Understanding user behaviour in activity-based offices

Abstract

Little is known about the factors that explain the differences in the ways that individuals use activity-based offices (ABOs). This study aimed to investigate whether person-related and situational factors are associated with self-reported use of workspaces and the perceived person-environment (P-E) fit in ABOs, independently of the job profile. Survey data were gathered in one organisation (N = 332) 7–11 months after an office re-design. Younger age, male gender, managerial position, and better work ability were associated with more frequent use of different workspaces. Workspace switching was perceived as more time-consuming by employees who worked at the office less, had a high workload, and were dissatisfied with ergonomics. All variables except gender were associated with the P-E fit. Person-related and situational factors appear relevant to workspace use and P-E fit, independently of job contents. Contextual, cultural, and office design differences should be considered when generalising these results.

Practitioner summary: This case study investigated individual differences in how activity-based offices are used. Being younger, male, a manager, or having good work ability were associated with using workspaces more actively. Person-related and situational factors appear relevant to how offices are used and perceived, in addition to job characteristics.

Abbreviations: ABO: activity-based office; P-E fit: person-environment fit; RQ: research question; SD: standard deviation; IN: interactive needs; CD: cognitive demands; OR: odds ratio; M: mean; ref.: reference category; CI: confidence interval; h: hour; PO: proportional odds

Keywords:

• Individual differences
• office ergonomics
• user needs analysis
• workplace design
• organisational ergonomics

Introduction

User behaviour in activity-based offices

Activity-based offices (ABOs) have become increasingly common and aim to facilitate new ways of working and to enable organisations to more flexibly and efficiently use office space. In ABOs, employees have no assigned desks; they are expected to switch between different workspaces (e.g. spaces for collaboration or concentration) depending on their task-related needs. The popularity of this office design could even be increased due to the Covid-19 pandemic if teleworking remains at above pre-pandemic levels in the future. Therefore, there is a growing need to understand how employees use, experience, and are affected by ABOs.

Previous quantitative research has mostly focused on the workplace level by comparing ABOs to traditional offices (e.g. Bodin Danielsson and Bodin 2009; Bodin Danielsson et al. 2014; De Been and Beijer 2014; Haapakangas et al. 2019). Engelen et al. (2019) reviewed this literature and concluded that the ABO was a ‘promising concept’ although it had some downsides (e.g. regarding privacy). Another line of research has taken a more individual approach, investigating the factors that explain the variation in the perceptions and effects of ABOs (e.g. Gerdenitsch, Korunka, and Hertel 2018; Haapakangas, Hallman, et al. 2018; Hoendervanger et al. 2018; Hoendervanger et al. 2022; Wohlers, Hartner-Tiefenthaler, and Hertel 2019). The latter approach is important because the ABO concept relies on specific assumptions regarding user behaviour, such as the task-based use of workspaces.

The ABO concept is supported by observations that switching workspaces more actively is associated with higher satisfaction with the work environment (Hoendervanger et al. 2016) and better well-being and self-rated productivity (Haapakangas, Hallman, et al. 2018). However, information on the frequency of switching workspaces alone is not sufficiently informative, because the right amount of switching depends on job characteristics and employees’ needs. In recent years, some researchers (e.g. Gerdenitsch, Korunka, and Hertel 2018; Hoendervanger et al. 2019; Hoendervanger et al. 2022) have applied the person-environment (P-E) fit theory (Edwards, Caplan, and Harrison 1998) to help clarify the interplay between employee’s activities, needs and their experiences of the work environment in ABOs. The P-E fit theory has been widely used in different areas of organisational research. Its core assumption is that a misfit between personal and environmental characteristics causes strain and stress (Edwards, Caplan, and Harrison 1998). Studies of ABOs have viewed employees being able to flexibly use different workspaces as a means of achieving better working conditions, i.e. a good fit (Gerdenitsch, Korunka, and Hertel 2018; Hoendervanger et al. 2022). A better fit has been associated with various positive outcomes in ABOs, including interaction (Gerdenitsch, Korunka, and Hertel 2018), environmental satisfaction (Gerdenitsch, Korunka, and Hertel 2018; Hoendervanger et al. 2019), performance (Hoendervanger et al. 2019), job satisfaction, and vitality (Wohlers, Hartner-Tiefenthaler, and Hertel 2019).

However, the potential of ABOs to support work performance is contradicted by the observed reluctance and difficulties involved in using the workspaces in the intended way (Appel‐Meulenbroek, Groenen, and Janssen 2011). Individuals vary greatly in how they use ABOs (Haapakangas, Hallman, et al. 2018) and frequent activity-based switching appears to be rare (Appel‐Meulenbroek, Groenen, and Janssen 2011; Hoendervanger et al. 2016). As a result, a misfit between work tasks and workspace is a common finding; for example, performing highly concentrative tasks in open workspaces (Appel‐Meulenbroek, Groenen, and Janssen 2011; Hoendervanger et al. 2019, 2022). In addition to task-related reasons, other motives can also affect workspace choices (Appel‐Meulenbroek, Groenen, and Janssen 2011; Hoendervanger et al. 2016; Kim et al. 2016). Furthermore, failures in office design can impede workspace switching if the number, availability, comfort, or usability of the workspaces do not meet their demand (Babapour Chafi, Harder, and Bodin Danielsson 2020; Babapour, Karlsson, and Osvalder 2018; Haapakangas, Hongisto, et al. 2018; Kim et al. 2016). Difficulties finding a suitable workspace and the time spent setting up and clearing a workstation are common complaints and may discourage workspace switching (Babapour, Karlsson, and Osvalder 2018; Hoendervanger et al. 2016; Kim et al. 2016; van der Voordt 2004). These results challenge the core assumption of the ABO concept (i.e. activity-based workspace use) and highlight a need to identify other factors that explain office use.

