Preliminary analyses
Bivariate correlations and descriptive statistics, including the means and standard deviations of the variables and the reliability coefficients of the scales, are shown in Table 1.
Table 1. Bivariate correlation matrix for variables
|
|
M
|
SD
|
α
|
|
1
|
|
2
|
|
3
|
|
4
|
|
5
|
|
6
|
|
7
|
|
8
|
|
9
|
|
1
|
Stressor
|
.55
|
.19
|
―
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2
|
Optimism
|
2.63
|
.47
|
.72
|
|
-.24
|
**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3
|
Sleep Pattern
|
2.50
|
.57
|
.65
|
|
-.10
|
|
.05
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4
|
Sleep Irregularity
|
2.40
|
.82
|
.82
|
|
.03
|
|
-.03
|
|
.38
|
**
|
|
|
|
|
|
|
|
|
|
|
|
5
|
Commitment
|
2.58
|
.73
|
.78
|
|
.21
|
**
|
-.04
|
|
.28
|
**
|
.47
|
**
|
|
|
|
|
|
|
|
|
|
6
|
Appraisal of Influence
|
3.55
|
.53
|
.82
|
|
.07
|
|
.03
|
|
-.03
|
|
.00
|
|
.14
|
**
|
|
|
|
|
|
|
|
7
|
Appraisal of Threat
|
2.41
|
.90
|
.93
|
|
.12
|
*
|
-.16
|
**
|
.27
|
**
|
.48
|
**
|
.62
|
**
|
.12
|
*
|
|
|
|
|
|
|
8
|
Controllability
|
2.72
|
.60
|
.56
|
|
-.08
|
|
.18
|
**
|
-.30
|
**
|
-.43
|
**
|
-.26
|
**
|
.12
|
*
|
-.31
|
**
|
|
|
|
9
|
Depression
|
2.16
|
.85
|
.88
|
|
.35
|
**
|
-.38
|
**
|
.12
|
*
|
.13
|
*
|
.26
|
**
|
.12
|
*
|
.30
|
**
|
-.13
|
*
|
|
|
10
|
Anxiety
|
2.13
|
.77
|
.85
|
|
.31
|
**
|
-.16
|
**
|
.08
|
|
.07
|
|
.25
|
**
|
.06
|
|
.10
|
|
-.08
|
|
.73
|
**
|
*p < .05, **p < .01; for stressor, “no” is coded as “0” and “experienced” is coded as “1.”
Evening-type sleep pattern was found to be positively associated with commitment (r = .28, p < .01) and appraisal of threat (r = .27, p < .01), but negatively associated with controllability (r = −.30, p < .01). Further, sleep irregularity was also found to be significantly associated with higher commitment (r = .47, p < .01), higher appraisal of threat (r = .48, p < .01), and lower controllability (r = −.43, p < .01). In the correlation analyses, sleep pattern and irregularity were found to have weak positive association with depression but no associations with appraisal of influence or anxiety.
The control variables were found to be significantly associated with stress responses. Specifically, stressors were positively related to depression (r = .35, p < .01) and anxiety (r = .31, p < .01), whereas dispositional optimism was negatively related to depression (r = −.38, p < .01) and anxiety (r = −.16, p < .01).
Testing multiple mediation models
We hypothesized that sleep habits would predict stress responses, such as depression and anxiety, through cognitive appraisals. Consequently, four multiple mediation models (two for each sleep habit and two for each stress response) were created using Preacher and Hayes’ (2008) procedure (available through SPSS macro) [38], which also allowed for covariates. This approach, using bootstrap analysis, enabled us to explore mediating relationships, thereby overcoming the limitations of previous methods (assuming normal distribution, or if it is impossible to test the indirect effect; for a review, see Preacher and Hayes 2008) [38]. Indirect effects were tested using a bootstrapping procedure, which involved running 5,000 samples with 95% confidence intervals (CIs).
Sleep pattern and the stress response
Table 2 presents the direct and indirect effects of the proposed mediators on the relationship between sleep pattern and the stress response. Stressors and optimism were added as covariates. Fig. 1 depicts the hypothesized model.
