to practice the intervention on employees experiencing a high level of stress. According to a “work environment and health” questionnaire (in 2016), one division within the organization showed notably large levels of stress. This division comprised 892 employees, divided into 12 departments. Initially, managers at all 12 departments were informed about the intervention, and one department showed a particular interest in participating as an intervention group (n = 183). According to the organization, this department had had a high workload for a long time and showed particularly high levels of stress in annual employee surveys, and this may have contributed to their interest in participating in the intervention. The control group was selected to have similar work tasks, be of approximately the same size (n = 161) as the intervention group, and have a similar degree of allowed flexibility (i.e., either flextime or non-regulated working hours). The inclusion criterion for individuals in the two groups was to have a work contract allowing FWA (i.e., flextime or non-regulated working hours). In both groups, the employees were located at different geographical locations.

2.3. Implementation

The intervention was delivered in two steps. First, the intervention group received education targeting individual work strategies in how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions. Second, workshops were arranged to develop common rules and routines for FWA within the work group, for example regarding expectations about availability (see Figure 2 for a detailed description of the contents of the intervention). The implementation covered the organizational, work group, and individual levels (Dellve & Eriksson, 2017). The Human Resource department and the top management of the organization took responsibility for allocating resources and deciding about the design and implementation of the intervention, in collaboration with the research group. The education mainly focused the individual level, while the workshops targeted the work-group level.

Figure 2. Flowchart describing the procedure in step one (education) and step two (workshops) of the intervention.

Step one: Education

The education was delivered by an external company specialized in personal efficiency. During the education, the participants received information and got practice in how technical functions (e.g., in Outlook and OneNote) can be used to handle emails, structure, and prioritize work tasks and minimize work interruptions. For example, participants were asked to prioritize their work tasks and then used the “task” function in Outlook to plan when the work tasks should be performed. The course leader also recommended removing notifications in Outlook and instead scheduling when emails should be handled during the working day to minimize interruptions. Finally, the participants allocated time each day in their calendar for reflections and planning the upcoming working day to increase work control. Seminars with the intervention group were held at three occasions: (1) a four-hour webinar on the use of technical tools (i.e., Outlook and OneNote); (2) an eight-hour on-site seminar about strategies to work more efficiently with technical tools; and (3) a four-hour webinar focusing on repetition and deepening of previously learned knowledge. Within two weeks after each occasion, the participants repeated and practiced what they had learned in pairs (one hour), followed by rehearsal in groups together with the course leader (1.5 h on Skype). After the education, the participants had the opportunity to email, call, or book a meeting with the course leader to ask questions regarding the work strategies learned during the education. Only a few of the participants used this support.

Step Two: Workshop

The department in which the intervention was conducted had nine work groups, each with approximately 20 employees and one manager, and each group was intended to have their own workshop. However, one group did not participate due to a booking error, and this workshop was canceled. The remaining eight groups conducted one in-person workshop (approximately six hours) led by an external moderator. The workshop started with a presentation by the moderator of the results from the survey performed in 2016 on work environment and health, as a basis for discussions. Contents of the workshop then followed a systematic process model (Högskolan Dalarna, 2018). In the first two steps, participants reflected and discussed: (i) what they believed to characterize a sustainably efficient work group (i.e., a group that could do the same amount of work in a more efficient way so that detachment from work after the workday would be easier); and (ii) factors contributing to or obstructing that a work group can be sustainably efficient. In two subsequent steps, the participants developed: (i) a “to-do list” of the most important initiatives to become a sustainably efficient work group; and (ii) based on the “to-do-list”, an action plan containing rules and routines in FWA. Subsequently, the work group managers were responsible for following up the action plan.

2.4. Background Characteristics

Background characteristics were retrieved from the baseline questionnaire (Table 1). Responses included gender (male, female or do not want to categorize), age, marital status (married/partnership or single), number of children living at home (full time or part time), educational level (primary school, high school, vocational school, or university), employment rate (100% or less), number of years employed in the organization, and work arrangement (flextime or non-regulated working hours).

Table 1. Background characteristics and outcome variables for the intervention and control group at baseline (i.e., ICT use outside regular working hours, productivity, expectations of availability, and clarity of expectations about availability).

