Descriptive Statistics
As a first step, the descriptive characteristics of the participants were assessed and compiled in Table 4.1. The descriptive characteristics were broken down in three categories, namely gender, age and chronotype.
The assessment showed, that out of the total 200 respondents, 129 (64,5%) were male and 71 (35,5%) were female. The age distribution of respondents was broken down into four groups. Out of the four age groups, the groups containing the majority of respondents, were age group 31-40 years old (57,5%) and age group 41-50 years old (23.5%). The analysis of the morningness scores, emerging from the rMEQ, among respondents revealed, that 36 (18%) participants classified as morning chronotype, 134 participants as neutral chronotype and 30 (15%) participants as evening chronotype.
Table 4.1.
Descriptive Characteristics of Participants
Variable Category Frequency Percentage %
Gender 1. Male 129 64,5%
2. Female 71 35,5%
Age 1. 20-30 years 32 16%
2. 31-40 115 57.5%
3. 41-50 47 23.5%
4. 50-64 6 3%
Chronotype 1. Morning 36 18%
2. Neutral 134 67%
3. Evening 30 15%
Validity and Reliability
To assess the reliability of the utilized scale, Cronbach Alpha values for each construct scale were computed. The rMEQ scale, which was used to measure the independent variable Morningness, consisted of five items and delivered a cronbach alpha value of 0.622. The dependent variable, work engagement, was measured with the UWES scale, which had a cronbach alpha value of 0.933. The scale used for measuring the control variable job satisfaction, provided a cronbach alpha value of 0.821. The personality dimensions, which were used as a moderator in the analysis, were all measured with the BFI-10 scale, which utilizes 2 items per personality dimension. Cronbach alpha values for each dimension were as follows:
Extraversion (0,614), Agreeableness (0.209), Conscientiousness (0.234), Neuroticism (0.4749), Openness (0.443). As outlined in chapter III, these low alpha values were anticipated.
Looking at the reliability values of the scales utilized to measure each construct, it becomes apparent, that the cronbach alpha values for the rMEQ as well as the BFI-10 are either slightly or in some cases significantly lower than usually accepted. For example, the rMEQ scale returned an alpha value of 0.622. Although this value is slightly lower than the usual accepted threshold of 0.7(George & Mallery, 2003), it is important to point out that a lower than 0.7 value is no exception when working with the Chinese version of the rMEQ (Carciofo et al., 2012, 2016; Qu et al., 2015). Possible explanations for these repeating occurring values could be the low number of items that make up the scale as well as the limited sample size of studies. However, these low alpha values shouldn’t necessarily be considered as a sign for a lack of reliability of the utilized scale, since earlier research found significant test-retest correlations between the original MEQ and the reduced MEQ scores as well as chronotype classifications (Carciofo et al., 2012). Consequently, established researchers in the field do see merit in using the reduced scale and continue to do so in their research(Carciofo, 2019; Carciofo et al., 2016; Qu et al., 2015). Additionally, it is necessary to point out that some scholars make the case that alpha values between .5 and .7 demonstrate moderate reliability and can therefore be accepted (Dall’Oglio et al., 2010; Field, 2009; Hinton, 2004). Lastly, in the present study the scale was also inspected in light of the inter-item correlations. The average of inter-item correlations is 0.25, which provides further support for the reliability of the scale (Piedmont, 2014).
The low alpha values of the BFI-10 were anticipated, since the scale measures each of the five-personality dimension with only two items. Since the amount of items in a scale have a significant influence on the Cronbach alpha values, it is common to find low alpha values when utilizing mentioned scale (Furnham, 2008; Gosling et al., 2003). The alpha values found
in the present study actually were in line with those of a previous large-scale study also utilizing the Chinese BFI-10 (Carciofo et al., 2016). Due to the low number of items in the scale, alpha values might not be a good indicator for the reliability of the measurement instrument. Previous research therefore often conducted test-retests with good results, providing support for the use of the utilized instrument (Carciofo et al., 2016; Rammstedt & John, 2007). The alternative scale, which could have been utilized is the BFI-44, which consists of 44 items. While this scale typically produces higher reliability values, the time needed to complete mentioned scale is significantly longer. Since the aim of this study was to research practicing professionals, which usually won’t have as much time at their disposal to participate in a low-scale research, the BFI-10 was the preferred choice.
