Path analysis with structural equation modelling is used to analyze the suitability of the model based on the selected variables. As recommended by Anderson and Gerbing (1988) 19, the measurement model to test the reliability and validity of the survey instrument is analyzed first, and by using AMOS version 20 the path analysis model is analyzed next. This path analysis with the Structural Equation Model (SEM) is the most constructive model while assessing the causal relationship between the variables as well as verifying the compatibility of the model used 20.

Structural equation modelling evaluates whether the data fit a hypothetical model. As per the result, Chi-square statistics with P = 0.001 is less than the recommended value (p=<0.05). Therefore, it demonstrates a good fit for the model. However, according to Joreskog and Sorbom (1993) 21, a sample size of over 200 (300 in this research), could affect Chi-Square statistics to indicate a significant probability level.

However, this model is considered for further interpretation in the goodness of fit measures 22. Common model-fit measures like chi-square/degree of freedom (χ2/df), the comparative fit index (CFI), root mean square error of approximation (RMSEA), the normated fit index (NFI), incremental fit index (IFI), and the Tucker Lewis index (TLI) are used to estimate the measurement model fit. Table 1 illustrates the estimates of the model fit indices from AMOS structural modelling.

According to Gerbing and Anderson (1992) 23, present study the criteria for a suitable model are as follows: RMSEA of 0.05 or lower; CFI of 0.90 or higher; and NFI of 0.90 or higher. The fit between the data collected and the proposed measurement model can be tested with a Chi-Square Goodness of Fit (GFI) test where the likelihood is greater than or equal to 0.9 indicating a good fit 24. The GFI (Table 1) of this study was 0.997more than the recommended value of 0.90 the other measures fitted satisfactorily; AGFI=0.973, RFI=0.960, CFI=1.000, TLI=1.061, IFI=1.006, NFI=0.951 with χ2/df= 9.000, RMSEA=0.001 25, RMR=0.019 and PGFI=0.111 indicates a good absolute fit of the model. The goodness of fit indices supports the model fit and these highlighted indices demonstrate the appropriateness of this structural model.

To test the hypothesis following assumptions are made.

The test of the above-mentioned hypothesis, the SPSS-AMOS result provides thirteen regression weights, unstandardized and standardized factor model structures (Figure 2, Figure 3) for every path and its significance. Table 2 demonstrates the unstandardized coefficients and associated test statistics. The amount of change in the dependent or mediating variable for each unit change in the variable predicting it is symbolized by the unstandardized regression coefficient. Table 2 also illustrates the standardized estimate, standard error (abbreviated S.E.), and the estimate divided by the standard error (abbreviated C.R. for Critical Ratio). Under the column P, the probability value associated with the null hypothesis is displayed.

Figure 2 Figure 3

The variable accessibility to the workplace increases by 1 unit and the health problem of the sanitary workers decreases by -0.131 units. The regression weight estimate, 0.131, has a standard error of about 0.147, predominantly the regression weight estimate by the estimate of its standard error gives z=0.131/0.147 = -0.887. Thus, the regression weight estimate is 0.887 standard errors above zero. The probability of getting a critical ratio is as large as 0.887 in the absolute value is 0.375. However, the regression weight for accessibility to the workplace in the prediction of health problems of the sanitary workers is not significantly different from zero at the 0.050 level (two-tailed). As a result, the hypothesis (H₁) that Accessibility to the workplace is a significant cause of health problems among sanitary workers is not supported.

The results of hypothesis testing for the respective paths are shown in Table 3.

 When the variable Public prejudice goes up by 1 unit and the health problem of the workers rises by 0.566 units. The regression weight estimate, 0.566, has a standard error of about 0.255. Separating the regression weight estimate by the estimate of its standard error provides z=0.556/0.255 = 2.224. In particular, the regression weight estimate is 2.224 standard errors above zero. The possibility of receiving a critical ratio as large as 2.224 in absolute value is 0.026. Therefore, the regression weight for Public prejudice in the prediction of health problems of the workers is significantly different from zero at the 0.026 (0.05%) levels (two-tailed). As a result, the hypothesis (H₂) that Public prejudice has a significant impact on the health problems of the workers is supported.

The component occupational security increased by 1 unit and the health problem of the workers increased by 0.060 units, and it has a standard error of about 0.200. Subtracting the regression weight estimate by the estimate of its standard error gives z=0.060/0.200= 0.302. Consequently, the regression weight for occupational security in the prediction of health problems of the workers is not significant at the levels because the absolute value 0.763 is larger than 0.05 levels (two-tailed). Therefore, the hypotheses (H₃) that occupational security has significant consequences on the health problems of the workers are not supported in the study area.

The variable Insufficiency of tool facilities goes up by 1 unit and the health problem of the workers goes up by 0.013 units. The regression weight estimate, 0.013 has a standard error of about 0.275. The regression weight estimate by the estimate of its standard error gives z=0.013/0.275=0.493.

The opportunity of getting a critical ratio as large as 0.493 and the absolute (p) value is 0.622. Especially, the regression weight for the Insufficiency of tool facilities in the prediction of health problems of the workers is not significantly different from zero at the 0.05 level (two-tailed). Therefore, the hypothesis (H₄) Insufficiency of tool facilities has a significant effect on the health problems of the workers of this study area is not supported.