	Intervention group				Control group
	Scale	n	%	Mean (SD)	n	%	Mean (SD)
Gender	Women	33	34.0		32	46.4
	Men	63	64.9		37	53.6
	Do not want to categorize	1	1.0
Age	Years	96		48.7 (10.5)	68		48.9 (9.2)
Marital status	Single	15	15.5		10	14.3
	Married/partnership	82	84.5		60	85.7
Children living at home	0 children	47	50.0		25	36.2
	One or more	47	50.0		44	63.8
Educational level	Primary school	2	2.1
	High school	13	13.4		15	21.4
	Vocational school	9	9.3
	University	73	75.3		55	78.6
Employment rate	100%	89	91.8		66	95.7
	Less than 100%	8	8.2		3	4.3
Years of employment	Years	96		13.1 (11.9)	70		13.3 (10.5)
Work arrangement	Flextime	14	14.6		6	8.6
	Non-regulated working hours	82	85.4		63	90.0
	Other				1	1.4
ICT use outside regular working hours	0–4	97		1.9 (1.2)	70		2.3 (1.2)
Perceived productivity	0–10	96		7.5 (1.6)	70		7.8 (1.6)
Expectations of availability	0–4	97		0.7 (0.7)	70		0.9 (0.8)
Clarity of expectations about availability	0–4	94		1.9 (1.1)	68		2.1 (1.1)

2.5. Outcomes

Baseline

Both the intervention and control group received the baseline questionnaire. In the questionnaire, we measured four outcomes, specifically ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability (see Supplemental Digital Content 1 for a detailed description of questions). ICT use outside regular working hours was measured by one question (Edvinsson et al., 2023): “do you use work-related ICT to work outside regular working hours, for example for answering emails, writing text messages or answering phone calls?,” with answers ranging from 0 (not at all) to 4 (to a very high extent). Perceived productivity was measured by one question (Bergsten et al., 2021; Haapakangas et al., 2018), “what score would you give your productivity at work during the preceding month?,” with answers ranging from 0 (not productive at all) to 10 (highly productive). Participants in the study were office-based employees with flexible work arrangements. Their job functions were to administer the national Swedish transportation system. For these participants, “productivity” was likely associated with executing work tasks efficiently and effectively.

Expectations of availability outside regular working hours was measured by an index (Cronbach Alpha = 0.82) based on the mean of five customized questions with answers ranging from 0 (not at all) to 4 (to a very high extent), with an example question being “to what extent do you experience expectations of availability outside regular working hours from your manager or colleagues?”. Clarity of expectations about availability outside regular working hours was measured by an index (Cronbach Alpha = 0.89) based on the mean of answers to three customized questions, each ranging from 0 (not at all) to 4 (to a very high extent); an example question being “How well does the following statement of expectations apply to yourself and your work situation: my colleagues are clear about their expectations regarding my answers to work-related questions.”

Follow-up One

At follow-up one (2 months after the education, see Figure 1), the intervention group was asked about their experience of the education using questions about participation in the education (yes or no), whether the education was relevant to their work, and if they were engaged in and satisfied with the education, with answers ranging from 0 (not at all) to 4 (to a high extent). To account for previous experience about education in personal efficiency, the intervention and control groups were both asked if they had participated in a similar education before (yes or no). To evaluate the effect of the education, we measured the following outcomes in both groups: the extent to which employees had changed their work strategies regarding how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions, with answer options ranging from 0 (not at all) to 4 (to a high extent) (see Supplemental Digital Content 1). In addition, both intervention and control groups received the baseline questions about ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability.

Follow-up Two

At follow-up two (4 month after the workshops, see Figure 1), the intervention group was asked about their experience of the workshop, specifically participation in the workshop, whether the workshop was relevant to their work, and if they were engaged in and satisfied with the workshop. We also asked if their working group had continued to work with the action plan that was developed during the workshop, with answers ranging from 0 (not at all) to 4 (to a high extent). To evaluate the effect of the workshops, we measured, in both the intervention and control group, the extent to which employees had changed their work strategies regarding how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions. In addition, both intervention and control groups were asked to answer again the questions in the baseline questionnaire about ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability. All questions included in the present study, except those being specific to the intervention (i.e., experience of the intervention and changed work strategies due to the intervention), have been validated previously by “think aloud interviews” and principal component analysis (Edvinsson, 2016).

