Stress as a Mediator of Work Engagement and Learning Organization on Employees’ Work Performance

: This study aims to determine and analyze the effect of work engagement and learning organizations simultaneously on work performance through employee stress at Pupuk Kalimantan Timur Corporation. Research sampling with purposive sampling technique, namely, the sample is carried out with the requirements set by the researcher totaling 1,370 employees from various work departments, both at the center and at the branch. This research uses a quantitative approach with Structural Equation Modeling (SEM) - Smart PLS. The results of the study are known how important work performance is as a critical indicator of success for both employees and the company, the importance of understanding stress from concept to application, especially related to how to manage stress, as well as the implementation of work engagement and learning organizations as concepts that affect work performance with stress as a mediator.


INTRODUCTION
Human resources are important assets in an organization.For an organization or company, having productive human resources will create work achievements that have an impact on the continuity and development of an organization or company.In a company there are many different departments and fields that are managed by each employee, with special skills and knowledge mastered, employees will be more optimal in working and advancing a company.With the dedication and loyalty of employees in carrying out their work, as well as utilizing the special skills and knowledge they have with the aim of advancing and developing an organization or company.That's where the company began to appreciate their efforts and hard work by assessing employee performance.Organizations are required to be able to adapt to change, employees must also be able to adapt to change.One way to improve employee performance is to make organizational changes.One of the most important steps in advancing a company is by restructuring the company to maximize company performance.Changes in organizational structure should increase the daily operational workload of employees in the department.This can affect the employee's work performance (Manurung, 2023).
Based on the results of discussions with the team, there is a phenomenon at PT Pupuk Kalimantan Timur where there is an organizational restructuring or structural change in the company.The change is a merger of two departments into the Human Resources and Organization Development Department (PSDMO).Pupuk Kaltim plans to expand its operations, especially in the methanol processing sector.This is indicated by the plan to build a new processing facility and prepare an industrial area in the Pupuk Kaltim industrial area.
This excessive workload can lead to fatigue and pressure, causing stress in employees.According to Cohen, Kessler, and Gordon (1997), stress is a condition in which a person feels pressured by environmental conditions and experiences biological and psychological conditions that affect health.Perception of stress as a positive experience (perceived control) is able to create selfconfidence, overcome pressure and even life difficulties, while negative experiences (perceived stress) are expressed in the form of irritability, depression, anger and nervousness as a form of negative emotions due to the inability to control stress (Cohen et al., 1997).
Based on research conducted by Djauhar, Baso, and Subhan (2022), which states that stress has a very positive effect on work performance.Changes or changes in organizational structure and company policies that often occur in the work environment can affect employee performance or instability of working conditions (Pratama, 2023).The organizational structure distributes roles more clearly and coordinates.The work performance of each employee is different, this is because each employee has different abilities and willingness to carry out their work, but if employees do not have achievements at work, then this will have an impact on the progress of an organization.
Employee work performance is basically the quality and quantity of work obtained by an employee in carrying out his duties according to the responsibilities given (Anggrainy, Darsono, and Putra, 2018).According to Manurung (2023) there are no criteria that can describe a person's work performance in an organization as a whole, precise and complete because work performance is multidimensional.
Based on the urgency and differences in the results of previous studies described above, researchers want to further examine the variables of work attachment and learning organizations and their influence on work performance, so researchers are interested in conducting research with the title "Stress as a mediator of Work Attachment and Learning Organization on Employee Work Performance".

State of the Art and Novelty
Based on the results of research conducted by Purnamasari (2019), it states that learning organizations have a positive and significant effect together with other variables, and based on the results of separate analysis conducted, learning organization variables have a positive and significant effect on employee work performance.Another study conducted by Rumijati (2020) obtained results that showed learning organizations affect employee work performance and the better the implementation of learning organizations will improve employee work performance, knowledge sharing and motivation.Aryanti, Sari, and Widiana (2020) state that employees who have attachment will provide their best performance for the organization and this is a very positive impact on the organization.This statement is evidenced in a study conducted at PT Golden Sari Lampung conducted by Attamimi, Hayati, and Karim (2022), which shows the results that work engagement has a positive effect on employee performance.In another study conducted by Qodariah (2019), in addition to work attachment having a positive effect on employee performance at PT Surveyor Indonesia, it also states that employee performance can be improved by improving work attachment.However, in contrast to the results of research conducted by Guan and Frenkel (2018), which states that work attachment has no effect on employee performance.

