Data analysis. We discarded data from two participants because a program error was found for one participant and the other participants misunderstood an instruction for the RAT and could not follow the appropriate procedure. Thus, we analyzed data only on the remaining 30 participants.
To represent physiological arousal level changes, we calculated the averages of SCL values for front rest, intermediate rest, and blocks of the RATs55. Reactivity in response to every block was calculated by subtracting baseline SCL from SCL during the RAT blocks, which combined nature exposure (absence or presence) and phone (presence or absence) conditions; front rest as the baseline SCL corresponded to blocks 1 and 2, and intermediate rest corresponded to the remaining blocks.
To confirm whether the individual differences in cell phone usage dependence affect the observed effect of exposure to nature on creativity, the PUMP score between the nature absence and presence conditions were analyzed for normality (Shapiro–Wilk normality test: meanabs = 127.20, Wabs = 0.92, pabs = 0.19; meanpre = 116.13, Wpre = 0.97, ppre = 0.89), equality of variances (F-test: F(1, 14) = 0.80, p = 0.69, 95% CI 0.27–2.40), and differences in means (T-test: t(28) = −0.93, p = 0.36, IC 95% = −35.41–13.28, d = 0.34), respectively. These results show that the two groups of PUMP scores followed a normal distribution with equal variance and no difference between the means was observed, indicating that any influence we observed between nature absence and presence could not be attributed to the individual inference of cell phone usage.
The angular transformed values of the ratio of correct RAT response and reaction time (key press reaction time in the thinking section of the correct answer) were analyzed using two-way analysis of variance (ANOVA) with the presence of a smart phone and exposure to nature as factors. We analyzed SCL using a three-way ANOVA with phone presence, exposure to nature, and RAT block as factors. The score of negative affect and positive affect was analyzed by a three-way ANOVA with measurement timing (marked as P1, P2, P3, and P4), phone presence, and exposure to nature as factors. Subjective ratings other than PANAS were analyzed using a two-way ANOVA as in the analysis of hit rate. All degrees of freedom were adjusted using Chi-Muller’s epsilon. All the multiple comparisons used Shaffer’s multiple-comparison procedure. The details of the nonsignificant results are available at http://osf.io/x93j6/?view_only=8c8ed1f4bd294f8c928c3b4226fd6907.
RAT performance. Figure 3 shows the ratios of correct responses (left); the main effect of phone presence was significant, F(1, 28) = 11.06, p < 0.01, η2 = 0.0625, whereas the main effect of exposure to nature was not significant, F(1, 28) = 0.30, p = 0.59, η2 = 0.0083. The two-way interaction for nature and phone presence was also not significant, F(1, 28) =1.46, p = 0.24, η2 = 0.0083. Figure 3 (right panel) shows that we observed no significant difference in either main effects or interactions in the results for reaction time. Irrespective of nature condition, the participants with a smart phone nearby generated more correct responses than did those who had only a mobile battery present. In short, the presence of a cell phone increased creativity task performance, supporting H1a.
SCL change. The main effect of the block was significant, F(1.35, 37.85) = 9.24, p < 0.01, η2 = 0.0751, whereas the main effects of phone presence, F(1, 28) = 2.38, p = 0.13, η2 = 0.0129, and exposure to nature, F(1, 28) = 1.05, p = 0.31, η2 = 0.0156, were not. The two-way interaction for the block and phone presence was significant, F(2.24, 62.6) =3.26, p < 0.05, η2 = 0.0086, and the simple main effect of block was significant in the phone-absent condition, F(1.66, 46.61) =12.54, p < 0.01, η2 = 0.1092; this latter finding indicated that SCL changed more in block 1 (6.53 μS) than in blocks 3 (3.57 μS) and 4 (2.26 μS; ps < .05); SCL in blocks 2 and 3 was also higher than that in block 4 (ps < .05). Whereas the simple main effect of the block was marginally significant in the phone-present condition, F(1.48, 41.42) =3.53, p = 0.05, η2 = 0.0498, there were no significant differences between the blocks. The simple main effect of phone presence was only significant in block 2, F(1, 28) = 5.89, p < 0.05, η2 = 0.0334, which indicated that SCL changed more in the absence of a smart phone than in its presence in that block (3.56 μS > 1.81 μS). In the phone-absence condition, the SCL changes decreased over time, but we did not observe this tendency when the phone was present. Meanwhile, there was no significant difference between nature conditions; thus, H3b was not supported. Figure 4 graphically presents the SCL changes.
Subjective affect ratings.Figures 5 and 6 presents the PANAS results for affect ratings in the presence or absence of a nature environment. No main effects or interactions were significant for positive affect, whereas for negative affect (one scale of the PANAS), the main effect of measurement timing was significant, F(2.58, 72.12) =10.05, p < 0.001, η2 = 0.0523; neither the other main effects nor the interactions were significant. Participants showed higher scores for negative affect after completing the first half of the tasks (P1 < P2, P4; P3 < P4, ps < 0.05) that decreased after rest (P2 > P3, p < 0.05); moreover, the negative affect increment in the first half of the tasks was larger than that in the second half (P1 = P3, P2 > P4, p < 0.05). These results did not reflect any difference due to the presence of a smart phone; thus, H1b was not supported.
Figure 7 displays the remaining subjective ratings for all participants. For valence, the main effect of exposure to nature was significant, F(1, 28) = 5.40, p < 0.05, η2 = 0.0903, but not that for phone presence, F(1, 28) = 0.65, p = 0.43, η2 = 0.0100, which reflected that participants felt more pleasure from the presence of nature than they did in its absence irrespective of phone presence conditions. No two-ways interaction was detected, F(1, 28) = 0.20, p = 0.66, η2 = 0.0031. For arousal, neither main effect nor interaction was significant. Therefore, H3c was partially supported.
For desire of knowledge, the main effect of phone presence was marginally significant, F(1, 28) = 3.38, p = 0.07, η2 = 0.0244, whereas the main effect of exposure to nature was not, F (1,28) = 0.26, p = 0.61, η2 = 0.0071. No two-ways interaction was detected, F(1, 28) = 1.62, p = 0.21, η2 = 0.0117. The participants’ desire for knowledge was more conscious when the phone was absent than when it was present, which indicated that the mental demand without a smart phone was higher, and the participants had to attempt to retrieve more knowledge to arrive at a correct answer.
For frustration, both the main effect of exposure to nature, F(1, 28) = 4.38, p < 0.05, η2 = 0.0782, and the main effect of phone presence, F(1, 28) = 4.54, p < 0.05, η2 = 0.0588, were noted. No two-ways interaction was detected, F(1, 28) = 0.10, p = 0.75, η2 = 0.0014. Participants showed more frustration in the absence of nature than in its presence and more frustration in the absence of a smart phone than in its presence. For the remaining subjective ratings, all main effects and interactions were nonsignificant. Figure 7 presents all the mean subjective ratings. In the overall rating of five items from the NASA task load index, mental load in the form of frustration was lower in the nature environments, which partially supported H3a.