Table 2. Results of the multiple mediation model for sleep pattern regarding depression and anxiety.
|
|
Depression
|
|
Anxiety
|
|
|
b
|
|
SE
|
|
|
t value
|
|
b
|
|
SE
|
|
|
t value
|
Partial effects of covariates on stress response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Stressor
|
1.19
|
|
|
.24
|
|
|
4.86
|
**
|
|
1.04
|
|
|
.24
|
|
|
4.35
|
**
|
|
Optimism
|
-.58
|
|
|
.10
|
|
|
-5.89
|
**
|
|
-.21
|
|
|
.10
|
|
|
-2.21
|
*
|
Effects of sleep pattern on mediators
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Commitment
|
.34
|
|
|
.07
|
|
|
5.30
|
**
|
|
.34
|
|
|
.07
|
|
|
5.30
|
**
|
|
Appraisal of Influence
|
.00
|
|
|
.05
|
|
|
-.10
|
|
|
.00
|
|
|
.05
|
|
|
-.10
|
|
|
Appraisal of Threat
|
.39
|
|
|
.08
|
|
|
4.85
|
**
|
|
.39
|
|
|
.08
|
|
|
4.85
|
**
|
|
Controllability
|
-.33
|
|
|
.05
|
|
|
-6.17
|
**
|
|
-.33
|
|
|
.05
|
|
|
-6.17
|
**
|
Direct mediating effects on stress response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Commitment
|
.09
|
|
|
.08
|
|
|
1.12
|
|
|
.27
|
|
|
.08
|
|
|
3.55
|
**
|
|
Appraisal of Influence
|
.13
|
|
|
.09
|
|
|
1.48
|
|
|
.06
|
|
|
.08
|
|
|
.70
|
|
|
Appraisal of Threat
|
.11
|
|
|
.06
|
|
|
1.75
|
|
|
-.15
|
|
|
.06
|
|
|
-2.44
|
*
|
|
Controllability
|
.05
|
|
|
.08
|
|
|
.61
|
|
|
-.01
|
|
|
.08
|
|
|
-.17
|
|
Effects of sleep pattern on stress response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Total
|
.23
|
|
|
.07
|
|
|
3.13
|
**
|
|
.14
|
|
|
.07
|
|
|
1.97
|
*
|
|
Direct
|
.17
|
|
|
.08
|
|
|
2.15
|
*
|
|
.10
|
|
|
.08
|
|
|
1.31
|
|
Indirect effects
|
b
|
|
SE
|
95% BCa CI
|
|
b
|
|
SE
|
95% BCa CI
|
Total
|
.06
|
|
|
.04
|
(-.02; .13)
|
|
.04
|
|
|
.03
|
(-.03; .11)
|
Commitment
|
.03
|
|
|
.03
|
(-.02; .08)
|
|
.09
|
|
|
.03
|
(.04; .16)
|
Appraisal of Influence
|
.00
|
|
|
.01
|
(-.02; .01)
|
|
.00
|
|
|
.00
|
(-.01; .01)
|
Appraisal of Threat
|
.04
|
|
|
.03
|
(-.00; .10)
|
|
-.06
|
|
|
.03
|
(-.12; -.01)
|
Controllability
|
-.02
|
|
|
.03
|
(-.08; .04)
|
|
.00
|
|
|
.03
|
(-.05; .06)
|
Model statistics
|
R
|
|
|
R2
|
|
|
F (7, 279)
|
|
R
|
|
|
R2
|
|
|
F (7, 279)
|
|
|
.53
|
|
|
.29
|
|
15.91
|
**
|
|
.40
|
|
|
.16
|
|
|
7.59
|
**
|
*p < .05, **p < .01.
Regarding the relationship between sleep pattern and depression, sleep pattern (a greater tendency to have an evening-type sleep pattern) positively predicted depression (b = .23, p < .01). With regard to cognitive appraisal, sleep pattern positively predicted commitment (b = .34, p < .01) and appraisal of threat (b = .39, p < .01), and negatively predicted controllability (b = -.33, p < .01).There was no significant relationship to appraisal of influence. The direct effects of cognitive appraisal types on depression were not significant. The direct effect of sleep pattern on depression was reduced with the inclusion of the mediators and covariates (b = .17, p < .05). The indirect effect of cognitive appraisals, both cumulatively and individually, showed that cognitive appraisals did not contribute to the prediction of depression (i.e., the 95% CI included zero).
In the results for the multiple mediation model for anxiety, sleep pattern positively predicted anxiety (b = .14, p < .01). Sleep pattern also positively predicted commitment (b = .34, p < .01) and appraisal of threat (b = .39, p < .01), whereas sleep pattern negatively predicted controllability (b = -.33, p < .01). There was no significant relationship to appraisal of influence. In addition, commitment and appraisal of threat significantly positively predicted anxiety. The direct effect of sleep pattern on anxiety became nonsignificant with the inclusion of the mediators (b = .10). Although the total effect was not significant (b = .04, 95% CI = -.03 to .11), commitment (b = .09, 95% CI = .04 to .16), and appraisal of threat (b = -.06, 95% CI = -.12 to -.01) contributed uniquely to the mediation effect (i.e., the 95% CI did not include zero). However, because appraisal of threat did not correlate with anxiety, we could not interpret the indirect effect of threat.
Sleep irregularity and the stress response
Table 3 shows the direct and indirect effects of the proposed mediators in the relationship between sleep irregularity and the stress response. Stressors and optimism were added as covariates. Fig. 2 depicts the hypothesized model.
