Reasons for this research
A study by DG&A Consulting Company reveals that only 37% of employees understand the company's goals and their significance, while just 20% recognize how their roles contribute to the organization Additionally, approximately one-third of employees express a desire to contribute, yet 20% show a lack of enthusiasm, and 50% remain disengaged or avoid responsibilities Research on HR satisfaction in local companies highlights key reasons why staff tend to leave after 2-3 years of employment.
Employees often find themselves in a state of confusion and lack of direction after spending time at a company, primarily due to unclear future goals, their roles, and expectations regarding behavior and decision-making Additionally, the organizational structure of local companies can stifle initiative, leading to diminished trust in both the employer and the company's future This erosion of confidence can result in employees considering leaving the organization The ongoing economic crisis further exacerbates these issues, highlighting the need for companies to prioritize the well-being and engagement of their workforce Adopting a new management approach that views employees as essential customers is crucial for fostering loyalty and encouraging them to contribute positively to the company’s success.
Since Vietnam's accession to the WTO, local companies aiming to expand or collaborate with foreign firms must not only hire highly skilled employees but also build their trust to prevent "brain-drain." It is crucial for organizations to identify and assess the factors influencing employee trust in their employers, enabling them to develop effective human resource management policies.
“High-involvement Work Practices, Procedural Justice and Trust in the Employer
The study conducted in Hochiminh City highlights the complex interplay between HR practices, procedural justice, and trust within organizations It reveals that both HR practices and procedural justice have direct and indirect relationships with trust, which are partially mediated by perceptions of organizational trustworthiness, specifically regarding the organization's perceived ability and intentions Additionally, the findings indicate that justice significantly enhances the predictive power of trust in organizational settings.
HR is less developed The implications of these findings for research and practice are discussed.
Objectives of the research
This thesis aims to explore the impact of high-involvement work practices and procedural justice on employee trust in employers in Ho Chi Minh City, focusing on their respective roles and effects.
- Firstly, investigating employee trust, defining factors impacting it and the interactive effects between them
- Secondly, defining the role of organizational trustworthiness and high-involvement work practices in order to examine their impact to trust in the employer
- Secondly, building and testing measurement scales of each factor affecting trust in the employer and that of trust in the employer
- Thirdly, using the CFA to find out suitable factors for the model
- Fourthly, defining the strength of those factors affecting trust in the employer in HCM City.
Subjective and Scope of the research
This research investigates the relationship between trust in organizations and the impact of high-involvement work practices (HIWP) and procedural justice on employer trust By analyzing data from a survey questionnaire, we identify three key areas: trust research, strategic HR management, and procedural justice literature The study aims to determine if HIWP and procedural justice enhance perceptions of organizational trustworthiness and trust in the employer Additionally, we assess the unique contributions of HIWP and procedural justice to organizational trust and explore their potential as functional equivalents.
Scope of the research is limited to middle managers and staffs from several organizations in HCM City This survey was carried out at the beginning of 2012.
Research Methodology
The thesis used quantitative research method to investigate and define the trust in the employer The 7-point-scale Likert method was used to measure the value of variables
For this is a discovery research, sampling method was non-probability one, with convenient form
The auto-reply questionnaire will be utilized to gather essential data for the quantitative analysis It will be distributed via email, including a link to the online version of the questionnaire, which is officially detailed in the Appendix of the thesis.
The thesis employed statistical methods to analyze the collected sample results, utilizing Cronbach’s alpha to select and validate measurement scale components Confirmatory Factor Analysis (CFA) was conducted to assess the fit of the data to the proposed measurement model Additionally, linear regression analysis was performed to determine the impact of various factors on trust in organizations and to quantify their coefficients in the regression equation.
Practical meaning of the research
The research has some following meanings:
- Firstly, the research result will give readers an overview of the role and impact of high- involvement practices and procedural justice on trust in the employer
This research identifies the measurement scales for assessing trust within organizations and the factors influencing employee trust in employers in HCM City Consequently, companies in HCM City can enhance their HR management policies by revising or developing effective management strategies and initiatives aimed at fostering employee trust and engagement, ultimately benefiting the organization.
-Thirdly, this is a discovery research, which is a foundation for further researches about other aspects of employee trust in organizations in Vietnam.
Thesis Structure
The thesis is structured into five chapters Chapter 1 – Introduction outlines the research's rationale, objectives, scope, methodology, and significance Chapter 2 – Literature Review and Research Model presents definitions of trust in employers, its antecedents, and their interrelations, along with the proposed research model Chapter 3 – Research Methodology details the development of the measurement scale, sampling methods, data collection processes, and statistical analysis techniques Chapter 4 – Research Results analyzes the collected data, focusing on the reliability and validity of the measurement scale and the results of statistical inference.
Chapter 5 – Conclusion and Proposals will give out some conclusions for trust in the employer and some limitations as well as proposals for further researches.
LITERATURE REVIEW AND RESEARCH MODEL 2.1 Introduction
Literature Review on Trust in the Employer
In recent years, trust has become a crucial focus in management research, as highlighted by Krammer and Tyler (1996) and Rousseau et al (1998) Behavioral studies over the decades have emphasized that trust is vital for organizational success A certain level of trust is necessary to facilitate coordinated actions both within organizations and across their boundaries, as noted by Kramer and Tyler (1996) and Sako (1992).
Trust facilitates relationships between and within organizations, reducing transaction costs and potentially becoming a source of competitive advantage (Chile and McMackin, 1996;
Trust is recognized as a valuable asset in organizational settings, as highlighted by Barney and Hansen (1994) However, empirical research on the mechanisms for building trust within organizations remains limited, as noted by Mayer and Davis (1999).
Building and maintaining employee trust is essential for organizations, as it significantly influences their effectiveness, efficiency, and overall performance, according to various studies (Whitney 1994).
Trust is crucial for promoting positive work-related behaviors, as highlighted by various studies (Kramer and Tyler 1996; Mayer and Davis 1999; Zand 1972; Konovsky and Pugh 1994; Kramer 1999) Employees who trust their organizations tend to remain longer, exert greater effort, and collaborate more effectively In contrast, a lack of trust can lead to decreased work effectiveness, counterproductive behaviors, and even employee turnover (Dirks and Ferrin 2001; Bies and Trip 1996; Robinson 1996) Consequently, perceptions of organizational trustworthiness can offer firms a significant competitive advantage (Barney and Hansen 1994).
Despite the growing body of literature on interpersonal trust, the concept of trust within organizations has received significantly less focus This article explores the factors that contribute to employees' trust in their organizations Research indicates that both individual predispositions towards trust and perceptions of an organization's trustworthiness can strengthen this trust However, empirical studies investigating these connections remain scarce.
Schoorman, Mayer, and Davis (2007) expanded on their foundational model of interpersonal trust, originally proposed by Mayer, Davis, and Schoorman (1995), to explore the factors influencing employees' trust in their employers However, there is a lack of field studies validating this model We argue that perceptions of organizational trustworthiness differ from interpersonal trust, necessitating an examination of both the organization's established processes and principles, as well as how its representatives embody these concepts (Barber 1983; Giddens 1990).
Organizational policies regarding human resource (HR) practices and their fair implementation play a crucial role in fostering trust within the workplace While some research has begun to examine how HR practices affect employee trust in their employers, the overall impact of policy bundles and their actual execution on organizational trust remains insufficiently investigated.
Researchers in organizational behavior generally conceptualize trust as faith in and loyalty to the leader (Marlowe and Nyhan, 1997; Mayer, Davies and Schoorman, 1995;
Trust is a crucial precursor to risk-taking behavior in organizations (Mayer et al.; Hartog, 2003) In the literature on organizational behavior, trust is defined by three key components: (a) trustworthiness, which involves a rational evaluation of another party's reliability based on direct evidence or reputation, as well as their competence and intentions to uphold agreements; (b) faith in the leader, which pertains to the psychological foundations of trust in leadership; and (c) loyalty to the leader, which encompasses the emotional connections and routines established within specific relationships (Hartog; Marlowe and Nyhan; Nooteboom and Six).
Trust is a crucial element in both personal and professional relationships, extensively studied in psychology and organizational communication In interpersonal contexts, trust plays a significant role among spouses, friends, and family members Similarly, in business management, it is vital for fostering relationships between managers and employees, highlighting its importance across various domains.
Trust is defined as the willingness to be vulnerable to another party, grounded in the belief that they are reliable, open, competent, and compassionate It involves a decision to rely on someone—be it an individual, group, or organization—despite the inherent relational risks, with the expectation of a neutral or positive outcome This reliance fosters a readiness to be vulnerable, as the trusting party faces the potential for negative outcomes if the other party proves untrustworthy The perceived trustworthiness of the trustee is crucial, as it influences the trustor's ability to make the cognitive leap necessary to establish trust.
Trust in an organization is distinct from interpersonal trust, primarily due to its focus and depth; individuals often find it ambiguous what specific aspects of their employer they are placing their trust in.
Research on organizational trust has primarily examined interpersonal dynamics, focusing on trust between employees and employers through direct working relationships (Cook and Wall 1980; Butler 1991) and interactions between staff and management at various organizational levels (Child and Rodrigues 2004) Additionally, Giddens (1990) connects organizational trust to the reliability of abstract principles, while Carnevale (1995) emphasizes the belief that institutions will act fairly, reliably, competently, and without threat Consequently, the foundation of organizational trust is built on these essential elements.
The collective characteristics of an administrative organization and its top management group are essential for ensuring continuity in activities and direction, even as individual actors change (Whitley 1987) Giddens (1990) highlights the importance of individuals in fostering trust within abstract systems, particularly those in roles that facilitate the building and maintenance of trust Consequently, trust in an employer is derived from the assessment and aggregation of various sources of evidence at multiple organizational levels (Rousseau et al 1998; Zaheer, McEvily, and Perrone 1998).