2.6. Statistical Methods

IBM SPSS Statistics 27 was used to perform all analyses. Descriptive statistics on frequencies and percents (categorical variables), and mean values and standard deviations (SD) (continuous variables), were used to describe the background characteristics of the two groups and the intervention group’s experience of the intervention. Baseline differences between the intervention and control groups were analyzed using Chi-square tests (proportions) and independent t-tests (continuous variables). To examine group differences in change between follow-up one and two of work strategies regarding how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions, a one way between-groups analysis of variance (ANOVA) was used. Cohen’s criteria (Cohen, 1969; Richardson, 2011) was used to interpret whether effect sizes (Partial Eta Squared) were small (.01), medium (.06) or large (.14).

Linear Mixed Models (LMM) for repeated measurements were used to examine the effects of the intervention on ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability. Separate models for each dependent variable were constructed with time (three levels: baseline (reference), follow-up one and follow-up two), group (two levels: intervention and control), and the interaction of these two (time × group) as fixed factors, and intercept and participants as random factors (model 1). The interaction of group and time (group × time) was used to investigate the differences in changes between groups over time. Goodness of fit was tested by Negative 2-Residual Log-likelihood (−2RLL) and Akaike’s Information Criterion (AIC), in a first-order autoregressive (AR1) covariance structure with homogenous variance, selected as being the most suitable for the data. These models were run on all participants having data at baseline and at least one of the two follow-ups. Missing data were considered as missing at random (MAR). Subsequently, the analyses were re-run with adjustments for gender, marital status, educational level, and children living at home (model 2). The effects of the intervention (B) were determined with 95% confidence intervals (CI), and the statistical significance level was set to 0.05. A sensitivity analysis was performed by re-running the analysis on employees with a high baseline extent of ICT use outside regular working hours, specifically 34 participants in the intervention group and 32 in the control group answering 3 or 4 on the scale from 0 (not at all) to 4 (to a high extent), to check whether those with extensive ICT use perceived a greater effect of the intervention than those with less ICT use. We also performed a sensitivity analysis on only those participants who completed all three questionnaires (i.e., baseline and both follow-ups), to investigate the extent to which missing values would affect the results. In this sensitivity analysis, 18 participants from the control group and three from the intervention group were excluded. Thus, the analysis included 94 participants in the intervention group and 52 in the control group.

3. Results

The inclusion of study participants is presented in Figure 3. In total, 288 eligible employees received the questionnaire at baseline and 217 responded. Unfortunately, a technical error occurred at baseline: 56 employees from the control group did not receive the questionnaire and were therefore excluded from the study. Other reasons for exclusion included having left the organization, being on parental leave/sick leave, not responding at baseline and to at least one follow-up questionnaire, and not having participated in the first step of the intervention (the education). The analyzed sample consisted of 167 employees (Intervention n = 97; Control n = 70) at baseline, 162 (Intervention n = 94; Control n = 66) at follow-up one, and 153 (Intervention n = 97; Control n = 56) at follow-up two.

Figure 3. Flowchart of participants through the study and numbers analyzed after application of inclusion criteria (respond to baseline questionnaire and at least one follow-up).

3.1. Background characteristics

Background characteristics and dependent variables at baseline are summarized in Table 1. The intervention and control groups differed significantly with respect to gender and educational level at baseline (p < 0.05). There were no other baseline differences between the groups. Significant differences at baseline were found for two of the dependent variables: ICT use outside regular working hours and expectations of availability; both were higher in the control group. There were no significant differences between the groups at baseline in perceived productivity and clarity of expectations about availability.

3.2. Experience of the intervention

Table 2 presents descriptive statistics concerning the participants’ experience of the intervention. Regarding the education (step one), previous participation in a similar personal efficiency education was reported by 27.4% of the intervention group and 33.3% of the control group. Overall, most of the participants in the intervention group were highly satisfied with the education in step one (76.6%), reported a high level of engagement (93.5%), and found the education to be relevant to their work (82.0%). Regarding the workshop in step two, 84.2% participated, of which the majority (90.1%) reported a high level of engagement. More than half of the participants (55.7%) were highly satisfied with the workshop, perceived it as relevant to their work (57.5%), and continued to work with the action plan on rules and routines for FWA (61% reported “at least to some extent”).