Hypotheses
The study presents several hypotheses regarding employee dynamics at PT Pupuk Kalimantan Timur.Hypothesis 1 posits that there is a positive and significant effect of work engagement on employee stress.Hypothesis 2 suggests that there is a positive and significant effect of learning organization on employee stress.Hypothesis 3 examines whether there is a positive and significant effect of job attachment on job performance.Hypothesis 4 proposes that there is a positive and significant effect of learning organization on employee performance.Hypothesis 5 asserts that there is a positive and significant effect of stress on the work performance of employees.Hypothesis 6 investigates whether there is a positive and significant effect of work engagement on work performance with stress as a mediator.Hypothesis 7 explores whether there is a positive and significant effect of learning organization on work performance with stress as a mediator.Finally, Hypothesis 8 considers whether there is a positive and significant effect of work engagement and learning organization on work performance with stress as a mediator.

Research Design
This research design uses a mixed method with a quantitative approach through the Structural Equation Model (SEM) analysis technique.The research plan in this case includes from hypothesis generation to the final analysis stage.Meanwhile, the SEM technique is used to determine and analyze the relationship or influence of several variables (factors and observable).
Hypothesis generation was carried out based on the initial survey conducted and previous literature review.Then a research design was developed to collect data.The next stage is data collection, analysis and interpretation.Through analysis and interpretation, the research statement can be answered.After the analysis and research questions are answered, report writing is carried out.

Research Variables
The study consists of two exogenous variables (work engagement and learning organization), one endogenous variable (work performance) and one mediator variable (stress).

Work engagement
In this research, the scale used is the Utrecht Work Engagement Scale (UWES) from the concept of work engagement according to Bakker and Schaufeli (2004) which consists of 3 aspects, namely: enthusiasm, dedication, absorption.The measuring instrument used used the concepts and theories of Bakker and Schaufeli (2004) and then tested using the Rasch model by Aryanti, Sari, and Widiana (2020).In previous research conducted, these items were answered using a Likert scale from 1-5.Based on the results of the analysis carried out by Aryanti, Sari, and Widiana (2020), 16 out of 18 items were obtained with an alpha reliability coefficient of 0.93.Meanwhile, in research conducted by Prahara and Hidayat (2019), the reliability test had a value of 0.834.

Learning organization
In this research, the Learning Organization uses the concept from Senge (2006), which has 5 dimensions, namely systems thinking, personal expertise, mental models, shared vision, and learning teams.This Learning Organization Instrument consists of 25 items, where these items are measured on a Likert scale from 1-5.Previous research conducted by Sepfita (2023) found a validity test value of 0.313 -0.720 and a reliability test of 0.879; Previous research conducted by Rumijati (2017) had a validity test of 0.503 -0.816 and a reliability test of 0.888.

Work performance
The work performance instrument in this research uses measuring tools and concepts from Manurung (2023).In his research, he explains that work performance has three aspects, namely, characteristic aspects, behavioral aspects, and managerial aspects.In this research, the instrument used to measure the dependent variable or work performance variable has 33 statement items.In previous research conducted, these items were answered using a Likert scale from 1-5.In previous research conducted by Amal (2023), the validity test was 0.311 -0.795 and the reliability test was 0.923.And in previous research conducted by Manurung (2023), the reliability test was 0.972 and the validity test was 0.187.The following is a blueprint of work performance.

Stress
The stress instrument used in this research refers to the concept of stress in the form of a questionnaire, adapted from the Perceived Stress Scale (PSS-10) constructed by Cohen et al. (1997), which consists of 2 (two) dimensions, namely perceived control (item 4,5,7,8) and perceived distress (1,2,3,6,9,10).The instrument in this research is used to measure employee stress variables which consist of 10 (ten) items.These items are measured using a Likert scale, namely 5 = Very Suitable (SS); 4 = Appropriate (S); 3 = Fairly Appropriate (CS); 2 = Not Appropriate (TS); 1 = Very Unsuitable (STS).Next, adding up all the item scores, the resulting scores will be categorized into three.(1) Mild stress with a total score of 0-13; (2) Moderate stress with a total score of 14-26; (3) Severe stress with a total score of 27-40 (Cohen et al., 1983).

