Descriptive statistics of the sample
After three months of recruitment, a sample size of N = 62 completed questionnaires was reached. However, we had to exclude 16 participants from data analysis due to unmet inclusion criteria, resulting in a final sample size of N = 46. All participants were German citizens. Most of the participants work in the IT industry (N = 40 or 86.9%), in large enterprises (N = 25 or 54.3%). However, although this result may be biased by the recruitment strategy, this distribution of industries and enterprise sizes confirms previous findings among virtual teams (2). Further results are provided in Table 2.
Table 2
Sociodemographic and Occupational Characteristics of the Sample (N = 46)
Variable
|
n
|
%
|
Gender
|
|
|
Female
|
11
|
23.9
|
Male
|
35
|
76.1
|
Age
|
|
|
18-20 years
|
1
|
2.2
|
21-30 years
|
10
|
21.7
|
31-40 years
|
18
|
39.1
|
41-50 years
|
7
|
15.2
|
51-60 years
|
9
|
19.6
|
61-70 years
|
1
|
2.2
|
Highest educational level
|
|
|
Secondary education
|
9
|
19.6
|
Higher education
|
37
|
80.4
|
Leadership responsibility
|
|
|
Yes
|
20
|
43.5
|
No
|
26
|
56.5
|
Work experience
|
|
|
1-2 years
|
13
|
28.3
|
3-5 years
|
16
|
43.8
|
6-10 years
|
7
|
15.2
|
≥ 11 years
|
10
|
21.7
|
Average working hours per week
|
|
|
35-40 hours
|
18
|
39.1
|
41-45 hours
|
18
|
39.1
|
≥ 46 hours
|
10
|
21.7
|
Descriptive statistics of the main variables
Means, standard deviations, Cronbach’s α, and zero-order correlations of the main variables are provided in Table 3.
Table 3
Means, Standard Deviations, Internal Consistencies, and Zero-Order Correlations
Variable
|
M
|
SD
|
Range
|
Min
|
Max
|
1
|
2
|
3
|
4
|
5
|
1. Boundarylessness
|
2.870
|
.966
|
1-5
|
1.000
|
5.000
|
(.866)
|
|
|
|
|
2. Virtuality
|
3.541
|
.524
|
1-5
|
2.625
|
4.875
|
.351*
|
(.561a / .731b)
|
|
|
|
3. Psychological detachment
|
3.011
|
.823
|
1-5
|
1.000
|
4.500
|
-.395**
|
-.215
|
(.844)
|
|
|
4. Perceived stress
|
2.490
|
.559
|
1-5
|
1.700
|
3.800
|
.092
|
.237
|
-.241
|
(.843)
|
|
5. Sleep quality
|
3.766
|
.706
|
1-5
|
2.000
|
5.000
|
.018
|
-.279
|
.282
|
-.367*
|
(.815)
|
Note. N = 46. One-tailed Spearman Correlation Coefficients were used. * p < .05. ** p < .01. Cronbach’s alphas are listed in parentheses on the diagonal. aInternal consistency for self-developed virtuality scale. bInternal consistency for virtuality scale after removing items on digital media usage
Descriptive statistics of virtual job demands and health promotion offers
We also evaluated job demands other than boundarylessness using another self-developed scale as well as free text answers. Six participants reported specific job demands, such as a need for clear and measurable objectives, lack of social exchange, required availability and high discipline, as well as different levels of media skills and technological equipment among virtual team members. Results from the descriptive analysis of the virtual job demands scale indicate that the participants of this study rather collaborated with virtual team members within the same time zone, but speaking different languages. Although social exchange among and integration of virtual team members seemed to pose greater challenges to virtual team members in this sample, results show high levels of satisfaction with communication and collaboration within the virtual teams. More details are provided in Table 4.