2.2.2 Antecedents to trust in the employer: trust, strategic HR management and organizational justice
In trust research, antecedents to trust can be categorized into two main types: the dispositional trust of the trustor and the perceived trustworthiness of the trustee Kramer (1999) identifies various bases of trust, distinguishing between disposition-based trust and trust derived from perceptions of individualized trustworthiness, such as history-based trust, as well as impersonalized trustworthiness, like category-based trust.
Dispositional trust refers to an individual's inherent tendency to trust others, shaping their expectations of trustworthiness (Rotter 1980) This trait is believed to significantly impact trust in institutions (Johnson and Swap 1982; McKnight, Choudhury, and Kacmar 2002) Furthermore, dispositional trust plays a crucial role in fostering trusting beliefs, particularly in uncertain or ambiguous situations (Gill, Boies, Finegan, and McNally).
Research Model and estimation indices for measuring trust in the employer
The proposed linear regression model examines the relationship between the dependent variable "Trust in the Employer" and independent variables including "High-Performance Work Practices (HIWP)", "Procedural Justice", and "Perceived Organizational Trustworthiness" Notably, "Perceived Organizational Trustworthiness" serves as a mediating variable that influences the connection between HIWP, Procedural Justice, and Employee Trust.
With mentioned-above hypotheses, we can summarize the research model as follows:
(+) Diagram 2.1 The hypothesized research model
The signals (+/-) in each arrow show the direction (directly/indirectly) of each dependent factor to the independent one (employees’ trust)
These hypothesized relationships will be verified and analyzed in the following survey
2.3.2 Estimation Indices for measuring trust in the employer:
From definitions about trust in the employer and its factors, estimation indices are built as in the Table below:
High Involvement Work Practices (HIWP)
Trust in the Employer Procedural Justice
Trust in the Employer - To what extend do you trust your organization?
High Involvement Work Practices - Information sharing and employee participation
- Training and family-friendly work practices
Procedural Justice - Fair formal procedure
Table 2.1 Estimation Indices for measuring Trust in the Employer
This chapter presents the research methodology, divided into two key sections: research design and statistical data analysis techniques The research design section covers the development of the measurement scale, sampling methods, and the data collection process Meanwhile, the statistical data analysis techniques section discusses the reliability testing of the measurement scale using Cronbach’s alpha, along with confirmatory factor analysis and linear regression analysis.
Research design
In the research design, we will mention about the used measurement scale, its reliability and suitability, sampling method, data collection tool and process
The approach to research in this study was quantitative, of which the applied methodology was a cross-sectional survey
In this research, a 7-point Likert scale was selected as the measurement scale for all variables, encompassing both dependent and independent factors, following the definition of the research model and the identification of measured variables.
This research utilized a convenient non-probability sampling method, with a sample size of approximately 200 participants, as detailed in the sampling section of this chapter.
The next step involved selecting a data collection tool, specifically a self-designed questionnaire detailed in the data collection section of this chapter After constructing the questionnaire and determining the required sample size, it was distributed for data collection The gathered data will be analyzed using statistical methods, with inferential statistics employed to present the research findings.
We will consider in details the choice method of measurement scale, sampling, data collection tool, process and dealing
This research examines the impact of high involvement work practices and procedural justice on employee trust in employers in Ho Chi Minh City It focuses on understanding individuals' attitudes in this context, utilizing closed questions for convenience in attitudinal research One of the key objectives is to investigate the factors influencing employee trust and their interactive effects To achieve this, a Likert measurement scale with optional answers is employed, allowing for the assessment of trust levels across various factors This approach enables the identification of employee satisfaction and trust levels, whether high or low, using 5 and 7-point scales Additionally, the interval nature of the Likert scale facilitates quantitative analysis, enabling the exploration of correlations and regression between dependent and independent variables.
However, to ensure the suitability of the measurement scale, Kumar (2005), it is necessary to deal with 02 following matters:
- Who will decide which measurement scale should be used for measuring the needed?
- How to know a certain tool is suitable for measuring the needed?
The answer for the first question is professional researchers in related fields, which, in this thesis, are the one on trust in the employer as follows:
Trust in the employer is a crucial factor, measured by asking respondents, "Overall, to what extent do you trust your organization?" This question was rated on a 7-point scale, where 1 indicates "to a very low degree" and 7 signifies "to a very high degree."