Population and Research Sample
In this study, the population used by researchers was all structural employees of PT Pupuk Kalimantan Timur, totalling 1,307 employees from various work departments at the head office and at branch offices.The sampling of this study is based on purposive technique, which means that the determination of the sample is carried out with certain conditions made by the researcher.The conditions include: 1) being an employee of PT Pupuk Kalimantan Timur with work agreement for an indefinite period of time, and 2) employees who are at the head office 3) employees who have worked for more than 3 years.After determining the sample based on the purposive technique, the number of respondents was obtained as many as 205 employees.

Data Collection Techniques
The data collection method in this study is divided into two, namely primary data collection and secondary data.The primary data used are questionnaire methods and focus group discussions (FGDs) conducted online via Zoom Meeting.Secondary data used in this study comes from book literature, scientific articles, journals, theses, previous similar studies, and organizational documents.

Data Analysis Method
The testing stages are carried out by testing the quality of the instrument, descriptive analysis, structural model test, descriptive analysis and hypothesis testing.Instrument quality test consists of reliability validity test.Researchers conducted convergent and discriminant validity testing.
Convergent validity is done by looking at the loading factor (the correlation value of the item score with the construct).The criterion for determining the decision is that if the loading factor value is higher, the more it interprets the factor matrix.The rule of thumb in this validity uses a loading factor value above 0.7 and Average Variance Extracted (AVE) above 0.5, even if the AVE value is above 0.5 it can still be said to be acceptable.Meanwhile, discriminant validity is carried out by looking at the cross loading of measurements on constructs.The decision criteria are if the AVE value of each construct is greater than the correlation between constructs.Meanwhile, reliability testing is carried out by means of composite reliability.Composite reliability measures the true value of the construct confidence value.The criterion is that the Cronbach's alpha or composite reliability value must be greater than 0.7, although the value of 0.6 can still be said to be reliable.
According to Sugiyono (2013), descriptive statistics function in describing or describing research objects through data provided by research subjects.Structural model testing aims to predict the cause and effect between latent variables.Prediction of this causal relationship is done by the bootstrapping process, and the t statistical test parameter.The structural model is evaluated in PLS using the R² of the dependent construct, the path coefficient value to test the significance between constructs in the structural model.The path coefficient value or structural model indicates significance in hypothesis testing.The indicated internal path or model coefficient score must have a t-statistic value higher than 1.96.The structural model in PLS is determined by R² for the dependent variable and the path coefficient value for the independent variable which is then evaluated for significance based on the t-statistic value of each path.Evaluation of the structural model can be seen through several indicators, including Path parameters are values that describe the strength of the relationship between combinations / variables.The signal or direction in the path must be in accordance with the predicted theory.Its significance can be seen from the t-test or critical ratio obtained from the bootstrapping or resampling method.
The coefficient of determination (R²) is the R² value to measure the level of change in exogenous variables on endogenous variables.The higher the R² value, the better the prediction of the research model.Small R² value = 0.19, medium = 0.33, and high = 0.67.Changes in the R² value can be used to see whether the effect of exogenous variables on endogenous variables an intrinsic effect has, which is measured by the effect size f².The acceptable effect size f² values are: 0.02, 0.15, and 0.35 with exogenous variables having small, medium, and large structural level effects.
Goodness of Fit (GoF).Considerations for determining appropriate model parameters according to Haryono, are small GoF = 0.10, medium = 0.25, and large = 0.36.The GoF value can be calculated manually using the formula GoF = √AVE x R² Predictive Relevance (Q²).The provision is that if the Q² value> 0 indicates that the model has a predictive relationship, while if Q² < 0 indicates that the model does not have a predictive relationship.Haryono (2017) says that the manual calculation to find the Q² value uses the formula Q² = 1-(1-R1²) (1-R2²) (....) (1-Rp²) hypothesis testing.This hypothesis test is carried out with a statistical t test and looks at the probability value.If the statistic > t table and the probability value is smaller than 0.05 then the hypothesis.Then the correlation analysis between dimensions is also carried out to determine the significant relationship between the dimensions of exogenous, endogenous and mediator variables.Qualitative data collection was also carried out in the form of FGDs.FGD aims to find the meaning or understanding of a group with a particular discussion.The characteristic of FGD is the interaction between researchers and their research subjects.

Results
Evaluation of the measurement model (outer model) assesses the validity and reliability of the model.The outer model with reflective indicators is evaluated through convergent validity and discriminant of latent construct forming indicators and composite reliability and Cronbach alpha for the indicator block (Ghozali, 2015).Based on the table above, it shows that the stress scale, work attachment, learning organization and work performance have a recommended loading factor of above 0.7, so it can be stated that all items have a good and recommended loading factor.