The theoretical framework of Wohlers and Hertel (2017) provides a useful basis for conceptualising the potential role of user behaviour in how employees are affected by ABOs. The model assumes that ABO features (such as flexible use of activity-related workspaces) affect perceived working conditions (such as privacy). The perceived working conditions, in turn, can have various positive or negative short- and long-term consequences for individuals, teams, and organisations (e.g. for well-being or performance). Different moderators (e.g. person- and task-related) are assumed to affect the relations between ABO features and working conditions, and between working conditions and consequences. Thus, it could be assumed that moderating effects of the consequences of ABOs could, to some extent, originate from individual variation in office use (i.e. the level of ABO features in the model). However, previous research has largely overlooked the factors behind this individual variation, both theoretically and empirically.

Factors investigated in this study

The purpose of this study was to identify any factors that explain the variation in individuals’ office use and perceived P-E fit, regardless of job characteristics. To our knowledge, only Hoendervanger et al. (2016) have addressed this theme using quantitative methods. They found that greater task variety, communication work, and mobility outside the workplace were related to more workspace switching. As no theoretical framework has outlined the factors explaining ABO use, the rationale of our study was to investigate any potentially relevant variables that have been theoretically or empirically associated with other outcomes of ABOs. Another principle was to identify previously overlooked factors and to include different types of variables, categorised as task-related, person-related (Wohlers and Hertel 2017), and situational factors. The investigated explanatory variables are reviewed briefly below.

In terms of task-related factors, collaborative and concentrative job characteristics represent key activities in the ABO concept (Wohlers and Hertel 2017) and have previously been associated with several outcomes (Hoendervanger et al. 2022; Wohlers, Hartner-Tiefenthaler, and Hertel 2019). However, earlier studies have tended to analyse collaboration and concentration (or individual work) as exclusive (Hoendervanger et al. 2016, 2022) or separate dimensions (van den Berg et al. 2020; Wohlers, Hartner-Tiefenthaler, and Hertel 2019). This may not sufficiently capture the work environmental needs of knowledge work, in which high collaborative and concentrative demands are often combined (Heerwagen et al. 2004). Thus, we investigated job profiles based on the combinations of interactive and cognitive job demands. Managerial position might also be relevant to office use. It most likely increases task variety, (cf. Wohlers and Hertel 2017) and thus, the need for various workspaces. Managers have perceived ABOs more positively (Sirola et al. 2022) but few studies have addressed managerial tasks.

Age and gender were included as person-related variables, as in the model of Wohlers and Hertel (2017). However, the directions of the expected associations were not clear. Wohlers and Hertel (2017) proposed that older employees switch workspaces more actively due to having better self-regulation skills. In contrast, empirical findings have associated older age with more negative perceptions of ABOs (Hoendervanger et al. 2018; van den Berg et al. 2020). Regarding gender differences, Wohlers and Hertel (2017) suggested that ABOs could affect women and men differently due to differences in territorial behaviour, privacy needs, and status needs. Such factors may relate to office use. As the empirical findings on gender effects in ABOs have been mixed (Bodin Danielsson et al. 2014, 2015; Bodin Danielsson and Theorell 2019), more studies on gender differences are needed. We also investigated work ability and satisfaction with ergonomics as factors that may be related to an individual’s ability to use workspaces in an activity-based way. Such factors have previously received little attention although the importance of ergonomics is generally known (Appel‐Meulenbroek, Groenen, and Janssen 2011; Hoendervanger et al. 2016; Kim et al. 2016; van der Voordt 2004).

As for situational factors, environmental distractions have often been mentioned as downsides of ABOs (Engelen et al. 2019), and may affect workspace choices (Appel‐Meulenbroek, Groenen, and Janssen 2011; Hoendervanger et al. 2016; van den Berg et al. 2020). As individuals differ in their tolerance of distractions (Maher and von Hippel 2005), we expected that perceived distractions would be associated with office use and P-E fit. We included office presence (i.e. the amount of time spent at the office) because it is conceptually closer to office use than external mobility, which Hoendervanger et al. (2016) associated with more workspace switching. Quantitative work demands were included as a stress-related variable; a few previous studies have associated this with ABOs (Haapakangas et al. 2019; Meijer, Frings-Dresen, and Sluiter 2009). As switching workspaces can be perceived as time-consuming, we assumed that employees with high workloads might be more reluctant to spend time and energy switching and thus, might experience a poorer P-E fit.

This study aimed to identify any factors that explain self-reported office use and perceived P-E fit regardless of job profile (i.e. combinations of interactive needs and cognitive demands). Office use was measured using three variables, similarly to Haapakangas, Hallman, et al. (2018): daily workspace switching, weekly workspace variety, and daily time spent looking for a workspace. P-E fit was included as the fourth outcome variable to obtain complementary information on whether the same explanatory variables contribute to perceiving workspaces as supportive or non-supportive.

The research questions were:

RQ1: How are task-related factors (job profile, managerial position), person-related factors (age, gender, work ability, satisfaction with ergonomics), and situational factors (office presence, distractions, workload) associated with office use and P-E fit in an ABO?

