The mediating role of work engagement: A survey data on organizational citizenship behavior

This dataset shows the results of research on factors that influence organizational citizenship behavior (OCB). Other variables included in the dataset are Psychological Capital (PC), Transformational Leadership (TL), and Work Engagement (WE). The dataset includes 156 respondents who are client counsellors (CC) in 19 correctional institutions throughout Indonesia. Data was collected with the help of an online questionnaire which was carried out in 3 waves. The dataset in this article also applies multi-group analysis to validate the drivers of organizational citizenship behavior in doing work which is divided into male and female employees.


Specifications
Organizational behavior Specific subject area Management, Human Resource Management, Human Capital  Type of data  Tables and Figures  How data were acquired Survey Questionnaire. Data was collected using a five-point Likert scale. All valid samples were processed with SPSS and Amos. Data format Raw, analyzed Parameters for data collection The dataset uses the sampling method with purposive sampling technique through the criteria of officers who interact directly with the inmates, and whose duty is to provide guidance and counselling. Description of data collection Primary data sources were collected from the distribution of online questionnaires (Google Form) involving 3 waves of data collection. The three waves of data were collected at an interval of 30 days in 2021. Data

Value of the Data
• The dataset provides evidence how organizational citizenship behavior create a more effective work environment. • The dataset provides insight that individual behavior and perceptions, which are the main drivers of extra behavior in organizations, can be support success. • The dataset is useful for predicting the psychological condition, leadership style, attitude, and behavior of employees on discretionary behavior. • The dataset can be used to explain how employee involvement in work can support realizing the right psychological and leadership style for employees towards discretionary behavior in the workplace. • The dataset is important for policy implementation and can encourage coordination of activities among team members and in generating a peer-to-peer environment in organizational learning activities.

Data Description
The data set includes questions related to four constructs through psychological capital adopted from research [1] , then for transformational leadership adopted from research [2] , then to measure work engagement adopted from research [3] , and to measure organizational citizenship behavior adopted from research [4] . Measurements on the indicators used using a Likert scale 1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, and 5 = strongly agree. The scale is used to measure indicators that show respondents' responses to the statements contained in the questionnaire.
Counsellor is part of the correctional officers whose job is to guide prisoners so that they can re-integrate socially into the community while the counselling process itself starts 1 year before the incarceration term ends. The total population of counsellors in Indonesia is 1565 officers, however the present study focuses only on counsellors for terrorism convict. Hence, the sample size is 156 officers. The entire demographic characteristics of the respondents in this study are illustrated in Table 1 . The characteristics of the respondents will be described based on gender, marital status, age, tenure, entry path, type of position, last education degree. Table 2 presents the results of data analysis based on respondents' answers to the questionnaire. The average value of respondents' responses to each indicator and the overall value of each research variable are also shown in Table 2 .

Materials and Methods
Data were collected using purposive sampling technique, with the criteria for selecting respondents were prison officers whose job was to interact directly with the inmates, and work 24 h through work shifts. Data were entered and coded in SPSS v.24 software, while the preliminary test was carried out in SPSS v.24. For the main analysis, Amos v.24 was used as part of Structural Equation Modeling (SEM).
Multivariate normality refers to detecting the shape of the data distribution on a variable multivariately and its correspondence is a normal distribution. In SEM, the multivariate normality test is carried out with a critical ratio value (c.r.) in the multivariate kurtosis section, the value of c.r. This is also called the Z-value. If the Z-value is greater than the critical value, then the data distribution is not normal, on the other hand, if the Z-value is less than the critical value, the data distribution is normal. The critical value can be determined based on the significance level of 0.05 (5%) which is 1.96.
The results of the normality test showed a multivariate c.r of 56.90 which was outside the range of -1.96 to + 1.96 at a significance level of 5%, so it can be concluded that the multivariate data were not normally distributed. This condition is actually not a problem because according to [5] , Maximum Likelihood Estimation (MLE) in SEM is efficient and unbiased either the multivariate normality assumption is met or not, and has been proven to remain robust in the event of a violation of the normality assumption.
Next, Table 4 shows the results of the construct validity test conducted through convergent validity with the rule of thumb that a construct is said to meet convergent validity if the indicator on the construct has a standardized regression weight (lambda/factor loading) value above 0.50. Furthermore, Table 5 shows the construct reliability test that was checked using the construct reliability value. A construct is said to be reliable if the construct reliability value is greater than 0.70 or the Average Variance Extracted (AVE) value is above 0.50. Research [5] explains, the rule of thumb construct reliability value must be greater than 0.70, but the construct reliability value greater than 0.60 is still acceptable as long as each indicator has met convergent validity.
The structural model stage begins with an evaluation of the structural model fit (goodness of fit) whose role is to ensure that the developed model is consistent with the data. The estimation results are presented in Fig. 1

below:
After getting 156 responses from respondents according to the research criteria, then all samples were processed with v.24 software to analyze validity, reliability, structural models, direct effect hypotheses, and indirect effect hypotheses.
Structural model analysis is done by examining the estimated parametric relationship between variables that represent each theoretical hypothesis on the direct effect path. This assumption can be accepted if the path parameter is significant in the predicted effect, meaning that the path parameter must be greater than zero for the positive direction and less than zero for the negative direction [5] . Furthermore, in the structural relationships test, hypothesis testing is carried out for the influence between variables, using the critical ratio (CR) value and the probability value (p-value). Whether or not there is a significant effect between variables using the provisions if the CR value 1.96 or the p-value 5% significance level, then it is decided that there is a significant effect between these variables. Likewise, the indirect effect path has the same provisions.
After testing the significance of the mediation effect, the final step is to know the nature of the mediation (described on Table 7 ). Detecting the nature of mediation can be seen from the     effect of mediation, if the direct effect of the exogenous variable on the endogenous variable is significant, and the indirect effect of the mediating variable also passes through the significant path, it is said to be partially mediation [6] . On the other hand, if the direct effect of the exogenous variable on the endogenous variable is not significant, while the indirect effect is through the mediating variable through a significant path, then it is said to be fully mediation or perfect mediation. Table 7 shows the role of employee involvement at work in order to support employee psychology and leadership style in influencing organizational citizenship behavior.

Ethics Statement and Informed Consent
This is a non-interventional research project. No ethical permission is necessary, according to the Research Ethics Committee of Universitas Airlangga. Informed consent has been acquired from all participants of the study.

Declaration of Competing Interest
The authors state that no financial or personal conflict of interest is deemed to affect the work reported in this article.

Data Availability
The