The independent variable is psychological contract breach, and job involvement is the dependent variable. Perceived organizational justice is the moderator between the psychological contract breach and job involvement. The perceived organizational justice consists of three dimensions which are distributive justice, procedural justice and interactional justice. Figure 3.1 shows the research framework.
\\
Figure 1. Research Framework
H2
H1 Psychological Contract
Breach
Job Involvement Perceived Organizational
Justice
Controls:
Company Type Job Position Annual Salary
H2
Hypotheses
Based on the previous literatures, the research purpose, and the research questions, hypotheses are listed below:
Hypothesis 1. Psychological contract breach is negatively correlated with job involvement.
Hypothesis 2. Perceived organizational justice moderates the relationship between psychological contract breach and job involvement.
Samples and Data Collection
This study utilized the quantitative approach and the data were collected by paper-based and online questionnaires. Since the study aimed to analyze the employees‟ behavioral under the unstable working environment the researcher tried hard to focus the target sample to the employees who have been through M&A process and still worked in the remained companies in Taiwan.
Until September, 2013, over 2,000 M&A cases happened in Taiwan and had created over 900 billion NT Dollars of transaction amounts which seemed to be a huge pool for researchers. However, under the M&A circumstances, lots of employees left the job willingness or unwillingness, and the enterprises were likely to regard the M&A strategy as confidential, which can made it a difficult task for researcher to approach large amount of the participants. The researcher approached to the employees or the managerial level of those companies which were known to have the experience of the M&A process without any industry constraint and requested them for helping the delivery of the paper based questionnaire. Also, researcher designed the on-line questionnaire and spread by the social networking website and board bulletin system to do the snowball sampling strategy.
The sample of this study was 277 in 510 responses. The researcher delivered 550 paper-based questionnaires to 13 companies which were known to have the experienced of M&A process and got 383 questionnaires back, the response rate is 70%, and 247 respondents (45%) fit the sample of the study. The detail of the distribution was listed in Appendix A. In addition, the researcher got 127 responses through on-line questionnaires and 30 responses fit the sample of this study.
Measurement Psychological Contract Breach
This study adopted the psychological contract breach and violation scale which was done by Robinson and Morrison (2000). The Cronbach alpha for the origin scale was .92.
Seven -point Likert scale (ranging from 1= „strongly disagree‟ to 7= „strongly agree‟) was used in this study to measure how well one‟s contract has been fulfilled by one‟s employer and the level of feeling violated. The sample item of perceived contract breach is as follows:
„I have not received everything promised to me in exchange for my contributions.‟ „I feel betrayed by my organization‟. A higher score represents the higher level of psychological contract breach.
Perceived Organizational justice
This study adopted the scale based on Moorman (1991). Twenty items with three dimensions are in this scale, including the distributive justice, formal procedures justice, and interactional justice. The reported reliabilities were all above .93. The respondents answered by the five-point Likert scale, which measures the extent to how they agree or disagree. The sample item of perceived organizational justice is as follows: „My work schedule is fair.‟ „I
think that my level of pay is fair.‟ The higher score is regarded as higher perceived organizational justice.
Job involvement
This research adopted the Job Involvement Questionnaire (JIQ) which was done by Kanungo (1982). The Cronbach alpha was .87 on the JIQ. The respondents answered the five-point Likert scale to measure the extent of how they agree or disagree to 10 questions.
The sample item of job involvement is as follow: „the most important things that happen to me involve my present job.‟ „I am very much involved personally in my job.‟ A higher score is regarded as higher level of job involvement.
Control variables
Looking through the previous literatures, many researchers specified some demographic variables that can influence job involvement. Therefore, the demographic variables were gathered through the personal information part of the questionnaire. In this study, according to the previous studies, we adopted the following demographics and chose company type (organization), job position, and annual salary as the control variables.
Company type. Brown (1996) stated that different type of organization form different
organizational culture and influenced employees‟ job involvement. The participants were asked to indicate the most proper way to identify their company type. Answering “domestic company” was coded as “1”, “foreign company” was coded as “2”.
Job position. Mckelvey and Sekaram (1977) proposed that higher position (manager) will
have higher job involvement. The participants were asked to indicate their job position during the M&A process happened. Answering “general employee” was coded as “1”, “junior manager” was coded as “2”, “middle manager” was coded as “3”, “senior manager” was
Annual salary. Ruh, White and Wood (1975) stated in their study that people have higher
salary will form higher job involvement. In this study, in order to make sure the author can compare the participants‟ salary in the same scale, the participants were asked to indicate the best way to describe their annual salary level. Answering “below NTD 500,000” was coded as “1”, “500,000 NTD~ 700,000 NTD” was coded as “2”, “700,000 NTD ~ 800,000 NTD”
was coded as “3”, “800,000 NTD ~ 1,000,000 NTD” was coded as “4”, and finally, “over 1,000,000 NTD” was coded as “5”.
