INTRODUCTION TO THE STUDY
Research background
Employees are extremely crucial to the organization since their value to the organization is essentially intangible and not easily replicated (Meaghan & Nick
Employee turnover intentions are a critical area of research in organizational studies, significantly impacting businesses (Mahmud, 2011) Numerous studies have explored this topic, highlighting its importance in understanding workforce dynamics and retention strategies (DeMicco & Giridharan, 1987; Dyke & Strick, 1990; Cantrell & Saranakhsh, 1991; Denvir & McMahon, 1992).
Muhammad et al (2011) highlighted a positive correlation between human resource management (HRM) practices and employee retention Despite this, limited research has explored the impact of HRM practices on employees' intentions to leave and their reactions to these practices.
In recent decades, Organizational Citizenship Behavior (OCB) has garnered significant attention from researchers As job complexity increases and employees face more unstructured tasks, the impact of their discretionary efforts becomes more pronounced, highlighting the disparity between high and low performers (Hunter et al., 1990) Despite the critical role of OCB, it is surprising that many studies have overlooked the influence of employees' behavioral responses on the relationship between Human Resource Management (HRM) practices and employees' intentions to leave their jobs (Guest et al., 2003; Morrison, 1996).
Organizations require employees who can anticipate and adapt to rapid business changes to maintain or improve effectiveness This adaptability is closely linked to employees' organizational citizenship behavior (OCB), which plays a significant role in understanding the relationship between human resource management (HRM) and performance.
A study by Lam et al (2009) explored the mediating role of Organizational Citizenship Behaviour (OCB) in the connection between perceived human resource management practices and employees' intention to leave The findings suggest that the impact of HRM practices, such as retention-oriented compensation and formalized training, on turnover intentions is complex and influenced by behavioral factors This insight could assist managers in developing more effective employee retention strategies However, the study's context in a Japan-China joint venture may not be applicable to Vietnam, and its limited sample size from a single company raises questions about the generalizability of the results regarding OCB's mediating role.
Hence, this study aims to investigate the mediating role of OCB on the relationship between the perceived human resource management practice and employees’ intention to leave in Vietnam context.
Research objective
This study investigates how organizational citizenship behavior mediates the relationship between perceived human resource management practices and employees' intention to leave The primary objective is to address the question of how these factors interact and influence employee retention.
Whether Organizational Citizenship Behavior mediates the relationship of perceived human resource management and employee intention to leave in organization in Vietnam?
Practical significances of the study
The practical relevance of this study is that it gives insight in how perceived HRM practice affects employees’ intention to leave that are becoming increasingly important for organizations.
The study's findings highlight the significance of implementing Human Resource Management (HRM) strategies to reduce employee turnover intentions Managers can leverage these insights to effectively utilize HRM as a strategic tool for managing and minimizing the likelihood of employees leaving the organization.
Research scope and approach
This study is conducted in Ho Chi Minh City, which is the biggest city in Vietnam.
Ho Chi Minh City is selected for this study due to its diverse workforce, which encompasses individuals from various organizations, offering a rich array of perceived human resource practices This diversity not only enhances the study's relevance but also makes it feasible to conduct online surveys within the constraints of limited time and resources.
This study focuses on employees who have been with the organization for a minimum of six months This duration allows respondents to fully understand the perceived human resource practices and provides sufficient time for their organizational behavior to influence the relationship between human resource management practices and their intention to leave the organization.
The research consists of two phases: an initial pilot study to assess questionnaire comprehension and evaluate measurement scales, followed by a main phase aimed at testing the hypothesis and validating the research model.
Research structure
Chapter 1: Introduction This chapter provides the research background,research objectives, research scope and approach.
Chapter 2: Literature review This chapter provides theoretical and empirical background behind the hypothesized research model.
Chapter 3: Research methodology This chapter details the methodologies used to conduct this research.
Chapter 4: Data analysis This chapter is conducted based on the collected data to test the hypothesis and answer research questions.
