Reasons for this research
Research by DG&A Consulting reveals that only 37% of employees understand their company's goals, while just 20% recognize how their roles contribute to the organization About one-third of employees express a desire to contribute, yet 20% are disengaged and 50% avoid responsibilities Many employees leave local companies within 2-3 years due to a lack of clarity regarding their roles and the company's objectives, which leads to confusion and misalignment Additionally, the organizational structure often stifles initiative, resulting in diminished trust in both the employer and their own positions The ongoing economic crisis has further eroded employee morale, highlighting the need for companies to prioritize employee satisfaction as a vital component of their success By viewing staff as essential customers, companies can foster loyalty and encourage employees to actively contribute to organizational goals.
Since Vietnam's accession to the WTO, local companies aiming to grow or partner with foreign firms must focus on attracting and retaining skilled employees while building their trust to prevent "brain-drain." It is crucial for organizations to identify and assess the factors that influence employee trust in their employer, allowing them to develop effective human resource management policies.
“High-involvement Work Practices, Procedural Justice and Trust in the Employer
An examination conducted in Ho Chi Minh City sheds light on the intricate relationships between HR practices, procedural justice, and trust within organizations The study reveals that perceptions of organizational trustworthiness—encompassing perceived ability and trustworthy intentions—partially mediate the connections between HR practices and procedural justice with trust Additionally, the findings indicate that procedural justice significantly enhances the predictive power of trust in organizational contexts.
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 their employers in Ho Chi Minh City, focusing on the relationship between these two factors and their influence on workplace dynamics.
- 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 article investigates the relationship between trust in organizations and the influence of high-involvement work practices (HIWP) and procedural justice on employer trust By analyzing data from a surveyed questionnaire, we identify three key areas of focus: trust research, strategic HR management, and procedural justice literature Our study aims to determine how HIWP and procedural justice contribute to perceptions of organizational trustworthiness and trust in the employer Additionally, we explore the unique effects of HIWP and procedural justice on organizational trust and their potential as functional equivalents in fostering trust within the workplace.
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
This thesis employed a quantitative research approach to explore and define trust in employers, utilizing a 7-point Likert scale to assess variable values As a discovery research study, it implemented a non-probability sampling method based on convenience.
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 The official version can be found in the Appendix of the thesis.
The study utilized statistical methods to analyze the collected sample data, employing Cronbach’s alpha to select and consolidate 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 calculate the coefficients of these factors within 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 highlights the factors influencing employee trust in HCM City Consequently, companies in the region can enhance their HR management policies by revising or developing strategies that foster employee trust and engagement, ultimately leading to greater contributions to 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 organized into five chapters, beginning with Chapter 1 – Introduction, which outlines the research's rationale, objectives, scope, and methodology, as well as its significance Chapter 2 focuses on the Literature Review and Research.
The Literature Review will define trust in the employer, explore its antecedents, and examine their relationships, while also presenting the proposed research model Chapter 3, Research Methodology, will detail the development of the measurement scale, sampling methods, data collection processes, and statistical analysis techniques used in the study Chapter 4, Research Results, will analyze and interpret the collected data, focusing on the reliability and validity of the measurement scale as well as the results of the inferential statistics.
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 critical focus in management research, highlighting its essential role in organizational behavior Research indicates that a certain level of trust is necessary for coordinated actions both within organizations and across their boundaries Trust not only facilitates relationships but also reduces transaction costs, potentially serving as a source of competitive advantage While the advantages of trust are well-documented, empirical studies on how to cultivate trust within organizations remain limited.
Trust in the employer is crucial for organizations to enhance effectiveness and performance, as numerous studies indicate that employee trust significantly influences organizational outcomes High levels of trust lead to increased employee retention, greater effort, and improved cooperation, while a lack of trust can result in decreased work effectiveness, counterproductive behaviors, and higher turnover rates Ultimately, perceptions of organizational trustworthiness can offer firms a competitive edge.