RQ2: Are person-related factors, managerial position, and situational factors associated with office use and P-E fit, regardless of job profile?

Materials and methods

Design and research ethics

This cross-sectional case study was conducted in an organisation that had redesigned its office into an ABO. The study was approved by the Ethical Review Board of the Finnish Institute of Occupational Health, Helsinki, Finland. Participation in the study was voluntary and based on informed consent.

The investigated ABO

The organisation was a Finnish pension provider that had implemented a strategy-driven workplace development project in its head office in 2016–2019. The organisation had renovated an eight-floor building into an ABO. The employees, including a small unit from a branch office, mostly moved to the renovated workspaces in two phases in May 2018 and May 2019, although there was some individual variation. The employees had previously worked in private offices. The organisation had several functions related to the administration of pensions and services to support work ability management and included customer service, administrative personnel, and experts from various fields. Working hours were flexible and telework was common.

The staff-only zone of the ABO covered four floors that had home zones assigned to teams. These floors comprised open workspaces, a coffee area (for informal interaction), rooms for phone calls and web meetings, project rooms, and meeting rooms for internal use. Separate meeting rooms for clients were located on the top floor. A quiet workspace, where talking was not allowed, was located on one of the floors and separated by a glass wall. Workstations, equipped with sit-stand desks, adjustable chairs, and screens on three sides, were non-assigned except in the case of a few people with specific needs or tasks. Each floor had its own colour scheme and theme. The lighting system was not renewed. Each floor had a ‘chair park’ providing ergonomic options. The building also had an allergen-free workspace. Employees were instructed to avoid strong fragrances. The design process included various participatory elements.

The organisation encouraged task-based workspace switching and using the whole office. The interaction was allowed in the open workspaces, but employees had to move elsewhere for longer conversations (e.g. meeting rooms, coffee areas). Those who were disturbed by conversations in the open areas were encouraged to move to a more suitable space. If an employee left a workstation for over 3 h, they had to adhere to the clean desk policy and take their items with them. According to a representative of the organisation, many employees chose the same workstations and did not actively switch workspaces. Moving to the quiet space was perceived as difficult and use of the space remained low.

Data gathering and participants

The organisation informed employees about the study in advance. The researchers sent an electronic survey twice to all employees working in the building: in April 2019 to those who had moved to the ABO first and in February 2020 to those who had moved later or had not responded earlier. Some employees participated in both surveys, but only their first responses were included. Most respondents had been working in the ABO for 7–11 months at the time they responded to the survey. However, the estimated range was 2–21 months due to new employment and individual variation in the time of moving or responding to the survey.

A total of 332 employees responded to the survey (response rate = 64.8%). The respondents’ mean age was 50.2 (SD = 9.2, range 26–66 years), and 72.5% were women. The vast majority had a university degree (79%). On average, the respondents had worked for 8.4 years (SD = 7.5) in their current job and for 13.7 years (SD = 10.8) in the organisation. The amount of telework ranged from no teleworking (23%) to working from home for over 20 h per week (16%), with a median of 3–9 h (approximately half a day to a whole workday) per week.

Questionnaire measures

Office use was measured by three items. Daily workspace switching was measured by the question ‘How many times during a workday do you normally switch between different workspaces?’ (0: I work in the same workspace the whole day–10: 10 or more times; Haapakangas, Hallman, et al. 2018).

Weekly use of different workspace types at the office was rated using six categories (not at all, 1–2, 3–9, 10–20, 21–30, 31–40 h; Ruohomäki, Sirola, and Lahtinen 2021). Nine workspace types (non-assigned desk, assigned desk, meeting rooms, individual web-meeting rooms, quiet space, phone booths, cafeteria for the whole building, coffee areas on different floors, half-closed sofa/armchair) were rated and the respondents could specify and rate one additional space. The score for weekly workspace variation was created by calculating the number of workspace types that were used at least 1–2 h per week (maximum score was 10).

The daily time spent looking for workspace was based on two questions (Haapakangas, Hallman, et al. 2018; Rolfö, Eklund, and Jahncke 2018). ‘Do you have to spend time looking for a suitable workspace?’ (yes/no), and ‘Please estimate how many minutes per day you usually spend looking for a suitable workspace’. The variables were combined by assigning a value of 0 min to the respondents who responded ‘no’ to the first question.

P-E fit was measured using the statement ‘The workspaces are well-suited for carrying out my work tasks’, evaluated on a five-point scale (1 strongly disagree, 5 strongly agree; Ruohomäki, Haapakangas, and Lahtinen 2013).

Interactive needs were measured using the statement ‘My work requires continual interaction with other employees’ (Ruohomäki, Sirola, and Lahtinen 2021) and a five-point scale (Not at all; Only a little; To some extent; Quite a lot; Very much). The cognitive demands and quantitative demands of work were measured using four-item scales from the second version of the Copenhagen Psychosocial Questionnaire (COPSOQ II; Pejtersen et al. 2010) translated into Finnish. Quantitative demands were rated on a five-point scale (Never/hardly ever; Seldom; Sometimes; Often; Always) and cognitive demands on a four-point scale (No, never; Rarely; Yes, sometimes; Yes, often). The responses to each scale were averaged.

A job profile variable was formed by first splitting interactive needs and cognitive demands into two categories (low ≤3; high >3) based on the median, which was 3.0 for both variables. Four job profile categories were then formed from the combinations of the low vs. high categories of these two variables (Table 1).

