THE EFFECT OF SOCIAL SECURITY ON EMPLOYEE PERFORMANCE WITH JOB SATISFACTION AS INTERVENING VARIABLES (Study from Administrative Section of Texas Public Health Facilities)

This study aims to determine “The Effect of Social Security on Employee Performance with Job Satisfaction as an Intervening Variable (Case Study in the Administrative Section of PT. Auto Padu Kuala Tanjung Batubara)”. The results of this study indicate that 1) it can be seen that tcount (7.314) > ttable (1.685), likewise with a significance value of 0.00 <0.05 it can be concluded that the first hypothesis is accepted, meaning that the social security variable(X) positive and significant effecton job satisfaction (Y1). 2) it can be seen that tcount (2.593) > ttable (1.685), and a significance value of 0.01 <0.05, it can be concluded that the second hypothesis is accepted, meaning social security(X) positive and significant effecton employee performance (Y2). 3) it can be seen that tcount (5.131) > ttable (1.685), and the significance value is 0.00 <0.05, it can be concluded that the third hypothesis is accepted, meaning job satisfaction (Y1)positive and significant effecton employee performance (Y2). 4) the direct effect of variable X on variable Y2 is 0.313. While the indirect effect through the Y1 variable is 0.765 x 0.619 = 0.473, the calculation results obtained show that the indirect effect through the Y1 variable is greater than the direct effect on the Y2 variable.


INTRODUCTION
The company's goals are said to be achieved not only depending on modern equipment, adequate facilities and infrastructure, but more depending on the human resources who carry out the work. The achievement of an organization is strongly influenced by the individual performance of its employees. Every corporate organization must always spur the performance of its employees in the hope of being able to achieve harmony in every part of the company, so that the expected goals are achieved. Performance is the result of work that can be achieved by a person or group of people in an organization, in accordance with the authority and responsibility of each in order to achieve the goals of the organization concerned legally, not violating the law, and in accordance with morals and ethics (Rachmawati, 2009) in (Hidayah, 2016). One way to spur employee performance in an organization or company is to further improve employee performance optimally, such as providing compensation, holding job training for new employees, getting special attention for employees with achievements such as giving right to social security to be able to meet the basic needs of a decent life and increase their dignity towards the realization of a prosperous, just and prosperous Indonesian society; and to provide comprehensive social security, the state has developed a National Social Security System for all Indonesian people.

Job satisfaction
According to Edy Sutrisno (2014: 73) Job satisfaction is an interesting and important issue, because it has proven to have great benefits for the interests of individuals, industry and society. For individuals, research on the causes and sources of job satisfaction allows for efforts to increase their happiness in life. For the industry, research on job satisfaction is carried out in an effort to increase production and influence costs through improving the attitudes and behavior of its employees. Furthermore, society will certainly enjoy the maximum capacity of the industry as well as increased human values in the context of work. With job satisfaction, an employee can feel whether his work is pleasant or unpleasant to do. Satisfied workers also tend to engage in organizational behavior that goes beyond their job and role descriptions, and helps reduce the workload and stress levels of members in the organization. Dissatisfied employees tend to be resistant in relation to leadership and engage in a variety of behaviors that are counterproductive.

Employee performance
Performance is the result of work that can be achieved by a person or group of people in an organization, in accordance with the authority and responsibility of each in order to achieve the goals of the organization concerned legally, not violating the law, and in accordance with morals and ethics (Rachmawati, 2009) in (Hidayah, 2016. One way to spur employee performance in an organization or company is to further improve employee performance optimally, such as providing compensation, holding job training for new employees, getting special attention for employees with achievements such as giving awards, and other forms of attention to all. his employees. The existence of activities will greatly affect the provision of compensation.

Data Types and Sources 1. Data Type
According to Sugiyono (2015), the types of data are divided into 2, namely qualitative and quantitative. This study uses data types in the form of qualitative and quantitative. a. Qualitative Data Qualitative data according to Sugiyono (2015) is data in the form of words, schemes, and pictures. The qualitative data of this research are the names and addresses of the research objects b. Quantitative Data Quantitative data according to Sugiyono (2015) is data in the form of numbers or qualitative data that is numbered.

