Home / Employee

Testing the Linear Relationship Between the Dependent Variable and the Independent Variable.


4.4.1 Model fit testing

Adjusted R2 = 0.675 means that 8 independent variables explain 67.5% of job satisfaction of Vietcombank employees in Ho Chi Minh City area .

Table 4.10: ANOVA table - model fit



Model

Total average

direction


df

Medium

square


F


Sig.

1

Regression

152,459

8

19,057

58,646

0.000 a


Remainder

69,541

214

0.325


Total

222,000

222


Maybe you are interested!

a. Independent: (constant), BC, DK, L_T, LD, PL, DN, DG, TT

b. Dependent: TM


The F-test used in the analysis of variance is a hypothesis test about the suitability of the overall linear regression model. The idea of ​​this test is about the linear relationship between the dependent variable and the independent variables. In the ANOVA analysis table, we see that the sig. value is very small (sig. = 0.000), so the regression model fits the data set and can be used.

Table 4.11: Regression results




Unstandardized coefficient

Standard coefficient

chemical


t


Sig.

Multicollinearity statistics

line

B

Standard error

Beta

Tolerance

VIF


-2.075E-16

.038


.000

1,000



TT

.268

.038

.268

7.009

.000

1,000

1,000

DG

.376

.038

.376

9,839

.000

1,000

1,000

DN

.308

.038

.308

8,046

.000

1,000

1,000

PL

.324

.038

.324

8,478

.000

1,000

1,000

LD

.285

.038

.285

7,442

.000

1,000

1,000

LT

.369

.038

.369

9,653

.000

1,000

1,000

DK

.141

.038

.141

3,695

.000

1,000

1,000

BC

.189

.038

.189

4,942

.000

1,000

1,000

a. Dependent variable: TM


In result 4.11, if sig. < 0.1 equivalent to 90% confidence and |t| > 2 then that factor is accepted, meaning it has an impact on the satisfaction trend. The regression results show that all 8 factors satisfy the condition. The regression coefficients all have positive signs showing that the factors in the regression model have a proportional impact on employee satisfaction.

4.4.2. Testing for multicollinearity

Based on the VIF value <10 (Table 4.11) , the independent variables do not have multicollinearity problems.


4.4.3 Test the linear relationship between the dependent variable and the independent variable.

The assumption to be tested is the assumption of linear relationship. The method used is a Scatterplot with the Standardized residual value on the vertical axis and the Standardized predicted value on the horizontal axis. Looking at the graph of Figure 4.1, we see that the residuals are randomly scattered in a region around the line passing through the zero coordinate without forming any shape. That means the hypothesis of linear relationship is not violated.




Figure 4.1: Scatterplot


4.4.4. Test for constant error variance

To perform this test, we will use the Spearman rank correlation test of the absolute values ​​of the residuals and the independent variables.

The test results show that the Sig. values ​​of all variables are greater than 0.05, so we cannot reject the hypothesis H 0 : the correlation coefficient of the population is equal to

0. Thus, the hypothesis of heteroscedasticity of the errors is rejected, i.e. the assumption

The variance of the error term is not violated.


Table 4.12: Spearman rank correlation coefficient test table






ABSE

Salary

Ladder

colleague

Chat board

Reply

Blade

Conditions

Notable

ABSE

Correlation Coefficient


1,000


-.016


.013


.020


.036


-.067


.105


.023


.079

Sig. (2-tailed)

.

.815

.852

.772

.589

.321

.118

.735

.240

N

223

223

223

223

223

223

223

223

223

4.4.5 Test for normal distribution of residuals

Assume normal distribution and residuals. Use histograms (PP plots) of the (standardized) residuals to test this assumption.



Figure 4.2: Histogram frequency graph


Figure 4.3: PP plot frequency graph


The results from the Histogam plot of the residuals from Figure 4.3 show that the distribution of the residuals is approximately normal (mean = 0, standard deviation Std.Dev = 0.982 is close to 1). This means that the assumption of normal distribution of the residuals is not violated.

The results from the PP Plot frequency chart from Figure 4.3 show that the observation points are not scattered too far from the expected straight line, so we can conclude that the assumption of normal distribution is not violated.


4.4.6 Testing the independence of residuals

Assumption of independence of errors (no correlation between residuals). We use the Durbin-Watson statistic (d) to test. According to the results of the second regression table, the value d=1.789 is in the range of 1.5 - 2.5. This means that d falls into the region accepting the hypothesis of no first-order serial correlation with each other. Therefore, the assumption of no correlation between residuals in the multivariate regression model is not violated.