Table 1. Distributions of responses in variables included in regression models.

Variable	Category	%
Task-related factors
Job profile	Both high	22.1
	High IN, low CD	15.8
	Low IN, high CD	24.9
	Both low	37.2
Managerial position	No	86.4
	Yes	13.6
Person-related factors
Age (years)	Under 46	33.1
	46–55	33.4
	Over 55	33.4
Gender	Women	72.5
	Men	27.5
Work ability	Below 8	28.0
	8	26.1
	9–10	45.9
Satisfaction with ergonomics	Satisfied	36.6
	Neutral	39.3
	Dissatisfied	24.1
Situational factors
Office presence (hours per week)	<20 h	28.6
	21–30 h	30.7
	Over 30 h	40.7
Office distractions	Agree	50.9
	Neutral	19.6
	Disagree	29.4
Quantitative demands	High	32.0
	Medium	27.0
	Low	41.0
Outcome variables
Daily workspace switching (no. of times)	0	60.4
	1–2	22.4
	3 or more	17.1
Weekly workspace variation (no. of types)	1–2	34
	3–4	46.1
	5 or more	19.9
Daily time spent looking for workspace (minutes)	0	61.3
	1–5	25.4
	Over 5	13.3
P-E fit	Good	61.9
	Poor/neutral	38.1

Managerial position, age, and gender were also included.

Work ability was assessed using a single item (‘What score would you give your current work ability?’) from the Work Ability Index (Ilmarinen 2006; Tuomi et al. 1998). This item correlates well with the whole index (Ahlstrom et al. 2010). We used a scale from 1 (‘Completely unable to work’) to 10 (‘Work ability at its best’).

Single statements were used to assess distractions (‘There are many distractions’; Hongisto et al. 2016) and satisfaction with ergonomics (‘Special needs have been taken into account in the workspaces, e.g. with regard to moving, hearing, vision, and allergies’). The scale was the same as that for the P-E fit.

Office presence was evaluated by asking ‘Please estimate how many hours you usually work in the following locations each week’ (Ruohomäki, Sirola, and Lahtinen 2021). Several locations, of which ‘At the workplace’ was analysed, were rated using the same six categories as for the weekly workspace variation.

Statistical methods

Statistical analyses were performed using Spearman’s correlation, ordinal regression with a logit link function, and logistic regression (IBM SPSS Statistics, Version 27, IBM Corporation). For regression analyses, we merged categories to avoid sparse data issues (Table 1). The number of categories was limited to three for consistency and statistical power and to preserve the ordinal nature of the data. For continuous variables, the classification was based on achieving as even a distribution of the responses across categories as possible, given that some variables had very skewed distributions. Of the categorical variables that measured the degree of agreement, the outermost categories were merged to maintain a logical interpretation (i.e. agree, neutral, disagree). However, P-E fit was analysed using logistic regression as the middle category had to be merged with the lowest categories due to a skewed distribution (Table 1). Separate unadjusted models were first estimated for all the independent variables (task-related, person-related, situational) to investigate their associations with office use and P-E fit. The models were then adjusted for job profile to address the second research question. The model estimates of the ordinal regression were transformed into odds ratios (ORs) by exponentiation. The proportional odds assumption of ordinal regression was tested using the test of parallel lines. In the few cases in which this assumption was violated, multinomial logistic regression was used.

Results

The descriptive statistics for all the variables are shown in Table 1 and the correlations in Table 2. Daily workspace switching ranged from zero to seven switches (M = 0.9, SD = 1.4), weekly workspace variation ranged from one to nine workspace types (M = 3.3, SD = 1.6), and the daily time spent looking for a workspace ranged from 0 to 30 min (M = 2.7, SD = 5.2).

Table 2. Associations between variables analysed in this study.

	1	2	3	4	5	6	7	8	9	10	11	12
1 Daily workspace switching	1
2 Weekly workspace variation	.457***	1
3 Time spent searching for workspace	.194**	.183**	1
4 P-E fit	.196***	.258***	−.103	1
5 Interaction needs	.210***	.206***	−.085	.313***	1
6 Cognitive demands	.215***	.273***	.096	.011	.286***	1
7 Age	−.184**	−.249***	−.090	−.206***	−.084	−.093	1
8 Work ability	.200***	.106	−.123*	.352***	.195**	.075	−.010	1
9 Satisfaction with ergonomics	.174**	.238***	−.129*	.523***	.210***	.042	−.166**	.278***	1
10 Office presence	.013	.027	−.141*	.134*	.164**	.112*	−.112*	.098	.063	1
11 Distractions	−.218***	−.167**	.115*	−.617***	−.206***	−.025	.120*	−.265***	−.432***	−.110*	1
12 Quantitative demands	.050	.064	.180**	−.277***	.040	.355***	−.010	−.228***	−.136*	.030	.241***	1

Spearman’s correlation coefficients, r_S, for original non-categorised variables are reported, excluding binary variables. The response scales are indicated in the ‘Questionnaire measures’ section.

*p < .05, **p < .01, ***p < .001.

Table 3 shows the results of the regression models for office use and Table 4 shows those for P-E fit. The statistically significant findings are reported below.

Table 3. Results of unadjusted and adjusted ordinal regression models for office use outcomes: daily switching, weekly workspace variation, and daily time spent looking for workspace.