Common Method Variance
Since the data collection method utilized in this study, was a self-report questionnaire, it was necessary to address the potential problem of common method variance. For this purpose, besides the usage of different kind of scales and amount of point Likert scales, Harman’s single factor test was run. The result of mentioned test revealed, that in the present dataset the maximum variance that is explained by a single factor is 27.38%. Consequently, it can be concluded, that this dataset doesn’t suffer from Common Method Bias issue (Podsakoff, Mackenzie, & Lee, 2003)
Variable Correlations
To explore the correlations among the examined variables, Pearson’s Correlation Analysis was run. The mean, standard deviation, correlation and reliability values were compiled in Table 4.2.
Pearson’s correlation analysis revealed, that the independent variable (morningness) was statistically significantly correlated to Conscientiousness (r=.297, p<.01), job satisfaction (r=.298, p<.05) and work engagement (r=.252, p<.01). Consequently, H1 is supported. The dependent variable (work engagement), besides the above mentioned correlation to morningness, was also statistically significantly correlated to Extraversion (r=.252, p<.01), Conscientiousness (r=.316, p<.01), Neuroticism (r=-.288, p<.01), Openness (r=.225, p<.01) and lastly Job Satisfaction (r=.598, p<.01).
Table 4.2.
Means, Standard Deviations, Correlations and Reliability of Variables
Mean S.D. 1 2 3 4 5 6 7 8
1. M 14.7 3.01 (0.622)
2. E 3.2 .89 .072 (0.614)
3. A 3.57 .65 .039 .048 (0.209)a
4. C 3.46 .70 .297** .177* .095 (0.234)b
5. N 2.84 .79 -.100 -.362** -.260** -.245** (0.474)c
6. O 3.31 .79 -.062 .295** .015 .107 -.060 (0.443)d
7. WE 3.2 .79 .252** .252** .091 .316** -.288** .225** (0.933)
8. JS 3.58 .77 .298* .162* .232** .301** -.284 .174* .598** (0.821)
Note. *p<.05; **p<.01
M= Morningness; E= Extraversion; A= Agreeableness; C= Conscientiousness; N= Neuroticism; O= Openness; WE= Work Engagement; JS= Job Satisfaction
Numbers in parenthesis is Cronbach alpha value
Note: The Cronbach alpha issue of a, b, c, d is explained in Validity and Reliability section
Analysis of Mean Differences among Work Engagement and Job Satisfaction
To analyze, whether there exist any significant differences in levels of work engagement among employees belonging to different chronotypes, one-way ANOVA was run.
Results demonstrated, that there was a statistically significant difference in levels of work engagement among employees belonging to different chronotypes (F(2,197) = 5.970, p = .003). A Tukey post hoc test revealed that work engagement was significantly lower among evening-type employees (2.8 ± .75, p = .002) compared to morning-type employees (3.5
± .92). There was no statistically significant difference in levels of work engagement between neutral-type employees and other chronotype employees. Consequently, H2 is supported.
Next, a one-way ANOVA was run to determine, whether or not there exist any significant differences in levels of job satisfaction among employees belonging to different chronotypes. The result was an observed statistically significant difference between morning-type and evening morning-type employees (F(2,197)=4.462, p=0.013). A Tukey post hoc test revealed that job satisfaction among morning-type employees was statistically significantly higher (3.8
± 0.7, p= .009) compared to evening-type employees (3.2 ± 0.9). There was no statistically significant difference between neutral-type employees and other chronotype employees.
Followingly, H3 is supported.
Table 4.3.
One-way ANOVA of Work Engagement and Job Satisfaction
Sum of Squares df Mean Square F p
Work Engagement
Between Groups 7.209 2 3.605 5.970 .003
Within Groups 118.947 197 .604
Total 126.156 199
Job Satisfaction
Between Groups 5.169 2 2.584 4.462 .013
Within Groups 114.109 197 .579
Total 119.278 199
Mediating Effect of Job Satisfaction
For mediation to be supported, four conditions must be met: 1) the independent variable must be related to the dependent variable, 2) the independent variable must be related to the mediator variable, 3) the mediator must be related to the dependent variable, while in the presence of the independent variable, and 4) the independent variable should no longer be a significant predictor of the dependent variable in the presence of the mediator variable (Baron
& Kenny, 1986). In Step one of the mediation model, the direct effect of morningness on work engagement, ignoring the mediator, was positive and significant (β =.0664, t(198)=3.65, p=<.001), indicating that people high in morningness are more likely to experience high levels of work engagement. Step two showed that the direct effect of morningness on the mediator, job satisfaction, was also positive and significant (β =.0765, t(198)=4.39, p=<.001), indicating that people high in morningness, are more likely to experience high levels of job satisfaction.
Step three of the mediation process showed that the mediator (job satisfaction), controlling for morningness, was significant, (β =.5903, t(198)=9.64, p=<.001). Step four of the analyses revealed that, controlling for the mediator (job satisfaction), the degree of morningness was not a significant predictor of work engagement, (β =.0212, t(198)=1.35, p=.1785). The indirect effect of morningness on work engagement was tested using non-parametric bootstrapping. In the present case the indirect effect (IE=.0452) is statistically significant: 95% CI=(.0212, .0731).
Consequently, it is confirmed that job satisfaction fully mediates the relationship between morningness and work engagement. As a result, H4 is supported. Further, the analyses revealed, that the proportion of the total effect of morningness on work engagement that operates indirectly is 68%. Followingly, 32% of the effect operates directly. Therefore, morningness accounts for 32% of the outcome of work engagement, while 68% is managed through
Job Satisfaction Morningness 0.0765 .0174 4.39 < .001 [.0422, .1108]
Step 3:
Work
Engagement Morningness 0.0212 .0157 1.35 .1785 [-.0098, .0522]
Job Satisfaction 0.5903 .0612 9.6377 < .001 [.4695, .7111]
Moderating Effect of Personality Dimensions
To assess the potential moderating effect of the big five personality dimensions on the relationship between morningness and work engagement, hierarchical multiple regression analysis was conducted. As a result, five regression tables were obtained, which each consisted of three models. The first model was utilized to control for the effect of study’s control variables, gender, age and job satisfaction. The second model goes further by adding the two independent variables, Morningness and one of the five personality dimensions, which is examined for a moderating effect. Lastly, in Model three the interaction of the independent variable and moderator was added to understand if and how the examined personality dimension moderates the relationship between morningness and work engagement. The regression analysis models are presented in Table 4.5, 4.6, 4.7, 4.8, 4.9.
As visible in Table 4, Extraversion (β=.173, p<.01) is positively correlated with work engagement, however looking at the interaction term of morningness and Extraversion, there is no support for a moderation effect of Extraversion. Similarly, Conscientiousness (β=.143, p<.05) and Openness (β=.147, p<.05) were also both found to be positively correlated to work engagement, but after considering the interaction term with morningness, were not found to have any moderating effect on the relationship between morningness and work engagement.
The same holds true for Neuroticism (β=-.121, p<.05), which in contrast, however, was negatively correlated with work engagement. There was no statistically significant correlation found between Agreeableness and work engagement. Consequently, H5 to H9 were rejected.
Table 4.5.
Results of Hierarchical Regression Analysis – Extraversion as the Moderator (N=200)
Variables Model 1 Model 2 Model 3
Controls
Gender -.100 -.113 -.113
Age .016 .012 .012
Job Satisfaction .591*** .549*** .555
Independent Variables
Morningness .045 .037
Extraversion .173** .175**
Independent Variables
M x E -.083
R2 .369 .400 .407
Adjusted R2 .359 .385 .389
D R2 .369 .031 .007
F 38.180 25.897 22.072
D F 38.180 5.057 2.223
Note. *p<.05; **p<.01; ***p<.001; Two-tailed tests of significance
Table 4.6.
Results of Hierarchical Regression Analysis – Agreeableness as the Moderator (N=200)
Variables Model 1 Model 2 Model 3
Controls
Gender -.100 -.088 -.083
Age .016 .012 .014
Job Satisfaction .591*** .587*** .579***
Independent Variables
Morningness .055 .054
Agreeableness -.047 -.053
Independent Variables
M x A -.036
R2 .369 .374 .375
Adjusted R2 .359 .357 .355
D R2 .369 .005 .001
F 38.180 23.139 19.282
D F 38.180 .733 .372
Note. *p<.05; **p<.01; ***p<.001; Two-tailed tests of significance
Table 4.7.
Results of Hierarchical Regression Analysis – Conscientiousness as the Moderator (N=200)
Variables Model 1 Model 2 Model 3
Controls
Gender -.100 -.100 -.090
Age .016 -.005 -.001
Job Satisfaction .591*** .544*** .547***
Independent Variables
Morningness .024 .014
Conscientiousness .143* .137*
Independent Variables
M x C -.048
R2 .369 .389 .391
Adjusted R2 .359 .373 .372
D R2 .369 .020 .002
F 38.180 24.712 20.665
D F 38.180 3.215 .651
Note. *p<.05; **p<.01; ***p<.001; Two-tailed tests of significance
Table 4.8.
Results of Hierarchical Regression Analysis – Neuroticism as the Moderator (N=200)
Variables Model 1 Model 2 Model 3
Controls
Gender -.100 -.083 -.073
Age .016 -.016 -.018
Job Satisfaction .591*** .544*** .542***
Independent Variables
Morningness .061 .059
Neuroticism -.121* -.127*
Independent Variables
M x N .078
R2 .369 .384 .390
Adjusted R2 .359 .368 .371
D R2 .369 .016 .006
F 38.180 24.222 20.591
D F 38.180 2.442 1.883
Note. *p<.05; **p<.01; ***p<.001; Two-tailed tests of significance
Table 4.9.
Results of Hierarchical Regression Analysis – Openness as the Moderator (N=200)
Variables Model 1 Model 2 Model 3
Controls
Gender -.100 -.108 -.106
Age .016 .001 -.006
Job Satisfaction .591*** .546*** .556***
Independent Variables
Morningness .072 .071
Openness .147* .157**
Independent Variables
M x O .064
R2 .369 .392 .396
Adjusted R2 .359 .376 .377
D R2 .369 .023 .004
F 38.180 25.011 21.067
D F 38.180 3.687 1.212
Note. *p<.05; **p<.01; ***p<.001; Two-tailed tests of significance
Discussion Relationships
As expected, the found correlation between morningness and work engagement turned out to be positive. While both concepts have been previously linked to the same personality traits (Bakker, Tims, et al., 2012; Randler, 2009), the revealed correlation now confirms a direct connection between the two constructs in a scientific fashion. After reviewing relevant literature, only one other study was found, linking morningness to one sub-dimension of work engagement (Waleriańczyk, Pruszczak, & Stolarski, 2019). The finding of the run ANOVA test further solidifies this correlation, by demonstrating, that morning-type employees experience significantly higher levels of work engagement than their evening-type employee counterparts.
Consequently, it can be assumed that employees are heavily influenced by their circadian rhythm in whether or not they feel vigorous, dedicated and absorbed in performing work tasks.
The same holds true for the found strong significant correlation between morningness and job satisfaction. As the connection between morningness and job satisfaction has been found before (Moreno et al., 2012), the finding of this study further provides support for the correlation. Further, the findings of the run ANOVA demonstrate that evening-type employees experience significantly lower levels of job satisfaction. The implications of the above findings have to be carefully considered, as they could propose, that a large proportion of working individuals feel unsatisfied with their job, not necessarily because of its contents or tasks, but simply because they aren’t allowed to perform during their peak cognitive performance time frame. Lastly, the finding that job satisfaction acts as a mediator in the examined relationship further contributes towards previous findings, indicating that job satisfaction can be a predictor for work engagement(Abraham, 2012; Johnson, 2000; Macey et al., 2011; Singh, 2017) and can act as a mediator in research examining work engagement (Rayton & Yalabik, 2014).
Looking at the large proportion of the effect of morningness on work engagement that is managed by job satisfaction, the close relationship between the two concepts once more becomes apparent.
The correlations found between the big five personality dimensions and the independent as well as dependent variable of the study were mostly in line with previous research findings as outlined in Chapter II. Solely, the finding surrounding Agreeableness differed, which in the present study didn’t show any correlation between said personality dimension and neither morningness nor work engagement. As anticipated, and in line with previous research (Bakker et al., 2012; Hogben et al., 2009), Conscientiousness was positively correlated with both
morningness and work engagement. Therefore, it was surprising to find that Conscientiousness does not act as a moderator in the relationship between morningness and work engagement. In general, it was unexpected to find that neither of the big five personality dimension constitute a moderator in the present research. However, this finding also implies that no matter the individual personality of an employee, they are all more or less equally affected by their circadian rhythm when it comes to work engagement. Consequently, the organizational implications of the far-reaching impacts of the morningness construct, should be even more carefully considered.
Additional Finding
Another significant correlation to highlight is the one found between morningness and age. To gain further insight into the correlation, another one-way ANOVA was run, to explore whether there exist any significant differences in levels of morningness among different age groups. As a result, it has been found, that there was a statistically significant difference between groups as determined by one-way ANOVA (F(2,197) = 6.831, p = .001). A Tukey post hoc test revealed that Morningness was significantly lower among respondents up to 30 years old (13.8 ± 2.9, p = .005) and respondents up to 40 years old (14.3 ± 3.0, p = .004) compared to respondents aged above 40 years old (15.9 ± 2.7). There was no statistically significant difference between the age groups 20-30 years and 31-40 years (p=.638).
While this observed correlation isn’t novel, it still further contributes to previous findings surrounding the concept of morningness in context with physiological conditions (Adan et al., 2012; Kim et al., 2010; Merikanto et al., 2012).