Perceived Organizational Trustworthiness: This is an independent factor Drawing on
Mayer and Davis’ (1999) measure of trustworthiness at the interpersonal level, we developed 10 trustworthiness items at the organizational level a Ability scale: includes 3 following items:
1 This organization is capable of meeting its responsibilities
2 This organization is known to be successful at what it tries to do
3 This organization does things competently b Benevolence/Integrity scale: includes 7 following items:
1 This organization is concerned about the welfare of its employees
2 Employees’ needs and desires are important to this organization
3 This organization will go out of its way to help its employees
4 This organization would never deliberately take advantage of its employees
5 This organization is guided by sound moral principles and codes of conduct
6 Power is not abused in this organization
7 This organization does not exploit external stakeholders
HIWP: This is an independent factor Nine items were used to measure the set of HIWP
(information sharing and employee participation, job security, performance management, training and family-friendly work practices) such as:
1 Specific goals are established for my job
2 My career progression is dependent on my performance relative to expected goals
3 I am consulted before decisions related to my work situation are reached
4 Employees are able to achieve a work/life balance
5 Adequate training is provided to ensure that employees are competent in their role
6 Appropriate levels of job security are offered to employees
7 There is an effort to locate opportunities for employees to apply their expanding knowledge and abilities
8 Employees are consulted about issues important to them
9 Employees can openly voice their opinions and concerns without fear of retribution
Procedural justice: This is an independent factor Procedural justice was measured with the five-item procedural justice scale developed by Niehoff and Moorman (1993) as follows:
1 Job decisions are made in an unbiased manner
2 Employees’ concerns are heard before job decisions are made
3 Job decisions are based on accurate and complete information
4 Job decisions are applied consistently across all affected employees
5 Employees can challenge or appeal job decisions made by management
The second question is crucial, as it involves two methods for establishing the suitability of a research tool: logical argument and statistical proof, with the latter being more persuasive Practical research has widely utilized the Likert measurement scale, which is recognized for its conformity and effectiveness.
Regarding reliability of the measurement scale, Cronbach’s alpha coefficient will be used to verify the reliability of variables used in the questionnaire
This research encompassed managers, office staff, and group leaders aged 18 to 60 from various organizations, including limited, joint stock, private, state, joint venture, and 100% foreign-owned companies.
To achieve the research objectives, a convenient non-probability research design was employed, deemed appropriate for this study This sampling method was selected because it allows respondents to easily answer questions while also being cost-effective and less time-consuming for data collection.
Questionnaires were distributed directly to friends and acquaintances, who were also encouraged to share them with their own networks until a sufficient number of responses were collected.
The sample size is determined by the objectives of the research and the relationships being examined (Kumar, 2005) A more complex research problem typically requires a larger sample size Additionally, a larger sample size generally leads to more accurate research results However, practical considerations such as financial resources and time constraints also play a crucial role in deciding the appropriate sample size.
The research model comprises six hypotheses and 25 variables, utilizing a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree) According to Nguyen Dinh Tho (2011), the standard formula for determining the sample size for Multiple Linear Regression (MLR) is \( n > P + 8p \) Consequently, the initial sample size calculation yields \( n > P + 8 \times 25 = 250 \).
This research determined a sample size of 200, which, while minimal due to financial and time constraints, adequately fulfills the requirements of the study.
The self-designed questionnaire was used to collect data in this research The benefits of using the questionnaire for data collection are as follows (Ranjit Kumar, 2005):
- Save cost, time and manpower
- Anonymity is highly ensured for the researcher and respondents are not necessary to meet each other
However, according to Bless (2006), the questionnaire also has some limitations as follows:
- The knowledge of respondents with used terminologies is limited
- The responding rate in the questionnaire is rather low
The questionnaire was developed to gather essential data, taking into account the demand for data collection and evaluating the strengths and weaknesses of various collection tools used in similar studies To ensure privacy, confidentiality was maintained, and all responses were kept anonymous.
Phases of designing the questionnaire:
Step 1: Base on the literature review and previous researches to build the initial questionnaire
Step 2: All items and instructions were translated from English into Vietnamese Back- translation into English by native speakers was done to ensure that the translated versions corresponded with the original English version This was the process of translating a document that has already been translated into a foreign language back to the original one – preferably by an independent translator
After back translation, the questionnaire underwent cognitive testing to assess how effectively respondents understood and answered the questions This evaluation focused on the question-response process, which includes four stages: comprehension, retrieval, judgment, and response It also considered the challenges respondents faced in formulating accurate answers The testing involved fewer than 10 individuals, including acquaintances, relatives, and colleagues, to ensure clarity and comprehension of the questionnaire.
After all, the questionnaire was revised and worked out as the final completed one
Statistical data analysis techniques
To carry out statistical jobs and analyze collected data, SPSS 16.0 software was used to verify the measurement scale reliability as well as do inferential statistic
Respondents’ age and working years were divided into groups for easy handling
The study categorized participants into four age groups: 20 to 30 years (Group 1), 31 to 40 years (Group 2), 41 to 50 years (Group 3), and 51 to 60 years (Group 4) Additionally, the duration of employment and tenure at the current company were also segmented into four groups: 1 to 5 years (Group 1), 6 to 10 years (Group 2), 11 to 15 years (Group 3), and over 15 years (Group 4).
Regarding gender of respondents, it was controlled as follows: 1 = Female, 2 = Male
Before handled and analyzed, data was screened and unsuitable answers have been eliminated
The purpose of data screening/cleaning is to:
(a) check if data have been entered correctly, such as out-of-range values
(b) check for missing values, and deciding how to deal with the missing values
(c) check for univariate outliers, check for mulitivariate outliers, and deciding how to deal with outliers
(d) check for normality, and deciding how to deal with non-normality
This research will explore the normal distribution of variables, a key concept in probability theory The normal (or Gaussian) distribution is characterized by its continuous probability distribution and features a bell-shaped probability density function, commonly referred to as the Gaussian function or the bell curve.
