Perception of empowerment provides the necessary resourcefulness in providing necessary conditions of working in life and organizational work-life [5]. Empowerment can motivate an individual to work with zeal and full capacity in achieving organisational goals [15]. The cognitive state of mind that triggers the state of confidence to work with full force termed as psychological empowerment [16]. Psycho- logical is the process of perception on the realities around the individual, and positive perception can lead to empowerment or otherwise.

Spreitzer [17] defined psychological empowerment as the cognition state resulted from the subjective positive trans- formation of perceptions, and an individual starts to work with full dedication and fully immersed in the work environment. Psychological empowerment consists of four structural processes namely meaning, competence, self-determination, and impact [15].

Meaning provides the objective of work and direction for a person’s actions [5]. Organizational objectives and individual objectives can sync in which an individual builds a strong bond based on common objectives and purposes between him or her and the organization [15]. The two sides can emerge as a common purpose that can lead to psychological empowerment [16].

Individual competency is the notion of having a true unbiased opinion of a person’s capacities with the right skills and set of knowledge [15]. This idea is associated with having self-efficacy and believing in oneself [5]. Having a positive belief helps an individual to become more dedicated in completing the task at hand and looking for new opportunities at the workplace [16]. The positive self-evaluation of one’s abilities and having the right abilities can influence an individual to work hard for the organization as the purposes match the individual’s precise capacities in their job performance.

However, having the right skills and objectives that match the organization is not enough. Individuals should be motivated to use their innate skills and objective at hand in performing with the provision of the right environment [16, 18]. Environment enablers are the autonomy and freedom at the workplace to make choices regarding work-related issues. Meaning, competence, and willingness to put the right efforts can empower individuals to build a sense of power and purpose [5]. When individuals motivated at their workplace, they will put more efforts in the right direction as they have the right motivation for the organization [16]. The perception of autonomy and freedom provides opportunities to manipulate and correct work procedures [15]. The impact is the outcome of individual efforts such as having a common purpose and putting the effort with dedication [19]. The positive impact can lead the individual to put more efforts [5].

The difference between high performing and low performing employees is the self-perceived psychological empowerment at the workplace [20]. If the psychological empowerment prevails at large in the organization, the organization can either perform well or vice versa. The general sense of empowerment leads to higher work performance and lower negative feelings towards the work and workplace.

Two tests were performed to evaluate the sample size for this study. This study used the table by Krejcie and Morgan [31] and determined that the sample size must be 378. Moreover, the GPower V2.0 software was also used to determine the sample size using multiple regression statistical analysis by setting the statistical significance at 0.05 with the statistical power at 0.95 for the null hypothesis rejection level and a medium level of effect size at 0.15. The results suggested that the sample size should be 130 samples. The sample frame of this study is the registered hospitals under MOH and the nursing staff who work at the registered hospitals. The list of hospitals retrieved from MHTC website, and this study selected the full-time registered nurses employed by 78 registered hospitals under MOH. A total of 550 survey questionnaires sent, and 467 respondents completed the questionnaires. However, only 432 questionnaires were used for the final analysis as 35 returned questionnaires were incomplete.

We used a total of twelve psychological empowerment items adopted from Spreitzer [15]. The estimations of deper- sonalization, emotional exhaustion, and personal unfulfillment were performed using the Maslach’s Burnout Inventory (MBI) with 4, 7, and 4 items, respectively. Maslach et al. [24] reported that the reliability coefficient for the scale is α = 0.91 for emotional exhaustion, α= 0.85 for depersonalization, and α= 0.87 for personal accomplishment.

The data collection in social sciences was done using the questionnaires, and this common method has resulted in common method variance and measurement errors [32]. The suggested handling of CMV was after data collection to identify and control the CMV. This study used the single factor Harman’s analysis for testing the CMV [33]. The single factor of this study was the limit of less than 40%; hence, there was no CMV for this study [32].

This study used Mardia’s online test for the multivariate normality on the web power website to test the multivariate normality of the data [34]. The web power online tool provided the results with skewness and kurtosis coefficients as well as the p-value for the data set. The data considered non-normal if the p-value for the Mardia’s multivariate coefficient is more than 0.05 [35]. The result confirmed that the data were normal.

PLS-SEM tested the models using two steps, namely measurement and structural model [36]. Cronbach’s alpha (α) and Composite Reliability (CR) are employed to confirm the intern consistency for the constructs. The recommended score for α and CR is 0.70 or above [37]. However, CR is considered a good indicator of internal reliability than α [36].

The Variance Inflation Factor (VIF) represents the inflation of variance due to the presence of multicollinearity within the constructs [37]. The Average Variance Extracted (AVE) value must be 0.50 or above for every construct [36]. The discriminant validity in PLS-SEM was corroborated with cross-loading, Fornell-Larcker, and Heterotrait-Monotrait ratio (HTMT) [36]. The Fornell-Larcker criterion needs to be above 0.700 to provide evidence of discriminate validity. However, HTMT is suggested to discriminate the validity in which the values must be less than 0.90 [36]. The measurement model is represented with r² that denotes the explanation power of the endogenous variables with the exogenous variables. The effect size (f²) and Q² are the estimates of the measurement model [37]. The effect size (f²) signifies the effect of each exogenous variable on the endogenous variable. Cohen [38] offered the guidelines for the mean values of f². The effect sizes of 0.30, 0.15, and 0.02 represent large, medium, and small effect sizes, respectively. The Q² values of 0.02, 0.15, and 0.35 indicate small, medium, and large predictive relevancies of the model, respectively [36].

Moreover, the Multi-Group Analysis (MGA) in PLS-SEM enables the researchers to identify the different pre-set groups in estimating the differences between the groups based on the pre-defined bases of the group [39]. MGA is efficient to confirm the differences between the groups within the dataset [37]. MGA enables the researchers to track the differences between the structural paths of the multiple groups in the dataset [39]. PLS-MGA determines whether there is any difference in the assessment group and categorises whether the dataset is homogenous or heterogeneous based on the different groups within the dataset [36]. The first step was to categorize the groups based on categorical variables of interest, such as age, gender, or income. Then, the path coefficients of multiple groups analyzed in determining whether the two groups are significantly different from each other based on the guidelines by Henseler et al. [39]. If the difference is significant, the cut-off value of permutation p-values should be less than or equal to 0.10 to establish a significant difference between the groups [37]. The differences that may exist within the dataset based on the characteristics of the sample may not be evident in the aggregated data and path coefficients of the subdivided group data; hence, MGA can be used to determine the differences between the data based on categorical bases [39].

Table 1 presents the profile of the respondents. The total of female respondents was 85%, and the nursing profession is considered a female profession. For the respondents’ age, the data show that 86% of the respondents were 31 years or more and 14% of them were less than 31. However, the majority of the respondents (68%) were married or once married. Education is the basic requirement in the nursing profession in which 78.9% of the respondents have a college degree in nursing, and 2.1% of them have a master's degree. The minimum qualification in the profession is a certificate. A total of 42.5% of respondents have working experience of the five years or more, and 58% of the respondents have six years or more experience. About 73% of the respondents were Malaysian citizens, and the remaining were non-Malaysians.

This study followed the recommendations by Hair et al. [36], as shown in Table 2. The α and CR for each construct are more than 0.871 and 0.869, respectively. Cronbach’s alpha value is the measure of inter-correlational approximation on the items for each construct. The Cronbach’s alpha values revealed that 0.963 is the minimum value and the other constructs’ values are above the prescribed limit of 0.70. The Cronbach’s alpha and CR values signify that the constructs are reliable before performing the next analysis. The AVE of all items for each construct must be above 0.50 to achieve convergent validity in approving the uni-dimensionality of each construct [36]. The items display that the constructs have acceptable convergent validity. The loading for every item and cross-loading should be verified to confirm the discriminant validity. The results show that the item loads on particular variables fulfill the notion of discriminant validity. Table 3 shows the results. Another test for discriminant validity is the Fornell-Larcker criterion. The results show that the values are in the acceptable range. Another recommended examination for discriminant validity is the HTMT ratio in which the HTMT values must be 0.90 or fewer to confirm the discriminant validity [37]. Table 3 shows that the study has evidence of discriminant validity.

The adjusted r² value for the three paths, namely psycho- logical empowerment and depersonalization, psychological empowerment and emotional exhaustion, and psychological empowerment and personal unfulfillment, are 20%, 32%, and 12%, respectively. The result explains that 20 per cent of the variance in depersonalization can be explained by psycho- logical empowerment as an input variable. Moreover, 32% per cent of the variance in emotional exhaustion can be explained by psychological empowerment as an input variable, and 12 per cent of the variance in personal unfulfillment can be explained by psychological empowerment as an input variable. The f ² values for psychological empowerment on deper- sonalization, emotional exhaustion, and personal unfulfillment are 0.260, 0.485, and 0.150, respectively. The effect of psychological empowerment on emotional exhaustion is large, whereas the effects on depersonalization and personal unfulfillment are small. The predictive relevance (Q²) values for the part of the model are 0.100, 0.212, and 0.079 for depersonalization, emotional exhaustion, and personal unfulfillment, respectively. The predictive relevance for personal unfulfillment is small, depersonalization is medium, and emotional exhaustion is medium.

Table 4 shows the standardized path coefficients, t-values, and significance level. The path coefficient values for psychological empowerment on depersonalization is β = -0.454 and p = 0.000, which provide the supports to accept H1. The result shows that the effect of psychological empowerment on the depersonalized type of burnout is negative but significant. The path coefficient forms psychological empowerment on emotional exhaustion which is β = -0.571 and p = 0.000, indicating a negative and significant effect of psychological empowerment on emotional exhaustion. Hence, H2 is supported. The path coefficient for psychological empower- ment on personal unfulfillment is β = -0.361 and p = 0.000. The effect of psychological empowerment on personal unfulfillment perception is negative and significant which supported H3. Table 4 shows the results of path coefficients.

Multiple group analyses were performed to compare the results for different groups based on age, gender, marital status, and work experience. One non-parametric test was utilized to evaluate the differences in the key relationship between the model based on the characteristics of the sample, such as age, gender, marital status, and work experiences (Table 5). Table 6 shows the path values for two groups with the differences within the groups and p-values as recommended by Henseler et al. [39]. P_MGA represents the p-values obtained by the use of multiple group analysis of PLS-SEM as the measure of significant difference among the groups [39].