Chapter 5: Discussion In this chapter, the author discusses the results; provide practical implication, and possible direction for future research.
Intended HR Actual HR Perceived HR Employees’ Organization
LITERATURE REVIEW
Human Resource Management
Human resource management plays a vital role in organizations, as employees are considered essential assets (Redman and Wilkinson, 2001) Consequently, understanding and addressing employees' intentions to leave has become a focal point for effective human resource management practices.
Nishii and Wright (2004) also investigated the relation between HR practices and firm performance They distinguish five levels; part of the model is described below:
Figure 2-1: Partial model by Nishii and Wright, 2004
Nishii and Wright (2008) describe employees' perceptions of Human Resource Management (HRM) as perceived HR practices, emphasizing that these perceptions are shaped by individual employee experiences The effectiveness of HR practices is influenced by how each employee interprets these practices.
HRM practices typically include providing extensive training and development opportunities, designing fair job and performance appraisal processes,developing incentive pay systems and so forth (Koch and McGrath 1996).
2.1.2 Human Resource Management effects on Intention To Leave
Numerous studies have highlighted the significant impact of Human Resource Management (HRM) practices, such as selection, training, compensation, and performance management, on employees' intention to leave For instance, Park, Ofori-Dankwa, and Bishop (1994) found that competitive pay can deter employees from leaving an organization This finding is supported by Shaw, Delery, Jenkins, and Gupta (1998), who analyzed data from 277 trucking companies and concluded that implementing high pay, attractive benefits, and electronic monitoring led to reduced employee turnover Additionally, research by Rahim, Magner, Antonioni, and Rahman (2001) indicated that employees' perceptions of distributive justice related to merit pay also play a crucial role in decreasing their intention to leave.
Despite the recognized connection between Human Resource Management (HRM) and organizational psychology, particularly through behavioral theories of commitment and motivation, researchers have largely overlooked the mechanisms behind these effects Studies by Guest et al (2003) and others emphasize the importance of understanding how employees' attitudes and behaviors influence the relationship between HRM practices and their intention to leave However, there remains a significant gap in research focusing on the specific behavioral mediating effects within this dynamic.
Organizational Citizenship Behaviour
Organizational Citizenship Behaviors (OCBs), defined by Organ in 1988 as discretionary actions that are not formally rewarded but enhance organizational effectiveness, have become a vital area of study in organizational research OCBs are understood to involve positive contributions to both colleagues and the organization, reflecting an active commitment to the workplace Contemporary theorists widely agree that these behaviors have a cumulative positive impact on overall organizational functioning, as highlighted by Wagner and Rush (2000).
In attempting to further define organizational citizenship behavior, Organ
(1988) highlights five specific categories of discretionary behavior.
• Altruism (e.g., helping new colleagues and freely giving time to others) is typically directed toward other individuals but contributes to group efficiency by enhancing individuals’ performance.
• Conscientiousness (e.g., efficient use of time and going beyond minimum expectations) enhances the efficiency of both an individual and the group.
• Sportsmanship (e.g., avoids complaining and whining) improves the amount of time spent on constructive endeavors in the organization.
• Courtesy (e.g., advance notices, reminders, and communicating appropriate information) helps prevent problems and facilitates constructive use of time.
• Civic Virtue (e.g., serving on committees and voluntarily attending functions) promotes the interests of the organization.
Supervisors cannot compel subordinates to engage in Organizational Citizenship Behavior (OCB), nor can employees expect formal rewards for such discretionary actions However, as noted by Organ (1997), supervisors often recognize and reward OCB through various means, including preferential treatment, performance ratings, and promotions Additionally, Organ's foundational work (1988) emphasizes that these behaviors are frequently driven by intrinsic motivations, stemming from an individual's internal desire for achievement, competence, and social belonging.