Table 4
Descriptive Statistics for Virtual Job Demands
Item
|
M
|
SD
|
Min
|
Max
|
Team members in different time zones
|
1.739
|
1.255
|
1
|
5
|
Team members with other mother tongues or dialects
|
2.848
|
1.414
|
1
|
5
|
Team members with different cultural backgrounds
|
2.935
|
1.357
|
1
|
5
|
Higher susceptibility to errors in communication
|
2.457
|
1.609
|
1
|
5
|
Higher susceptibility to conflicts among virtual team members
|
2.391
|
1.105
|
1
|
5
|
More difficulties due to diverse cultural and linguistic backgrounds
|
1.935
|
.952
|
1
|
5
|
More difficulties due to geographical distance
|
2.196
|
1.046
|
1
|
5
|
Restricted possibility of social exchange
|
3.130
|
1.128
|
1
|
5
|
More difficulties of socially integrating virtual team members
|
3.087
|
1.226
|
1
|
5
|
More difficulties in performance assessment
|
2.500
|
1.169
|
1
|
5
|
Difficulties to build trust in virtual team members
|
2.304
|
1.133
|
1
|
5
|
Satisfaction with communication within virtual team
|
4.087
|
.812
|
2
|
5
|
Satisfaction with collaboration within virtual team
|
4.152
|
.868
|
1
|
5
|
Note. N = 46. All scales are five-point Likert scales, ranging from one to five.
Regarding organisational health promotion, 54.3% of all participants reported that their employers provided personnel development or health promotion offers, but only 39.1% of those who were offered this opportunity made use of it. The most frequently mentioned measures include teambuilding activities (45.7%), soft skill training (39.1%), and self-management training (37.0%).
Relationships between boundarylessness, virtuality, psychological detachment, and health outcomes
An ordered logit model was estimated to investigate whether different levels of boundarylessness (“low”, “moderate”, “high”) predict different levels of sleep quality (“poor,” “medium,” “good”), testing hypothesis H1. The predictors did not account for a significant amount of variance in the outcome, likelihood ratio c2(2) = .588, p = .745. The ordered logit model for hypothesis H2a could not confirm the hypothesised association between boundarylessness and perceived stress (c2(2) = .589, p = .745). Hypothesis H2b, a negative relation between perceived stress and sleep quality, could be confirmed. The tested ordered logit model indicated that different levels of perceived stress (“low”, “moderate”, “high”) accounted for a significant amount of variance in sleep quality, likelihood ratio c2(2) = 7.667, p =.022. Low levels of perceived stress, b = 1.747, SE = .726, OR = 5.737, p = .016, and moderate levels of perceived stress, b = 1.561, SE = .708, OR = 4.764, p = .027, predicted better sleep quality. Good model fit and a satisfied assumption of proportional odds could be confirmed (c2(2) = 1.284, p =.526). These results indicate that for virtual team members of this sample who perceived low or medium levels of stress, the odds of being more likely to enjoy good sleep quality were more than 4-5 times higher than those who perceived higher levels of stress. To test hypothesis H2c, a mediation analysis based on model 4 was conducted using Hayes’s PROCESS macro for SPSS. Following hypotheses H1-H2b already tested, of which two could not be confirmed, the assumed mediation effect of perceived stress on the relation between boundarylessness and sleep quality could not be confirmed either (with indirect effects ab1 = .0130, SE = .1128, 95% CI [-.2062, .2542] and ab2 = -.0635, SE = .1216, 95% CI [-.3619, .1391]). Hypothesis H3a had to be rejected due to lack of significance (c2(2) = 1.315, p = .518), indicating no significant relation between the degree of virtuality and perceived stress. Virtuality was not found to be a significant moderator either, therefore denying H3b (c2(6) = 3.878, p = .693) and H3c (c2(6) = 8.572, p = .199). The assumed significant relation between psychological detachment and perceived stress (H4a) could not be confirmed (c2(2) = 1.351, p = .509). Both of the assumed moderation effects of psychological detachment could not be confirmed (c2(6) = 3.127, p = .793 for H4b and c2(6) = 12.077, p = .060 for H4c). However, post hoc analyses revealed a significant association of psychological detachment and sleep quality (c2(2) = 6.099, p = .047), indicating a positive relation between these two variables. Only lower psychological detachment (b = -1.615, SE = .678, OR = .199, p = .017) predicted sleep quality. The model’s goodness of fit, as well as the assumption of proportional odds (c2(2) = .630, p = .730), could be confirmed. Accordingly, we may assume that poor psychological detachment from work decreased the odds of enjoying good sleep quality among this sample of virtual team members. More details are provided in Table 5 and Table 6.