Data Source
According to Sugiyono (2012: 193) the types of data are divided into two, namely: a. Primary data is a data source that directly provides data to data collectors. In this study, the primary data was in the form of data from questionnaires and interviews conducted by researchers. b. Secondary data is a source that does not directly provide data to data collectors, for example through other people or through documents.

Data collection technique
The data collection technique used is by:

Questionnaire
In this questionnaire, a closed question model will be used, namely questions that have been accompanied by alternative answers before so that respondents can choose one of the alternative answers. The processing of data in this study uses a Likert Scale.According to Sugiyono (2013: 132) "Likert scale is used to measure attitudes, opinions and perceptions of a person or group of people about social phenomena". which has been filled in by the respondent needs to be scored. The following is the weight of the rating on the Likert scale.

Interview
According to Sugiyono (2015: 231) interviews are a data collection technique if the researcher wants to conduct a preliminary study to find problems that must be studied, but also if the researcher wants to know things from respondents that are more in-depth.

Library Studies
Literature study, according to Nazir (2013)

Validity Test
Validity testing using the SPSS version25.00 with criteria based on the calculated r value as follows: a) If r count > r table orr count < -r table then the statement is declared valid. b) If r count <r table orr count > -r table then the statement is declared no valid. This test was carried out on 40 respondents, then df = 40-k = 38, with α = 5%, an r table value of 0.312 was obtained (Ghozali, 2016), then the calculated r value would be compared with the r table value as shown in table 4.5 below :  Table 4.5 shows that all statement points, both the social security variable (X), employee performance (Y2) and job satisfaction (Y1) have a higher r count than the r table value, so that it can be concluded that all statements for each variable are declared valid .

Reliability Test
Reliability is an index that shows the extent to which a measuring device can be trusted or relied on. According to Sugiyono (2013) A factor is declared reliable if the Cronbach Alpha is greater than 0.6. Based on the results of data processing using SPSS 25.00, the following results are obtained:   (2020) Based on the reliability test using Cronbach Alpha, all research variables are reliable/reliable because Cronbach Alpha is greater than 0.6, the results of this study indicate that the measurement tools in this study have fulfilled the reliability test (reliable and can be used as a measuring tool).

Test the Classical Assumptions of Equation 1
The testing of the classical assumptions with the SPSS 25.00 program carried out in this study includes: a. Normality test The Normality Test aims to test whether in the regression model, the confounding or residual variables have a normal distribution (Ghozali, 2016). Data normality testing can be done using two methods, graphics and statistics. The normality test for the graphical method uses the normal probability plot, while the normality test for the statistical method uses the one sample Kolmogorov Smirnov test. The normality test using the graphical method can be seen in the following figure:

Figure 4.1 Normal P Plot
Data that is normally distributed will form a straight diagonal line and plotting the residual data will be compared with the diagonal line, if the distribution of the residual data is normal then the line that describes the actual data will follow the diagonal line (Ghozali, 2016). The test results using SPSS 25.00 are as follows:  Source: Data processed from attachment 4 (2020) From the output in table 4.7 it can be seen that the significance value (Monte Carlo Sig.) of all variables is 0.875. If the significance is more than 0.05, then the residual value is normal, so it can be concluded that all variables are normally distributed.

b. Heteroscedasticity Test
The heteroscedasticity test aims to test whether from the regression model there is an inequality of variance from the residuals of one observation to another. A good regression model is one that has homoscedasticity or does not have heteroscedasticity. One way to detect the presence or absence of heteroscedasticity is with the Glejser test, in the glejser test, if the independent variable is statistically significant in influencing the dependent variable then there is an indication of heteroscedasticity occurring. Conversely, if the independent variable is not statistically significant in influencing the dependent variable, then there is no indication of heteroscedasticity. This is observed from the significance probability above the 5% confidence level (Ghozali, 2016

Simple Linear Regression Testing
Multiple linear regression testing explains the role of social security (X) on job satisfaction (Y1). Data analysis in this study used multiple linear regression analysis usingSPSS 25.0 for windows. The analysis of each variable is explained in the following description: Source: Data processed from attachment 4 (2020) Based on these results, the multiple linear regression equation has the formulationY1 = a + b1X + ɛ, so the equation is obtained: Y1 = 2.165 + 0.898 X + ɛ The description of the multiple linear regression equation above is as follows: a. The constant value (a) of 2.165 indicates the magnitude of job satisfaction (Y1) if social security (X) is equal to zero. b. The regression coefficient value of social security (X) (b1) is 0.898 indicating the large role of social security (X) on job satisfaction (Y1). This means that if the social security factor (X) increases by 1 value unit, it is predicted that job satisfaction (Y1) will increase by 0.898 units.