Convergent Validity Test Results
Based on the measurement model test, the validity results of all variables are obtained by looking at the landing factor > 0.5.The test results are presented in the table as follows: The results of construct testing in the table above obtained the recommended coefficient value which is above 0.5.Thus, this scale has good validity and is recommended.

Composite Reliability Test Results
The reliability test was carried out using the Composite Reliability and Cronbach Alpha tests by looking at the composite reliability or reliability coefficient and Cronbach alpha > 0.7, it means that the construct has good reliability.Likewise with variance extracted > 0.5.Based on the table above, it can be seen that the results of testing composite reliability or reliability coefficients contained in the work attachment variable have a score value of 0.974, the learning organization variable has a value of 0.981, the stress variable with a value of 0.949 and the work performance variable has a value of 0.987 where the recommended coefficient is 0.7.so it can be concluded that the construct has good reliability, so it can be recommended in this research.

Model Fit Evaluation Results
Evaluation of the fit model in this study was carried out using 5 (five) testing models including root mean square error of approximation (RMSEA) with a value of <0.08, normal fit index (NFI) with a value ≥0.9, non-normed fit index (NNFI) with a value ≥ 0.9.

Hypothesis Test Results
Hypothesis testing in this study was carried out to determine whether the research hypothesis proposed in the research model was accepted or rejected.Hypothesis testing can be seen with the standardized solution and T-Value.The following hypothesis test results in this study can be seen in the figure below: For Hypothesis 1, Ho states that there is no effect of Job Attachment on Work Performance, while Ha suggests that there is an influence of Job Attachment on Work Performance.The analysis results show  = -0.11with t = -1.12,where t < 1.96, leading to the acceptance of Ho and rejection of Ha.This indicates that there is no effect of Job Attachment on Work Performance, and Hypothesis 1 is rejected.For Hypothesis 2, Ho posits that there is no effect of Learning Organization on Work Performance, whereas Ha asserts that there is an influence.The analysis reveals  = 0.94 with t = 17.88,where t > 1.96, resulting in the rejection of Ho and acceptance of Ha.Therefore, there is an influence of Learning Organization on Work Performance, and Hypothesis 2 is accepted.Hypothesis 3 states that there is no effect of Job Attachment on Stress (Ho) versus an effect (Ha).The results show  = 0.05 with t = 1.92,where t < 1.96, leading to the acceptance of Ho and rejection of Ha, proving that there is no effect of Job Attachment on Stress and Hypothesis 3 is rejected.Hypothesis 4 suggests that there is no effect of Learning Organization on Stress (Ho) versus an effect (Ha).The analysis results show  = 0.65 with t = 6.46,where t > 1.96, resulting in the rejection of Ho and acceptance of Ha.This indicates an effect of Learning Organization on Stress, and Hypothesis 4 is accepted.Lastly, Hypothesis 5 posits that there is no effect of Stress on Work Performance (Ho) versus an effect (Ha).The results indicate  = -0.01 with t = -0.44,where t < 1.96, leading to the acceptance of Ho and rejection of Ha.Thus, there is no effect of Stress on Work Performance, and Hypothesis 5 is rejected.For Hypothesis 6, Ho states that there is no effect of Job Attachment on Job Performance through Stress, while Ha suggests that there is an effect.The analysis results show  = 0.00 with t = 0.41, where t < 1.96, leading to the acceptance of Ho and rejection of Ha.This indicates that there is no effect of Job Attachment on Job Performance through Stress, and Hypothesis 6 is rejected.Hypothesis 7 posits that there is no effect of Learning Organization on Job Performance through Stress (Ho) versus an effect (Ha).The results reveal  = -0.01 with t = -0.04,where t < 1.96, leading to the acceptance of Ho and rejection of Ha.Therefore, there is no effect of Learning Organization on Job Performance through Stress, and Hypothesis 7 is rejected.For Hypothesis 8, Ho states that the model of the effect of Job Attachment and Learning Organization on Job Performance through Stress does not fit the empirical data, while Ha suggests that it does fit with the empirical data.Based on the fit index results, it is concluded that the value gets the fit criteria, thus Ho is rejected.The conclusion obtained is that Ho is rejected, and Ha is accepted, this proves that the model of the effect of work attachment and learning organization on work performance through stress fits the empirical data and declared Hypothesis 8 Accepted.0.99 Fit 0.9 1.00 Fit 0.9 1.00 Fit 0.9 1.00 Fit