Table 5
Ordinal Logistic Regressions of Associations Between Predictors and Sleep Quality
Effect
|
OR
|
95% CI
|
|
|
LL
|
UL
|
Boundarylessness
|
|
|
|
Low
|
1.462
|
.430
|
4.968
|
Medium
|
.892
|
.221
|
3.607
|
High
|
Ref.
|
|
|
Virtuality
|
|
|
|
Low
|
2.147
|
.583
|
7.909
|
Medium
|
.430
|
.111
|
1.670
|
High
|
Ref.
|
|
|
Detachment
|
|
|
|
Low
|
.199
|
.053
|
.752
|
Medium
|
.634
|
.163
|
2.469
|
High
|
Ref.
|
|
|
Perceived Stress
|
|
|
|
Low
|
5.737
|
1.383
|
23.831
|
Medium
|
4.764
|
1.190
|
19.068
|
High
|
Ref.
|
|
|
B x V
|
|
|
|
Low
|
.128
|
.005
|
3.445
|
Medium
|
.441
|
.042
|
4.614
|
High
|
Ref.
|
|
|
B x D
|
|
|
|
Low
|
2.504
|
.130
|
48.327
|
Medium
|
8.298
|
.915
|
75.189
|
High
|
Ref.
|
|
|
Note. N = 46, OR = odds ratio, CI = confidence interval, LL = lower limit, UL = upper limit, Ref. = reference category. BxV = interaction term of boundarylessness and virtuality. BxD = interaction term of boundarylessness and psychological detachment.
Table 6
Ordinal Logistic Regressions of Associations Between Predictors and Perceived Stress
Effect
|
OR
|
95% CI
|
|
|
LL
|
UL
|
Boundarylessness
|
|
|
|
Low
|
.661
|
.194
|
2.248
|
Medium
|
.631
|
.155
|
2.557
|
High
|
Ref.
|
|
|
Virtuality
|
|
|
|
Low
|
.504
|
.139
|
1.828
|
Medium
|
.556
|
.147
|
2.104
|
High
|
Ref.
|
|
|
Detachment
|
|
|
|
Low
|
1.978
|
.564
|
6.931
|
Medium
|
1.033
|
.269
|
3.963
|
High
|
Ref.
|
|
|
B x V
|
|
|
|
Low
|
1.931
|
.082
|
48.183
|
Medium
|
.582
|
.061
|
5.607
|
High
|
Ref.
|
|
|
B x D
|
|
|
|
Low
|
1.654
|
.111
|
24.656
|
Medium
|
.666
|
.095
|
4.660
|
High
|
Ref.
|
|
|
Note. N = 46, OR = odds ratio, CI = confidence interval, LL = lower limit, UL = upper limit, Ref. = reference category. BxV = interaction term of boundarylessness and virtuality. BxD = interaction term of boundarylessness and psychological detachment.
Differences in degrees of virtuality
In addition, we conducted median split t-tests to examine differences between groups among the sample. Splitting the data into higher and lower degrees of virtuality revealed that those 50% of participants who reported a higher degree of virtuality in their teamwork also reported significantly higher values (t(44) = -2.327, p = .025, d = .685) of boundarylessness (M = 3.200, SD = .966, n = 22) than those who worked in a team of a lower degree of virtuality (M = 2.567, SD = .880, n = 24).
Differences in levels of boundarylessness
Moreover, splitting the data into higher and lower levels of perceived boundarylessness resulted in a significant difference with regard to psychological detachment (t(44) = 1.138, p = .003, d = .928). Participants who experienced lower levels of boundarylessness reported higher levels of psychological detachment (M = 3.330, SD = .742, n = 24) compared to participants who experienced higher boundarylessness also reporting lower psychological detachment (M = 2.631, SD = .764, n = 22).
Differences in levels of psychological detachment
We also found significant differences in the levels of psychological detachment between participants with leadership responsibility and those without t(44) = 2.071, p = .044, d = .620). Compared to participants without leadership responsibility (M = 2.798, SD = .815, n = 26), supervisors reported higher levels of psychological detachment (M = 3.288, SD = .766, n = 20). All results from median split t-tests are provided in Figure 2.