Despite the growing literature on interpersonal trust, the concept of trust within organizations has received significantly less attention Understanding what fosters trust in employers is crucial, as both dispositional trust and perceptions of organizational trustworthiness play a role in enhancing this trust While Schoorman, Mayer, and Davis have expanded on their foundational model of interpersonal trust to identify factors influencing employee trust in their employers, there is a lack of field studies validating this model It is essential to recognize that perceptions of organizational trustworthiness differ from interpersonal trust, necessitating an examination of institutional processes and principles, as well as how organizational representatives embody these concepts.
Organizational policies and the fairness of human resource (HR) practices play a crucial role in shaping trust between employees and employers While some research has begun to examine how HR practices influence this trust (Whitener 1997; Gould-Williams 2003), the overall impact of comprehensive policy bundles and their implementation on organizational trust remains underexplored.
Trust in organizational behavior is defined as faith in and loyalty to a leader, serving as a crucial precursor to risk-taking behavior It encompasses three key elements: (a) trustworthiness, which is based on rational assessment of a leader's competence and intentions; (b) faith in the leader, reflecting psychological sources of trust; and (c) loyalty, which involves emotional connections and established routines within specific relationships.
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 Conversely, in business management, it is vital for fostering relationships between managers and employees Understanding trust as a key dimension enhances the quality of these relationships across various settings.
Trust is defined as a willingness to be vulnerable to another based on the belief in their reliability, openness, competence, and compassion It involves the decision to rely on another party, whether a person, group, or firm, despite the inherent relational risk, with the expectation of achieving at least neutral or positive outcomes This reliance fosters vulnerability, as the trusting party risks negative outcomes if the other proves untrustworthy The perceived trustworthiness of the trustee is crucial, enabling the trustor to make the necessary cognitive "leap of faith" to establish trust.
Organizational trust differs from interpersonal trust in its focus and complexity, as individuals may not clearly define what they trust about their employer Research has explored this concept by examining interpersonal relationships within the workplace, including interactions between employees and management at various levels Scholars like Giddens have linked organizational trust to the reliability of abstract principles, while Carnevale emphasizes the expectation that institutions will act fairly and competently Ultimately, trust in organizations is influenced by the perceived reliability and integrity of their practices.
Administrative organizations and top management groups possess collective characteristics that transcend individual traits, ensuring continuity in activities and direction during personnel changes (Whitley, 1987) Giddens (1990) highlights the crucial role of individuals in fostering trust within abstract systems, particularly those who manage the interfaces where trust is established and sustained Consequently, trust in an employer is derived from the assessment and aggregation of diverse evidence sources at various 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, two main antecedents to trust in an organization are identified: the dispositional trust of the trustor and the perceived trustworthiness of the trustee Kramer (1999) categorizes trust into two broad bases: dispositional trust and trust derived from perceptions of individualized trustworthiness, such as history-based trust, and 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 & Swap, 1982; McKnight, Choudhury, & Kacmar, 2002) and is particularly influential in fostering trusting beliefs during uncertain circumstances (Gill, Boies, Finegan, & McNally).
According to Kee and Knox (1970), trusting beliefs are influenced by dispositional trust, which acts as a filter that affects how individuals interpret the actions of others, even when prior experiences are present (Govier 1994).
A recent meta-analysis suggested that propensity to trust may drive and shape the
The concept of a "cognitive leap" in trust suggests that individuals may extend their trust beyond what their past experiences would typically justify Research by Colquitt, Scott, and Lepine (2007) highlights that trust predisposition plays a crucial and independent role in shaping trust levels, even when trustworthy information is available.
This research emphasizes the significance of perceived organizational trustworthiness in fostering employee trust in their employer, as supported by previous studies (Barber 1983; Kramer 1999; Schoorman et al 2007) Trust is fundamentally rooted in the concept of trustworthiness, highlighting its cognitive dimensions and importance in organizational dynamics.