Questionnaire Design
The questionnaire for this research adopted the scales which were all developed maturely. All of the scale‟s items are valid. The original questionnaires were in English version. Considering the participants of this research are Taiwanese, it is more adequate for them to answer the questionnaire in Chinese version. Therefore, the questionnaire was translated into traditional Chinese version. Since the independent and dependent variables will be answered by the same person and the same period of time, it may cause some common method variance (CMV) problem. To avoid the CMV problem, Author has made some effort on it.
First of all, based on the suggestion mentioned in the Common Method Variance in behavioral research: A critical review of the literature and recommended remedies (Podsakoff, MacKenzie, Jeong, & Podsakoff, 2003), the purpose and the variables in this study was not explained in the questionnaire. Second, the researcher adopted the reverse items in the questionnaire to prevent the respondents from filling the questionnaire randomly. Third, different point Likert scale was used in different variables. For instance, psychological contract breach adopted seven-point Likert scale while perceived organizational justice and job involvement adopted five-point Likert scale.
In order to enhance the response rate of the questionnaire, the researcher told the participants that the questionnaires are anonymous and we will not reveal the exact name of the company they represents. The researcher also provided the cash coupon to some participants and held the raffle draw on the on-line survey.
Translation and Back Translation
The original questionnaire was written in English. Since the participants in this research are Taiwanese, the researcher translated the questionnaire into Chinese. In order to make sure the Chinese translation maintain the original meaning of English, the author conducted the back translation.
The researcher found two friends, who have good English proficiency to help the translation and back translation. One of them was asked to translate the English version questionnaire into Chinese and another one was asked to back translate the Chinese version questionnaire into English. The main purpose is to ensure whether the Chinese versions are in the same meaning with the English version or not. The researcher also asked the thesis advisor to confirm both versions to make sure the words are translated into the consistent meaning.
Pilot Test
This research is a quantitative approach. The study did the pilot test before collecting the data. The researcher distributed the pilot questionnaire to 55 people to check the reliability of the questionnaire and examine the relationships between all variables. The researcher delivered 35 questionnaires to three target companies and 20 from the on-line questionnaire.
The reliability was tested by Cronbach alpha value. In pilot test of this study the Cronbach‟s alpha for each instrument were: 0.93 for psychological contract breach, 0.97 for
all three main variables were reliable with an acceptable cronbach‟s alpha value of above 0.7 (Nunnally & Berstein, 1994).
Data Analysis
This quantitative study was constructed by delivering and analyzing the questionnaire.
We use Statistical Package for the Social Sciences (SPSS) 22 version to analyze the data we collected from our target group. The analysis method included t-test, reliability and validity, descriptive statistics, Pearson correlation analysis confirmatory factor analysis and hierarchical regression analysis.
Two sample t- test
For this research, the researcher found it significant that people under M&A or not makes different to their job involvement. The researcher compared job involvement of another group of employees that had no M&A experience in their career with the participants in this study. As respected, the job involvement was significant difference between two groups of participants. In addition, the author also do the two sample t-test of the PCB and POJ, it showed no significant between two groups of participants. It indicated that employees under different kinds of circumstances perceived same feeling of psychological contract breach and organizational justice in their workplace, but the employees in different circumstances have different job involvement. The result of the t-test is listed in appendix B.
Reliability
The reliability was the way to test the internal consistency and stability. Reliability is a concept that determines the precision of the measurements. The degree of precision can be indicating by Cronbach‟s alpha coefficient to show the stability and consistency of measurements. The higher value of cronbach alpha means a higher consistency of each
Cronbach alpha value is over 0.7. In this study, the Cronbach‟s alpha of psychological contract breach is 0.94. The Cronbach‟s alpha of perceived organizational justice serving as a mediator was .98. and the three dimension are all above 0.94. The Cronbach‟s alpha of job involvement was .90. The detail was listed in Table 3.1.
Table 3.1
Reliability Test Results
Variables Number of items Cronbach‟s Alpha
Psychological contract breach 9 .94
Perceived Organizational Justice 20 .98
Job involvement 10 .90
Descriptive statistics
The study adopted descriptive statistics (means, standard deviations, percentages and frequencies) to describe the characteristics of the participants. By doing the percentage and frequencies report, we knew the participants‟ distribution pattern. Therefore, the descriptive statistics in this study contained the result of the control variables information which includes company type, job positions and annual salary.
Pearson Correlation Analysis
The study adopted Pearson Correlation Analysis to see the relationship between variables. By the method, we knew the positive and negative relationships among the variables. The Pearson‟s r ranged from -1 to +1. A correlation of +1 represents the perfect positive linear relationship between two variables while -1 represents the perfect negative linear relationship between them. A correlation of 0 means there is no linear relationship between two variables. What‟s more, different correlation coefficient means different level of
correlation. (e. g., below .4 of correlation coefficient means low correlation, .4 to .7 is medium correlation, and above .7 means high correlation).
Confirmatory Factor Analysis (CFA)
Before testing the hypothesis, confirmatory factor analysis (CFA) was conducted to assess the distinctiveness of the measurement used in the study by using AMOS 20. The validity of the questionnaire will be tested through CFA to confirm whether the data suit a hypothesized measurement model based on theory or previous researches. This research adopted indicators that had been proposed in previous study to examine the construct validity of the psychological contract breach, job involvement and perceived organizational justice.
The most commonly test is the Chi-square (χ²) test, which is to determine the whole model fit and shows the difference between expected and observed covariance matrices. The values lower than 3 infer a good model fit (Jöreskog & Sörbom, 1993). Since the Chi-square test can have a better power in a sample below 200 but with some bias (MacCallum, Browne,
& Sugawara, 1996), some other statistical tests will also be used in this study to check the goodness of fit of the theoretical value and experimental value. Root Mean Square error of approximation (RMSEA), which is used to determine the fit in the large sample sizes. When RMSEA value below 0.08 implied good model fit. The value range from 0.05 to 0.1 shows a medium model fit and the value higher than 0.1 indicates a bad model fit (Browne, Cudeck, Bollen, & Long, 1993). Comparative Fit Index (CFI) and Tuckers-Lewis index (TLI)/Non Norm Fit Index (NNFI) takes sample size into account based on Normed Fit Index (NFI).
NFI, represents a ratio of the difference in the value for the model and the independent model.
If these three values are higher than .9, it can be a desirable index (Bentler and Bonett, 1980;
Tucker and Lewis, 1973). Finally, when the value of Goodness of Fit Index (GFI) higher than .8 indicate the observed data fit in the experimental distribution well (Carmines &
Table 3.2
Summary of Fit Indexes
Fit Indexes Threshold References
Chi-Square/ df <2 or 3 Jöreskog and Sörbom (1993)
RMSEA <.08 Browne et al. (1993);
CFI >.9 Bentler and Bonett (1980)
NFI >.9 Bentler and Bonett (1980)
TLI/NNFI >.9 Tucker and Lewis (1973)
GFI >.9 or >.8 Carmines and McIver (1981)
Note. Author organized the above table.
In our analysis, the results of CFA for each variable were shown in Table 3.3. The perceived organizational justice presented good fit indexes while some of the fit indexes, especially in χ²/df and RMSEA, in psychological contract breach and job involvement are not successfully to reach the standard proposed by Byrne (1998).
Table 3.3
Results of Confirmatory Factor Analysis (N=277)
χ² df χ²/df RMSEA CFI NFI IFI TLI GFI
PCB 147.062 25 5.64 .13 .95 .94 .95 .93 .91
POJ 404.768 167 2.39 .07 .97 .94 .97 .96 .87
JI 172.595 33 4.94 .12 .91 .90 .92 .88 .89
Notes. RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index;
NFI = Normed Fit Index; IFI = Incremental Fit Index; TLI = Tucker–Lewis Index, GFI = Goodness of Fit Index.PCB represents psychological contract breach, POJ represents perceived organizational justice, and JI represents Job Involvement.
Hierarchical Regression Analysis
Based on Baron and Kenny (1986), the moderation effect can test by hierarchical regression analysis. We adopted this method to examine the relationships between variables
variable, hierarchical regression analysis was used to examine the hypothesis in this study to see how the independent variables impact the dependent variables.
Besides, the moderating effect was also tested by hierarchical regression analysis. Three steps were done to test the moderating effect of perceived organizational justice and its three dimensions between psychological contract breach and job involvement. The first step was to enter the control variables. Next, the psychological contract breach was entered to see the impact on the dependent variable. Finally, perceived organizational justice (moderator) was entered each time in order to test the moderating effect.
Common Method Variance Test
Since all measures used in the current study were completed by single participant and used a similar response format, Harman‟s one-factor test (Podsakoff, et al., 2003) was conducted to detect potential bias caused by common method variance (CMV). If a single factor emerges or one general factor explains most of the covariance in the independent and dependent variables, it is reasonable to conclude that a significant CMV is present (Podsakoff et al., 2003). We entered all scale items into a principal components analysis and examined the unrotated factor solution. Six factors possessing an eigenvalue greater than 1.0 emerged, which accounted for 76.11% of variance. The first factor accounted for 47.26% of variance, which showed that the items loaded on a general single factor. Although it is not greater than 50% which mentioned form Podsakoff et al. (2003), but it still can be considered to have some CMV problems.