The following chapter will illustrate the various research methods used in the study.
Firstly, the reader will be given an overview of the final research framework and the hypotheses resulting of the previous chapters. After explaining the reasons behind choosing the quantitative research design of the study, nearer information about the specific sample will be outlined. In a next step, the research instrument, which is the questionnaire, will be illustrated. In order to do so, the various dimensions measured in it, will be named and also the ways in which to measure them validly and reliably. Followingly, the way in which the data will be collected and analyzed will be discussed. Lastly, the research procedure will be outlined which summarizes the process the study at hand followed.
Research Framework
The research framework of this study is based on the research purpose and research questions outlined in Chapter I. Consequently, the research framework at hand consists of one independent variable (Employee Chronotype) and one dependent variables (Work Engagement). The mediator (job satisfaction) will be examined, in regards to whether it significantly effects the relationship between independent and dependent variable. The moderator (Big Five Personality Dimensions), consisting of five dimensions, will be examined in regard to whether or not Conscientiousness and Agreeableness strengthens the relationship between dependent and independent variable. Additionally, it will be looked at if there exists any significant relationship between the remaining personality dimensions and the relationship between morningness and work engagement.
Figure 3.1. Research Framework
Research Hypotheses
Based on the presented literature, the research questions and the research purpose, the following hypotheses were formulated:
H1: Morningness is positively related to work engagement.
H2: Employees with a Morningness-Chronotype experience higher levels of work engagement than Evening-chronotype employees.
H3: Employees with a Morningness-Chronotype experience higher levels of job satisfaction than Evening-chronotype employees.
H4: Job satisfaction mediates the relationship between morningness and work engagement.
H5: Agreeableness strengthens the positive relationship between morningness and work engagement.
H6: Conscientiousness strengthens the positive relationship between morningness and work engagement.
H7: Neuroticism weakens the positive relationship between morningness and work engagement.
H8: Extraversion strengthens the positive relationship between morningness and work engagement.
H9: Openness to new experience strengthens the positive relationship between morningness and work engagement.
Research Design
The study at hand aims to investigate whether or not there exists a statistical significance in the relationship between the above outlined variables. The data necessary to conduct such an analysis, will be collected in forms of numbers, using a survey questionnaire, from an adult working population. Since the research at hand requires the collection of answers from a large number of respondents and follows a linear research path. Studies characterized by these elements are recommended to follow a quantitative research design (Neuman, 2014).
Sample
To examine the relationship between employee chronotypes and work engagement, a sample was chosen which is subject to fixed work times and which starts their workday comparably early. The sample, which fulfills such criteria is found among white-collar office
workers in Taiwan. Workdays in Taiwanese service industry typically start around 8 a.m., regardless of the department in the company. Therefore, this sample is supposed to provide clear differences in the measurement results between the two different chronotypes.
Another reason for choosing a sample from this industry, is that organizational processes, such as employee evaluations and promotion decisions tend to follow a rather traditional approach, which is suspected to favor morning persons. Consequently, the insights gained from this research could additionally provide valuable insights into the fairness and effectiveness of current employee evaluation practices across this industry.
The sampling approach utilized in this research constitutes a mix of judgement sampling and convenience sampling (Lewis-Beck, Bryman, & Futing Liao, 2004). The majority of respondents was attained by having professional contacts, from companies fitting the target criteria, share the questionnaire, throughout the company via exclusive messaging channels. By doing so, it can be assumed that the respondents in fact meet the sampling criteria. In a second step, some respondents were recruited through an online forum. Due to the sample being predominantly recruited through direct professional contacts, the final sample consisted of a demographic closely matching the actual workforce composition. As such, this should be considered a strength of the sample of the present study and will allow for actionable insights gained from the statistical analyses.
The specific sampling criteria for respondents were as follows:
• Respondent is working full-time and has been with the current company for at least 3 months
• Respondent is working in an office environment and is performing predominantly intellectual labor
• Respondent is subject to a fixed starting work time in the morning
Instrument
The research instrument used in this study was a self-reported survey questionnaire. The participants filled out the questionnaire, by choosing an answer option that most closely matched their feeling or attitude towards a given question. The utilized instrument consists of 48 items, which are subdivided in four sections, which are:
I. Chronotype
II. Work Engagement
III. Big Five Personality Dimensions IV. Job Satisfaction
V. Demographic data
The various measures used in section I to IV of this questionnaire, were taken from the work of established researchers, which have been peer-reviewed and proven to possess good reliability and validity.
Chronotype
To assess an employee’s chronotype and to see whether if he or she classifies as a morning or evening person, the reduced version of the Morningness-Eveningess Questionnaire (rMEQ) shall be used. The rMEQ is based on the widely used and accepted Morningness-Eveningness Questionnaire (MEQ), developed by Horne and Östberg (Horne & Ostberg, 1976).
