Boundary Work Tactics and Their Effects on Information and Communication Technology Use After Hours and Recovery

Sample

We conducted our main study in summer 2019. Participants were recruited via the authors’ professional and personal networks as well as various professional social media. In total, 249 professionals (50.2% female; mean age 38.75 years, SD = 12.29; 49.8% with managerial responsibilities) participated in the online study. For 73.1% of the participants, there were no children younger than 18 years living with them. Participants’ current employment breaks down as follows: employed (74.3%), working student (10.8%), civil servant (8.0%), self-employed (6.5%), and other (0.4%). To further validate the measure, checking construct validity and the tactics’ position in their nomological net, we conducted an additional validation study (see Table S1 in the Electronic Supplementary Material (ESM 1) for details on the constructs’ measures and their expected relationships). This sample of 175 workers (47.4% female, mean age: 48.4 years, 36.6% with managerial responsibilities, 70.3% without children living with them, 84% employed, 8.6% self-employed, and 7.6% civil servant) was recruited using a professional survey provider.

Measures

As the main study was conducted in German, scales were translated and back-translated, if necessary, by the authors. If not indicated otherwise, items were measured with a response scale ranging from 1 (never) to 5 (always).

Preliminary Analyses

We first tested the factor solution of the tactics, conducting an explorative factor analysis (principal axis factoring) with varimax rotation, using SPSS version 27. Examining the Kaiser’s criteria suggested seven factors with eigenvalues above 1, with three factors marginally showing values above 1. Considering the scree-plot as well as the small eigenvalues rather suggested a three-factor or four-factor solution. Among these two, the three-factor solution represented the most conceptually interpretable as well as statistically distinct solution, as the three factors corresponded to the three motive-oriented types of tactics, with Factor 1 corresponding to preventive tactics, Factor 2 to restrictive tactics, and Factor 3 to rejecting tactics. Yet, as depicted in Table S2 in ESM 1, six items either showed cross-loadings to another factor above .30 or loadings below .40 onto their corresponding factor. Following recommendations by Howard (2016), we removed these items and reran the analysis with 22 items. As shown in Table 2, factor loadings now ranged between .42 and .84 without any cross-loading above .30. This three-factor solution accounted for 38.48% of the total variance.

Table 2 Factor loadings for motive-oriented types of tactics (22 items)

Items	Preventive	Restrictive	Rejecting
Note. Results for factor loadings from explorative factor analysis (principal axis factoring) with varimax rotation with 22 items and N = 249. Results for factor loadings from the CFA of the validation study are shown in parentheses, with N = 175. Loadings below .30 are not displayed. a Item was discarded for hypothesis testing due to its factor loading below .40 in the CFA results of the validation study. The final scale consists of the remaining 21 items. Items marked with (r) were reversed for the analyses.
I use different phone numbers for work-related matters and for personal matters.	.56 [.52]
I use different e-mail accounts for work-related matters and for personal matters.a	.44 [.35]
I use different ICT for work-related matters and for personal matters.	.55 [.46]
I indicate to my work contacts via my status (e.g., online or absent) when I am available in my leisure time and when I am not.	.42 [.51]
I inform my work contacts by means of absence notes that I am not available during my leisure time.	.58 [.65]
I communicate to customers or colleagues that I prefer not to be contacted via ICT in my leisure time for work-related matters.	.58 [.89]
I communicate to supervisors that I prefer not to be contacted via ICT in my leisure time for work-related matters.	.62 [.91]
I communicate to my work contacts that in my leisure time, contact via ICT is only possible in an emergency.	.52 [.84]
In my leisure time, I limit myself to certain ICT (smartphone only, laptop only, etc.) for receiving and editing work-related messages.		.42 [.64]
In my leisure time, I put away ICT on which I am contacted professionally at certain times (e.g., during certain social interactions, sports, dinner).		.46 [.70]
In my leisure time, I limit myself to using written messages (no phone calls) for work-related communication because it gives me flexibility in determining how and when to respond.		.60 [.76]
In my leisure time, I limit work-related ICT use to certain time slots.		.75 [.76]
In my leisure time, I check work-related messages only at certain times.		.64 [.77]
In my leisure time, I limit the work-related use of ICT to a certain duration.		.77 [.78]
In my leisure time, I decide whether to take work-related calls depending on who is calling.		.43 [.69]
In my leisure time, I make decisions about responding to work-related messages based on urgency through the use of an answering machine or mailbox.		.44 [.66]
In my leisure time, I decide which work-related messages to read or process based on urgency.		.55 [.74]
In my leisure time, I read work-related messages. (r)			.84 [.95]
In my free time, I check work-related messages. (r)			.78 [.92]
I take ICT, on which I am contacted professionally, with me in my leisure time. (r)			.67 [.77]
I take ICT, on which I am contacted professionally, with me on vacation. (r)			.70 [.65]
In my leisure time, I take work-related calls. (r)			.55 [.79]