Table 2. Employees’ experience of the intervention, measured about 2 months after the education (follow-up one) and about 4 months after the workshop (follow-up two).

	Intervention group
	Scale	n	%
Participated in the education	Yes	97	100
	No	0	0
Participated in the workshop	Yes	80	84.2
	No	15	15.8
Experience of the education	0–4
You were engaged	To a minor extent (0–1)	1	1.1
	To some extent (2)	5	5.4
	To a high extent (3–4)	87	93.5
It was relevant to your work	To a minor extent (0–1)	4	4.3
	To some extent (2)	13	13.8
	To a high extent (3–4)	77	82.0
You feel satisfied with the education	To a minor extent (0–1)	8	8.5
	To some extent (2)	14	14.9
	To a high extent (3–4)	72	76.6
Experience of the workshop	0–4
You were engaged	To a minor extent (0–1)	2	2.5
	To some extent (2)	6	7.5
	To a high extent (3–4)	72	90.1
It was relevant to your work	To a minor extent (0–1)	12	15.1
	To some extent (2)	22	27.5
	To a high extent (3–4)	46	57.5
You feel satisfied with the workshop	To a minor extent (0–1)	16	20.3
	To some extent (2)	19	24.1
	To a high extent (3–4)	44	55.7
You still work with the action plan	To a minor extent (0–1)	31	38.8
	To some extent (2)	24	30.0
	To a high extent (3–4)	25	31.3

3.3. Effects of the Intervention on Work Strategies

Table 3 summarizes descriptive statistics concerning intervention effects on work strategies. We found statistically significant differences between the intervention and control groups regarding perceived change in work strategies. At follow-up one (about 2 months after the education), the intervention group had changed their work strategies to a higher extent than the control group regarding how to handle emails (F (1,159) = 92.5, p < 0.001, η²_p = 0.37), how to structure work tasks (F (1,158) = 63.6, p < 0.001, η²_p = 0.29), how to prioritize work tasks (F (1,159) = 60.1, p < 0.001, η²_p = 0.28), and how to minimize work interruptions (F (1,155) = 37.1, p < 0.001, η²_p = 0.19). At the second follow-up (about 4 months after the workshops), the intervention group still reported larger changes than the control group, but the differences were somewhat smaller than at follow-up one: how to handle emails (F (1,151) = 71.9, p < 0.001, η²_p = 0.32), how to structure work tasks (F (1,149) = 42.5, p < 0.001, η²_p = 0.22), how to prioritize work tasks (F (1,152) = 41.2, p < 0.001, η²_p = 0.21), and how to minimize work interruptions (F (1,148) = 27.0, p < 0.001, η²_p = 0.16). According to Cohen’s criteria, differences between groups were large (effect size ≥ 0.14) for all elements of the work strategies both at follow-up one and two.

Table 3. Descriptive statistics describing the change in work strategies after the education (follow-up one) and workshop (follow-up two), presented as the mean change (SD) for the intervention and control group.

	Intervention group			Control group
Variables	Scale	n	Mean (SD)	n	Mean (SD)
During the last six months, have you changed your work strategies regarding…	0–4
How to handle emails
Follow-up one		94	2.7 (1.3)	66	0.8 (1.1)
Follow-up two		96	2.4 (1.2)	56	0.7 (1.1)
How to structure work tasks
Follow-up one		93	2.6 (1.6)	66	1.2 (1.1)
Follow-up two		95	2.3 (1.2)	55	1.0 (1.2)
How to prioritize work tasks
Follow-up one		94	2.6 (1.2)	66	1.1 (1.1)
Follow-up two		97	2.3 (1.1)	56	1.1 (1.2)
How to minimize work interruptions
Follow-up one		91	2.1 (1.2)	65	0.9 (1.1)
Follow-up two		95	1.8 (1.1)	54	0.8 (1.1)

Note. The scale ranged from 0 (no change at all) to 4 (changed to a large extent).