“based on a cognitive process which discriminates among persons and institutions that are trustworthy, distrusted, and unknown’ (Lewis and Weigert 1985)
Perceived organizational trustworthiness is multi-dimensional, encompassing aspects such as trustworthy intentions and ability Trustworthy intentions are further divided into benevolence and integrity At the organizational level, these dimensions translate into "organizational ability," which refers to the collective competencies and characteristics that enable effective functioning, and "organizational benevolence," which reflects the genuine care for stakeholders' well-being.
Research Model and estimation indices for measuring trust in the employer
The proposed linear regression model aims to analyze "Trust in the Employer" as the dependent variable, with "High-Performance Work Practices" (HIWP), "Procedural Justice," and "Perceived Organizational Trustworthiness" as independent variables Notably, "Perceived Organizational Trustworthiness" serves as a mediating variable that influences the relationship 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 outlines 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 details 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 employed to measure all variables, both dependent and independent Following this, a convenient non-probability sampling method was utilized to define the sample, resulting in 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 Following the development of the questionnaire and determination of the required sample size, it was distributed for data collection The gathered data will be analyzed statistically, utilizing inferential statistics 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 investigates the impact of high involvement work practices and procedural justice on employee trust in employers in Ho Chi Minh City By utilizing closed questions in attitudinal research, the study effectively assesses individuals' attitudes toward various aspects of their work life One of the key objectives is to explore the factors influencing employee trust and their interactive effects The use of a Likert measurement scale for responses allows for a nuanced understanding of employee trust levels, categorizing them as satisfied or unsatisfied, and high or low on a 5 or 7-point scale Additionally, the interval nature of the Likert scale enables quantitative analysis of the collected data, facilitating the examination 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:
Measuring trust in the employer is crucial, and to assess this, we asked respondents the question, “Overall, to what extent do you trust your organization?” Participants rated their trust 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 suitability of a research tool can be established through two primary methods: 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 in data collection.
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 a diverse group of participants, including managers, office staff, and group leaders from various types of organizations, such as limited companies, joint stock firms, private enterprises, state-owned entities, joint ventures, and fully foreign-owned companies, all ranging in age from 18 to 60 years.
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 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 research goals and the relationships being examined (Kumar, 2005) A more diverse research problem typically requires a larger sample size, which can enhance the accuracy of the 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 includes 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 formula for calculating the sample size for Multiple Linear Regression (MLR) is n > P + 8p, leading to an initial sample size calculation of n > P + 8 * 25 = 250.
This research was conducted with a sample size of 200, which was determined to be the necessary minimum due to financial and time constraints, while still effectively meeting the research objectives.
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
To meet the demand for data collection, a questionnaire was developed by evaluating the strengths and weaknesses of various collection tools used in related research The survey ensured confidentiality and anonymity for all participants' responses.
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
Following back translation, the questionnaire underwent cognitive testing to assess its effectiveness in eliciting accurate responses from survey participants This evaluation focused on whether respondents comprehended the questions and could provide precise answers Cognitive testing analyzed the question-response process, which involves four stages: comprehension, retrieval, judgment, and response, while also considering the challenges respondents faced in formulating accurate answers This step was conducted with fewer than 10 individuals, including acquaintances, relatives, and colleagues, to ensure clarity and understanding of the questionnaire.