The variable visiting the hospital is higher by 1 unit and the health problem of the workers increases by 0.210 units and it has a standard error of about 0.085. Taking away the regression weight estimate by the estimate of its standard error gives z=0.210/0.085=2.476. Then, the regression weight for visiting the hospital in the calculation of the health problems of the workers is significant (p=0.013) at the 0.05 level (two-tailed). Hence, the hypothesis, (H₅) that visiting the hospital has significant results on the health problems of the workers is supported.

When the variable accessibility to the workplace has gone up by 1 unit and worker’s level of job satisfaction decreases by 0.056 units. The regression weight estimate, 0.056, has a standard error of about 1.022. Separating the regression weight estimate by the estimate of its standard error provides z=0.056/1.022 = -2.524. Especially, the regression weight estimate is 2.524 standard errors above zero. The possibility of receiving a critical ratio as large as 2.524 in absolute value is 0.012, less than 0.05. Thus, the regression weight for accessibility to the workplace in the prediction of worker’s level of job satisfaction is significantly different from zero at the 0.012 (0.05%) levels (two-tailed). Consequently, the hypothesis (H₆) that Accessibility to the workplace has significant influences on workers' level of job satisfaction in this study region is supported.

The element of public discrimination (prejudice) of the people goes up by 1 unit and the worker's level of job satisfaction goes up by 0.011 units. It has a standard error of about 0.022. Dividing the regression weight estimate by the estimate of its standard error gives z=0.011/0.022=0.522. As a result, the regression weight for public discrimination of the workers in the prediction of worker’s level of job satisfaction is not significant at the 0.05 (p=0.601) level (two-tailed). Thus, the hypothesis (H₇), that public discrimination of people has a significant effect on worker’s level of job satisfaction in this study area is not supported.

The variable occupational safety of the workers rises by 1 unit the worker’s level of job satisfaction increases by 0.011 units and the standard error is about 0.017 units. Separating the regression weight estimate by the estimate of its standard error gives z=-0.011/0.017=-0.659. Accordingly, the regression weight for occupational safety of the workers in the prediction of worker’s level of job satisfaction is not significant (p=0.510) level. Therefore, the hypothesis (H₈), Occupational safety of the workers has a significant role in worker’s level of job satisfaction in the study area is not supported.

The variable insufficiency of tool facilities increases by 1 unit and the worker’s level of job satisfaction decreases by -0.004 units. The regression weight estimate, 0.004, has a standard error of about 0.023, mainly the regression weight estimate by the estimate of its standard error gives z=0.004/0.023 = -0.185. Therefore, the regression weight estimate is 0.185 standard errors above zero. The probability of getting a critical ratio as large as 0.185 in absolute value is 0.853and it is more than 0.05 (two-tailed). But the regression weight for insufficiency of tool facilities in the prediction of worker’s level of job satisfaction is not significantly different from zero at the 0.050 level (two-tailed). Accordingly, the hypothesis (H₉) insufficiency of tool facilities has a significant responsibility on workers' level of job satisfaction is not supported.

The factor variable sanitary workers visiting the hospital has increased by 1 unit and worker’s level of job satisfaction increases by 0.029 units and it has a standard error of about 0.012. Therefore, subtracting the regression weight estimate from the estimate of its standard error gives z=0.029/0.012=2.440. As a result, the regression weight for sanitary workers visiting the hospital in the prediction of worker’s level of job satisfaction is significant. Because the absolute value is 0.015 and it is less than 0.05 (two-tailed). Hence, the hypothesis (H10) that sanitary workers’ visiting the hospital has a significant impact on worker's level of job satisfaction is supported in the study area.

The variable Worker’s health problem goes up by 1 unit the worker’s level of job satisfaction increases by 0.030 units. The regression weight estimate, 0.030, has a standard error of about 0.011. Hence, separating the regression weight estimate by the estimate of its standard error provides z=0.030/0.011 = 2.821. Particularly, the regression weight estimate is 2.821 standard errors above zero. The possibility of receiving a critical ratio as large as 2.821 in absolute value is 0.005 and it is less than 0.05. Thus, the regression weight for the Worker's health problem in the prediction of the worker's level of job satisfaction is significantly different from zero at the 0.005 (0.05%) levels (two-tailed). Therefore, the hypothesis (H₁₁) Worker’s health problem has a significant influence on their level of job satisfaction is supported.

The variable worker's health problem gets higher by 1 unit and their behaviour of alcoholism decreases by -1.264 units and it has a standard error of about 0.834. Hence, the regression weight estimate by the estimate of its standard error gives z=1.264/0.834=-1.516. Therefore, the regression weight for workers' health problems in the prediction of their behaviour of alcoholism is not significant (p=0.130) because the absolute value is higher than 0.05 level (two-tailed). Hence, the hypothesis, (H₁₂, that workers' health problems have a significant impact on their behaviour toward alcoholism is not supported.

The worker’s job satisfaction level rise by 1 unit and their behaviour of alcoholism increases by 43.552 units and the standard error is about 18.656 units. So, separating the regression weight estimate by the estimate of its standard error gives z=43.552/18.656=2.334. As a result, the regression weight for workers' job satisfaction level in the prediction of their behaviour of alcoholism is significant (p=0.020) level, since the absolute value is higher than 0.05 level (two-tailed). Therefore, the hypothesis (H₁₃), that Worker’s job satisfaction level has a significant effect on their behaviour of alcoholism in the study area is supported.