Farh et al (1997) identified two distinct types of citizenship behaviors within Organizational Citizenship Behavior (OCB) measures: active positive contributions, such as helping others, and passive behaviors characterized by the avoidance of harmful actions, like not infringing on others' rights The latter, often referred to as omission, reflects a moral obligation to refrain from negative behaviors that could impact individuals or the organization.
“Do no harm,” or more specifically “Do no harm through action” (Baron, 1998)
Organ (1988) emphasized the critical role of organizational citizenship behavior (OCB) in ensuring an organization's survival, asserting that OCB enhances both employee efficiency and overall productivity, thereby contributing to the organization's effective functioning Katz (1964) further warned that an organization relying solely on prescribed behaviors is a fragile social system In their 1966 work, Katz and Kahn highlighted that an organization's success significantly relies on extra-role behaviors, indicating that managers require employees who exceed their contractual obligations (cited in Funda, 2010).
Organizations require proactive employees who can anticipate and monitor swift changes in the business landscape, enabling them to effectively address these shifts and maintain or improve overall effectiveness.
Research model and hypotheses
2.3.1 The effects of HRM on OCB
Empirical research indicates that two key HRM practices significantly influence employees' intention to leave: compensation and training Studies by Schnake and Dumler (1997), Shaw et al (1998), and Ngo et al (1998) highlight this relationship, as cited in Lam et al (2009).
In 1986, Tajfel and Turner emphasized the importance of self and social identity among employees, highlighting that compensation and training can enhance these aspects by ensuring that valuable employees feel fairly rewarded and supported in their career development (Lam et al., 2009) Furthermore, Blau (1964) suggested that when employees receive benefits from their organizations, they may feel a sense of obligation, leading to reciprocation through extra-role behaviors that ultimately benefit the organization.
Numerous empirical studies, including those by Park et al (1994), Rahim et al (2001), and Shaw et al (1998), have demonstrated a negative correlation between attractive compensation and employees' intention to leave their jobs Specifically, Shaw et al (1998) identified high pay as the most significant predictor of employees' likelihood to depart from their positions.
This study investigates the connection between compensation systems and employees' intention to leave, particularly focusing on how Organizational Citizenship Behavior (OCB) mediates this relationship It aims to determine whether retention-oriented compensation and formalized training enhance employee engagement in OCB.
The litterature reveals that little research has been conducted to posit the relationship among HRM practices, OCB, and intention to leave.
Recent studies indicate a positive correlation between enhanced perceptions of justice and organizational citizenship behavior (OCB), while also revealing a negative relationship between pay-for-performance and OCB This paradox suggests that equity-based compensation systems may deter OCB engagement, highlighting the need for further investigation into the interplay between compensation, OCB, and employee retention We propose that retention-oriented compensation could foster greater OCB involvement Additionally, research by Tuiji et al (2007) supports the notion that human resource management practices can enhance OCB Thus, our first hypothesis is formulated accordingly.
Hypothesis 1: The use of retention-oriented compensation is positively related to the engagement of OCB.
Numerous studies indicate that effective training significantly enhances organizational performance (Koch and McGrath, 1996; Ferratt et al., 2005) Formalized training programs are designed to encourage Organizational Citizenship Behavior (OCB) among employees (Organ et al., 2006) Furthermore, such training equips employees with the skills necessary for advancement within the organization (Noe et al., 1997) Consequently, it is recommended that employees proactively engage in discretionary behaviors to enhance their knowledge, skills, and abilities in preparation for future promotions This leads to the following hypothesis:
Hypothesis 2: An emphasis on providing formalized training is positively related to the engagement of OCB.
2.3.2 The effect of Retention Oriented Compensation on Intention to Leave
Research indicates a strong connection between compensation and employee attraction and retention, as highlighted by Steers and Porter (1991) Murphy (2003) found that organizational compensation significantly affects employees' decisions regarding their commitment to the company and their likelihood of turnover Additionally, Roya et al (2011) demonstrated that strategic compensation practices contribute to the fulfillment of the psychological contract, which in turn enhances affective organizational commitment Employees who experience a fulfilled psychological contract and possess high levels of affective commitment are less inclined to leave their organization, suggesting that retention-oriented compensation may be inversely related to employees' intentions to depart.