Table 2 Factor loadings for motive-oriented types of tactics (22 items)

View as image HTML

Besides, we conducted confirmatory factor analyses (CFA) with Mplus (Muthén & Muthén, 1998-2017) with our validation sample, which suggested to further remove the item (preventive tactic) “I use different email accounts for work-related matters and for personal matters” due to loadings below .40 (see Table 2). The resulting, final three-factor model showed a better fit (χ² = 347.89, df = 180; RMSEA = 0.07; CFI = 0.93, SRMR = 0.08) than a one-factor model (χ² = 591.30, df = 179; RMSEA = 0.12; CFI = 0.84, SRMR = 0.13). We also examined the three motive-oriented tactics’ intercorrelations with relevant constructs in their nomological net with the validation sample. As shown in Table 3, correlations with the related construct of general boundary work tactics were mostly low to medium in size, with the highest correlation of r = .57. Besides, the three motive-oriented tactics showed differential correlations with segmentation preferences and segmentation enactment as we expected in our description of the tactics’ categorization. For example, restrictive tactics showed the lowest correlations with segmentation preferences (r = .19) as well as segmentation enactment (r = .21), while rejecting tactics showed the highest correlations (r = .38 for segmentation preferences; r = .52 for segmentation enactment). Together, these results suggest discriminant and convergent validity, indicating that the three motive-oriented tactics represent distinct types with differential relationships in their nomological net. Hence, we proceeded with this three-factor solution in our main study. The final measure with 21 items is shown in Table S3 in ESM 1. Table 4 depicts the correlations between the study variables in our main study.

Table 3 Validation study: descriptive statistics and correlations among motive-oriented tactics and variables in their nomological net

Variable	M	SD	α	1	2	3	4	5	6	7	8	9
Note. Pearson correlations among motive-oriented tactics and variables in their nomological net, with their corresponding 95% confidence interval in square brackets. α = Cronbach’s α; WLB = work–life balance. N = 175. *p < .05, **p < .01.
1. Preventive tactics	3.05	1.17	.87	—
2. Restrictive tactics	3.25	1.11	.91	.60** [.48; .71]	—
3. Rejecting tactics	3.60	1.05	.91	.35** [.21; .47]	.10 [−.07; .26]	—
4. Temporal tactics	3.95	0.81	.79	.28** [.17; .39]	.25** [.12; .39]	.40** [.25; .54]	—
5. Physical tactics	3.81	0.86	.90	.27** [.14; .39]	.17* [.03; .31]	.55** [.44; .65]	.73** [.63; .81]	—
6. Behavioral tactics	2.96	1.31	.95	.41** [.27; .55]	.47** [.32; .59]	.02 [−.13; .17]	.34** [.20; .47]	.21** [.07; .34]	—
7. Communicative tactics	3.27	1.23	.86	.57** [.46; .68]	.54** [.39; .66]	.27** [.12; .42]	.35** [.22; .47]	.42** [.28; .55]	.44** [.30; .56]	—
8. Segmentation preferences	3.88	0.99	.91	.34** [.20; .47]	.19* [.05; .33]	.38** [.25; .51]	.40** [.27; .53]	.51** [.40; .62]	.20** [.06; .34]	.46** [.33; .59]	—
9. Segmentation enactment	3.27	1.07	.94	.36** [.22; .48]	.21** [.07; .35]	.52** [.39; .62]	.55** [.44; .65]	.72** [.63; .79]	.24** [.09; .39]	.41** [.28; .54]	.56** [.43; .67]	—
10. Satisfaction with WLB	3.87	0.93	.96	.18* [.05; .31]	.07 [−.07; .21]	.38** [.25; .50]	.49** [.37; .60]	.55** [.42; .66]	−.01 [−.16; .15]	.16* [.02; .29]	.11 [−.03; .26]	.41** [.28; .55]