Table 3. Results of the multiple mediation model for sleep irregularity regarding depression and anxiety.
|
|
Depression
|
|
Anxiety
|
|
|
b
|
|
SE
|
|
|
t value
|
|
b
|
|
SE
|
|
|
t value
|
Partial effects of covariates on stress response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Stressor
|
1.07
|
|
|
.24
|
|
|
4.39
|
**
|
|
.96
|
|
|
.24
|
|
|
4.07
|
**
|
|
Optimism
|
-.56
|
|
|
.10
|
|
|
-5.61
|
**
|
|
-.20
|
|
|
.10
|
|
|
-2.06
|
*
|
Effects of sleep irregularity on mediators
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Commitment
|
.38
|
|
|
.05
|
|
|
8.17
|
**
|
|
.38
|
|
|
.05
|
|
|
8.17
|
**
|
|
Appraisal of Influence
|
.02
|
|
|
.04
|
|
|
.51
|
|
|
.02
|
|
|
.04
|
|
|
.51
|
|
|
Appraisal of Threat
|
.48
|
|
|
.06
|
|
|
8.58
|
**
|
|
.48
|
|
|
.06
|
|
|
8.58
|
**
|
|
Controllability
|
-.28
|
|
|
.04
|
|
|
-7.30
|
**
|
|
-.28
|
|
|
.04
|
|
|
-7.30
|
**
|
Direct mediating effects on stress response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Commitment
|
.10
|
|
|
.08
|
|
|
1.28
|
|
|
.29
|
|
|
.08
|
|
|
3.66
|
**
|
|
Appraisal of Influence
|
.13
|
|
|
.09
|
|
|
1.54
|
|
|
.06
|
|
|
.08
|
|
|
.75
|
|
|
Appraisal of Threat
|
.12
|
|
|
.07
|
|
|
1.76
|
|
|
-.15
|
|
|
.06
|
|
|
-2.26
|
*
|
|
Controllability
|
.00
|
|
|
.08
|
|
|
.03
|
|
|
-.04
|
|
|
.08
|
|
|
-.50
|
|
Effects of sleep irregularity on stress response
|
|
|
|
|
|
|
|
|
|
|
|
|
Total
|
.12
|
|
|
.05
|
|
|
2.24
|
*
|
|
.06
|
|
|
.05
|
|
|
1.10
|
|
|
Direct
|
.03
|
|
|
.07
|
|
|
.40
|
|
|
.01
|
|
|
.06
|
|
|
.12
|
|
Indirect effects
|
b
|
|
SE
|
95% BCa CI
|
|
b
|
|
SE
|
95% BCa CI
|
Total
|
.10
|
|
|
.04
|
(.02; .18)
|
|
.05
|
|
|
.04
|
(-.02; .13)
|
Commitment
|
.04
|
|
|
.03
|
(-.02; .10)
|
|
.11
|
|
|
.03
|
(.05; .18)
|
Appraisal of Influence
|
.00
|
|
|
.01
|
(-.01; .02)
|
|
.00
|
|
|
.00
|
(-.01; .01)
|
Appraisal of Threat
|
.06
|
|
|
.03
|
(-.00; .12)
|
|
-.07
|
|
|
.03
|
(-.14; -.01)
|
Controllability
|
.00
|
|
|
.03
|
(-.05; .05)
|
|
.01
|
|
|
.03
|
(-.04; .07)
|
Model statistics
|
R
|
|
|
R2
|
|
|
F (7, 279)
|
|
R
|
|
|
R2
|
|
|
F (7, 279)
|
|
|
.52
|
|
|
.27
|
|
14.51
|
**
|
|
.39
|
|
|
.15
|
|
|
7.15
|
**
|
*p < .05, **p < .01.
Sleep irregularity positively predicted depression (b = .12, p < .05), commitment (b = .38, p < .01), and appraisal of threat (b = .48, p < .01), and negatively predicted controllability (b = -.28, p < .01). There was no significant relationship to appraisal of influence. The direct effects of the cognitive appraisal types on depression were not significant, and the direct effect of sleep irregularity on depression became nonsignificant with the inclusion of the mediators and covariates (b = .03). Analysis of the indirect effect (b = .10, 95% CI = .02 to -.18) showed that cognitive appraisals as a group contributed to the prediction of depression. However, no individual type of cognitive appraisal showed a unique mediating effect (i.e., the 95% CI included zero).
For the mediation model for sleep irregularity and anxiety, sleep irregularity did not predict anxiety (b = .06). This was despite the fact that there were significant relationships between sleep irregularity and cognitive appraisals and between cognitive appraisals and anxiety, as well as indirect effects. Because we could not confirm the preconditional relation between sleep irregularity and anxiety, we could not interpret the mediation effects.