Hypothesis 3: Retention oriented compensation may negatively related to employees’ intention to leave.
2.3.3 The effect of Formalized Training on Intention to Leave
Training enhances employee performance and bridges the gap between initial education and job-specific skills, as noted by Bartel (1995) and De Grip, Heijke, & Willems (1998) It also helps prevent skills obsolescence due to technological and organizational changes (Bishop, 1997) Furthermore, trained employees exhibit greater versatility in their roles (Groot & Maassen van den Brink, 2000) Consequently, effective job training is crucial for career development and may influence employees' intentions to leave their positions, leading to the hypothesis that formalized training is negatively related to turnover intentions.
Hypothesis 4: Formalized training may negatively related to employees’ intention to leave.
2.3.4 The effect of OCB on intention to leave
Research by Chen, Tjosvold, and Fang (1998) indicates a negative relationship between organizational citizenship behavior (OCB) and employee turnover Supporting this, Dorothea (2012) found that higher employee engagement correlates positively with OCB, suggesting that engaged employees are more likely to demonstrate individual OCB This behavior reflects an employee's sense of belonging and intention to remain with the organization Additionally, employees engaging in OCB may anticipate rewards such as promotions or salary increases, further reducing their likelihood of leaving (Organ et al., 2006) This leads to the development of the third hypothesis.
Hypothesis 5: OCB may negatively relate to employees’ intention to leave.
2.3.5 The mediating role of OCB
Numerous studies have demonstrated that effective human resource management (HRM) practices can significantly reduce employee turnover (Huselid, 1995; Huang, 1997; Rahim et al., 2001; Boselie and Wiele van der, 2002) Additionally, research by Organ et al (2006) suggests that HRM practices foster employee engagement in organizational citizenship behavior (OCB), which, in turn, contributes to lower turnover rates This leads to the hypothesis that OCB may serve as a mediating factor in the relationship between HRM practices and employee retention.
Formalized training H2: + perceived HRM practices (i.e retention oriented compensation and formalized training) and employees’ intention to leave.
Hypothesis 6: OCB mediates the negative relationship between retention- oriented compensation and employees’ intention to leave.
Hypothesis 7: OCB mediates the negative relationship between formalized training and employees’ intention to leave.
Figure 2.2 below shows the hypothesized relationships among perceived HRM practices, OCB and intention to leave
RESEARCH METHODOLOGY
Research design
This study consists of two phases: a pilot study, which includes two sub-phases, and the main study In the first sub-phase of the pilot study, in-depth interviews were conducted with a small group of five target participants to evaluate the content and meaning of the terminology used in the measurement scales This feedback led to modifications in the draft questionnaire for the subsequent pilot study phase The second phase of the pilot study utilized an online survey method with a larger group of 50 target participants to assess the internal consistency of the measurement scales, response rate, and quality of responses.
Following the pilot study, the main research was carried out utilizing a combination of online surveys via Google Docs and face-to-face interviews This approach aimed to evaluate the measurement scales, test the hypotheses, and validate the proposed research model.
Research process
This study has been conducted in a process showed in Figure 3.1.
Questionnaire development
A questionnaire, also known as a measuring instrument, is a structured set of questions designed to gather information from respondents (Malhotra, 1996) It serves three key purposes: first, it converts essential study-related information into a format that respondents can easily answer; second, it fosters a positive dynamic to encourage participation and completion; and third, it aims to minimize response errors (Malhotra, 1996).
The questionnaire in this surveys were translated and back-translated intoVietnamese in line with cross cultural translation procedures (Harkness &
In the study by Schoua-Glusberg (1998), a committee approach was employed for translation, involving multiple translators who independently translated the same questionnaire, following Brislin's (1980) recommendations During a reconciliation meeting, these translators compared their translations, resolved discrepancies, and reached a consensus on a final version that incorporated the strengths of the independent translations The author utilized a committee of two translators for this process.