The normal distribution is the most significant probability distribution in statistics due to its foundational role in the central limit theorem, which indicates that the mean of a large sample of random variables tends to be normally distributed, regardless of the original distribution's shape This property allows for extensive applications in sampling Additionally, the normal distribution is analytically manageable, enabling the derivation of numerous results in explicit form.
Skewness quantifies the degree of asymmetry in a distribution A distribution is considered symmetric when its left and right sides mirror each other around the central point.
Kurtosis quantifies the shape of a data distribution, indicating whether it is peaked or flat compared to a normal distribution High kurtosis signifies that the data set has a pronounced peak around the mean, a swift decline, and heavy tails.
The histogram is an effective graphical technique for showing both the skewness and kurtosis of data set
For univariate data Y 1, Y 2, , Y N , the formula for skewness is:
The mean represents the average value, while the standard deviation measures the data's dispersion, with N indicating the total number of data points In a normal distribution, the skewness is zero, and symmetric data typically exhibits skewness values close to zero.
For univariate data Y 1, Y 2, , Y N , the formula for kurtosis is:
(3.3) where is the mean, is the standard deviation, and N is the number of data points
The kurtosis for a standard normal distribution is three
Correlation, as defined by Bobko (2001), is a statistical method that quantifies the strength and degree of association between two variables This relationship is represented by the correlation coefficient, which ranges from -1 to 1.
1 Perfect correlation: When both the variables change in the same ratio, then it is called perfect correlation
2 High degree of correlation: When the correlation coefficient range is above 0.75, it is called high degree of correlation
3 Moderate correlation: When the correlation coefficient range is between 0.50 to 0.75, it is called in moderate degree of correlation
4 Low degree of correlation: When the correlation coefficient range is between 0.25 to 0.50, it is called low degree of correlation
5 Absence of correlation: When the correlation coefficient is between 0.0 to 0.25, it shows that there is no correlation
There are many techniques to calculate the correlation coefficient
In SPSS, the analysis menu offers a bivariate analysis option that utilizes Pearson correlation for continuous variables If the correlation between two variables exceeds 0.08, we will remove the variable that is most similar to the other.
3.2.5 Verifying the measurement scale reliability:
This research aims to establish and validate the reliability of measurement scales for each factor influencing trust within organizations To achieve this, we will utilize the Cronbach’s alpha coefficient as a key tool in our analysis.
Cronbach’s alpha will check the reliability of measured variables Those which do not ensure the reliability will be deleted from the measurement scale
The literature review identifies three key factors influencing trust in employers: high-involvement work practices, procedural justice, and perceived organizational trustworthiness It is essential to assess the reliability of the measurement scale for each of these factors.
This research verifies the reliability of the measurement scale using Cronbach’s alpha coefficient to eliminate unsuitable variables According to Hoang Trong and Chu Thi Mong Nguyet (2005), a Cronbach’s alpha value between 0.8 and nearly 1 is considered excellent, while a value from 0.7 to nearly 0.8 is deemed acceptable.
In this research, only factors with a Cronbach’s alpha greater than 0.7 were deemed reliable and retained Additionally, factors with a corrected item total correlation exceeding 0.4 were also included in the final analysis.
The Cronbach’s alpha coefficient is calculated as follows:
N: number of variables taken for analysis
: Variance of the (i) observed variables
: Variance of the general variable
We will, in turn, verify the Cronbach’s alpha coefficient of each factor’s measurement scale of trust in the employer and that of employee trust in organizations
After deleting unreliable variables, the remaining variables will be considered about their suitability by analyzing CFA CFA will help us know whether the data fit the hypothesized measurement model
Factor analysis in statistical analysis is divided into two main categories: exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) EFA is used to uncover patterns in data without imposing constraints, assuming that all common factors influence every observed variable, and that these factors may be correlated or uncorrelated In contrast, CFA is a theory-driven approach that allows researchers to impose meaningful constraints on the factor model, enabling the testing of specific hypotheses regarding a particular factor structure.
This research evaluates the suitability of measured variables for organizational models in HCM City through confirmatory factor analysis using Amos 16.0 Incompatible variables were eliminated to proceed with linear regression analysis Various fit indices were compared, with the Chi-square statistic traditionally used to assess model fit, although several authors, including Bentler and Bonnett (1980), have noted potential issues with this approach.
Bentler 1990, Bollen 1989, Mulaik, James, Van Alstine, Bennett, Lind & Stilwell, 1989)
In evaluating model fit, several indices are considered, including the Comparative Fit Index (CFI) and the Tucker-Lewis Fit Index (TLI), both of which provide robust measures independent of sample size Traditionally, values in the high 0.80s to 0.90s indicate a good fit, though recent research suggests that a threshold of 0.95 or higher is preferable Additionally, the Root Mean Square Error of Approximation (RMSEA) is utilized to account for degrees of freedom, with ideal values below 0.05, acceptable values between 0.06 and 0.08, and poor fit indicated by values above 0.10 Ultimately, a model is deemed suitable if TLI and CFI values are ≥ 0.9, CMIN/df ≤ 2, and RMSEA ≤ 0.08, as noted by Tho & Trang (2008).
