The Role of Training in Mediating The Effect of Competence, Compensation on Employee Performance

This study aimed to develop a conceptual model of the relationship between competence, compensation, and employee performance hammering training on the performance of employees of public health centres in Indragiri Hulu Regency, Riau Province. Data analysis was carried out using Structural Equation Model (SEM) Smart PLS. The main findings in this study are competency and compensation which affect the employee performance. Training plays a role in mediating the effect of competence and compensation on employee performance. The acceptance of the hypothesis that claimed the role of mediation from training shows the importance of training as one of the strategies in improving employee performance. 127 orang pegawai puskesmas di Kabupaten Indragiri Hulu. Teknik analisa data menggunakan Structrual Equation Model (SEM) Smart PLS. Temuan utama pada penelitian ini adalah kompetensi dan kompensasi berpengaruh terhadap kinerja pegawai. Diklat berperan dalam memediasi pengaruh kompetensi dan kompensasi terhadap kinerja pegawai. Diterimanya hipotesis yang menyatakan peran mediasi dari diklat memperlihatkan pentingnya diklat sebagai salah satu strategi dalam meningkatkan kinerja pegawai.


INTRODUCTION
The public health centre is a technical implementation unit of the Health Office that carries out various programs in the field of health. Problems may emerge due to the minimum employee performance in achieving multiple programs related to the results of work based on quality, quantity, working time, and cooperation to achieve the goals set by the organization (Sutrisno, 2019). Veingerl Cic et al., (2020) is more likely to use the word performance in referring to the word performance expressing results that a person or group can achieve in an organization. Table 1  Efforts to improve performance are qualitative and quantitative results of the employees' performance. An organization is considered good in implementing the program based on the employees' performance level both individually and in groups, which is predetermined by the instruments used to measure the performance. Thus, performance is a reflection of the results achieved by a person or group of people. Schreibauer et al., (2020) stated that individual performance with institutional performance or corporate performance is closely related. In other words, if the performance of employees (individual performance) is good, then the performance of the public health centre (institutional performance) is also good. The performance of public health centre employees in Indragiri Hulu Regency in implementing the program can be seen in Table 1. Based on table 1, it appears that the realization of the UCI (Universal Child Immunization) immunization program from 2016 to 2020 has not reached the target. The achievement of the UCI (Universal Child Immunization) immunization program in 2020 was only 27%. The achievement of cervical and breast cancer early detection programs continuously increases annually. However, the achievement is still too far from the target of the program's accomplishments in 2016 to 2020.
Achievements for the toddler weighed program in Indragiri Hulu District conducted by public health centre also have not reached the target. In 2019 and 2020, there was a decrease in the achievement of the target from 67.13% to 61.13%. The achievement of the toddler program is stated in the provisions of BGM (Below the Red Line) in 2020 of 36.37%. In 2020, for infants aged 0-6 months who breastfeed exclusively did not reach the target with a triumph of 64.83%. Furthermore, based on table 1, we can see that the performance realization of public health centre employees has not reached the target set. The work program has not been achieved at the public health centre in Indragiri Hulu Regency because it faces various problems, including internal problems in the public health centre and issues of community participation in supporting public health.  