The survey included a questionnaire divided into three sections that evaluated perceived human resource practices: Compensation, Training, and Organizational Citizenship Behavior Following Babbie's (1998) guidelines, brief instructions were provided at the start of each section Participants rated all items using a Likert-type scale, where 1 indicated "strongly disagree."
Table 3-1: Measurement scales of independent variables
OCB Organizational Citizenship Behaviour (Cronbach’s Alpha: 0.771)
OCB1 Conducts personal business on company time (e.g., trading stocks, shopping, going to barber shops)
OCB2 Views sick leave as a benefit and makes excuses for taking sick leave
OCB3 Willing to help colleagues solve work-related problems
OCB4 Willing to assist new colleagues in adjusting to the work environment
OCB5 Takes credit, avoids blame, and fights fiercely for personal gain
OCB6 Complies with company rules and procedures even when nobody is watching and no evidence can be traced
OCB7 Uses illicit tactics to seek personal influence and gain with harmful effect on interpersonal harmony in the organization.
OCB8 Uses company resources to do personal business (e.g., company phones, copy machines, computers, and cars)
OCB9 Does not mind taking on new or challenging assignments
OCB10 Takes one’s job seriously and rarely makes mistakes
OCB11 Willing to cover work assignments for colleagues when needed
OCB12 Actively attends company meetings
OCB13 Eager to tell outsiders good news about the company and clarify their misunderstandings
OCB14 Often speaks ill of the supervisor or colleagues behind their backs
OCB15 Makes constructive suggestions that can improve the operation of the company
COM1 Individual performance in the job is important in determining wages of employees in your company
COM2 Bonuses determination in your company is based primarily on age and seniority of employees
COM3 Wages and bonuses are fairly determined in your company
COM4 The standard of bonuses determination in your company is satisfactory
COM5 The standard of wages determination in your company is satisfactory
COM6 The wages level in your company is high in comparison to other organizations
TRA1 The contents and variety of training programmes offered in your company are satisfactory
TRA2 The amount and duration of training programmes offered in your company are satisfactory
TRA3 In your company, many ‘non-technical’ training programmes have already been systematized
TRA4 Compared to other organizations, extensive training programmes are provided for employees in their jobs in your company
TRA5 In your company, training programmes are conducted by professionals and experienced managers or trainers
TRA6 In your company, many ‘technical’ training programmes have already been systematized
Table 3-2: Measurement scales of dependent variable
ITL Intention To Leave (Cronbach’s Alpha: 0.913)
ITL1 Is there any likelihood that you would resign from the organization?
How likely is it that you would not continue to be a staff of the organization during your tenure of employment with your present company?
ITL3 Do you have any intentions of resigning from the organization?
The pilot study
After translating the questionnaire, the author conducted a pretest with five MBA students currently employed in an organization, ensuring they would not participate in the main study Feedback focused on refining word choices and phrases to enhance clarity and sensitivity, enabling respondents to provide accurate ratings reflecting their experiences Following these revisions, the author finalized the questionnaire for the main study.
In phase 2 of the pilot study, the author utilized the finalized questionnaire from phase 1 to conduct interviews with participants A total of 60 paper questionnaires were distributed, yielding 56 responses However, 6 questionnaires were discarded due to missing answers, resulting in a refined dataset for analysis.
The evaluation of the questionnaire was conducted by analyzing the collected data to determine whether the study subjects comprehended the questions accurately and were willing to provide the necessary information To assess the reliability and validity of the measurement scales, Cronbach’s alpha was utilized on the collected data Following these preliminary results, the author made essential adjustments to finalize the questionnaire for the main study.
Main study
The targeted population for this study comprised of employees who has been working for an organization at least 6 months.