After calculation, if such above indices are not satisfied, we should delete some items of each variable, which are smaller than 0.4, until they are satisfactory with the measurement indices
RESEARCH RESULT 4.1 Eliminating unsuitable answers
Screening the data
The research required only the respondents' year of birth, leading to the removal of specific dates from two responses for simplicity and consistency Ultimately, only the year of birth was retained for analysis.
In the current company, there were 09 instances where employees provided both the month and year of their start date, even though only the year was required Consequently, only the year of employment was retained in these cases.
In the Methodology chapter, we performed a normal distribution analysis, examining the coefficients of Skewness and Kurtosis The results indicated that the Skewness and Kurtosis values for the three independent variables—HIWP, procedural justice, and perceived organizational trustworthiness—along with the dependent variable, trust in the employer, were satisfactory This suggests that these variables fluctuated within the acceptable limits outlined in the previous chapter, confirming that they are normally distributed.
Table 4.1 Normal Distribution of Perceived Organizational Trustworthiness
POT1 POT2 POT3 POT4 POT5 POT6 POT7 POT8 POT9 POT10
Std Error of Skewness 172 172 172 172 172 172 172 172 172 172 Kurtosis -.401 005 -.494 -.245 -.744 -.241 -.166 -.338 -.050 -.226 Std Error of Kurtosis 342 342 342 342 342 342 342 342 342 342
Table 4.2 Normal Distribution of High-Involvement Work Practices
Table 4.3 Normal Distribution of Procedural Justice
PJ1 PJ2 PJ3 PJ4 PJ5
HIWP1 HIWP2 HIWP3 HIWP4 HIWP5 HIWP6 HIWP7 HIWP8 HIWP9
Std Deviation 1.402 1.371 1.393 1.314 1.419 1.387 1.282 1.396 1.307 Skewness -.083 -.244 -.278 -.366 -.379 -.324 -.137 -.359 -.262 Std Error of
Table 4.4 Normal Distribution of Trust in the Employer
Correlation between variables
As presented in the Chapter of Methodology, after defining the normal distribution of variables, we will, next, define the correlation between variables by a coefficient
From the result, we can see that the correlation coefficient of all variables were less than 0.80, which means their correlation can be acceptable and not any items were eliminated
Table 4.5 Correlation coefficient of Perceived Organizational Trustworthiness
POT1 POT2 POT3 POT4 POT5 POT6 POT7 POT8 POT9 POT10
Table 4.6 Correlation Coefficient of High-Involvement Work Practices
Table 4.7 Correlation Coefficient of Procedural Justice
PJ1 PJ2 PJ3 PJ4 PJ5
PJ1 Pearson Correlation 1 485 ** 356 ** 259 ** 182 * PJ2 Pearson Correlation 485 ** 1 448 ** 318 ** 172 * PJ3 Pearson Correlation 356 ** 448 ** 1 496 ** 221 **
** Correlation is significant at the 0.01 level (2-tailed)
* Correlation is significant at the 0.05 level (2-tailed).
Analysis of reliability and suitability of the measurement scale
4.5.1 Measurement scale of each factor of employee trust:
The Cronbach alpha analysis result of factors of employee trust is summarized as follows:
The Cronbach’s alpha of high – involvement work practices was 0.726, acceptable We found that if deleting any items of this factor, the Cronbach’s alpha coefficient will reduce
Although there were some corrected items total correlations less than 0.4 such as HIWP1
Specific goals are set for my job, and my career advancement relies on my performance in relation to these goals Additionally, employees are encouraged to maintain a healthy work/life balance and can express their opinions and concerns freely without fear of retaliation These factors will undergo confirmatory factor analysis to determine their validity for retention.
Table 4.8 Reliability Statistics of HIWP
Cronbach's Alpha Based on Standardized Items N of Items
Table 4.9 Item Total Statistics of HIWP
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Cronbach's Alpha if Item Deleted
The Cronbach’s alpha for procedural justice was found to be 0.703, indicating an acceptable level of reliability Notably, removing the last item, PJ5, which states “Employees can challenge or appeal job decisions made by management,” would increase the coefficient to 0.722 Additionally, the corrected item-total correlation for PJ5 was below 0.4 Therefore, we opted to eliminate this item and retain only the first four.
Table 4.10 Reliability Statistics of PJ
Cronbach's Alpha Based on Standardized Items N of Items
Table 4.11 Item-Total Statistics of PJ
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Cronbach's Alpha if Item Deleted
Table 4.12 Reliability Statistics of PJ after deleted
Cronbach's Alpha Based on Standardized
Table 4.13 Item-Total Statistics of PJ after deleted
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Cronbach's Alpha if Item Deleted
The Cronbach’s alpha of perceived organizational trustworthiness was 0.763, acceptable
Removing the first item, POT1, which states "This organization is capable of meeting its responsibilities," results in an increased coefficient of 0.765, while its corrected item total correlation falls below 0.4 Consequently, we have opted to eliminate this item.