4.4.7 Impact of factors on satisfaction

Table 4.13. Impact assessment



Standardization factor

Impact (%)

Promotion opportunities

0.27

11.86

Job Evaluation

0.38

16.65

Colleague Relationship

0.31

13.61

Benefits

0.32

14.34

Leader

0.28

12.59

Salary and bonus

0.37

16.33

Working conditions

0.14

6.25

Nature of work

0.19

8.36

Among the factors affecting the level of job satisfaction, the factors "job evaluation" and salary have the strongest impact (accounting for over 16%). The factor "benefits" affects about 14.34% of the change in the level of job satisfaction of bank employees. Next are the factors "colleague relationships", "leadership" and "promotion opportunities" affecting the change in the level of job satisfaction of bank employees from 11.86% to 13.61%. "Nature of work" and "conditions"


"Working conditions" are two factors that have low impact on changes in job satisfaction levels of bank employees (from 6.25% to 8.36%).

4.5 Testing hypotheses in the model:

Table 4.14: Summary of results of tested hypotheses



Symbol

Hypothesis

After inspection


H1

The more satisfied employees are with the bank's salary and bonus system, the more satisfied they are with their jobs.

Accept


H2

The more satisfied employees are with training and advancement opportunities, the more satisfied they are with their jobs.

Accept


H3

The more satisfied employees are with their coworkers, the more satisfied they are with their jobs.

Accept


H4

The more satisfied employees are with their working conditions at the bank, the more satisfied they are with their jobs.

Accept


H5

The more satisfied employees are with the nature of their jobs, the more satisfied they are with their jobs.

Accept

H6

The more satisfied employees are with their bank benefits, the more satisfied they are with their jobs.

Accept

H7

The more satisfied employees are with their leaders, the more satisfied they are with their jobs.

Accept

H8

The higher the employee's job performance rating, the more satisfied he or she is with his or her job.

Accept


The initial research model has 8 hypotheses to be tested: H1, H2, H3, H4, H5, H6, H7, H8. All 8 hypotheses suggest a positive relationship between the factors and "General satisfaction". The results after running factor analysis still have 8 accepted factors, however, some variables have been removed from the initial factors because they do not meet the conditions. At this time, the hypothesis is adjusted to 8 new hypotheses to be tested: H1, H2, H3, H4, H5, H6, H7, H8. After testing the multivariate regression, based on the Sig value in the " coefficient " table, with a confidence level of 90%, there are 8 variables remaining corresponding to eight qualified hypotheses with positive values.

4.6 Testing for differences in employee groups based on individual characteristics


Factors affecting job satisfaction of Bank employees will be tested for differences between attributes to satisfaction level.


4.6.1 Gender:


The results of testing the difference in employee satisfaction with a significance level of 10% are shown in the table (Appendix 6) . Thus, the satisfaction level of employees in the Bank does not differ by gender. That is, between male and female employees, the level of job satisfaction is only considered in other aspects.


4.6.2 Marital status


The test results with a significance level of 10% show that marital status affects the assessment of employee satisfaction in the following aspects: job satisfaction, job nature and job evaluation (Appendix 7) . Married people have higher satisfaction levels than unmarried people in all three aspects (Table 4.13). Thus, married people often want a more stable life, which shows that married people are more satisfied with their current jobs than unmarried people.


Table 4.15: Statistical table describing satisfaction level by marital status


Marriage

N

Medium

Standard deviation

Average error

Overall satisfaction

Married

147

3,7704

0.63806

0.05263

Unmarried

76

3,6151

0.66763

0.07658

Salary and bonus

Married

147

3.6825

0.62827

0.05182

Unmarried

76

3,6228

0.65170

0.07476

Promotion

Married

147

3.5551

0.69274

0.05714


Unmarried

76

3,4158

0.83890

0.09623

Colleague

Married

147

3,7789

0.63739

0.05257


Unmarried

76

3,8322

0.62635

0.07185

Nature of work

Married

147

3,6599

0.56465

0.04657

Unmarried

76

3,4430

0.59476

0.06822

Benefits

Married

147

4,0782

0.63354

0.05225


Unmarried

76

4,1546

0.53536

0.06141

Leader

Married

147

3.6735

0.57055

0.04706


Unmarried

76

3,6184

0.62838

0.07208

Condition

Married

147

3,7959

0.57265

0.04723


Unmarried

76

3,7730

0.59467

0.06821

Job Evaluation

Married

147

3,4626

0.64926

0.05355

Unmarried

76

3,2961

0.71147

0.08161

(Source: author's processing)


4.6.3 Age


Test of Homogeneity of Variances: means to check if there is a difference in variance between groups. According to the results, the significance level is chosen as 10%. We see that Sig in the Test of Homogeneity of Variances Table of 4 variables "General satisfaction", "leadership", "conditions" and "job evaluation" are all small.

Comment


Agree Privacy Policy *