Discussion
Based on the hypothesis testing conducted, it shows that hypothesis 1 states that Ho is accepted, and Ha is rejected, which means that there is no effect of work attachment on work performance.Acceptance of Ho indicates that in the context of this study, work engagement has no significant effect on employee performance at PT. Pupuk Kalimantan Timur.This is in line with the findings of Guan and Frenkel (2018), that work engagement cannot affect employee work performance.
The results of hypothesis 2 test found that Ho was rejected, and Ha was accepted, which indicates that the learning organization has a significant influence on the work performance of PT PKT employees.The better the application of the learning organization concept, it will encourage employees to continue to develop their competencies and abilities to be able to produce more optimal performance with the role of knowledge sharing and motivation (Rumijati, 2020).These results are also in line with research which states that learning organizations have a positive and significant influence on employee performance (Purnamasari, 2019).
The results of hypothesis 3 test obtained in this study show Ho is accepted and Ha is rejected, which means that there is no effect of job attachment on stress.This finding indicates that in this research sample, job attachment is not directly related to the level of stress experienced by PT PKT employees.Whether employees' job attachment is high or low does not have a significant influence on their stress levels.
Hypothesis testing conducted on hypothesis 4 Ho is rejected and Ha is accepted, this result indicates that the learning organization in the Company's organization has a positive and significant influence on the stress of PT PKT employees.Learning organizations require employees to continue to learn, adapt and develop competencies to achieve company targets.This condition can cause work stress among employees if not followed by a supportive work environment.
The results of hypothesis 5 test obtained in this study show Ho is accepted and Ha is rejected, which means that there is no effect of stress on work performance.This finding illustrates that PT PKT employees who experience stress have decreased work performance.The influence between these two variables is in accordance with Cohen et al. (1997) theory, a person who experiences stressful conditions produces physical and emotional reactions in the workplace.
This condition produces physical and emotional reactions in that person, giving rise to 2 aspects in the form of perceived control and perceived distress, he suggests that the results when experiencing stressful conditions do not always have a positive impact, but also a negative impact in the form of perceived distress The results of hypothesis 6 testing found in this study that Ho is accepted, and Ha is rejected, which means that there is no effect of job attachment on work performance through stress.In other words, stress does not act as a mediator that affects work engagement on work performance.Although work engagement can have an impact on work performance, the path of influence is not through stress.This result is interesting because previous research has identified the role of stress as a mediating variable in linking work engagement with work performance.
The results of hypothesis 7 test obtained in this study show that Ho is accepted, and Ha is rejected, which states that learning organization has no significant effect on work performance through stress.Stress is not an important factor that connects learning organization to work performance.Learning organization approach does not automatically improve work performance through stress management.Then, the results of hypothesis 8 test found in this study show that Ho is rejected, and Ha is accepted, which means that the model of the influence of work attachment and learning organization on work performance through stress is proven to have a fit model with empirical data from the results of testing 5 indices.
The pressure experienced at work is interpreted as a positive experience (perceived control) in everyone.Likewise, employees of PT Pupuk Kalimantan Timur are able to commit and adapt when experiencing structural changes in all departments.Work engagement and learning organization are important things to achieve eustress and more optimal work performance.It is not only stress and work engagement, but it requires a learning organization that continues to be improved.Employees who are able to show work performance are individuals who constantly develop their capacity in new and broad mindsets are nurtured, so that individuals continue to learn how to learn together (Senge, 2006).Thus, work performance will increase better because stress, work attachment and learning organization are important factors that strengthen superior work performance.

CONCLUSION
This study reveals that work engagement and learning organizations significantly impact work performance at PT. Pupuk Kalimantan Timur, with employee stress acting as a mediator.The research underscores the critical role of managing stress and effectively implementing work engagement and learning organization practices to enhance work performance.Additionally, it highlights the importance of these concepts in achieving success for both employees and the company.

Table 2
Measurement Model Evaluation Test Results Stress

Table 5
Measurement Model Evaluation Test Work Achievement

Table 1 .
Convergent Validity The test results are presented in the following table:

Table 8 .
Direct, Indirect dan Total Effect

Table 9 .
The Results of Model Fit Test of Work Attachment and Learning Organization on Job Performance through Stress