3.4. Effects of the Intervention on ICT Use, Perceived Productivity, Expectations of Availability, and Clarity of Expectations

Descriptive statistics on ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability are reported in Table 4 for the intervention and control groups, and the effects of the intervention are shown in Table 5. We did not find any statistically significant effect of the intervention. Although estimates were generally in favor of the intervention group in terms of less ICT use and increased perceived productivity (for both the adjusted and the unadjusted model at both follow-ups), 95% CIs included zero.

Table 4. ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability at baseline, follow-up one and follow-up two, presented as mean (SD) for the intervention and control groups.

		Intervention group		Control group
Variables	Scale	n	Mean (SD)	n	Mean (SD)
ICT use outside regular working hours	0–4
Baseline		97	1.9 (1.2)	70	2.3 (1.2)
Follow-up one		94	1.7 (1.1)	66	2.2 (1.1)
Follow-up two		96	1.7 (1.2)	56	2.1 (1.0)
Perceived productivity	0–10
Baseline		96	7.5 (1.6)	70	7.8 (1.6)
Follow-up one		93	7.5 (1.6)	66	7.5 (1.6)
Follow-up two		95	7.4 (1.8)	56	7.5 (1.7)
Expectations of availability	0–4
Baseline		97	0.7 (0.7)	70	1.0 (0.8)
Follow-up one		94	0.7 (0.7)	66	0.8 (0.7)
Follow-up two		96	0.6 (0.7)	56	0.8 (0.8)
Clarity of expectations about availability	0–4
Baseline		94	1.9 (1.1)	68	2.1 (1.1)
Follow-up one		91	1.9 (1.3)	65	2.1 (1.0)
Follow-up two		90	1.8 (1.3)	56	2.1 (1.2)

Table 5. Effects of the intervention on changes in ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability at follow-up one and two.

			Model 1. Unadjusted				Model 2. Adjusted
				95% CI				95% CI
Variables	Scale	n	B	Lower	Upper	p	B	Lower	Upper	p
ICT use outside regular working hours^a	0–4
Follow-up one		I = 94, C = 66	−0.10	−0.39	0.18	0.49	−0.14	−0.43	0.16	0.36
Follow-up two		I = 96, C = 56	−0.09	−0.37	0.18	0.51	−0.11	−0.41	0.19	0.47
Perceived productivity^b	0–10
Follow-up one		I = 93, C = 66	0.31	−0.18	0.79	0.22	0.29	−0.22	0.79	0.27
Follow-up two		I = 95, C = 56	0.37	−0.20	0.93	0.20	0.36	−0.23	0.95	0.23
Expectations of availability^a	0–4
Follow-up one		I = 94, C = 66	0.10	−0.09	0.28	0.30	0.10	−0.09	0.29	0.31
Follow-up two		I = 96, C = 56	0.08	−0.09	0.26	0.34	0.12	−0.07	0.31	0.22
Clarity of expectations about availability^b	0–4
Follow-up one		I = 91, C = 65	0.10	−0.30	0.51	0.62	0.16	−0.24	0.56	0.43
Follow-up two		I = 90, C = 56	−0.08	−0.50	0.34	0.71	−0.12	−0.56	0.32	0.60

Effects (B) with 95% confidence intervals (CI) and p-values are presented for the interaction between group and time (group × time). Model 2 is adjusted for gender, marital status, educational level, and children living at home.

Note. I: intervention group and C: control group.

^a Negative value of effect favor the intervention group.

^b Positive values of effect favor the intervention group.

3.5. Sensitivity analysis

The results of analyses performed on employees with high levels of ICT use at baseline (n = 66 of which 34 belonged to the intervention group and 32 to the control group), were consistent with the original analysis of all employees in showing no statistically significant effects of the intervention on ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability. In general, the sensitivity analyses on participants who completed all three questionnaires (i.e., baseline and both follow-ups; n = 94 in the intervention group and n = 52 in the control group) showed no major differences compared to the original analyses (see supplemental digital content 2). The only notable difference was that expectations about availability were reduced to a greater extent in the control group in the sensitivity analysis than in the complete analysis, while there were no changes in the intervention group.