Table 4.14 Reliability Statistics of POT
Cronbach's Alpha Based on Standardized Items N of Items
Table 4.15 Item-Total Statistics of POT
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Cronbach's Alpha if Item Deleted
Table 4.16 Reliability Statistics of POT after deleted
Table 4.17 Item-Total Statistics of POT after deleted
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Cronbach's Alpha if Item Deleted
Based on Cronbach’s alpha coefficient analysis, we will retain all items related to high-involvement work practices, procedural justice, and perceived organizational trustworthiness, except for PJ5 “Employees can challenge or appeal job decisions made by management” and POT1 “This organization is capable of meeting its responsibilities,” as their corrected item total correlations were below 0.4, indicating that their removal would enhance the overall Cronbach’s alpha Although the four high-involvement work practices variables—HIWP1 “Specific goals are established for my job,” HIWP2 “My career progression is dependent on my performance relative to expected goals,” HIWP4 “Employees are able to achieve a work/life balance,” and HIWP9 “Employees can openly voice their opinions and concerns without fear of retribution”—also exhibited low corrected item total correlations, their elimination would negatively impact the Cronbach’s alpha, necessitating further analysis before making a final decision.
4.5.2 Measurement scale of trust in the employer:
The reliability of the factor, which focuses solely on the question "Overall, to what extent do you trust your organizations?", cannot be assessed using Cronbach's alpha coefficient; however, it is implicitly regarded as reliable.
Confirmatory factor analysis
The CFA structure consists of three key factors: Perceived Organizational Trustworthiness (POT), High-Involvement Work Practices (HIWP), and Procedural Justice (PJ) Following the reliability testing using Cronbach’s alpha, two variables, POT1 and PJ5, were removed, resulting in the POT factor being measured by 9 observed variables, HIWP by 9, and PJ by 4 The reliability of these factors was affected by random measurement error, as indicated by the associated error term Each observed variable was regressed onto its corresponding factor, revealing a correlation among the three factors.
Before any discussion of how we might go about testing this model, let’s take a few minutes first to dissect this model and list its component parts as follows:
1 There were three factors, as indicated by the four ellipses labelled POT, HIWP and PJ
2 The three factors were correlated, as indicated by the two-headed arrow
3 There were 22 observed variables, as indicated by the 22 rectangles; they represented item pairs from the POT, HIWP, PJ subscales of the Trust in the Employer (Marsh, 1992a)
4 The observed variables loaded on the factors in the following pattern: POT2 – POT10 load on factor 1, HIWP1 – HIWP9 load on factor 2, PJ1 – PJ4 load on factor 3
5 Each observed variable loaded on one and only one factor
6 Errors of measurement associated with each observed variable (e1-e22) were uncorrelated
After calculation, the hypothesized model was as follows:
Diagram 4.1 The CFA model after calculation
Chi-square/df = 2.307 ; GFI = 783 ; TLI = 245 ; CFI = 326 ; RMSEA = 081
These were standardized factor loadings, the squared multiple correlation coefficient for each observed variable, and a Chi-square statistic of model fit
The model fit is inadequate, indicated by a CMIN/df value of 2.307, which exceeds the acceptable threshold of 2 Additionally, the TLI and CFI values are below 0.9, and the RMSEA is 0.081, slightly above the ideal limit of 0.08 This suggests the presence of variables that do not align well with the model The challenge lies in identifying which variables have the least impact on the factor, allowing for their removal to enhance the fit indices.
The analysis revealed that in the POT factor, variables POT2 to POT6 were identified, while the HIWP factor included variables HIWP1 to HIWP5 and HIWP9 Additionally, the PJ factor contained variables PJ1 and PJ2 However, the factor loadings for these variables were minimal, indicating their limited impact on the POT, HIWP, and PJ factors Consequently, these variables were removed from the model.
The model, then, was re-calculated as follows:
Diagram 4.2 The CFA model after re-calculation
The model demonstrated a strong fit, evidenced by a Chi-square/df statistic of 1.382 and a low RMSEA of 0.044, indicating that the model aligns well with the data.
Chi-square/df = 1.382 ; GFI = 970 ; TLI = 917 ; CFI = 950 ; RMSEA = 044
The revised model now consists of three factors, incorporating the variables POT7, POT8, POT9, HIWP6, HIWP7, HIWP8, PJ3, and PJ4 In the following section, a linear regression analysis will be conducted to assess the impact of these variables on trust in the employer.
Linear regression analysis
In the previous section, we identified three key factors influencing trust in the employer through Cronbach’s alpha coefficient and confirmatory factor analysis: high-involvement work practices, perceived organizational trustworthiness, and procedural justice However, certain variables within these factors were excluded due to their incompatibility with the model and will not be considered in the linear regression analysis.