After all, the questionnaire was revised and worked out as the final completed one
Step 3: Before officially delivering the questionnaire to the public for interview, we conducted the pilot study, which was a small experiment designed to test logistics and gather information prior to a larger study, in order to improve the latter’s quality and efficiency A pilot study could reveal deficiencies in the design of a proposed experiment or procedure and these could then be addressed before time and resources were expended on large scale studies
We carried out the pilot study on about 50-60 persons working at different organizations in different fields
3.1.4 Data collection process: The software Forms – Google Docs was used to design the questionnaire in the internet This one was directly or indirectly sent to the surveying subjects
This questionnaire could be found at the Appendix of this research
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: 20-30 years (Group 1), 31-40 years (Group 2), 41-50 years (Group 3), and 51-60 years (Group 4) Additionally, work experience and tenure at the current company were segmented into four categories: 1-5 years (Group 1), 6-10 years (Group 2), 11-15 years (Group 3), and over 15 years (Group 4) Gender was recorded with a binary classification, where 1 represents Female and 2 represents 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 focuses on the normal distribution of variables, a key concept in probability theory The normal distribution, also known as the Gaussian distribution, features a continuous probability density function that is characterized by its bell-shaped curve, commonly referred to as the bell curve.
The normal distribution is the most significant probability distribution in statistics due to its wide applications and analytical tractability It is derived from the central limit theorem, which indicates that the mean of a large number of random variables from the same distribution tends to be normally distributed, regardless of the original distribution's shape This property makes it particularly useful in sampling and allows for the derivation of numerous results in explicit form.
Skewness measures the degree of asymmetry in a distribution, indicating how much a dataset deviates from a symmetric shape A distribution is considered symmetric when its left and right sides mirror each other around the central point.
Kurtosis measures the peakedness or flatness of a data distribution compared to a normal distribution Data sets with high kurtosis exhibit 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:
In statistics, the mean represents the average of a dataset, while the standard deviation measures the data's dispersion, and N indicates the total number of data points For a normal distribution, the skewness is zero, suggesting that symmetric data should also exhibit a skewness 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 used to assess the strength and degree of association between two variables This relationship is quantified using 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, to analyze the correlation between continuous variables, you can use the bivariate analysis feature with Pearson correlation If the correlation coefficient 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 trust factor within an organization To achieve this goal, we will utilize the Cronbach’s alpha coefficient as a key tool for assessment.
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 Therefore, 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 of 0.8 to nearly 1 is considered excellent, while a value from 0.7 to nearly 0.8 is deemed acceptable In this study, only factors with a Cronbach’s alpha greater than 0.7 were retained as reliable Additionally, the corrected item total correlation was evaluated, with only those factors having a correlation greater than 0.4 remaining in the 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 identify patterns in data without imposing constraints, assuming that common factors influence all observed variables and can be either 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 the factor structure.
This research conducted in HCM City utilized confirmatory factor analysis with Amos 16.0 to assess the suitability of measured variables for the model Incompatible variables were eliminated to prepare for linear regression analysis Various fit indices were compared, including the Chi-square statistic, which has known limitations (Bentler & Bonnett, 1980; Bentler 1990; Bollen 1989; Mulaik et al., 1989) Incremental fit indices such as the Comparative Fit Index (CFI) and the Tucker-Lewis Fit Index (TLI) were also evaluated, with values in the high 0.80s to 0.90s traditionally indicating a good fit, though Hu and Bentler (1999) suggest a threshold of 95 or higher Additionally, the Root Mean Square Error of Approximation (RMSEA) was used to account for degrees of freedom, with values below 0.05 being ideal and those above 0.10 indicating poor fit (Browne & Cudeck, 1992) In conclusion, a model is deemed suitable if it meets criteria of TLI and CFI values ≥ 0.9, CMIN/df ≤ 2, and RMSEA ≤ 0.08, as stated 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
Linear regression analysis was employed to determine the impact of various factors on trust in organizations, specifically focusing on the coefficients associated with these factors in the regression equation In this model, trust in the employer serves as the dependent variable, while high involvement work practices, procedural justice, and perceived organizational trustworthiness are identified as independent variables, with the latter also acting as a mediating variable.