Table 3 Validation study: descriptive statistics and correlations among motive-oriented tactics and variables in their nomological net

View as image HTML

Table 4 Descriptive statistics and correlations among motive-oriented tactics and study variables

Variable	M	SD	α	1	2	3	4	5	6	7	8
Note. Pearson correlations among motive-oriented tactics and other study variables, with their corresponding 95% confidence interval in square brackets. α = Cronbach’s α. ICT use = work-related ICT use after hours; Detachment = psychological detachment. N = 249. †p < .1, *p < .05, **p < .01.
1. Job demands	3.39	0.75	.80	—
2. Preventive tactics	2.67	0.97	.76	.25** [.11; .37]	—
3. Restrictive tactics	2.87	0.86	.81	.13* [.01; .25]	.26** [.14; .36]	—
4. Rejecting tactics	2.72	0.99	.83	−.16* [−.27; −.04]	.26** [.13; .37]	.08 [−.08; .22]	—
5. ICT use	3.42	0.96	—	.17** [.06; .28]	−.19** [−.31; −.07]	−.13* [−.27; .01]	−.45** [−.57; −.34]	—
6. Detachment	2.93	0.87	.90	−.31** [−.43; −.18]	.04 [−.10; .17]	.11 [−.03; .23]	.35** [.23; .46]	−.26** [−.38; −.13]	—
7. Relaxation	3.52	0.78	.89	−.12† [−.27; −.03]	.20** [.08; .33]	.18** [.03; .30]	.19** [.07; .31]	−.11† [−.24; .04]	.38** [.25; .49]	—
8. Mastery	3.28	0.77	.86	.03 [−.11; .17]	.09 [−.05; .21]	.11† [−.03; .25]	.00 [−.13; .13]	−.03 [−.15; .10]	.04 [−.11; .16]	.27** [.11; .39]	—
9. Control	3.96	0.82	.93	−.15* [−.28; −.01]	.07 [−.05; .19]	−0.03 [−.16; .09]	.12† [−.01; .25]	.00 [−.15; .14]	.23** [.10; .35]	.51** [.40; .59]	.30** [.18; .41]

Table 4 Descriptive statistics and correlations among motive-oriented tactics and study variables

View as image HTML

Hypothesis Testing

We tested our hypotheses with path analyses using Mplus. We specified the proposed paths between the three motive-oriented tactics, ICT use, and recovery experiences in Model 1. Recovery experiences were allowed to covary. In Model 2, we additionally controlled for paths from job demands to all endogenous variables. We show the results in Table 5.