Variable	Category	Daily workspace switching								Weekly workspace variation								Daily time spent looking for workspace
		Unadjusted				Adjusted				Unadjusted				Adjusted				Unadjusted				Adjusted
		OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p
Task-related factors
Job profile	Both high	2.99	1.65	5.43	<0.001					2.56	1.46	4.50	0.001					0.89	0.49	1.61	0.69
	High IN, low CD	1.61	0.82	3.17	0.17					1.74	0.93	3.24	0.083					0.41	0.19	0.88	0.021
	Low IN, high CD	1.36	0.75	2.49	0.32					1.94	1.13	3.33	0.016					1.13	0.65	1.98	0.66
	Both low (ref.)
Managerial position	Yes	3.17	1.72	5.85	<0.001	2.14	1.11	4.11	0.023	3.37	1.84	6.2	<0.001	2.71	1.42	5.16	0.002	PO viol.				0.79	0.39	1.59	0.51
	No (ref.)
Person-related factors
Age	56 or older	0.35	0.2	0.63	<0.001	0.36	0.20	0.66	<0.001	0.33	0.2	0.56	<0.001	0.36	0.21	0.61	<0.001	0.65	0.38	1.14	0.14	0.63	0.35	1.11	0.11
	46–55	0.88	0.52	1.5	0.65	1.01	0.58	1.76	0.98	0.61	0.37	1.02	0.059	0.72	0.42	1.22	0.22	0.92	0.54	1.58	0.77	0.90	0.51	1.58	0.71
	Under 46 (ref.)
Gender	Female	0.62	0.38	1.01	0.054	0.60	0.36	0.99	0.044	0.55	0.35	0.87	0.011	0.58	0.36	0.93	0.025	0.90	0.55	1.46	0.67	0.88	0.54	1.45	0.62
	Male (ref.)
Work ability	9–10	2.68	1.5	4.81	0.001	2.19	1.20	4.00	0.011	1.38	0.83	2.27	0.21	1.21	0.72	2.04	0.47	0.70	0.41	1.20	0.20	0.75	0.43	1.32	0.32
	8	2.35	1.23	4.49	0.010	2.40	1.23	4.67	0.010	1.54	0.87	2.71	0.14	1.7	0.95	3.05	0.073	1.3	0.72	2.35	0.39	1.40	0.76	2.57	0.28
	Below 8 (ref.)
Satisfaction with ergonomics	Dissatisfied	0.46	0.26	0.83	0.010	0.45	0.24	0.84	0.012	PO viol.				0.36	0.20	0.64	<0.001	2.33	1.3	4.16	0.004	2.44	1.33	4.45	0.004
	Neutral	0.57	0.35	0.94	0.029	0.68	0.40	1.13	0.14					0.76	0.47	1.23	0.26	1.72	1.02	2.90	0.042	1.69	0.98	2.90	0.058
	Satisfied (ref.)
Situational factors
Office presence	31–40 h	1.05	0.62	1.8	0.85	0.81	0.46	1.43	0.47	1.27	0.77	2.08	0.35	1.14	0.68	1.91	0.62	0.46	0.27	0.78	0.004	0.46	0.27	0.80	0.006
	21–30 h	1.21	0.69	2.11	0.51	0.87	0.47	1.59	0.64	1.69	1	2.86	0.052	1.40	0.80	2.45	0.23	0.47	0.27	0.82	0.009	0.47	0.26	0.85	0.013
	20 h or less (ref.)
Distractions	Agreed	0.43	0.26	0.71	0.001	0.45	0.26	0.76	0.003	0.61	0.38	0.98	0.042	0.62	0.38	1.02	0.062	PO viol.				PO viol.
	Neutral	0.54	0.29	1	0.051	0.53	0.28	1.02	0.058	0.53	0.29	0.97	0.040	0.51	0.28	0.95	0.033
	Disagreed (ref.)
Quantitative demands	High	1.33	0.78	2.24	0.29	1.14	0.65	2.01	0.65	1.32	0.81	2.14	0.26	1.16	0.69	1.94	0.57	PO viol.				2.40	1.36	4.23	0.003
	Medium	1.15	0.66	1.99	0.62	1.25	0.71	2.21	0.44	1.38	0.83	2.3	0.21	1.47	0.87	2.49	0.15					1.81	1.02	3.23	0.044
	Low (ref.)

IN: interactive needs; CD: cognitive demands.

The adjusted models include job profile as a covariate. Outcome variables had three categories (see Table 1). In cases where the proportional odds assumption was violated (PO viol.), multinomial logistic regression was used, and results are reported in the text. Bold values indicate statistically significant odds ratios with a p-value less than 0.05.

Table 4. Results of unadjusted and adjusted logistic regression models for P-E fit, with poor or neutral fit as reference category.

Variable	Category	Unadjusted				Adjusted
		OR	95% CI		p	OR	95% CI		p
Task-related factors
Job profile	Both high	2.12	1.12	4.01	0.022
	High IN, low CD	3.15	1.44	6.89	0.004
	Low IN, high CD	0.94	0.53	1.67	0.84
	Both low (ref.)
Managerial position	Yes	2.78	1.29	5.99	0.009	2.99	1.25	7.18	0.014
	No (ref.)
Person-related factors
Age	56 or older	0.33	0.18	0.61	<0.001	0.37	0.20	0.69	0.002
	46–55	0.36	0.20	0.67	0.001	0.47	0.25	0.88	0.018
	Under 46 (ref.)
Gender	Female	0.62	0.36	1.04	0.070	0.63	0.37	1.10	0.10
	Male (ref.)
Work ability	9–10	4.61	2.61	8.14	<0.001	5.38	2.91	9.95	<0.001
	8	3.80	2.01	7.19	<0.001	4.61	2.34	9.11	<0.001
	Below 8 (ref.)
Satisfaction with ergonomics	Dissatisfied	0.08	0.04	0.15	<0.001	0.08	0.04	0.16	<0.001
	Neutral	0.25	0.14	0.47	<0.001	0.22	0.11	0.42	<0.001
	Satisfied (ref.)
Situational factors
Office presence	31–40 h	2.21	1.28	3.81	0.005	2.08	1.17	3.72	0.013
	21–30 h	1.46	0.83	2.56	0.19	1.22	0.67	2.25	0.52
	20 h or less (ref.)
Distractions	Agreed	0.09	0.04	0.17	<0.001	0.10	0.05	0.21	<0.001
	Neutral	1.00	0.38	2.60	1.0	1.36	0.50	3.71	0.55
	Disagreed (ref.)
Quantitative demands	High	0.28	0.16	0.49	<0.001	0.25	0.14	0.46	<0.001
	Medium	0.62	0.35	1.11	0.11	0.61	0.33	1.13	0.12
	Low (ref.)

IN: interactive needs; CD: cognitive demands.

Adjusted models include job profile as a covariate. Bold values indicate statistically significant odds ratios with a p-value less than 0.05.

Office use

In terms of task-related factors, employees with a combination of high interactive needs and high cognitive demands used workspaces more actively, both daily and weekly, than the reference group. Similar results were observed among managers in comparison to non-managers. High cognitive demands in combination with low interactive needs were associated with higher weekly workspace variation, whereas the combination of high interactive needs and low cognitive demands was associated with spending less time looking for a workspace.

Regarding person-related factors, the oldest employees (55+) switched workspaces less often daily and used a lower variety of workspaces during the week than those aged under 46. Women used a lower variety of workspaces during the week than men. A similar association emerged for daily workspace switching when the job profile was controlled for.

Better work ability was associated with more daily workspace switching, whereas dissatisfaction with ergonomics was associated with less daily switching. Multinomial logistic regression showed that dissatisfaction with ergonomics was related to lower odds of using over five workspaces weekly (OR = 0.15, 95% CI: 0.06–0.41). It was also associated with more time spent looking for a workspace.

As for situational factors, employees who worked at the office for over 20 h per week spent less time daily looking for a workspace than those who worked at the office less often.

Perceiving distractions was associated with less daily workspace switches and using fewer workspaces during the week. Multinominal logistic regression showed that perceiving, as opposed to not perceiving, distractions were associated with higher odds of spending over 5 min looking for a workspace daily (OR = 5.30, 95% CI: 1.79–15.7).

Multinominal logistic regression showed that high quantitative demands were associated with spending over 5 min looking for a workspace (OR = 3.1, 95% CI:1.5–6.8), whereas medium quantitative demands were associated with spending 1–5 min looking for a workspace (OR = 2.5, 95% CI: 1.3–4.6).

Person-environment fit

All factors were associated with the fit, except gender. In terms of job profile, high interactive needs were associated with a good P-E fit, particularly if combined with low cognitive demands. Managerial position, lower age, better work ability, and greater satisfaction with ergonomics were associated with higher odds of a good P-E fit. As for situational factors, differences were observed between the outermost categories of the explanatory variables. That is, working at the office for over 30 h per week was associated with a higher P-E fit, whereas perceiving distractions and having high quantitative demands were related to lower odds of a good P-E fit, in comparison to reference categories.

Adjusted models

The observed associations remained statistically significant when job profile was adjusted for, with very few exceptions. Regarding satisfaction with ergonomics and distractions, the associations with office use ceased to be significant in some but not all categories (Table 3).

Discussion

This study sought to identify any factors that explain individual variation in how ABOs are used and perceived. In line with the results of earlier studies (Appel‐Meulenbroek, Groenen, and Janssen 2011; Hoendervanger et al. 2016), most employees used workspaces passively and individual variation was great. The investigated task-related, person-related, and situational factors were associated with at least one office use outcome, and all except gender were related to perceived P-E fit (RQ1). Furthermore, these associations were independent of job profile (RQ2), supporting the assumption that office use is only partly task-based. In sum, employees used workspaces more actively if they were younger, were male, had better work ability, and were satisfied with office ergonomics. Office use appeared more difficult (in terms of having to spend more time looking for a workspace) among employees who worked less at the office, had high quantitative work demands, and were dissatisfied with ergonomics. Our results suggest that ABO design should be improved to accommodate a wider range of user needs, and should enable employees with different abilities and characteristics to achieve a good P-E fit.

Person-related and situational factors

The results concerning age complement earlier observations of older employees having more negative perceptions of ABOs (e.g. Hoendervanger et al. 2018; Pullen 2014; van den Berg et al. 2020). Our study suggests that such results may be partly related to more passive workspace switching, in contrast to the behaviour expected by Wohlers and Hertel (2017). Older workers appear to have more individual needs regarding office ergonomics (Afacan 2015; Gonzalez and Morer 2016) which may be difficult to accommodate in shared office spaces. As for gender differences, women switched workspaces less often than men, both daily and weekly. Other studies have shown that some workers prioritise social motives over task-based reasons in workspace choices (Hoendervanger et al. 2016; Bäcklander et al. 2021) and it seems possible that social motives could be more important for women. Gender differences could also reflect different task characteristics, as women and men are not evenly represented in all jobs. As previous findings (Bodin Danielsson et al. 2014, 2015; Bodin Danielsson and Theorell 2019) and hypotheses (Wohlers and Hertel 2017) on gender differences are mixed, more research is needed on whether and how gender is relevant to office use.