The MEQ, consisting of 19 items questioning individuals sleeping timeframes and preferred time for activities, is commonly used to assess individuals’ chronotypes. It has been found to be valid across various populations and countries (Lee, Kim, Lee, Jang, Kim, & Duffy, 2014;
Roveda, Vitale, Montaruli, Galasso, Carandente, & Caumo, 2017; Taillard, Philip, Chastang,
& Bioulac, 2004). Due to concerns about length and general structure a reduced 5-item version has been developed (rMEQ). The shorter version developed by Adan and Almirall (1991), only uses five items of the original MEQ and has been widely adopted due to its simplicity and convenience. Further it has been evaluated across various countries (Adan & Almirall, 1991;
Randler, 2013; Raoof, Asaad, & Al-Hadithi, 2014). Evaluations of the rMEQ have found Cronbach’s alpha values between 0.68 and 0.78 (Caci et al., 2009; Danielsson, Sakarya, &
Jansson-Fröjmark, 2019; Jankowski, 2013) The original Chinese translation used in this study was created by Li et al. (2011). The Cronbach α coefficient for mentioned translation was assessed to be 0.769.
Work Engagement
The measurement tool used to assess work engagement is a short questionnaire called Utrecht Work Engagement Scale (UWES). The UWES is a self-report questionnaire that assesses the three core dimensions of work engagement an individual experiences: “vigor, dedication, and absorption” (Schaufeli et al., 2002, p. 74; Schaufeli & Bakker, 2010). Studies researching work engagement across different countries ran confirmatory factor analysis and were able to confirm the hypothesized three-factor structure to the data (Rothmann, 2003;
Schaufeli et al., 2002; Yi-wen & Yi-qun, 2005). However, some other studies failed to reproduce these results (e.g. Sonnentag, 2003). Since the three dimensions, constituting work engagement, have been repeatedly found to have moderate to high correlations, Schaufeli and Bakker (2003) proposed that in some cases, it might be beneficial to just utilize the total score for work engagement. The scale, which will be used to assess work engagement in this study, also developed by Schaufeli et al. (Schaufeli, Bakker, & Salanova, 2006), incorporates the three engagement dimensions, which were found to be moderately strong related.
The UWES questionnaire, developed by Schaufeli and Bakker (2003) consists of 17 items but has been subject to revisions, which resulted in a UWES-9 version consisting of only 9 items. With confirmed factorial validity and good internal consistency (α = .89) as well as test-retest reliability (Schaufeli, Bakker, & Salanova, 2006), the UWES-9 represents a valuable tool to measure levels of work engagement. Due to its comparably short and simple nature it should also be easily accepted by respondents.
The Chinese translation used in this study, stems from the official UWES test manual (Schaufeli & Bakker, 2003, p. 59).
Big Five Personality Dimensions
To assess an employee’s score across the various personality dimensions, the 10-item version of the Big Five Inventory (BFI) shall be used. The original BFI (John, 1990) is a widely adopted as well as validated questionnaire to assess the Big Five factors of personality. With a Cronbach alpha value of (α = .83) and good convergent validity with corresponding scales (Denissen, Geenen, Aken, Gosling, & Potter, 2008; Goldberg, 1992; McCrae & John, 1992), it should be deemed a solid tool to assess the various personality dimensions. To be able to collect data on individual’s personality dimensions even quicker and in a more convenient way for the respondent, a shorter version of the BFI has been developed(Gosling, Rentfrow, & Swann,
2003). This ten-item version of the BFI has been found to be valid and reliable across different populations and cultures (Kim et al., 2010; Rammstedt, Kemper, Klein, Beierlein, & Kovaleva, 2013, 2017). However, it needs to be pointed out that low alpha values are quite common when working with the Chinese version of this measure and should therefore be expected (Carciofo et al., 2016;Li et al., 2015; Yu et al., 2017). The low alpha values are most likely due to the low number of items per personality dimension. In order to address this potential reliability issue, previous studies performed test-retests of the BFI-10 and its extended BFI-44 version (Carciofo et al., 2016; Rammstedt & John, 2007), as well as confirmatory factor analysis to confirm the construct validity of the measure and received satisfactory results (X. Li et al., 2015).
Job Satisfaction
To assess the levels of job satisfaction among employees, the Michigan Organizational Assessment Questionnaire Job Satisfaction Subscale (MOAQ-JSS) (Cammann et al., 1979) was utilized. The MOAQ-JSS is a five-point Likert-type scale comprising only three items. This is in sharp contrast to alternative job satisfaction measures, which are typically much longer.
However, for the scope of the research at hand, this measure was deemed a good fit, as quick completion times of the questionnaire are assumed to translate into higher completion rates of the survey questionnaire. For the present research, the completion time factor was specifically considered as the sample are all practicing professionals, which take extra time out of their workday to provide their responses.
The MOAQ-JSS was examined and found to be a reliable and construct-valid measure of job satisfaction (Bowling & Hammond, 2008). The Chinese translation used for this questionnaire, was developed by Jiang et al. (2012).