Dignath, 2021;Harangus et al., 2020;Tukhtamishevich, 2021 The results of research on the influence of Compensation on training Effect Leonu, 2017;Parmin, 2017Mas'adah et al., 2020 There are differences in the results of research on the influence of Competency on employee performance Effect Rai & Nandy, 2021;Berberoglu, 2018;Araujo & Taylor, 2012;Yang et al., 2012;Shah & Prakash, 2018;Tognazzo et al., 2017;Sains, 2019;Ren et al., 2015;Ataunur & Ariyanto, 2016;Hassanin, 2014;Rahman, 2015;Indriani, 2018No effect Ahrul, 2013Ratnasari, 2016;Adam & Kamase, 2019 There are differences in the results of the study on the effect of compensation on the performance of employees Effect Marín & Sánchez, 2003;Appelbaum & Mackenzie, 1996Chong & Law, 2016Ren et al., 2014;Gilles et al., 2016 ;Do, 2016;Yang, 2016;Baledi & Saed, 2017;Jean et al., 2017;Sukidi & Wajdi, 2017;Ramadhan et al., 2015;Komara & Neliawati, 2014;Wardhana, 2018No effect Rizal et al., 2014Anthonia. et al., 2019;Hameed et al., 2014 The research gaps, based on the mapping of the research conducted, describe on table 2. Based on the table 2 , previous research showed that a group of researchers found that the variables of competency and compensation significantly affected the performance, but several other studies found that both competency and compensation variables had no significant effect on performance. Based on the controversy of the results of this study, it issued the truth and the existing belief that competence and compensation variables affect the employee performance. The description of the phenomenon and empirical studies that have been unexplained about the research problem is the unachieved target of nine work programs set by the centre of public health caused by the lack of maximum performance of the employees.
Researchers conducted theoretical studies and justified the social exchange theory as an umbrella theory to solve research problems, which is the performance of employees who have not met the target of the work program planned. The Social Exchange Theory is based on the idea that people view their relationships in economy context and they count and compare the sacrifices with the rewards produced by the continuous relationship (West & Turner, 2009). The reason researchers take social exchange theory as an umbrella theory is because social exchange theory states that employees tend to develop high-quality relationships based on who they interact with, how they interact, and how they experience it. The relationship with the problem of this research is when the public health center employees treated in a fair and respectful manner, they tend to consider the relationships in terms of social exchanges rather than economic exchanges. Furthermore, they tend to reciprocate by putting extra effort into work or dedication to more work and are willing to become more involved in work so as to improve their performance. Researchers have not found research that uses training variables as a solution to the research gap between competence, compensation and employee performance. Therefore, the researcher tries to take the place of using the training variable as one of the novelties of solving this research problem. Based on the problem related to the phenomenon of gap and research gap, the conceptual model is as follows: Based on the phenomenon of the gap, research gap, and conceptual framework, the hypothesis obtained are: H1. Competence has a significant effect on employee performance H2. Compensation has a significant effect on employee performance H3. Training mediates the effect of competence on employee performance H4. Activity mediates the effect of compensation on employee performance

