2.1 Theoretical considerations on business performance
A crucial parameter expressing business performance is growth of productivity. The most important factors affecting improved productivity include employees' job satisfaction and personal relationship among them - both peer and hierarchical relationship. The reward factors include training, education and salary.
Geweke(1982, 1984), using linear feedback system, analyzed the relationship between productivity and salary. Milliea(1998, 1999) adopted Geweke's linear feedback and conducted a similar research. This linear feedback method separates linear dependence between the two time series factors into bi-directional and enables us to examine contemporaneous association as well as the effect of one on another.(Geweke. 1982) Friendlander classified factors of job satisfaction into 3 categories: social/technical environment, fundamental job aspect and stability through development(Friedlander. 1963), whereas Vroom claims that the factors are supervision, work group, job contents, salary, opportunities for promotion and work hours.(Vroom. 1964) Locke, on the other hand, lists 9 factors: job itself, salary, promotion, supervision, stability, benefits, job conditions, coworkers and management policies.(Locke. 1973) Adlerfer classifies the factors into 5 categories: salary, fringe benefits, respect from superiors, respect from coworkers and growth.(Adlerfer. 1967) Education/training varies from different industries; it shows very big time series difference. The importance of training/education is much bigger in the industry that involves rapid and big change in technology. The difference in training among the subcategories of the manufacturing industry is also remarkable in every sense. Black, S. and Lynch, I.(1996) compared the productivity of U. S. manufacturing industry at two distinct times and identified the significant relationship between education/training and productivity. They also showed that education/training at one point will have a positive effect on the productivity at another point after a certain period of time. It is essential that each company investigate into the relationship between productivity and job satisfaction on one hand and organization-internal elements such as work characteristics, employees' achievement directional, job environment(peer relationship, hierarchical relationship and supervision), reward factors(salary, promotion and education/training) on the other hand. The identified relationship will help the businesses establish their strategies to improve their competitiveness in the rapidly changing and developing industry in the world.
2.2 Research hypotheses
As mentioned above, the present research aims to present helpful and practical suggestions to help automobile manufacturers keep developing from the survey intended to identify how job satisfaction and productivity improvement are influenced by a variety of factors including accomplishment-orientation, work characteristics, job environment(communication, hierarchical relationship, peer relationship, supervision and training/education), reward factors(salary and promotion). To that purpose, we have set up 8 hypotheses for test as follows.
Hypothesis 1 : Each member's work-orientation will affect his work performance.
Hypothesis2 : Each member's work characteristics will affect his salary.
Hypothesis 3 :Training/education will affect the vertical factor of the work environment.
Hypothesis 4 : Each member's work characteristics will affect the horizontal factor.
Hypothesis 5 : The horizontal factor will affect job satisfaction and productivity improvement.
Hypothesis 6 :salary will affect job satisfaction and productivity improvement.
Hypothesis 7 : Each member's work characteristics will affect the vertical factor.
Hypothesis 8 : The vertical factor will affect the horizontal factor.
2.3 Empirical analysis and hypotheses testing
The current paper reports the results of the survey conducted to analyze the factors affecting job satisfaction and productivity improvement. Table 1 summarizes the organization of the survey. To test the research hypotheses and models, we evaluated the reliability and validity of the collected data and analyzed structural equation models using Amos of SPSS.
Table 1. Survey questions
2.3.1 Reliability and validity of themeasured variables
We have performed a factor analysis displayed in Table 2 in order to group the common properties of the determinant factors of vertical factor, education/training, horizontal factor, achievement directional and 24 variables relating job characteristics. The analysis shows that only 7 factors whose total dispersion is up to 75 % are selected. Table 3 also illustrates the grouping of the factors varimax method into 7 groups. The 7 groups are named as follows: Vertical Factor, Horizontal Factor, Salary, Achievement Directional, Work Characteristic, Education/training and Job Satisfaction/Productivity Improvement in that order. These 7 groups were set up as the variables of the structural equation model.
We use the Varimax method in order to eliminate the problem of multicollinearity that might result from the independence of the factors by regressing those variables.(Noh. 2002) The result of the factor analysis of the validity of the seven Group factors with 24 questions, it shows that all the items under measurement are within their factors
We have regressed factors and performed a factor analysis for the total of 24 items and found that all the items of measurement are involved in the original set of factors and that 74.18% of total dispersion are accounted for. Thus, it might be safe to conclude that the result of the factor analysis well meets the requirements for the validity of convergence and differentiation of variables.
Table 2. Total dispersion against variables
To test the reliability of the grouped variables we have measured Cronbach's alpha indexes of the variables Table 4 exhibits the results and figures of each item under measurement. There might be various methods to analyze the reliability of items. If we use a scale of many items, we usually calculate the split-half reliability for each item and the Cronbach's alpha that represents their average.<Table 4> show that all the values range from 0.7 to 0.8, which accounts for their good reliability.
Table 3. Constituent matrix of variables
Table 4. Cronbach's alpha ofvariables
2.3.2 Analysis of the model and test of hypotheses
2.3.2.1 Analysis of the model
We have performed an analysis of the structural equation model using the variance-covariance matrix maximum likelihood method in order to test the set of research hypotheses and the research model. The overall suitability of the established model can be tested by calculating a set of goodness of fit tests. As illustrated in Table 5, GFI(Goodness of Fit Index) represents how well a given model explains the variance and covariance. In general, a given model is considered very good if the figure is greater than 0.9. Thus, the current model can be counted as 'good', since its GFI is 0.891.
Code numbers of the table5 are Survey question number(table 1) We have examined a set of models and decided on the structural equation model displayed in Figure 1. The path coefficient of exogeneous variables and endogeneous ones is summarized in table 5. These variables are shown to exert reciprocal influence to one another.
Table 5. Path coefficient of the structural equation model
Figure 1. The Path Coefficient of the structural equation model
Table 6. Regression Weight of the Structural Equation Model
2.3.2.2 Test of hypotheses
To test hypotheses we appeal to the standard path coefficient and t-values using Amos of SPSS. The main results of the analysis are summarized in Table 6 and below.
First, a alternative hypothesis can be adopted for Hypothesis 1, since t≥1.96, which shows that an individual's achievement directional affects his or her work characteristics.
Second, a alternative hypothesis can be adopted for Hypothesis 2, since t≥1.96, which means that an individual's work characteristic affects his or her salary.
Third, a alternative hypothesis can be adopted for Hypothesis 3, since t≥1.96, which shows that education/training influences the vertical factor.
Fourth, a alternative hypothesis can be adopted for Hypothesis 4, since t≥1.96, which shows that an individual's work characteristic also influences the horizontal factor.
Fifth, a alternative hypothesis can be adopted for Hypothesis 5, since t≥1.96, which shows that the horizontal factor will affect job satisfaction and productivity improvement.
Sixth, a alternative hypothesis can be adopted for Hypothesis 6, since t≥1.96, which shows that reward factors(promotion and salary) affect productivity improvement.
Seventh, a alternative hypothesis can be adopted for Hypothesis 7, since t≥1.96, which shows that an individual's work characteristic influences the vertical factor.
Finally, a alternative hypothesis can be adopted for Hypothesis 8, since t≥1.96, which shows that the vertical factor influences the horizontal factor.