Base on the literature review and results of Pearson correlation coefficient as mentioned- above, we will take all independent variables in the confirmed regression model by Enter method
To evaluate the impact of various factors on trust within organizations, we conducted a three-step analysis The first step involved examining the relationship between control variables and trust in organizations (TE) In the second step, we assessed the connections between control variables, High-Intensity Work Practices (HIWP), Procedural Justice (PJ), and TE Finally, the third step focused on the relationship between control variables, Perceived Organizational Trust (POT), and TE This comprehensive approach enabled us to test Hypotheses 1, 2, and 4, which highlight the significant effects of HIWP justice and trustworthiness on organizational trust.
RESEARCH RESULT 4.1 Eliminating unsuitable answers
Screening the data
Research indicated that only the respondents' year of birth was necessary However, two responses included both the date and year of birth To streamline the data, the specific dates were removed, leaving only the year of birth for analysis.
In the analysis of employment start dates at the current company, there were nine instances where both the month and year were provided, despite only the year being required Consequently, only the year of employment was retained for these cases.
In the Methodology chapter, we performed a normal distribution analysis, assessing the coefficients of Skewness and Kurtosis for three independent variables: HIWP, procedural justice, and perceived organizational trustworthiness, along with the dependent variable of trust in the employer The results indicated that the Skewness and Kurtosis values for each variable fell within acceptable limits, confirming that these variables 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
POT1 Pearson Correlation 1 391 ** 343 ** 131 079 -.013 042 183 ** 082 119 POT2 Pearson
POT5 Pearson Correlation 079 230 ** 215 ** 557 ** 1 474 ** 397 ** 045 101 159 * POT6 Pearson
POT8 Pearson Correlation 183 ** 128 130 039 045 247 ** 526 ** 1 473 ** 335 ** POT9 Pearson
Table 4.6 Correlation Coefficient of High-Involvement Work Practices
Table 4.7 Correlation Coefficient of Procedural Justice
PJ1 PJ2 PJ3 PJ4 PJ5
PJ3 Pearson Correlation 356 ** 448 ** 1 496 ** 221 ** PJ4 Pearson Correlation 259 ** 318 ** 496 ** 1 285 **
** 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 for high-involvement work practices was found to be 0.726, indicating an acceptable level of reliability Analysis revealed that removing any items from this factor would decrease the Cronbach's alpha coefficient Notably, some items, including HIWP1, exhibited corrected item-total correlations below 0.4.
Specific goals are set for my role, and my career advancement relies on my performance in relation to these expectations 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 further examination in a confirmatory factor analysis to determine their relevance and potential elimination.
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, which is acceptable However, removing the last item, PJ5, "Employees can challenge or appeal job decisions made by management," would increase the coefficient to 0.722, as its corrected item-total correlation was below 0.4 Consequently, we chose 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 for perceived organizational trustworthiness was measured at 0.763, indicating an acceptable level of reliability Notably, removing the first item, POT1, which states, “This organization is capable of meeting its responsibilities,” would raise the coefficient to 0.765, as its corrected item total correlation was significantly below 0.4 Consequently, we opted to eliminate this item from the assessment.
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, which states, “Employees can challenge or appeal job decisions made by management,” and POT1, “This organization is capable of meeting its responsibilities,” due to their low corrected item total correlations below 0.4, indicating that their removal would enhance the Cronbach’s alpha of their respective factors While 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”)—exhibit low corrected item total correlations, their elimination would negatively impact the overall Cronbach’s alpha, necessitating further analysis before making a final decision.
4.5.2 Measurement scale of trust in the employer:
Due to the singular variable of trust in organizations, specifically measured by the question, "Overall, to what extent do you trust your organizations?", we are unable to assess its reliability using Cronbach's alpha coefficient; thus, 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 assessment of Cronbach’s alpha reliability, two variables, POT1 and PJ5, were removed, resulting in the POT factor being measured by nine observed variables, the HIWP by nine, and the PJ by four 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 respective factor, and the analysis revealed significant correlations 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 suboptimal, indicated by a CMIN/df of 2.307, which exceeds the acceptable threshold of 2 Additionally, both TLI and CFI values are below 0.9, and the RMSEA is 0.081, slightly above the ideal cutoff of 0.08 This suggests the presence of incompatible variables within the model The challenge lies in identifying the variables that have the least impact on the factor, allowing for their removal to enhance the overall fit indices.