Coefficient of Determination (R2)
The coefficient of determination is used to see how much the independent variable contributes to the dependent variable. The greater the value of the coefficient of determination, the better the ability of the independent variable to explain the dependent variable. If the determination (R 2 ) the greater (closer to 1), it can be said that the influence of variable X is large on job satisfaction (Y1). The value used in viewing the coefficient of determination in this study is in the adjusted R square column. This is because the value of the adjusted R square is not susceptible to the addition of independent variables. The value of the coefficient of determination can be seen in Table 4.10 below: Source: Data processed from attachment 4 (2020) Based on table 4.10 it can be seen that the value of the adjusted R square is 0.574 or 57.4%. This shows if social security (X) can explainjob satisfaction(Y1) of 57.4%, the remaining 42.6% (100% -57.4%) is explained by other variables outside this research model.

Test the Classical Assumptions of Equation 2
As for testing the classical assumptions with the SPSS program25.00 which was carried out in this study included: a. Normality test The Normality Test aims to test whether in the regression model, the confounding or residual variables have a normal distribution (Ghozali, 2016). Data normality testing can be done using two methods, graphics and statistics. The normality test for the graphical method uses the normal probability plot, while the normality test for the statistical method uses the one sample Kolmogorov Smirnov test. The normality test using the graphical method can be seen in the following figure:

Figure 4.2 Normal P Plot
Data that is normally distributed will form a straight diagonal line and plotting the residual data will be compared with the diagonal line, if the distribution of the residual data is normal then the line that describes the actual data will follow the diagonal line (Ghozali, 2016). The test results using SPSS 25.00 are as follows: Source: Data processed from attachment 4 (2020) From the output in table 4.11 it can be seen that the significance value (Monte Carlo Sig.) of all variables is 0.850. If the significance is more than 0.05, then the residual value is normal, so it can be concluded that all variables are normally distributed.

b. Multicollinearity Test
The multicollinearity test aims to determine whether there is a correlation between the independent variables in the regression model. The multicollinearity test in this study was seen from the tolerance value or variance inflation factor (VIF). The calculation of the tolerance value or VIF with the SPSS 25.00 program for windows can be seen in Table 4.12 below:  Source: Data processed from attachment 4 (2020) Based on table 4.12 it can be seen that the tolerance value of social security (X) is 0.415, job satisfaction (Y1) is 0.415 where everything is greater than 0.10 while the VIF value of social security (X) is 2.408, job satisfaction (Y1) of 2.408, all of which are less than 10. Based on the calculation results above, it can be seen that the tolerance value of all independent variables is greater than 0.10 and the VIF value of all independent variables is also less than 5, so there is no correlation symptom in the independent variables. So it can be concluded that there are no symptoms of multicollinearity between independent variables in the regression model.

c. Heteroscedasticity Test
The heteroscedasticity test aims to test whether from the regression model there is an inequality of variance from the residuals of one observation to another. A good regression model is one that has homoscedasticity or does not have heteroscedasticity. One way to detect the presence or absence of heteroscedasticity is with the Glejser test, in the glejser test, if the independent variable is statistically significant in influencing the dependent variable then there is an indication of heteroscedasticity occurring. Conversely, if the independent variable is not statistically significant in influencing the dependent variable, then there is no indication of heteroscedasticity. This is observed from the significance probability above the 5% confidence level (Ghozali, 2016). The results of data processing using SPSS 17.00 show the results in the following