Trust is a crucial value within an organization, represented as a single item The independent variables, namely HIWP, POT, and PJ, are determined by calculating the average value of all items associated with each confirmed factor.
4.7.1 Relation between variables to trust in the employer:
The impact of each independent and control variables to trust in the employer as follows:
Step 1: Testing the impact of control variables to TE:
Table 4.18 Model Summary of control variables
Std Error of the Estimate
Table 4.19 Coefficients of control variables
As we can see, the control variables have not related to trust in the employer for their p- values are not significant (>0.05)
Step 2: Testing the impact of HIWP and PJ to TE:
Table 4.20 Model Summary of HIWP and PJ
Std Error of the Estimate
Table 4.21 Coefficients of HIWP and PJ
High-Involvement Work Practices (HIWP) and Procedural Justice significantly enhance employees' trust in their employer, as indicated by their significant p-values Consequently, both hypothesis 2, which states that HIWP are positively related to employees' trust in the employer, and hypothesis 4, which asserts that Procedural Justice is positively related to employees' trust in the employer, are supported.
Step 3: Testing the impact of POT to TE:
Table 4.22 Model Summary of POT
Std Error of the Estimate
Perceived Organizational Trustworthiness significantly influences employees' trust in their employer, supporting the acceptance of Hypothesis 1: "Perceived Organizational Trustworthiness is positively related to employees’ trust in the employer."
However, their relation to the dependent variable is not the same as shown in the indices of
R and R square The higher was the impact of HIWP and PJ, and that of POT to TE is lower
Linear regression analysis indicates that perceived organizational trustworthiness does not serve as a mediating factor between trust in the employer and high-involvement work practices, as demonstrated in the table below.
Table 4.24 Some regression indices of three models
In the analysis, the β values for HIWP and PJ in model 2 were 0.235 and 0.322, respectively However, in model 3, which included POT, these values decreased to 0.200 and 0.285 Additionally, the p values for HIWP and PJ in model 2 were 0.001 and 0.000, indicating a significant impact on trust in the employer In model 3, the p value for HIWP increased to 0.004, while the p value for POT was not significant at 0.085.
0.005) So we can conclude that Perceived Organizational Trustworthiness has not supported the research model as a mediating variable (Hypothesis 3 and hypothesis 5)
The regression analysis revealed a significant interaction between High-Performance Work Practices (HIWP) and procedural justice, supporting hypothesis 6 Both HIWP and procedural justice exhibited significant positive beta weights, while their interaction showed a negative beta weight This suggests that the relationship between HIWP and trust is more pronounced among employees who perceive lower levels of procedural justice within their organization.
4.7.2 Checking assumptions in linear regression:
The linear regression model using the OLS method relies on specific assumptions, and its significance is contingent upon the satisfaction of these assumptions To ensure the model's reliability, it is essential to investigate any potential violations of the necessary assumptions in linear regression.
The first assumption is that there is a linear correlation, which is assessed using a scatterplot diagram featuring standardized residual values on the y-axis and standardized predicted values on the x-axis The analysis of the diagram reveals that the residual values remain consistent and do not deviate from the predicted values, indicating that the assumption of linear correlation has not been violated.
The next assumption pertains to the variance of unchanged residuals To evaluate this assumption, we calculated the Spearman correlation coefficient between the absolute values of the residuals and the independent variables The significance value of the 95% reliable correlation coefficients indicates that we do not have sufficient evidence to reject the null hypothesis (H0), suggesting that the absolute values of the residuals are independent of the independent variables Consequently, the assumption regarding the variance of unchanged residuals remains valid.
Minimum Maximum Mean Std Deviation N
Centered Leverage Value 010 156 035 024 200 a Dependent Variable: TE
Table 4.26 Correlation of variance of unchanged residual
ABScuare POT_mean HIWP_mean PJ_mean
- Following is the assumption about normal distribution of residual In the Histogram, we can see that the residual was normally distributed with mean = -3.82 and the std deviation
Looking at the Q-Q plot chart, we can see that the observed values focus closely to the diagonal of expected values, which means standardized residuals are normally distributed
The independence of residuals is a crucial assumption that must be examined According to the Durbin-Watson table for a sample size of N = 200 and three independent variables, the critical values are dU = 1.704 and dL = 1.643 If the calculated d value falls within the range of (dU, 4 - dU) or (1.704, 2.290), it indicates that the residuals are independent In this case, the statistical result shows d = 1.971, confirming the independence of the residuals.
Table 4.27 Durbin-Watson coefficient of the regression model
Model R R Adjusted Std Error Change Statistics Durbin-
- Finally, we will consider the linear collinearity violation of the model The tolerance of three variables is rather high, more than 0.5 while VIF coefficient is rather low, less than 2
A VIF coefficient below 10 indicates that the assumption of linear collinearity in the model can be rejected Therefore, the linear regression model established by equation 4.1 adheres to the essential assumptions of linear regression.
Table 4.28 Tolerence and VIF coefficients of the regression model
B Std Error Beta Tolerance VIF
POT_mean 123 071 126 1.731 085 762 1.313 a Dependent Variable: TE