Sampling techniques offer various methods to minimize data collection by focusing on a specific sub-group instead of the entire population (Saunders and Thornhill, 2000) In this study, a non-probability sampling design utilizing convenience sampling was deemed suitable for data gathering.
Sample size: In this study, the author used the theory proposed by Hair et al.
In accordance with Tabachnick and Fidell's (1989) guidelines, which recommend a minimum sample size of 50 to 100 participants, this study aimed for a sample of 150 due to the inclusion of 30 items in the questionnaires Following the 5-to-1 ratio for subjects per predictor in regression analysis, the author successfully gathered 182 valid responses, exceeding the required sample size.
The study utilized a convenience sampling method, specifically snowball sampling, which involves building a sample through referrals (O’Leary, 2004) While Faugier & Sargeant (1997) acknowledged the personal bias and distortion associated with snowball sampling, they emphasized that the trust established over time can enhance the sincerity of responses (Foote Whyte, 1982) and ultimately improve the validity of the data collected from these hidden populations.
The main study's fieldwork took place from June 6 to June 20, 2013, initially planned as an online survey However, after only 36 respondents participated in the first week, the author shifted to face-to-face interviews, primarily involving postgraduate students from the University of Economics in Ho Chi Minh City, along with some acquaintances Ultimately, the author collected 179 responses, resulting in 182 qualified responses after data cleaning, which were then analyzed.
Methods of data analysis
The survey utilized both online and face-to-face interviews to gather data For online interviews, data cleaning involved verifying participants met predefined criteria, such as being employed full-time and having worked in an organization for a minimum of six months In contrast, face-to-face interviews required the author to manually check for any missing ratings, as this process is not automated like in online interviews Additionally, the author reviewed reverse-scored questions to ensure they were answered logically.
The author employed Cronbach's Alpha to evaluate the internal consistency of the measurement, a crucial step in statistical analysis as highlighted by Schumacker and Lomax (2004) Churchill (1979) notes that this reliability coefficient is one of the most commonly used indicators for assessing the internal level of measurement, necessitating its calculation before proceeding with further analysis The alpha coefficient ranges from 0.0, indicating no internal reliability, to 1.0, representing perfect internal consistency Nunnally (1978) recommends that a Cronbach’s alpha value of 0.7 or higher is essential for achieving acceptable internal consistency.
The author conducted confirmatory factor analysis (CFA) before conducting hypotheses test in order to evaluate the model fit or whether the data fit the hypothesized measurement model.
Confirmatory Factor Analysis (CFA) is a theory-testing model widely used in social research, as noted by Kline (2010) and Stevens (1996) It is particularly suitable when the dimensionality of a variable set is already established through prior research, allowing researchers to assess whether the proposed factor-loading pattern fits a new sample from the same population CFA primarily evaluates a priori hypotheses and is heavily influenced by existing theory, with all aspects of the model specified in advance (Brown, 2006) In contrast, Exploratory Factor Analysis (EFA) aims to identify underlying factors based on data, maximizing explained variance, and should be reserved for truly exploratory areas where no previous analyses exist (Gorsuch, 1983).
In this study, the author established a hypothesized model grounded in a comprehensive literature review and an analysis of relevant theories Consequently, Confirmatory Factor Analysis (CFA) was determined to be more appropriate for this research than Exploratory Factor Analysis (EFA).
To evaluate model fit, several indices are considered, including chi-square (χ²), which serves as an absolute fit index assessing how well the estimated covariances align with the measured variables (Kline, 2011) Additionally, the comparative fit index (CFI; Bentler, 1990), the Jo¨reskog-Sö rbom goodness-of-fit index (GFI), and the root mean square error of approximation (RMSEA; Steiger, 1990) are important metrics These indices reflect how well a research model fits compared to a null or independence model, where correlations among observed variables are assumed to be zero The CFI and GFI are regarded as the most accurate representations of the population value for a single model, while RMSEA is a "badness of fit" index, with a value of zero indicating the best fit; higher values suggest a poorer fit (Kline, 2005) A summary of reported fit indices and their desired ranges is provided in Table 3-3.