Data Collection
The data from the sample was collected through an online questionnaire, to facilitate the accessibility and timing for the respondents. These two issues were of particular concern, as the sample consists entirely of actively practicing professionals. Since respondents belonging to this category, are usually short on time for non-direct work-related activities, it was attempted to create the answering process in a convenient way. The questionnaire was made available through the google forms platform and was available to be answered for a window of around three months. The usage of an online questionnaire further made sure that the gained data was transferred without fault to the utilized statistics software. The majority of respondents were
recruited through personal contacts in companies fitting the sampling criteria. Followingly, the access link to the questionnaire, was shared with professional contacts via LINE app and email, which then was forwarded throughout the companies via exclusive messaging channels. As a result, the researcher should be confident, that the questionnaire was predominantly accessed by individuals meeting the sample criteria. To further ensure the sample fit, the questionnaire contained several screening questions. Since gathering enough data from working professionals is always challenging, another data collection approach was utilized. Some respondents were recruited through an online forum. To prevent against the potentially lower sample criteria fit emerging from such an approach, the questionnaire contained a series of a screening questions, making sure the respondents met the agreed upon sample criteria.
In total, the questionnaire received 283 responses, of which 200 were evaluated to be valid.
Data Analysis
The collected data shall be analyzed with the help of the statistics software SPSS 23.0.
Exploratory Data Analysis (EDA)
Firstly, to gain an overview of the collected data, the descriptive statistics will be computed. By doing so, insights will be gained among the gender, age and chronotype distribution of the sample. This will allow to establish an understanding of how the workforce in the targeted industry is composited.
Additionally, Means, Standard Deviations, Correlations and Cronbach alpha values of the employed scales will be compiled to further visualize the data and gain insights over the reliability of the data set.
Correlation Test
In order to examine the relationships between the various variables inspected in the research at hand, the Pearson product-moment correlation test will be utilized. While the main focus lies, in determining the relationships between independent and dependent variables, also possible correlations with and between the remaining variables will be explored. This is mainly due to the explorative nature of the study.
The results of the Pearson product-moment correlation will also reveal whether or not Hypothesis H1 is supported.
ANOVA
Next, a one-way analysis of variance will be used to determine whether or not H2 & H3 are supported or not. Since both of these hypotheses are concerned with statistically significant differences between the means of different groups, a one-way ANOVA will be the most suitable analysis. If any statistically significant differences between the groups’ levels of work engagement and job satisfaction are found, a Tukey post hoc test will be used to determine which specific groups differ from each other. To have three distinct groups to compare, the respondents are coded into three different chronotype groups according to their score on the rMEQ.
Mediating Analysis
To check, whether or not job satisfaction acts as a mediator in the relationship between morningness and work engagement, a mediation analysis with the help of PROCESS in SPSS will be run. Version 3.5. of PROCESS will be used for this purpose. The result of the
mediation analysis will allow us to check whether or not H4 is supported.
Hierarchical Regression Analysis
To test if and how the five personality dimensions, Conscientiousness, Agreeableness, Neuroticism, Extraversion and Openness function as a moderator in the relationship between chronotypes and work engagement, a hierarchical multiple regression analysis will be utilized.
Therefore, this analysis will serve as a way to test if H5, H6, H7, H8 & H9 are supported or not.
The Issue of Common-Method Variance (CMV)
Common-Method Variance (CMV) is “variance that is attributable to the measurement method rather than to the constructs the measures represent” (Podsakoff, Mackenzie, & Lee, 2003, p. 879) This is a common issue that can occur when utilizing self-report questionnaires for data collection, especially when measuring more than one construct in the same instrument.
A possible negative consequence is, that correlations between variables might be created, simply by the respondents tendency to answer in a consistent manner throughout the questionnaire (Chang et al., 2010).
To avoid this issue in the present study, several steps have been taken as suggested in previous research(Chang et al., 2010; Podsakoff, 2003). Firstly, the instrument is composed of several independent and widely proven to be reliable and valid research instruments. Secondly,
the potential of CMV is minimized by using different scales for each measured dimension in the instrument. Consequently, respondents aren’t tempted to answer in the same fashion throughout the questionnaire. Lastly, after the data collection, a post-hoc Harman one-factor analysis will be used to make sure whether or not the occurring variance can be attributed to a single factor.
Research Procedure
Initially, on the basis of the researcher’s interest, the research topic was identified through extensive literature review. After having identified the specific research topic, the theoretical framework for the study was created. As a next step, the different scales to assess the measures have been compiled, combined and adjusted to create the measurement instrument used in this study. As a next step, sample participants received the questionnaire with additional information on the topic of the study and the time it will take them to complete it. The questionnaire included all the scales to measure the various researched dimensions. Participants were given a timeframe of around three months to complete the questionnaire. After the data was collected and assessed, it was analyzed with the help of SPSS. After analyzing the collected data, the results and potential implications of the findings were be discussed. Lastly, the researcher arrived at a conclusion and provided possible suggestions for future research. The research process is illustrated below:
Figure 3.2. Research Procedure
Literature Review Research Topic Identification
Theoretical Framework Participants Identification
Instrument Development Data Collection
Data Analysis
Research Results & Discussion Conclusions & Recommendation