Previous research has overlooked individual limitations to flexibly using workspaces, despite employees’ ability to create a good P-E fit being viewed as central to the benefits of ABOs (Gerdenitsch, Korunka, and Hertel 2018; Hoendervanger et al. 2022; Wohlers, Hartner-Tiefenthaler, and Hertel 2019). Our study contributes to this research gap by showing that particularly lower satisfaction with ergonomics but also lower work ability were related to more passive use of the ABO. Furthermore, both factors had a strong association with a poorer P-E fit. Flexible workspaces have usually been viewed as a job resource (e.g. Bäcklander et al. 2019; Fincke et al. 2020; Wohlers and Hertel 2017), but specific needs or poorer work ability could prevent some employees from utilising them. Flexible workspaces might even become a job demand in such cases. Although knowledge of the health effects of ABOs is still limited (Engelen et al. 2019), results from traditional offices suggest that office conditions can even affect the risk of disability retirement (Nielsen, Emberland, and Knardahl 2021). More research on the interplay between ergonomic needs, work ability, and office conditions is clearly needed, even though the topic is challenging as it involves very individual and complex factors.

In terms of situational factors, distractions were related to all the outcome variables and had the highest correlation with P-E fit. These findings are in concordance with earlier observations that ABOs often fail to optimally support privacy and concentrative work (Engelen et al. 2019; Hoendervanger et al. 2022). Also in line with earlier accounts (Haapakangas, Hongisto, et al. 2018; Hoendervanger et al. 2019, 2022; Wohlers and Hertel 2017), the insufficient use of quiet workspaces may partly explain our findings. Such behaviour might reflect a discrepancy between work demands and the number and availability of suitable workspaces (Hoendervanger et al. 2022; Haapakangas, Hongisto, et al. 2018), distance or difficult access to appropriate workspaces (Haapakangas, Hongisto, et al. 2018), or personal preferences, such as sitting near colleagues (Hoendervanger et al. 2016). In this case, the eight-floor building had only one quiet area, which may have weakened its usability (e.g. distance, difficulty moving work tools, no knowledge of availability). Assigning home zones for teams may have exacerbated the problem in this ABO, as moving to another floor highlighted physical separation from colleagues. However, not having home zones might lead to other challenges, such as a decreased sense of community due to difficulties locating colleagues (Haapakangas et al. 2019).

Office presence was not related to daily or weekly workspace use, which contradicts the results of Hoendervanger et al. (2016). However, employees who worked less often at the office spent more time looking for a workspace, as also previously observed by Bosch-Sijtsema, Ruohomäki, and Vartiainen (2010). Such employees might have poorer situational awareness of conditions at the workplace or may need time to re-organize their activities when changing location. These issues may partly explain a poorer P-E fit. A reverse explanation is also possible that employees who experience a poor P-E fit at the office may prefer to work elsewhere. Nevertheless, this association is worth further investigation because the increase in teleworking during the Covid-19 pandemic may have caused a permanent shift in employees’ preferences towards working less at the office (Eurofound 2021; Ruohomäki et al. 2020).

Higher quantitative work demands were associated with increased time spent looking for workspace and a poorer P-E fit. Other studies have observed that moving to an ABO may negatively affect the quantity of performed work or perceived quantitative demands even 6 (Meijer, Frings-Dresen, and Sluiter 2009) or 12 months (Haapakangas et al. 2019) after moving. As most of the respondents of our study had worked in the ABO for 7–11 months, these results could still reflect an adaptation phase. Although various factors can contribute to a high workload, the associations between quantitative demands, work ability, distractions, office use, and P-E fit may indicate a non-optimal interplay between job demands, environmental stressors, and the individual’s ability to use office spaces adaptively. Perceiving workspace switching as more time-consuming may also indicate stress rather than an actual increase in time needed because time might be perceived as more valuable when one’s workload is heavy (Block 2014; Droit-Volet and Meck 2007).

Task-related use of ABOs

Our results also provide new information on how task characteristics are related to the use and perception of ABOs. Unlike earlier studies (e.g. Hoendervanger et al. 2016, 2022; Wohlers, Hartner-Tiefenthaler, and Hertel 2019), we investigated interactive and cognitive job demands in combination to better capture the nature of knowledge work (Heerwagen et al. 2004). The combination of high interactive and high cognitive demands was related to more active workspace use and a higher P-E fit than that of low interactive and low cognitive demands. These results align with the assumption that task variety moderates the effects of ABOs (Hoendervanger et al. 2016; Wohlers and Hertel 2017). Although high cognitive demands also contributed to more active office use on a weekly level, the overall pattern of results agrees with earlier conclusions that ABOs support collaborative work the best (Engelen et al. 2019).

A managerial role was related to more active workspace use and a higher P-E fit. These results could similarly reflect higher task variety (Wohlers and Hertel 2017). It is also possible that as managers are involved in implementing the ABO, they intentionally lead by example and adapt to activity-based working more quickly. Recent findings by Sirola et al. (2022) support this explanation: They found more positive perceptions of the implementation process and ABO workspaces among managers. The managerial role has seldom been investigated in ABO research and could be an important moderator to consider.