METHOD Population and Sample
The population involved in this study was the public health centre employees in Indragiri Hulu Regency, Riau Province, as many as 1266 employees. Zainal (2011:244) stated that there is no absolute prosisions in determining the number of samples that should be involved in a research project, but as an illustration, if the population members are above 1000, then the samples taken were 10-15 percent of the population. Therefore, the sample in this study was 15 percent of 1266 people, which is 127 people. Thus the sampling method used was simple random sampling.

Measurement
This study employed quantitative methods, in which the instrument used was the primary questionnaire data from the respondents. The variables in this study were exogenous variables and endogenous variables. Endogenous variables belong to the category of dependent variables that are affected by exogenous variables or independent variables. In this study, the dependent variables are employee performance (KIN) and Training (DIK), while the independent variables are Competency (KOMP) and Compensation (KOMPEN).
The questionnaire in this study consisted of 38 statements. Employee performance consists of 6 indicators and 12 statements developed from (Riau Governor Regulation, 2019). The compensation consists of 3 indicators and 6 statements developed from Wilson, (2012). The competency consists of 5 indicators and 10 statements designed by (Spencer & Spencer, 1993). Meanwhile, the training consists of 5 indicators measured by ten statements that have been developed by (Rae 2005).

Analysis using Structural Equation Model Partial Least Square (SEM-PLS)
This study used SMART PLS version 3.2.8 software to analyze the data collected. PLS or Partial Least Square is a variant of structural equation modelling (SEM)-based test. PLS has two forms of test models called measurement models and structural models. The measurement model consists of validity and reliability tests, while structural models consists of causality or hypothetical tests with predictive models.
Measurement Model Analysis (Outer Model) consists of convergents' validity based on the correlation between component scores, individual reflexive size is high if correlated more than 0.70 compared to measurable variables. In addition, the Measurement Model Analysis (Outer Model) consists of a Discriminant Validity test calculated based on the cross-loading value of the manifest variable against each latent variable. If the correlation between latent variables and each indicator (manifest variable) is more significant than the correlation with other latent variables, it is better than other latent variables. Construct reliability test was also conducted to see the ave value with composite reliability. AVE in reliability testing is a tool to measure the actual value of the reliability of a construct (Hair et al., 2012). Furthermore, the Structural Model Analysis (Inner Model) consists of 1) Coefficient of Determination (R2), or R Square which is useful to predict and see how much influence the variable X simultaneously contributes (together) to variable Y (Hair et al., 2012); 2) Predictive Relevance (Q2), which measures the observed values resulting from the model and parameters estimated if the values are 0.02 (small), 0.15 (medium), and 0.35 (large) (Hair et al., 2012); 3) Effect Size (f2), which is to find the goodness of fit model. When the f2 value obtained is 0.02, it indicates weak effects of latent variables, f2 value of 0.15 belongs to the category of good effects, while f2 value of 0.35 is a strong attraction (Hair et al., 2012). 4) Hypothesis testing, testing using t-test. If p-value 0.05 (alpha 5%), 0.1 (alpha 10%), and 0.01 (alpha 1%), is considered significant, and vice versa.

FINDING AND DISCUSSION Measurement Model Analysis (Outer Model) a. Convergent Validity Test
The results of the concurrent validity test from the instrument data in this study can be seen in Table 3 below: Based on the concurrent validity test results in table 3, if the loading factor is < 0.5, it will be excluded from the calculation and re-estimated. But if all loading factors have been > 0.5, we can use all indicators for the following analysis step; from table 3, we can see the loading factors are all > 0.5, then all endogenous and exogenous factors that make up the construct are valid. The results of the discriminant validity of the research instrument data can be seen in table 4 below: Table 4 above can know that the model already has a good discriminant validity if the loading value of the latent variable indicator is greater than the correlation of other variables. The cross-loading value in this study on each indicator is more significant than other latent variables, indicating that the variable has good discriminant validity.

Measurement Model Analysis (Inner Model) a. Coefficient of Determination (R 2 )
Based on table 6, the R Square adjusted value for employee performance variables is 0.872, it indicates that the percentage of the effect of competency and compensation variables on employee performance is 87.20%, while the remaining 12.80% is affected by other factors not involved in this study. Furthermore, the value of R Square adjusted for training variable is 0.742, meaning that the percentage of the effect of competency and compensation variables for training is 74.20%, while the remaining is affected by other factors not studied in this study.

b. Predictive Relevance (Q 2 )
The value of Q 2 is equal to the coefficient of determination (R-Square). A Q-Square (Q 2 ) value of 0 indicates the model has sizeable predictive suitability; if the Q 2 value is smaller than 0, it indicates the model has a small predictive fit. In other words, if the Q 2 value is significant, we can say that the model has conformity with the data. The Q2 value is calculated as follows:: Q 2 = 1-(1-R 1 2 )(1-R 2 2 )…(1-R n 2 ) Q 2 = 1-(1-0.872)(1-0.742) Q 2 = 1-(0.128)(0.258) Q 2 = 1-0.033 Q 2 = 0. 967 The results showed that the value of Q 2 is 0.967, which means that the values observed can already be explained by the structural model developed in the study and 0.033 caused by other factors not studied in this study.

c. Effect Size (F2)
Effect size (f2) is the goodness of the model. F2 is used to determine whether the predictor of a variable is at a weak, medium, or strong level structure.
Based on table 7, the effect size f2 of compensation has a weak value towards performance, but competence has a medium f2 effect size on employee performance.

Competency has a Significant Impact on Employee Performance
The hypothetical test results in Table 8 show that the received H1 was due to the small P-Value of Alpha 0.05, which means that competence significantly affects employee performance. The results of this study are in line with the results of other studies that state  (Ramo et al., 2009;Chan, 2006;Zainol & Al Mamun, 2018;Araujo & Taylor, 2012;Shah & Prakash, 2018;Tognazzo et al., 2017;Otoo, 2019;Ren et al., 2014;Ataunur & Ariyanto, 2016;Rahman, 2015;Indriani, 2018;Astarina et al., 2021;Indriani, 2018;Adam & Kamase, 2019). Respondents' assessment of the highest competency variable is in indicators of confidence, values, skills, experience, personality characteristics, and intellectual abilities.
Beliefs and values reflect how public health centre employees have principles, objectives, and social standards in carrying out their duties and responsibilities. In addition, intellectual ability shows the ability of employees to think, reason, and solve problems in carrying out the work. In addition, although academic ability has a high index value, it ranks last in competency indicator, meaning there needs to be an increase in the intellectual capacity of public health centre employees to be able to improve performance.
The highest respondents' assessment is on the salary indicator, followed by incentives and insurance. Salary reflects the regular and fixed rewards earned by public health centre employees every month. In addition, insurance is a risk countermeasure for losses arising from an event. Insurance has a high index value but still ranks last in the compensation indicator, meaning that public health centre employees need to be covered by insurance for health and retirement to improve performance.

Training mediates the influence of Competency on Employee Performance
The hypothetical test results shown in Table 8 show H3 because of the small P-Value of Alpha 0.05, which means training mediates the influence of competence on employee performance. Based on the results of the research, it proves that competence can improve the performance of employees through training. The results of this study are in line with the results of other studies that state that training has a significant effect on employee performance (Athar & Shah, 2015;Olajide & Obialo, 2020;Abogsesa & Kaushik, 2018;Kum , et al., 2014;Khan et al., 2014;Imran & Tanveer, 2015;Mahadevan & Yap, 2019;Motlokoa et al., 2018;Diab, 2015;Ibrahim et al., 2017).
Respondents' assessment of the highest training variable is in the training indicator followed by the instructor's methods, attitudes, skills, facilities, and length of time. The content of the training reflects the program given to the trainees. Besides, the length of time shows how long employees will follow the implementation of the activity. This length of time has an index value with a high category but still ranks last on the trailing indicator, meaning the length of time indicates the length of time of the program provided. In addition, the material delivered can be absorbed well by employees to improve performance.

Training mediating the Effect of Compensation on Employee Performance
The hypothetical test results obtained in table 8 show H4 due to the small P-Value of Alpha 0.05, which means that training mediates the effect of compensation on employee performance. Based on the study results, compensation proves that it can improve employees' performance through training. The results of this study are in line with the results of other studies that state that training has a significant effect on employee performance (Athar & Shah, 2015;Olajide & Obialo, 2020;Abogsesa & Kaushik, 2018;Kum , et al., 2014;Khan et al., 2014;Imran & Tanveer, 2015;Mahadevan & Yap, 2019;Motlokoa et al., 2018;Diab, 2015;Ibrahim et al., 2017).
Respondents' assessment of the highest employee performance variables is on the service orientation indicator followed by commitment, cooperation, integrity, and discipline. The service orientation reflects how public health centre employees interact with patients in carrying out their duties. Besides, the field shows how the employee's obedience follows the rules that apply in the public health centre. The area has a high index value but still ranks last in employee performance indicators, meaning there needs to be an improvement in the field of public health centre employees to improve the employees' performance.

CONCLUSION
Based on the study results, the following conclusions are obtained (1) Competence has a significant effect on employee performance; (2) Compensation has a significant effect on employee performance; (3) Training mediates the effect of competence on employee performance; and (4) Training to mediate the effect of compensation on employee performance.
Based on this research, the public health center in Indragiri Hulu Regency can improve employee performance by enhancing education and training by providing training facilities regularly and evenly to all employees according to the employee's job description. Researchers advise further research to include mediation or other moderation variables to fill the research gap between compensation, competence to employee performance.