Table 5 Results from path analyses and relative weight analyses

Variable	Work-related ICT use after hours				Psychological detachment				Relaxation				Mastery				Control
	Est.	SE	t	RRW	Est.	SE	t	RRW	Est.	SE	t	RRW	Est.	SE	t	RRW	Est.	SE	t	RRW
Note. Estimates are standardized results from path analyses, specifying all paths simultaneously in Model 1 and Model 2, respectively. Est. = estimates; RRW = raw relative weights; ICT use = work-related ICT use after hours. N = 249. †p < .1, *p < .05, **p < .01.
Model 1
Preventive tactics	−0.06	0.06	−1.13	.01	−0.09	0.06	−1.43	.01	0.13	0.07	1.97*	.02	0.07	0.07	0.94	.01	0.06	0.07	0.86	.00
Restrictive tactics	−0.08	0.06	−1.25	.02	0.10	0.06	1.62	.00	0.13	0.07	1.97*	.02	0.10	0.08	1.24	.01	−0.05	0.07	−0.74	.00
Rejecting tactics	−0.43	0.06	−6.96**	.19	0.31	0.07	4.67**	.10	0.15	0.07	2.01*	.03	−0.03	0.07	−0.45	.00	0.14	0.07	1.93†	.02
ICT use	—	—	—		−0.12	0.07	−1.79†	.04	0.00	0.07	0.04	.00	−0.02	0.07	−0.30	.00	0.06	0.07	0.87	.00
Model 2
Preventive tactics	−0.11	0.06	−1.81†	.01	−0.01	0.06	−0.17	.01	0.18	0.07	2.74**	.03	0.07	0.08	0.86	.01	0.11	0.07	1.63	.00
Restrictive tactics	−0.09	0.06	−1.44	.03	0.12	0.06	2.11*	.00	0.15	0.07	2.26*	.03	0.10	0.08	1.24	.01	−0.03	0.06	−0.50	.01
Rejecting tactics	−0.39	0.06	−6.21**	.17	0.27	0.07	3.86**	.08	0.12	0.08	1.57	.02	−0.03	0.08	0.02	.00	0.11	0.07	1.48	.01
ICT use	—	—	—		−0.08	0.06	−1.19	.03	0.03	0.07	0.44	.00	−0.02	0.07	−0.30	.00	0.09	0.07	1.25	.00
Job demands	0.15	0.06	2.51*	.03	−0.27	0.06	−4.34**	.08	−0.17	0.07	−2.41*	.02	0.00	0.08	0.02	.00	−0.17	0.07	−2.48*	.02

Table 5 Results from path analyses and relative weight analyses

View as image HTML

In regard of work-related ICT use after hours, rejecting tactics had a negative effect (β = −0.43, p < .01), while preventive tactics (β = −0.06, p = .26) and restrictive tactics (β = −0.08, p = .21) were unrelated to ICT use (Model 1). When controlling for job demands, the effect of preventive tactics became marginally nonsignificant (β = −0.11, p = .07; Model 2). Hypothesis 1c was supported, while Hypotheses 1a and 1b were not.

Similarly, rejecting tactics (β = 0.31, p < .01) were positively related to psychological detachment in Model 1, while preventive (β = −0.09, p = .15) and restrictive tactics were not (β = 0.10, p = .11). However, when controlling for job demands in Model 2, not only rejecting tactics (β = 0.27, p < .01) but also restrictive tactics (β = 0.12, p < .05) showed a positive effect on psychological detachment. Thus, Hypotheses 2b and 2c were supported, but only when controlling for job demands. Hypothesis 2a was rejected.

Regarding relaxation, preventive (β = 0.13, p < .05), restrictive (β = 0.13, p < .05), and rejecting tactics (β = 0.15, p < .05) showed a positive effect in Model 1. In Model 2, the positive effect of preventive (β = 0.18, p < .01) and restrictive tactics remained (β = 0.15, p < .05), while rejecting tactics did not predict relaxation anymore (β = 0.12, p = .12), supporting Hypotheses 3a and 3b but not Hypothesis 3c.

In regard of mastery, no significant effects were found, neither in Model 1 nor Model 2. Regarding control, only rejecting tactics showed a marginally nonsignificant, positive effect in Model 1 (β = 0.14, p = .054), while preventive (β = 0.06, p = .39) and restrictive tactics did not (β = −0.05, p = .46). However, in Model 2, the positive effect of rejecting tactics became nonsignificant (β = 0.11, p = .14). Hypotheses 4 and 5 were rejected.

Finally, to examine RQ1, we compared the effects of the three motive-oriented tactics on work-related ICT use after hours and recovery experiences. The findings suggest that their raw relative weights (Tonidandel & LeBreton, 2015) differ, depending on outcome (Table 5): Rejecting tactics were the strongest predictor of work-related ICT use after hours and psychological detachment. Preventive and restrictive tactics were the strongest predictors of relaxation when controlling for job demands.

