Experiment 1 was conducted in the laboratory. We manipulated the environment with the help of virtual 3D-environments, designed by TriCAT GmbH2. Such virtual environments provide a complex sensory experience in order to manipulate the environment in a controlled, but still realistic, manner.
3.2 Participants
2 www.tricat.net
one. The remaining 132 participants had a mean age of 23.58 (SD = 3.61, range from 18 to 35), majority (n = 91) was female. Participants were randomly assigned to both environment conditions (n = 66 work vs. n = 66 non-work). A majority of participants (n = 79) indicated that they had at least a part time job, the remaining 53 participants were students. Volunteers were paid 8€ for participation.
3.3 Procedure and manipulation of the environment
After being seated and signing an informed consent statement, participants started the experiment with a 5-minute free exploration through the virtual environment (work vs.
non-work) from the first-person perspective. Participants were asked to empathize with their environment, thinking about how they would spend their time if they were there in actuality. Participants navigated through the environment with help of the integrated keyboard and the computer mouse. It was possible to sit down on the virtual furniture by means of a mouse click. Presentation of the virtual environments and all of the tasks and assessments were conducted on a laptop (HP 15.4” HP EliteBook 8530p).
The virtual environment was programmed in Unity R Pro © (Unity R Pro, 2016). It consisted of a flat office building with three rooms, surrounded by a large terrace and a park. The office building and the outside park were connected via a sliding door. In the work-associated environment conditions, participants were only allowed to stay within the office building, navigating through the three office rooms. Office rooms were furnished with a desk, chairs, white boards and flip charts. In the non-work-associated environment condition, participants were only allowed to stay outside, navigating through the terrace and park area. The park area consisted of lawns with benches, a tiled terrace with garden furniture and trees (see Figure 1).
Figure 1. Screenshots of the work (left) and non-work (right) environments.
minutes the participants were asked to sit down virtually on either a desk chair (work-associated environment) or a park bench (non-work-(work-associated environment) and to open a virtual tablet by clicking on it. The questionnaire was presented on the virtual display of the tablet within the virtual environment. This meant that while participants filled out the survey on the tablet, they still had the possibility to raise their vision seeing the virtual environment surrounding them.
After opening the tablet, the experiment followed the general procedure of assessing decision making3, the moderators and the control variables. In Experiment 1, one control variable was added to the general procedure: a questionnaire assessing immersion and any dizziness or nausea experienced in the virtual environment (Presence questionnaire by Witmer & Singer, 1998).
3.3 Results and discussion
Manipulation of environments
Regarding the closed question asking for associations of the environment with work vs.
non-work, the manipulation worked out as expected. Participants in the work-associated environment did associate the environment more with work (M = 4.55, SD = 1.46) compared to participants in the non-work-associated environment (M = 1.92, SD = .95), t(130) = 12.232, p < .001. And the other way round, participants in the non-work environment associated the environment more with leisure (M = 4.52, SD = 1.62) compared to participants in the work environment (M = 1.79, SD = .69), t(130) = -12.58, p <.001. This also held true for the additional measurement at the end of the experiment (both p <. 001 in the expected direction). In addition, the free association task indicated successful manipulation. One participant is missing in the analyses as he/she did not insert a free association. Participants in the work-associated environment mentioned more work-related words (M = -.10, SD = .18) than participants in the non-work-associated environment (M = .14, SD = .19), t(129) = 7.40, p < .001.
Effects of the environment on work-related vs. non-work-related decision making
3 In addition KONT-P and ZST were assessed in a randomized manner, see section 2.1.
In Hypothesis 1a we assumed that participants would make more risky decisions in a work-associated environment (compared to a non-work-associated environment) when the decision making was work-related. We did not find any differences in work-related decision making: the decision making of participants in the work-associated
environment (M = 3.21, SD = .55) was not riskier than the decision making of
participants in the non-work-associated environment (M = 3.12, SD = .73), t(130) = .82, p = .413. Regarding Hypothesis 1b, we also didn’t find any differences in non-work-related decision making: the decision making of participants in the work-associated environment (M = 3.13, SD = .57) was not any less risky than the decision making of participants in the nonworkassociated environment (M = 3.21, SD = .72), t(130) = -.70, p = .488.
Conditional effects of the environment
We next examined the conditional effects of the environments on work-related vs. non-work-related decision making, this time including mood in a moderation model (Model 1 as suggested by Hayes, 2012). Conditional effects (bCE) of the environment on decision making are only reported when the bootstrap confidence interval (bootstrap = 1000) of the interaction does not include zero.
Non-work-related decision making in work-associated environments was not affected by a positive mood. However, we did find a conditional effect of environment on work-related decision making for a low (one standard deviation below the mean) positive mood: Participants experiencing a low positive mood showed more risky behaviour in work-related decision making when they were in a work-associated environment (see Table 1, Figure 2) compared to the non-work-associated environment. Note: the effect of a positive mood has to be treated with caution, as the confidence interval of the coefficient includes zero (see Table 1).
.25 .06 .41 2.89 3 128 .038
Note. Sample size n = 132; R = coefficient of correlation, R2 = coefficient of determination, MSE = mean squared error, F = F-Test statistics, Df = degrees of freedom, p = significance value, b = unstandardized beta coefficient, se = standard error, t = t-test statistic, CIl = lower confidence interval, CIh = higher confidence interval
Figure 2. Conditional effects of the environment on work-related decision making (from 1, no risk taking, to 6, high risk taking); separated for work-associated and non-work-associated environment. Lines show low (black), medium (grey), and high (dashed) positive mood (a). Lines marked with an asterisk * show significant effects.
We did not find any significant conditional effect of environment on non-work-related decision making when mood was included in the model (all p > .143).
Control variables
We did not find any differences between the work and the non-work condition regarding gender, age, level of education, and current state of employment (all p >
.112). Regarding participants who reported being currently employed, there was one difference in professional status. Participants assigned to the work condition rated the
1 2 3 4 5 6
Work Non-Work
Work-related decision making
Environment
Low positive mood
*
Medium positive mood
High positive mood
b se t p CIl CIh
Constant 3.19 .35 9.00 .000 2.49 3.90
Environment -1.21 .54 -2.23 .028 -2.28 -.14
Positive mood .00 .02 .06 .952 -.030 .04
Interaction .06 .03 2.08 .040 .003 .11
However, means in both groups were quite low (note: with a range of 1 to 5) thus this result is negligible. We found no difference for ‘I have decision-making power’ (p = .535).
4. Experiment 2