Goodness-of-fit indices Desired range of value for a good fit
Chi-square/degree of freedom ratio χ2/df 2 to 5
Goodness of fit indices GFI ≥0.90
Root mean square Error of Approximation RMSEA ≤0.08
Source: Hair et al., 2006; Tabachnick & Fidell, 2007; Kline, 2011; Hu & Bentler, 1999
3.6.4 Analysis of Variance (one-way and two-way ANOVA)
The author employs one-way ANOVA to assess the equality of means across different gender and age groups Subsequently, a two-way ANOVA is conducted to validate the impact of age on intrinsic motivation within the context of gender differences.
Regression analysis is a statistical technique used to estimate relationships among variables, focusing on minimizing collinearity issues (Aiken and West, 1991) Prior to conducting the analysis, the mean values of each predictor were calculated, and centered variables were utilized in the regression model This linear multiple regression model was employed to test the study's hypotheses, ensuring that all criteria for optimal regression estimation were thoroughly tested and confirmed.
Multiple regression analysis employs three primary strategies: standard multiple regression, sequential (hierarchical) regression, and statistical (stepwise) regression (Tabachnick & Fidell, 2007) In standard multiple regression, all independent variables are included in the regression equation simultaneously, with each assessed as if it were the last to enter Conversely, sequential regression allows researchers to specify the order of independent variable entry, evaluating each based on its contribution at the point of entry Lastly, statistical regression determines the order of variable entry based purely on statistical criteria.
The regression strategy enables researchers to effectively manage the progression of the regression process, making sequential regression the preferred choice for theoretical applications and for testing specific hypotheses, as noted by Tabachnick and Fidell.
The present study utilized a sequential (hierarchical) multiple regression strategy to test a series of theoretical hypotheses This approach allows for the explicit examination of the proportion of variance attributable to specific independent variables, after accounting for variance from other variables already included in the model Unlike standard regression, which primarily assesses relationships among variables and evaluates multiple correlation, this method focuses on model-building to effectively test the proposed hypotheses.
3.6.6 Testing mediation with regression analysis
Mediation is a hypothesized causal chain in which one variable affects a second variable that, in turn, affects a third variable The intervening variable, M, is
Independent variables (X) Dependent variables (Y) c the mediator It “mediates” the relationship between a predictor, X, and an outcome Graphically, mediation can be depicted in the following way:
Paths a and b represent direct effects, while the mediational effect, where X influences Y through M, is referred to as the indirect effect This indirect effect signifies the part of the relationship between X and Y that is facilitated by M.
A mediating relationship is significant when it satisfies four conditions (Howell, 2002, citing Baron & Kenny, 1986):
1 There is a significant relationship between the independent variable and the mediator (path a).
2 There is a significant relationship between the mediator and the dependent variable (path b).
3 There is a significant relationship between the independent variable and the dependent variable (path c).
4 When the mediator and the independent variable are used simultaneously to predict the dependent variable, the previous significant part between the independent variable and the dependent variable (path c) is now greatly reduced (partial mediation) or ideally nonsignificant (full mediation).
To test for mediation, as recommended by Cohen, Cohen, West, and Aiken,
(2003), hierarchical multiple regressions were used as below:
First step: Enter all independent variables (Xi) into the prediction of the dependent variable (Y) to determine the total effect (βtxi) of each independent variable.
Second step: Enter all independent variables (Xi) and all mediating variables
In analyzing the impact of independent variables on a dependent variable (Y), the total effect (βmj) of each mediating variable is assessed alongside the coefficients (βxi) of the independent variables The difference between the total effects (βtxi) and the coefficients (βxi) reveals the extent to which these effects are mediated A nonsignificant coefficient (βxi) indicates full mediation, suggesting the independent variable has no direct effect on the dependent variable Conversely, if the coefficient is reduced yet remains significant, this points to partial mediation of the independent variable's effect on the dependent variable (Howell, 2002).