Implications for Research

Our study contributes to research in several ways. First, we provide a comprehensive overview of boundary work tactics found in previous studies (e.g., T. D. Allen et al., 2021; Cousins & Robey, 2015; Sayah, 2013). Organizing these different tactics in (1) preventive, (2) restrictive, and (3) rejecting tactics, we propose a conceptual framework that helps to distinguish the variety of tactics according to their underlying motives. Our studies’ results provide support that diverse tactics can be ascribed to these three types of motives. With this motive-oriented perspective, our research complements previous studies that mostly focused on contentual categories (e.g., Fonner & Stache, 2012; Lirio, 2017). Besides, by showing that the three motive-oriented tactics relate differently to segmentation preferences and segmentation enactment, our research initially acknowledges that boundary work tactics may serve different purposes – with rejecting tactics facilitating impermeable boundaries and restrictive tactics allowing for more permeability, which reflects boundary work on the segmentation–integration continuum (Ashforth et al., 2000; Kreiner et al., 2009). With its parsimonious, theory-based framework, our conceptualization has the potential to align the various existing categorizations because additional tactics could be assigned to our framework. As our studies further provide initial support that the three tactics relate differently to boundary conditions such as segmentation preferences and job demands, a fruitful pathway for future studies may be to build on this conceptual framework and to examine how such boundary conditions interact with the use and efficacy of boundary work tactics. As such, segmentation preferences and job demands might not only predict the use of tactics but also determine if and how the tactics relate to various outcomes. Consequently, we could gain a more theory-based understanding of which tactics might be more effective for segmenters versus integrators or when job demands are high versus low.

Second, we contribute to research by capturing a variety of distinct availability-related boundary work tactics in our studies. While previous studies investigated rather broad tactics, which did not necessarily refer to concrete actions (Carlson et al., 2016) or only addressed selected specific tactics (Park et al., 2020), our research portrays a diversity of specific boundary work tactics and provides a scale to measure them. Thereby, our validation study provides first insights on how the motive-oriented tactics relate to constructs in their nomological net. Together, our approach to examine distinct tactics enables to add to our understanding of what specific strategies may be effective in managing boundaries and fostering recovery (Hu et al., 2021; Rudolph et al., 2021).

Third, our study offers initial insights on the relationship between different types of availability-related boundary work tactics, work-related ICT use after hours, and recovery experiences. While previous findings on boundary work tactics mostly resulted from qualitative studies (e.g., T. D. Allen et al., 2021; Sayah, 2013), we shed light on the relationships of the use of tactics with recovery in a quantitative study design.

In more detail, our results show that rejecting tactics are negatively related to work-related ICT after hours, indicating that tactics aiming at stricter segmentation are most effective in reducing work-related ICT use after hours compared to the other tactics. In addition, our findings suggest that several tactics can help to foster recovery. Specifically, our results show that rejecting tactics are positively related to psychological detachment. Thus, substantiating previous qualitative results (Sayah, 2013), rejecting tactics might help to not only behaviorally but also to mentally disengage from work. Besides, we found that restrictive tactics are positively related to psychological detachment when controlling for job demands. This finding indicates that when employees decide to not engage with work for some time after hours, for example, by putting away their work-related ICT during certain activities, their psychological recovery process can start. Besides, our study showed a positive effect of both preventive and restrictive tactics on relaxation. In contrast, rejecting tactics did not predict relaxation when job demands were controlled for. Together, these findings indicate that not only tactics building very strong boundaries can support recovery experiences but also tactics that allow for some permeability can be beneficial. Particularly, restrictive tactics might represent an important alternative strategy to promote recovery when the use of rejecting tactics is not realistic or wanted, for example, when employees are working flexibly or prefer to integrate their life domains (Kreiner et al., 2009; Sayah, 2013).

Results on mastery and control were less conclusive, showing rather nonsignificant relationships in our analyses. This finding suggests that boundary work tactics might be more crucial to foster psychological detachment and relaxation. Rejecting tactics seem to be most effective to promote psychological detachment and thus, to help to reduce psychological load from work demands (Headrick et al., 2023). In addition, restrictive and preventive tactics seem helpful to increase relaxation, a recovery experience that rather helps to gain psychological resources (Headrick et al. 2023).