The analysis revealed that within the POT factor, variables POT2 to POT6, and within the HIWP factor, variables HIWP1 to HIWP5 and HIWP9, along with variables PJ1 and PJ2 from the PJ factor, exhibited minimal factor loading Consequently, these variables were removed from the model due to their insignificant impact on the POT, HIWP, and PJ factors.
The model, then, was re-calculated as follows:
Diagram 4.2 The CFA model after re-calculation
The model demonstrated strong fit indices, with a Chi-square/df statistic of 1.382 and an RMSEA of just 0.044, indicating excellent alignment with the data.
Chi-square/df = 1.382 ; GFI = 970 ; TLI = 917 ; CFI = 950 ; RMSEA = 044
The revised model now comprises three factors, incorporating the variables POT7, POT8, POT9, HIWP6, HIWP7, HIWP8, PJ3, and PJ4 In the subsequent section, a linear regression analysis will be conducted to assess the impact of these variables on employee trust in the employer.
Linear regression analysis
In the previous analysis, we identified three key factors influencing trust in the employer: high-involvement work practices (HIWP), perceived organizational trustworthiness (POT), and procedural justice (PJ), using Cronbach’s alpha coefficient and confirmatory factor analysis Certain variables were excluded from the model due to incompatibility and will not be included in the linear regression analysis Notably, the trust in the organization was assessed as a single item, while the independent variables (HIWP, POT, and PJ) were measured by the average values of all items within 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 posits that HIWP positively correlates with employees' trust in the employer, and hypothesis 4, which suggests that Procedural Justice is positively related to this trust, 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 has a significant positive relationship with employees' trust in their employer, supporting Hypothesis 1 However, the relationship between these variables and the dependent variable differs, as indicated by the indices.
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 evidenced by the data presented 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, but decreased to 0.200 and 0.285 in model 3 with the inclusion of POT The p values for HIWP and PJ in model 2 were significant at 0.001 and 0.000, indicating a strong influence on employer trust; however, HIWP's p value increased to 0.004 in model 3, while POT's p value was not significant (0.085 > 0.005) This suggests that Perceived Organizational Trustworthiness does not act as a mediating variable in the research model, supporting the rejection of Hypotheses 3 and 5 Additionally, the regression results revealed a significant interaction between HIWP and procedural justice, consistent with Hypothesis 6, showing that while both HIWP and procedural justice have positive beta weights, their interaction is negative, indicating a stronger relationship between HIWP and trust among employees who perceive lower procedural justice.
4.7.2 Checking assumptions in linear regression:
The linear regression model using the Ordinary Least Squares (OLS) method relies on specific assumptions for its validity To ensure the model's reliability, it is crucial to assess whether these assumptions are met, as any violations can compromise the significance of the results.
The initial assumption is that there is a linear correlation between variables This is assessed using a scatterplot diagram, which displays 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 is upheld.
To assess the assumption of unchanged residual variance, we calculated the Spearman correlation coefficient between the absolute values of the residuals and the independent variables The significance value at a 95% confidence level indicated no basis for rejecting the null hypothesis, suggesting that the absolute residuals are independent of the independent variables Consequently, the assumption regarding the variance of unchanged residuals holds true.
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
The Q-Q plot indicates that the observed values align closely with the diagonal of expected values, suggesting that the standardized residuals are normally distributed Additionally, the independence of residuals must be assessed using the Durbin-Watson statistic For a sample size of 200 and three independent variables, the critical values are dU = 1.704 and dL = 1.643 Since the calculated value of d = 1.971 falls within the range (1.704, 2.290), we can conclude that the residuals are independent of one another.
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, confirming that 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