DATA ANALYSIS
Data cleaning
The author gathered 251 responses, and after screening for incomplete answers, 182 fully qualified responses were retained for data analysis.
Profiles of qualified respondents
In a survey of 182 qualified respondents, 39.6% were male and 60.4% were female The age distribution showed that 44.0% were between 22-24 years old, 32.4% were aged 26-30, and 23.6% were over 30 A significant majority, 75.3%, were employees, while 24.7% held managerial positions Regarding tenure, 28.0% had been with their current company for one year or less, 21.4% for 2-3 years, and 17.0% for more than three years Employment sectors included 17.0% in state-owned companies, 32.4% in private companies, 12.6% in joint-stock companies, 23.6% in foreign companies, and 2.7% in joint ventures, with 11.5% not specifying their company type These findings are summarized in Table 4-2.
Table 4-2: Socio-demographic Characteristics of The Qualified Samples
Socio-demographic variables Frequency Percentage (%)
Normality analysis
Skewness and kurtosis, which indicate the asymmetry and peakedness of frequency distributions, significantly influence inferential statistics, particularly affecting the mean, standard deviation, and correlation coefficient estimates (Schumacker & Lomax, 2004) As shown in Table 4-3, the skew and kurtosis measures for all items fall below the critical thresholds of 3 and 5, respectively (Kline, 2005) This finding suggests that there are no significant issues with the normality of the variables, supporting the assumptions required for multiple regression analysis Consequently, the author is able to proceed with the subsequent analytical steps.
Reliability of the measurements
Reliability analysis demonstrated that all four variables achieved a Cronbach Alpha greater than 0.70, indicating acceptable reliability according to Nunnally (1978) Specifically, Organizational Citizenship Behaviour was assessed using 15 items (OCB1 to OCB15) and yielded a Cronbach Alpha of 0.771 The Perceived Retention-Oriented Compensation System, measured by 6 items (COM1 to COM6), showed a Cronbach Alpha of 0.780 Perceived Formalized Training, evaluated with 6 items (TRA1 to TRA6), exhibited a high Cronbach Alpha of 0.939 Lastly, Employees’ Intention to Leave was measured by 3 items (ITL1 to ITL3), resulting in a Cronbach Alpha of 0.913 These results indicate that the reliability of all factors is strong enough for further analysis.
Table 4.3 summarized Cronbach Alpha measures across variables More details on measurement reliability test results can be found in Appendix B.
Table 4-3: Summary of Cronbach Alpha Measures across Variables
Perceived retention-oriented compensation systems COM 0.780
Employees' intention to leave ITL 0.913
Confirmatory factor analysis
The initial confirmatory factor analysis (CFA) model with 30 observed variables required refinement for improved data fit, as indicated by the GFI, CFI, and TLI indices Analysis of standardized regression weights revealed that seven items exhibited factor loadings below 0.500, indicating a lack of convergent validity Consequently, items COM2, OCB1, OCB6, OCB10, OCB12, OCB13, and OCB15 were removed from the model The revised model demonstrated enhanced performance across all indices, including GFI, CFI, TLI, and RMSEA Additionally, based on modification indices, new correlation links were established, resulting in a modified first-order CFA model with 23 items that proved to be robust and suitable for further data analysis.
The circles represent the latent constructs and the rectangles represent the observed measures (variables).
Table 4-4: Assessing Fit Indices – Hypothesized Model
Goodness-of-fit indices Desired
Chi-square/degree of freedom ratio χ2/df ≤2 2.284
Goodness of fit indices GFI ≥0.90 0.725
Root mean square Error of Approximation RMSEA ≤0.08 0.084
Table 4-5: Standardized Regression Weights – Hypothesized Model
The circles represent the latent constructs and the rectangles represent the observed measures (variables) All parameters are standardized and have the significance at p