4. Discussion

In the present study, we examined the extent to which work strategies, ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability changed among office-based employees with flexible work arrangements (FWA) after a participative workplace intervention comprising an education, with follow-up measurements after 2 months, and workshops, with follow-up measurement after 4 months. We found that perceived changes in work strategies were larger in the intervention group compared to the control group, with large effect sizes at both follow-ups. This indicates that the participants regarded the intervention to be effective in influencing how they handled emails, how they structured work tasks, how they prioritized work tasks, and how they minimized work interruptions. Whether the changes in strategies were for the better or for the worse is, however, not clear, since the question concerned only whether work strategies had changed or not. Still, as the participants were highly satisfied with the intervention, we expect that the changes were perceived as favorable. This large effect on work strategies may relate to the participative approach used for developing the intervention (Bjärntoft et al., 2021); this approach has previously shown potential for positive effects on behavior change (Johnson et al., 2020; Van Eerd et al., 2010). Furthermore, the intervention targeted multiple organizational levels and included both an education (individual level) and workshops (group level), which may increase the likelihood of individuals actually changing work strategies (Dellve & Eriksson, 2017).

For a sustainable change, it is likely important to also discuss how to use the new work strategies at the workplace, both individually and in the work group. Previous research (Prochaska & Velicer, 1997) has pointed to compliance as an important step in achieving behavioral change, described as a process where individuals establish their new behavior and become confident in a behavior change. It has been argued that an intervention effect should be maintained for at least 6 months before a long-lasting behavioral change is achieved (Prochaska & Velicer, 1997). However, the present intervention did not include support in keeping the new work strategies even after the intervention, and only 31% of the participants worked continuously with the action plan. This highlights the need for continued support in the process, preferably by integrating the changed strategies into existing systematic work environment management. Also, we did not examine the impressions of colleagues and management of changes in routines and work demands within the intervention group, nor how productivity and availability were affected. Doing so could have increased understanding of the intervention results, and we suggest including data from ‘peripheral’ staff in the organization in further studies.

Even though work strategies changed considerably among employees in the intervention group, we could not confirm a statistically significant effect on intermediate outcomes (i.e., ICT use outside regular working hours, perceived productivity, expectations of availability, and clarity of expectations about availability). These non-significant results may be explained in several ways. One explanation may be failure in implementation. Although results indicate that both parts of the intervention were implemented, the implementation was apparently not sustainable. Also, the follow-up time may have been too short to see any effects on these outcomes. Another explanation may be that the intervention did not address factors that are key in changing the desired outcomes, and other initiatives could have generated a greater effect. One example could be to reduce job demands, which might minimize the risk of employees using work-related ICT outside regular working hours (Dettmers et al., 2016; Kühner et al., 2023). Second, it may be important to focus interventions on employees perceiving problems with ICT use after regular working hours rather than rolling it out in a group where many may not need it much; or at least ask employees whether the ICT use is voluntary or not. Work-related ICT use after regular working hours can be a personal choice to easier combine work and private life for some individuals, and therefore not all employees may be motivated to decrease the level of ICT use (Schlachter et al., 2018).

The overall situation was relatively good before the intervention, with quite low levels of ICT use outside regular working hours, low expectations of availability, and high levels of perceived productivity (see Table 4), which makes any further improvement unlikely. However, the sensitivity analysis on employees reporting a high degree of ICT use outside regular working hours did not indicate any effects different from those found for all participants in the intervention group. Also, the sensitivity analysis on participants who completed all three questionnaires (i.e., baseline and both follow-ups) showed no major differences compared to the original analyses, except for a more pronounced decline in expectations of availability in the control group. One explanation could be that among participants who did not answer all questionnaires and were therefore not included in the sensitivity analysis, expectations of availability were unchanged after the intervention to a greater extent than among those who did participate in all three data collections. This would lead to a larger decrease of expectations in the control group of the sensitivity analysis than in the control group in the original analysis.

Our results may be useful for other organizations offering FWA. However, ICT use outside regular working hours may be higher in other organizations (Eurofound, 2020; Reinke & Ohly, 2021; Schlachter et al., 2018), and the same intervention may therefore have more marked effects in other samples.