Theoretical implications

In terms of theoretical implications, our findings highlight the complexity of the factors behind the variation in individuals’ ways of using ABOs. These differences in office use may be more important for understanding the effects of ABOs on employees than current theoretical accounts have realised (e.g. Wohlers and Hertel 2017). The assumption that individual differences in office use contribute to individual variation in the perceptions and effects of ABOs is, however, compatible with the framework of Wohlers and Hertel (2017) as well as the P-E fit approach. Future studies could specify the relations depicted by Wohlers and Hertel (2017) by, for example, testing whether the relation between office use and long-term outcomes, such as well-being (Haapakangas, Hallman, et al. 2018), is moderated by task- and person-related variables, and whether P-E fit could mediate relations between office use and short- and long-term consequences.

Second, the factors identified in our study offer potential moderators that could also be relevant to other associations in the framework of Wohlers and Hertel (2017), such as those between perceived working conditions and the long-term effects of ABOs. Although the moderating role of the managerial position has been previously recognised (Sirola et al. 2022), work ability and ergonomic needs present a new category of person-related factors that have been overlooked in the past.

Third, our study adds to the knowledge on the potentially demanding aspects of ABOs (e.g. Babapour, Karlsson, and Osvalder 2018; Kim et al. 2016; van der Voordt 2004). In the light of the P-E fit theory, individuals’ abilities and resources (e.g. energy, skills, work ability, time) affect their interaction with the environment, such as switching behaviour in ABOs. While some ABO studies have approached the P-E fit more specifically as a need-supply fit (e.g. Gerdenitsch, Korunka, and Hertel 2018), we suggest that the concept of demands-abilities fit (Edwards et al. 2006) could provide complementary information on ABOs’ benefits and risks, particularly among more vulnerable groups (e.g. those with special needs). Evidence from other contexts suggests that a need-supply misfit may be preceded by a demands-abilities misfit (Edwards, Caplan, and Harrison 1998). Our results also contribute to the use of the Job Demands-Resources model (Bakker and Demerouti 2007) to conceptualise ABOs’ working conditions. Future studies could pay more attention to specifying the conditions in which certain ABO features become resources vs. demands, as workspace switching, for example, might be either a resource or a demand, depending on work ability.

Finally, the assumption of task-based workspace use appears too simplistic to characterise user behaviour in ABOs. There is a need for theoretical development to conceptualise the wider motivational basis and other prerequisites of user behaviour, to account for accumulating observations of non-task-based choices in ABOs (e.g. Appel‐Meulenbroek, Groenen, and Janssen 2011; Babapour, Karlsson, and Osvalder 2018; Babapour Chafi, Harder, and Bodin Danielsson 2020; Hoendervanger et al. 2016; Kim et al. 2016).

Practical implications

ABO design should not be based solely on assumptions regarding task-based behaviour at the workplace. Although employees need training in learning to work in an ABO, designing offices to incorporate versatile user needs should also be part of the solution for a better P-E fit. Inspiration could be drawn from the idea of a universal design, i.e. an environment ‘usable by all people to the greatest extent possible without the need for adaptation and specialized design’ (Mace 2004). Including user-centric and participative methods in the design and implementation process should help ensure that user needs are widely addressed. Monitoring workspace use with, for example, occupancy sensors, and gathering feedback provide ways in which to identify the obstacles to flexible workspace use. Managers and HR departments should work together with occupational health services if possible, to find tailored solutions to individual situations related to special needs and work ability.

Limitations

Some accuracy was lost due to having to combine categories for statistical analyses. As office use tends to have very skewed distributions (Haapakangas, Hallman, et al. 2018; Hoendervanger et al. 2016), considerably larger samples are needed to more accurately define investigated relations, including finer profiling of job characteristics. Office use could also be measured objectively by wearable technology (Montanari et al. 2017) or data from organisations’ occupancy sensors.

This study did not investigate causal relations. Some relations could be opposite to the discussed explanations or explained by unknown third variables. This study revealed several interesting questions for future research, but further studies are needed to investigate the observed associations and (causal) mechanisms behind them in more detail.

Only one organisation was investigated, and the use (Haapakangas, Hallman, et al. 2018; Hallman, Mathiassen, and Jahncke 2018; Hoendervanger et al. 2016) and perceptions (Brunia, De Been, and van der Voordt 2016) of ABOs may vary greatly between workplaces. In this case, the results were likely influenced by difficult access to the quiet space, the inclusion of home zones, and the preceding office type (private rooms), which do not apply to all ABOs. Caution is required when generalising these results as applying to other organisations and sectors, due to potential differences in contextual and cultural factors and office design.

Conclusions

This study showed that individuals differ greatly in how they use ABOs. The results challenge the underlying assumption regarding task-based office use by showing that various person-related and situational factors are also relevant to how the office is used and perceived. Thus, office designers and researchers should pay more attention to individual needs related to age, work ability, and ergonomics. As only one organisation was studied, the results may not be generalisable to all workplaces. Nevertheless, the study highlights the need to develop office design to accommodate heterogeneous user needs and to take employees’ different abilities and characteristics into account to achieve a high P-E fit.

Acknowledgements

The authors thank the participating organization and its contact person and employees, who provided valuable data for this study, and Maria Hirvonen M.Sc. (statistics) for her advice on statistical methods.

Disclosure statement

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

Data availability statement

Data not available due to legal restrictions.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by the Academy of Finland under Grant number 317279.