Lastly, our findings indicate that reducing work-related ICT use after hours might not be the decisive factor to promote recovery: Work-related ICT use after hours did not predict any recovery experiences when job demands were controlled for. Instead, job demands showed positive effects on work-related ICT use after hours, as well as negative effects on psychological detachment, relaxation, and control. This finding underpins previous notions from scholars that ICT use is not detrimental for employee recovery per se (Heissler et al., 2022; Reinke & Ohly, 2021) but rather a result of high job demands and consequently, of work-related thoughts. Thus, work-related ICT use after hours could rather be seen as a behavioral strategy to deal with high job demands (Heissler et al., 2022). Besides, job demands affected the relationships of tactics with recovery in our study. When controlling for job demands, restrictive and preventive tactics became more effective, while rejecting tactics became less helpful. When job demands were not controlled for, rejecting tactics overall seemed to be more effective than the other tactics. These findings indicate that different boundary work tactics might be used in response to different levels of job demands. Hence, job demands need to be considered in future research as a potential antecedent of the use of boundary work tactics as well as a moderator that might influence the tactics’ effects on recovery.

Practical Implications

Our results provide support that using availability-related boundary work tactics can foster employee recovery, particularly psychological detachment and relaxation. Hence, organizations should support their employees in engaging in boundary work tactics, for example, by developing employment agreements or by offering trainings in which employees learn about boundary work tactics and practice to implement them in their daily life. Thereby, employees should first become aware of their boundary management preferences, which might range from segmentation to integration. Next, they should learn about preventive, rejecting, and restrictive tactics and how those can match their preferences. As a result, employees may become more conscious about their boundary management and learn to actively apply those tactics that are suitable for their preferences and needs.

Limitations and Directions for Future Research

Our study has several limitations. First, the use of cross-sectional data does not allow for any causal inferences. Hence, it should be a fruitful pathway for future research to measure job demands, boundary work tactics, work-related ICT use after hours, and recovery in a longitudinal study design. Since recovery experiences are likely to vary within individuals among time (Sonnentag & Fritz, 2007), studies could examine whether those within-person fluctuations are affected by the person’s use of boundary work tactics in a daily diary study. Besides, reciprocal relationships between the use of boundary work tactics and recovery experiences should be examined (Heissler et al., 2022): Boundary work tactics might be a behavioral response to low recovery experiences or high demands. For example, when individuals experience low detachment, they might engage in tactics that help them to overcome this.

We further suggest that future studies should examine the role of boundary conditions influencing the efficacy of boundary work tactics. Beyond the aforementioned potential role of job demands and segmentation preferences as moderators, other availability-related demands such as availability expectations and norms (Thörel et al., 2022) might affect whether the use of boundary work tactics actually translates into improved recovery and well-being.

Although our categorization of boundary work tactics and the corresponding measurement scale were based on a comprehensive literature review, future studies could expand our focus on availability-related tactics and add other tactics such as transition rituals (Ashforth et al., 2000; Kreiner et al., 2009). Besides, while our studies provide initial support for the scale’s validity, it should be noted that model fit in the CFA was acceptable yet not good (Schermelleh-Engel et al., 2003), possibly because of the relatively small sample size. Alternatively, this might indicate that the scale should be refined, for example, by combining content-oriented categories with motive-oriented categories. Thus, we strongly suggest testing and validating the scale in different samples and with an extended set of criterion variables, such as work-related thoughts or affective well-being (e.g., Thörel et al., 2022).

Conclusion

Our research has three important implications. First, the diversity of availability-related boundary work tactics can be assigned to three motive-oriented types of tactics: preventive, restrictive, and rejecting tactics, which acknowledges that boundary work occurs along the segmentation–integration continuum. Second, our study offers insights into the relationships between different tactics and recovery, indicating that all three motive-oriented tactics contribute to recovery, with rejecting and restrictive tactics predicting psychological detachment and preventive and restrictive tactics predicting relaxation when job demands are controlled for. Finally, our results add to previous research questioning work-related ICT use after hours as a stressor itself and suggest rather considering the role of job demands. Together, this research provides new insights about the structure and effects of availability-related boundary work tactics for professionals’ work-related ICT use after hours and recovery.