4.1. Study Strengths and Limitations

The main strength of the present study is the use of a participative approach based on a survey on work environment and health in 2016 (unpublished data), specifically prior to the intervention, which identified the problems targeted in the intervention, followed up by focus groups interviews to collect employee suggestions of improvements in FWA (Bjärntoft et al., 2021). Furthermore, both the contents and implementation of the intervention included several levels within the organization (i.e., organizational, work group and individual level) (Dellve & Eriksson, 2017). Another strength of the study is the quite homogenous sample; the intervention and control groups had similar work tasks and belonged to the same division within the organization.

Our study also faced some limitations. By mistake, the baseline questionnaire did not reach 56 participants in the control group, who had to be excluded from the study. Several additional participants were also excluded due to the study’s exclusion criteria. Therefore, the groups ended up being quite small, which reduced power and compromised representativeness. Another limitation may be the selection process of the participants. The organization decided to focus on one division with high levels of stress, in which one department showed a particular interest in participating. This “self-selection” likely contributed to a result that is not fully representative to the organization in general, since the intervention group may have been more motivated to implement changes than other parts of the organization. Furthermore, as the control group was aware of the intervention, it is likely that the differences between the intervention and control groups were reduced due to contamination, even if the control group reported a very small change in works strategies over time (follow-up one and two), which suggests that contamination bias is likely limited. Unfortunately, we have no information on the extent to which the two groups knew each other or talked to each other about the intervention.

Another limitation may be that ICT use was measured using only a single item addressing the extent to which employees used work-related ICT to work outside regular working hours. However, a more detailed response scale measuring the number of hours per day engaged in work-related ICT use outside regular working hours would probably have been more accurate. Multi-item indices would also be preferable to capture employees’ ICT use more comprehensively. Thus, a more extensive questionnaire could have been helpful. However, the size of the questionnaire was limited for pragmatic reasons: the organization wished for a shorter version of the ‘work environment and health’ questionnaire from 2016, which should be possible to complete in 15-20 min. In addition, our study did not include any formal process evaluation [29], which prevented us from analyzing organizational changes outside the intervention that may have had an impact on the results. Finally, the investigated organization also contributed to the funding of the study, which may have caused pressure on the employees to respond in a socially desirable way. However, involvement of the organization, also financially, is an integrated consequence of collaborative research, and loyalty issues of employees vis-a-vis the organization are difficult to avoid. The involvement of the organization in the research project was controlled in an agreement between the organization and the researchers, clarifying that the organization did not have any influence on measurement procedures, results, and reporting of the study.

4.2. Practical Implications

Our findings suggest that a participative workplace intervention focusing on changing work strategies in FWA may, indeed lead to changed work strategies at an individual level. It may also lead to clearer rules and routines for FWA within the work group. The occurrence of FWA is increasing, especially after the COVID-19 pandemic since many organizations in the office sector have introduced hybrid solutions allowing more work from home than before the pandemic. Therefore, our findings may be useful regarding how to support employees in identifying alternative work strategies in organizations offering FWA. Our findings suggest that organizations should support working groups in creating an action plan containing rules and routines in FWA. To secure sustainable change, the manager should work regularly with the action plan in the work group and support employees in maintaining new work strategies.

In conclusion, we found that combining education in work strategies regarding how to handle emails, how to structure work tasks, how to prioritize work tasks, and how to minimize work interruptions with workshops addressing common rules and routines for FWA is effective in changing employees’ work strategies, but may not affect ICT use outside regular working hours, perceived productivity, expectations of availability, or clarity of expectations about availability. Our results suggest that studies of other interventions in FWA and in populations with greater needs of changing their working conditions are required. It is also important in future studies to understand the long-term health effects of interventions in FWA.

Acknowledgments

The authors wish to thank the Swedish Transport Administration and all included participants for their collaboration in the study. We also want to thank Bengt Pontén, Mansur Köyluoglu, Johanna Edvinsson, Camilla Zetterberg, Helena Jahncke, Johan Larsson, and Lina Persson Hammarström for their valuable contribution to the design of the intervention and the data collection.

Conflict of interest

The authors declare no conflict of interest.

Additional information

Funding

This research was supported by a grant from the Swedish Transport Administration [Dnr. 2017/528], and the Swedish Research Council for Health, Working life and Welfare [Forte Dnr. 2009-1761].