Multiple Linear Regression Testing
Multiple linear regression testing explains the role of social security (X) and job satisfaction (Y1) on employee performance (Y2). Data analysis in this study used multiple linear regression analysis using SPSS 25.0 for windows. The analysis of each variable is explained in the following description: Source: Data processed from attachment 4 (2020) Based on these results, the multiple linear regression equation has the formulation:Y2 = a + b1X + b2Y1 + ɛ, so the equation is obtained: Y2 = 1.377 + 0.339X + 0.571Y1 + ɛ The description of the multiple linear regression equation above is as follows: a. The constant value (a) of 1.377 indicates the level of employee performance (Y2) if social security (X) and job satisfaction (Y1) are equal to zero. b. The regression coefficient value of social security (X) (b1) is 0.339 indicating the large role of social security (X) on employee performance (Y2) assuming the variable job satisfaction (Y1) is constant. This means that if the social security factor (X) increases by 1 value unit, it is predicted that employee performance (Y2) will increase by 0.339 value units assuming constant job satisfaction (Y1). c. The regression coefficient value of job satisfaction (Y1) (b2) is 0.571 indicating the magnitude of the role of job satisfaction (Y1) on employee performance (Y2) assuming the social security variable (X) is constant. This means that if the job satisfaction factor (Y1) increases by 1 unit value, it is predicted that employee performance (Y2) will increase by 0.571 value units assuming social security (X) is constant.

Coefficient of Determination (R2)
The coefficient of determination is used to see how much the independent variable contributes to the dependent variable. The greater the value of the coefficient of determination, the better the ability of the independent variable to explain the dependent variable. If the determination (R 2 ) the greater (closer to 1), it can be said that the effect of variable X is large onjob satisfaction(Y1).
The value used in viewing the coefficient of determination in this study is in the adjusted R square column. This is because the value of the adjusted R square is not Source: Data processed from attachment 4 (2020) Based on table 4.15, it can be seen that the value of the adjusted R square is 0.764 or 76.4%. This shows that job satisfaction (Y1) and social security (X) can explain employee performance (Y2) by 76.4%, the remaining 23.6% (100% -76.4%) is explained by other variables outside the model this research.

Hypothesis testing a. t test (Partial)
The t statistical test is also known as the individual significance test. This test shows how far the influence of the independent variables partially on the dependent variable.
In this study, partial hypothesis testing was carried out on each independent variable as shown in Table 4.16 below:  Source: Data processed from attachment 4 (2020) a. Hypothesis test of the effect of social security(X)on employee performance (Y2) The form of hypothesis testing based on statistics can be described as follows: Decision Making Criteria: a) Accept H0 If tcount < ttable or -tcount> -ttable orSig value. >0.05 b) Reject H0 If tcount ≥ ttable or -tcount ≤ -ttable orSig. < 0.05 From table 4.17, a tcount value of 2.593 is obtained. With α = 5%, ttable (5%; nk = 38) a ttable value of 1.685 is obtained. From this description it can be seen that tcount (2.593) > ttable (1.685), and its significance value is 0. 01 < 0.05, it can be concluded that the second hypothesis is accepted, meaningsocial Security(X) positive and significant effecton employee performance (Y2). b. Hypothesis test of the effect of job satisfaction (Y1) on employee performance (Y2) The form of hypothesis testing based on statistics can be described as follows: Decision Making Criteria: a) Accept H0 If tcount < ttable or -tcount> -ttable orSig value. >0.05 b) Reject H0 If tcount ≥ ttable or -tcount ≤ -ttable orSig. < 0.05 From table 4.17, a tcount value of 5.131 is obtained. With α = 5%, ttable (5%; nk = 38) a ttable value of 1.685 is obtained. From this description it can be seen that tcount (5.131) > ttable (1.685), and its significance value is 0. 00 < 0.05, it can be concluded that the third hypothesis is accepted, meaningjob satisfaction (Y1)positive and significant effecton employee performance (Y2).

b. Path Analysis
In order to prove that whether a variable is capable of being a variable that mediates the relationship between the independent variable and the dependent variable, a direct and indirect effect calculation will be carried out between the independent variable and the dependent variable. If the indirect effect of the independent variable on the dependent variable through the intervening variable is greater than the direct effect of the independent variable on the dependent variable, then this variable can be the variable that mediates between the independent variable and the dependent variable (Ghozali, 2016). To carry out direct and indirect calculations, it is carried out from the standardized values of the regression coefficients equations I and II as follows:   Furthermore, the value of standardized coefficients beta will be entered into the path analysis image as follows:  Figure 4.3 the path analysis shows the direct effect of variable X on variable Y2 of 0.313. While the indirect effect through the Y1 variable is 0.765 x 0.619 = 0.473, the calculation results obtained show that the indirect effect through the Y1 variable is greater than the direct effect on the Y2 variable. These results can be seen in table 4.20 below: