To analyze the influence of Work motivation factor towards job satisfaction and to know which job satisfaction factors that have the highest influence on job satisfaction, writer is using the multiple regression to analyze the influence of each work motivation factors towards job satisfaction. Furthermore, writer will use reliability and validity test and four classic assumption tests to measure the data.
Justification of the Data
Validity and Reliability Test
To justify the data from the questionnaire, writer will use validity and reliability test. The purpose of validity test it know whether the measurement items used able to measure the variable (Ghozali, 2001). ...view middle of the document...
If the histogram shows that the data is normally distributed, it means that the data is normal. The data have normal distribution if the histogram shows a bell shaped curve. This means that the histogram will look symmetric. Normal probability plot is another tool to measure the normality of the data. These tools will plot the data and if the data show a straight diagonal line and this means, the data fulfill the normality assumption. However, the accuracy of the graphic analysis depends on the researcher. Researchers have their own perception of graphic. Thus, writer will use the statistical test to measure normality. Writer will measure the skewness and kurtosis value of the data. If the skewness value is less than -1.0 or higher than +1.0 it means that the data is highly skewed (measures of shape, 1992, p.1). If the value of the excess of kurtosis is more than +1.0 or less than 1.0 it means that, the distribution of the data is not normal.
Autocorrelation test is a test to measure the correlations between residual of two sets of data in different time (Anderson, Sweeney, William, 2008). . This means residual of an observation have an effect on another observation. Autocorrelations problem violated the regression tools assumption. To measure this autocorrelation, Anderson, Sweeney, and Williams (2008) has suggested to use durbin watson test. According to Anderson, Sweeny, and Williams (2008), if the results of the durbin watson shows the value around 2 it means there is no autocorrelations. Results of the durbin watson test that show the value closer to 0 mean that there is positive correlations and if the closer to durbin watson test shows the value closer to 4 it means negative autocorrelations happened.
Multicollinearity is a event when an independent variable have an effect on another independent variable (lind and Marchal and Whaten, 2008). If this multicollinearity happen, researcher are going to have a difficulty to know the influence of the independent variable towards dependent variable. However, According to Lind, Marchal, Whaten (2008) the multiple regression’s ability to explain the variance of dependent variable is not impaired by this multicollinearity problem. Multicollinearity will become a problem if a research wants to measure the relationship between each independent variable to dependent variable because the results of the multiple regressions may not show the true relationship between variable (lind and Marchal and Whaten, 2008). Researchers usually use correlation coefficient to measure whether multicollinearity happens between independent variables. However, Lind, Marchal, Whaten (2008) has pointed out that it is nearly impossible to select variables that are completely unrelated. Lind, Marchal, and Whaten(2008) has suggested that VIF test is a more precise test to measure if multicollinearity problem is available or not. Here is the formula for VIF test: