Introduction

Knowledge sharing is the hallmark of the organizational systems, and organizational success significantly rests on the internal knowledge sharing of the organization (Bies et al., 1997). Knowledge management system (KMS) rise in the organization indicates the importance of knowledge sharing, empowering the efficacy, and progress (Moon & Lee, 2014). Knowledge sharing is expedited by rewards, improving the social network, and having a strong knowledge sharing culture (Peng, 2013; Richman & Leary, 2009). Knowledge sharing remains the focus of new organizational development policies; still, many workers are unenthusiastic towards knowledge sharing (Serenko & Bontis, 2016). Knowledge sharing remains low in the organization based on the employee fears of the cost of control, status, and hesitant attitude towards knowledge sharing even with supervisors (Zhao et al., 2016). In the last decade, knowledge hiding remains the focus of organizational research (Riaz et al., 2019). Personal dynamics also clues towards the knowledge hiding, like distrust among fellow employees’ starts knowledge hiding (Heizmann & Olsson, 2015; Peng, 2013). Nevertheless, situational factors also encourage knowledge hiding, like handling the problematic or complex conditions, time restrictions, and knowledge sharing reflected not promising (Serenko & Bontis, 2016). The knowledge hiding base on the circle of distrust and hints to the fall of creativity hinders the overall organizational structure from performing (Moon & Lee, 2014). It leads to service sabotage; service sabotage is the unsatisfactory performance of the FLEs internally or externally caused (Harris & Ogbonna, 2006).

The antecedents of knowledge hiding persist the personal mode and personality that endorse knowledge hiding (Heizmann & Olsson, 2015). However, the growing stress to perform in service organization (Serenko & Bontis, 2016) and handling cumbersome customer demands encourage negative sentiments (Kern & Grandey, 2009). The abusive supervision remains the prime cause of the employees’ deviant workplace behaviours (Khalid et al., 2018). Abusive supervision deteriorates the relationship among colleagues. The rise of the marketing era and push for customer services instigates the customer pressure faced by the FLEs (Kern & Grandey, 2009). The customer mistreatment also challenges the FLEs right to perform and may lead to an adverse reaction among the FLEs (Kim & Smith, 1993) and create possible service sabotage in the end.

Public sector organizations (PSO) are the largest service providers and are recognized as having ill-informed service orientation and high incidents of customer mistreatment from FLEs (Armeli et al., 1998). Abusive supervision is allegedly extraordinary in the PSOs, and worker response to abusive supervision is commonly reported (Shao & Skarlicki, 2014). Knowledge sharing or overall organizational performance remains under strain with these PSOs (Seba et al., 2012). Service sabotage is high in PSOs across the globe and reportedly high among developing countries (Harris & Ogbonna, 2006). However, the felt obligation at the personal level can tackle the knowledge hiding occurrences in case of abusive supervision and customer maltreatment (Connelly & Zweig, 2015; Kern & Grandey, 2009; Nga & Feldman, 2015).

In addressing the above-discussed concerns, this study aims to reconnoiter the FLEs RVA caused by the ABS and CMT and the FLEs RAV affecting the knowledge hiding behaviours. Furthermore, FTO among the FLEs reduces the knowledge hiding behavior. The study results enrich the knowledge hiding literature in the service sector by incorporating antecedents of knowledge hiding behaviors and reducing the causes of the knowledge hiding among FLEs.

The following section is about the literature of revenge attitude, abusive supervision, customer mistreatment, and felt obligation, influencing the different kinds of knowledge hidings within the service organizations. The section after the literature review describes the method taken for this very study based on the literature review and the hypotheses development. The analysis and results are reported in “Data analysis” and “Discussion,” respectively. In the last “Conclusion,” conclusions with the future research opportunities and study limitations are reported.

Literature review

Revenge attitude

Handling human social relationships is sophisticated and reciprocated; negative and positive emotions are part of the human social relations outcomes (Kim & Smith, 1993). Negative emotions evolved during the human social relationship is “revenge” based on the perception of harm drive from the other and caused anger as negative emotion (Bies et al., 1997). Achieving self-justice and self-initiated retaliation are the significant causes of the revenge attitude (Grandey et al., 2007). Perception of personal injury from others is the initiating stage of revenge conduct. In organizational settings, revenge instigated workplace stealing and associated other harmful actions towards everyday business (Bradfield & Aquino, 1999). Revenge attitude established towards the organization broadly, like a specific individual within an organization, like towards supervisor, or even for customers (Bies et al., 1997). Revenge attitude activated with wrong allegations, abating position, harsh criticism, and overtly deriding subordinates by the managers (Černe et al., 2014). Revenge theory proposed the undesirable and disruptive characteristics of revenge (Grandey et al., 2007). However, revenge attitude as positive side comforts to stop power misuse by the powerful within organizational settings (Černe et al., 2014). This study defines revenge as punishing and causing harm to their targets (supervisor/customers) because of the harm they (employees) have received (Yeh, 2015).

The research demonstrated that tough negative emotions like rage and frustration are the ultimate source of provoked revenge FLEs because these emotions stimulate employees’ sentiment to fight back (Yeh, 2015). When employees feel negative emotions formed inside them, this negatively affects their enthusiasm (Grandey et al., 2007). Employees who feel that they are treated negatively will take revenge if they think that there is no possible way to restore justice (Crene et al., 2014). The prime purpose of revenge remains to reinstate justice, restore well-being, and reject injustice perceived by the deprived one (Yeh, 2015). FLEs faced negative attitudes from supervisors and customers (Eisenberger et al., 2001). Revenge attitude comes to rescue one from the feelings of frustration and fatigue (Černe et al., 2014).

Abusive supervision

Antecedents and consequences of abusive supervision (ABS) remains a relevant debate in the organizational literature (Tepper, 2007). ABS is connected to unruly towards subordinates, destructive feedback, unfair criticisms, and not recognizing FLEs hard work at random (Zellars et al., 2002). Based on the interactional justice facet of justice theory, Tepper proposed the theory of abusive supervision (Tepper, 2000).

ABS negatively influenced the assistants’ self-esteem and caused a reduction in the efficiency of organizational performance (Zellars et al., 2002). In the USA, 14% of employees face the ABS, and the cost of ABS was about $23.8 billion every year (Liu et al., 2018). However, ABS caused a reduced perception of organizational justice, commitment to the job, and general job satisfaction among the subordinates (Tepper, 2007)).

Furthermore, ABS intensifies the reduced work engagement, hostile workplace relations, burnout, and intention to leave the job (Liu et al., 2018). ABS triggered a frazzled workplace, and the consequences of interactive problems clue the unfavorable conduct among subordinates (Tepper, 2000; Zellars et al., 2002). Employees’ feelings of vulnerability increase, and employees attempting to reduce aggression engage with a vengeance (Khalid et al., 2018). However, ABS causes the drain of motivation, reduces workplace well-being, and promotes violent and work deviance behaviors (Liu et al., 2018). The above discussion stirred us to propose the following hypothesis:

  • H1a: Abusive supervision has a positive relationship with employees’ revenge attitude.

Customer mistreatment

An increase in customer-driven marketing culture witnessed around the globe, causing the FLEs to face problematic behaviors from their corresponding customer (Kern & Grandey, 2009). The customer is continuously involved in undesirable conduct towards FLEs, asking for unreasonable demands, and tangled with intimidating actions (Grandey et al., 2007). The CMT described poor-quality behavior exhibited in verbal aggression, ambiguous demands, hostility, unfair/unfriendly conduct, or customer shouting, cursing, and disdainful looks towards workers (Kern & Grandey, 2009).

A vast power distance exists between the customer and FLEs (Grandey et al., 2007). The literature stresses that the FLEs need to behave with submissive, amenable, and plasticity towards the customer. It became a norm to better control natural behaviors at the workplace to serve customers (Shao & Skarlicki, 2014). FLEs were increasingly facing emotional fatigue due to aggressive and discriminating behaviour from their respective customers. These vulnerabilities increase workplace hitches and lessen workplace efficiency; it leads to negative workplace behaviors (Kern & Grandey, 2009; Yeh, 2015). The FLEs voices seldom heed, or action was taken against the customer mistreating with the FLEs (Shao & Skarlicki, 2014). FLEs start taking personal actions to tackle customer mistreatment (Kern & Grandey, 2009). The emotional involvement of the FLEs promotes service sabotage (Harris & Ogbonna, 2006). In an attempt to restore a just workplace and protect the self, FLEs develop negative revenge attitude behaviors (Grandey et al., 2007). FLEs develops the need to retaliate as the customer violates the social norms, no action is taken by the management, and the employee is left alone to act (Kern & Grandey, 2009). Oppressive customer behaviors trigger the coping strategies among the FLEs, and it leads to a reduction in service quality or even FLEs decline to serve the customer (Shao & Skarlicki, 2014). The FLEs engaged in a personal coping strategy to tackle the tyrannical customer and develop a revenge attitude, harming customer interest, not providing full service, or even engage in emotional arguments with the customers (Heizmann & Olsson, 2015). The above leads us to propose the following hypothesis:

  • H1b: Customer mistreatment has a positive influence on the employees’ revenge attitude.

Felt obligation

Sense of responsibility promotes the enhanced dutifulness and obligation to behave appropriately in organizational settings (Amayah, 2013). Nga and Feldman (2015) postulated that reciprocity promotes positive psychology among the employee, such as the employee starts taking superior work responsibility to reduce the engagement in negative workplace behaviors. Employees learn the value enrichment and display the accepted workplace behaviors as the notion of affective commitment (Armeli et al., 1998). Eisenberger et al. (2001) suggested that the pro-social workplace values harness the positive workplace behaviors. The above leads us to propose the following hypothesis:

  • H1c: Felt obligation has a negative influence on the employees’ revenge attitude.

Knowledge hiding behaviors

Knowledge hiding behaviors are counterproductive work behaviours, and the discussion of knowledge hiding (KH) commenced from knowledge management in organizations (Connelly et al., 2012). Scope of knowledge management preliminary focuses on promoting a knowledge sharing environment (Serenko & Bontis, 2016). Knowledge hiding scantly debated in the literature as the deviant organizational behaviors instigated by the organizational employee and precise strategies that promote the climate of knowledge sharing at the workplace (Connelly & Zweig, 2015). Knowledge hiding is the deliberate concealment of knowledge from another person who has demanded it (Connelly et al., 2012). Different knowledge hiding behaviors prevails in organizations that vary with the objective and how knowledge hiding occurs. It can be evasive hiding, rationalize hiding, playing dumb, or lack of sharing (Serenko & Bontis, 2016).

Evasive knowledge hiding

Employees get involved in the hiding of relevant information with the misleading objective. Ingenuity is the key in this type of knowledge hiding behavior, and confusing or imprecise provision of requested information knowledge sharing occurs (Connelly & Zweig, 2015). However, partial information is delivered so that the information withheld intending to pass the complete information in the future (Heizmann & Olsson, 2015). This facet of knowledge hiding is termed evasive knowledge hiding (Connelly & Zweig, 2015). It is also regarded as the partial knowledge delivery or partial knowledge hiding and keeping in mind the complete information will be delivered when asked.

This kind of knowledge hiding is antisocial and associated with the motivation of deceiving from the perspective of the one requesting the knowledge (Peng, 2013). Evasive knowledge hiding is rather destructive, and the perpetrator dodges the knowledge enquirers. Knowledge hiding happens based on diverse motivations and antecedents (Heizmann & Olsson, 2015). That kind of knowledge hiding is an antisocial demeanor, which motivates to deceive the other person. It is always dependent on the target who asks for the knowledge or information (Connelly et al., 2012). Knowledge hiding is triggered by the fear and or loss of prestige by sharing the material requested.

Nevertheless, the organizational rewards are associated with team performance; many employees are unwilling to share complete information unless requested (Peng, 2013). Knowledge hiding behavior is established on the rudeness of the supervisors or customers at the workplace (Connelly & Zweig, 2015). As the supervisors are unable in promoting, the team sense and treat fellow employees unfavorably. That promotes the revenge attitude instigating the evasive knowledge hiding attitude. Therefore, the study hypothesis is as following:

  • H2a: Revenge attitude has a positive effect on the service worker’s evasive knowledge hiding.

Playing dumb

Knowledge hiding is based on a trick to escape knowledge delivery when knowledge is demanded (Peng, 2013). When requested, the knowledge accumulator displays ignorance and plays a trick not to deliver the requested material (Connelly et al., 2012). That type of knowledge hiding behavior is labeled as playing dumb (Connelly & Zweig, 2015). It is a kind of escaping approach to avoiding knowledge delivery to the demanding party. People elude knowledge delivery when people disbelief somebody or perceive that knowledge delivery may cause the issue to the knowledge emancipator (Khalid et al., 2018). Deceptively behaving to provide the information on the later stage with no actual intent or pretending to know information (i.e., playing dumb) (Connelly & Zweig, 2015), many FLEs display the attitude that they are not aware of the requested information from their supervisors (Khalid et al., 2018). The problematic relationship with supervisors and colleagues suggests using playing dumb as a strategy to avoid knowledge sharing (Connelly et al., 2012). Therefore, we hypothesis the following:

  • H2b: Revenge attitude has a positive effect on the service worker’s playing dumb knowledge hiding.

Rationalized knowledge hiding

Often, an employee holds the critical information not with the objective to deceit but holds the knowledge-based on ones’ incapability to render the information or accuse the third party of their inability to provide the complete demanded information (Connelly et al., 2012). Knowledge hoarder has clear validations available for withholding the knowledge from the knowledge seeker, based on the knowledge provider incapability grounded on authority, or based on the organizational formal power structure not permitting to deliver the requested knowledge (Heizmann & Olsson, 2015), but suggesting to forward the request the right person (Khalid et al., 2018).

Moreover, it signifies that a third party asks or restrict him/her not to deliver the demanded knowledge for formality or other relevant justifications (Connelly & Zweig, 2015). The rationalized knowledge hiding is based on justified reasons and a lack of positional capacity to supply requested information (Heizmann & Olsson, 2015). The rationalized knowledge is hiding, not inspired by the negative emotions developed at the workplace (Connelly et al., 2012). However, the negative emotions developed towards the organization or supervisor may not be valid for holding knowledge. FLEs offers valid reasons for not offering requested knowledge and justifying that the workplace relationship is not the sole reason for not rendering the knowledge. Therefore, we hypotheses the following:

  • H2c: Revenge attitude has a negative effect on the FLEs’ rationalized knowledge hiding.

Felt obligation influence on knowledge hiding

Felt responsibility is the inherent awareness to respond dutifully to other (Amayah, 2013), the idea of duty based on the norm of reciprocity, as one feels to return the favourable conduct with appropriate behaviors (Kern & Grandey, 2009). The conscientiousness of duty encourages the pro-social behaviors towards the organization for value enrichment (Nga & Feldman, 2015). Perceived organizational support (POS) explains the felt obligation among the employees as a return to the organization for indebtedness with a display of more significant affective commitment (Kern & Grandey, 2009). POS promotes the sense of reciprocity that positive attitude from organization returned by the employees positively.

Moreover, felt obligation positively influences affective commitment and the employee’s in-role performance at the workplace (Armeli et al., 1998). However, felt obligation diminishes the counterproductive behaviors among workforces (Eisenberger et al., 2001); it promotes positive psychology at the workplace. It became evident that felt obligation is the personal pro-social behaviors towards the organization, causing positive outcomes for the organizations from the employees (Nga & Feldman, 2015). The instigation of felt responsibility at the employee end reduce the employees’ inclination to engage in knowledge hiding and facilitates the conducive working climate (Kern & Grandey, 2009). Therefore, we suggest the following direct effects of the felt obligation on the knowledge hiding dimension:

  • H3a: Felt obligation has a negative influence on the service worker’s evasive knowledge hiding.

  • H3b: Felt obligation has a negative influence on the service worker’s playing dumb knowledge hiding.

  • H3c: Felt obligation has a negative influence on the service worker’s rational knowledge of hiding.

Moderating effect of work experience and gender

Behavioral assertiveness significantly diverges one’s behaviour based on individual factors. The individuals’ work experience and gender moderate the fundamental relationships between attitude and behaviors (Moon & Lee, 2014). The work experience instigates a higher understanding of the work climate and builds close ties with the work culture and peers. The higher work experience builds the sense of compliance with work routines and engages in pro-organizational norms (Nga & Feldman, 2015).

Gender also plays a significant role in individual behaviors in general social settings. FLEs gender can play a significant role in the knowledge hiding behaviors and promotes compliance behaviors. Therefore, it is vital to explore the contributing factor of respondents’ experience and gender on the revenge attitude and knowledge hiding behaviours as postulated in the study. Consequently, this study proposes the following hypotheses:

  • H1MGA: There is a significant categorical moderating effect of respondent experience on the relationship among model constructs.

  • H2MGA: There is a significant categorical moderating effect of respondents’ gender on the relationship among model constructs.

Research methodology

Research design

The current study assumed a deductive research approach with the quantitative research design (Sekaran & Bougie, 2016). The cross-sectional study design was adopted for the study to evaluate the hypotheses proposed for the study (Creswell, 2009). Quantitative research design helps evaluate the causal relationship among study variables with operational study concepts with a structured approach and a sufficient sample size to evaluate the relationship between the study constructs (Hair et al., 2019). Cross-sectional survey-based data was collected for the study from frontline employees working in a public sector establishment in Pakistan.

Data collection and sample selection

Sample size requirement estimated by using GPower 3.0 software (Faul et al., 2007). We calculate the sample size for the multiple regression with a significance level at 0.05, the statistical power of 0.95, having a medium effect size of 0.15, and having four exogenous variables. One hundred and twenty-nine sample sizes are required for the study to perform a multiple regression analysis. However, the recommended sample size for the SmartPLS 3.1 was 200 (Hair et al., 2019). The population for the study was the police officers working as FLEs in Punjab Police, Pakistan. Snowball sampling was utilized as a non-probability sampling technique and received the 234-filled questionnaire back. Subsequently, removal of incomplete and non-usable, a total of 201 questionnaires were exploited for final analysis. The data collection was performed for 2 months, from October 2019 to December 2019. Study respondents were informed about the study, and approval was taken as the respondent’s signature on the survey form.

Measurement and scales

The questionnaire for this study premeditated exhausting sensible and straightforward wordings so that the respondents could readily understand and answer the questionnaire items. For the abusive supervision, 15 items scale was altered from the work of Tepper (2000). Customer mistreatment was assessed by using the 15 items scale developed by Shao and Skarlicki (2014). Revenge attitude gauged consuming the seven-item scale of Bradfield and Aquino (1999). Felt obligation evaluated with the seven items from the work of Eisenberger et al. (2001). Evasive knowledge hiding measured with validated four-item, playing dumb estimated with four items, and rationalized hiding measured with three items from the work of Connelly et al. (2012). A seven-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree was used to gather data for each construct of the research model.

Assessment of common method variance

Social science research methods are associated with common method bias caused due to a single source and single point of time data collection method (Podsakoff et al., 2003). Harman’s (1976) one-factor test was the recommended remedying to assess the impact of CMV on the study’s constructs (Podsakoff et al., 2003). Using one-factor Harman’s test confirms that CMV is not a critical issue for study as the highest factor accounts for 18.6% variance and less than the suggested limit of 50%. Moreover, an inter construct correlation of less than 0.900 shows the excellent convergence of no issue of CMV (Podsakoff et al., 2003).

Multivariate normality

Multivariate normality is not the apparent requirement to exploit the SEM-PLS as it is a non-parametric analysis tool (Hair et al., 2019). However, data normality was tested by the procedure endorsed by Peng and Lai (2012), an online tool of web power employed to evaluate the dataset normality. The test results confirm that the data set was not normal, and Mardia’s multivariate coefficient p-values come less than 0.05 (Cain et al., 2017).

Data analysis method

Partial least squares structural equation modeling (PLS-SEM) with the SmartPLS software 3.1 was exploited to inspect the study data empirically. PLS-SEM is a multivariate analysis tool to assess path models with the latent constructs having composites (Hair et al., 2019). PLS-SEM empowers the researcher to work with the non-normal and small data set (Chin, 2010). Furthermore, the casual-predictive nature of the PLS-SEM assists of work with complex models having composites and work without having the assumption of goodness-of-fit estimation than the covariance-based SEM (Chin, 2010). Two-step techniques set-out for PLS-SEM data analysis, and the first measurement was performed on the model to test the reliability and validity of the study constructs (Hair et al., 2019). The second stage is performed with the structural model associations and examination of study hypotheses with significance levels (Chin, 2010). Model estimation performed with r2, Q2, and the effect size f2 describes the path effect from exogenous construct to endogenous construct (Hair et al., 2019).

Furthermore, multi-group analysis (MGA) of PLS-SEM empowers the researchers to decide about the variances in pre-defined groups in the data set under investigation (Henseler et al., 2015). The MGA is a handy technique to evaluate the variances between the groups within the dataset (Hair et al., 2019). The MGA supports evaluating the variations between the structural paths of the various groups that exist in the data (Henseler et al., 2015). The first step was to produce groups created on the categorical variables of interest like age, gender, or income. Then, the path coefficients of the groups were examined, and to report, the two groups significantly varied from each other or not based on the procedures endorsed by Henseler et al. (2015). The differences within the dataset based on the features of the subjects may not be evident in combined data. Path coefficients of the group data can validate the statistical variance by using MGA to establish the statistically significant differences amongst data based on the categorical features of the subjects (Henseler et al., 2015).

Data analysis

Descriptive statistics

Descriptive of the study respondents’ were offered in Table 1. An outstanding share of respondents was male, with 74.6% of the total study respondents, and the rest are female. The study respondents having experience between 0 and 3 years are 16.4% of the total sample. Moreover, respondents having experience between 4 and 5 years are 27.8% of the total respondents. The respondents having experience between 6 and 10 years are 55.7% of the total sample. However, a large segment of the respondents have the rank of constables and accounted for 52.2%, and respondents that were head constables are 10.4%. Respondents have the rank of the inspector 3.9%, sub-inspector is 16.9, and the assistant sub-inspector is 16.4% of the total sample, respectively. The majority of the study respondents are on the pay scale of BPS7, with 51.7% of the total sample. Respondents in BPS 9 are 9.9%, BPS 12 is 16.9%, BPS 14 is 16.4, and BPS 16 is 4.9% of the total sample.

Table 1 Profile of the respondents
Full size table

Validity and reliability

Following the endorsement of Hair et al. (2019), we estimate and demarcate the Smart PLS results. The study’s constructs reliabilities were assessed with alpha (α) and composite reliability (CR), and each construct shows that reliabilities come to be more than 0.519 and 0.776, respectively. Values of (α) and (CR) for each construct is well above the threshold of 0.50 (Hair et al., 2019). The results are reported in Table 2. The results indicate that the study constructs are consistent and reliable. AVE for each construct must be above the value of 0.50 for instituting the convergent validity to provide evidence of the uni-dimensionality for each construct (Hair et al., 2019). Variance inflation factor (VIF) for each construct reported in Table 2, all the VIF values are less than 3.3, establishing the lack of multi-collinearity problems among the study constructs. Items show that constructs have acceptable convergent validity (see Table 2). The item loading and cross-loading reported establishing the construct discriminant validity. The study construct has acceptable discriminant validity (see Annexure Table 6).

Table 2 Reliability analysis
Full size table

Additionally, the Fornell-Larcker criterion (1981) and hetrotrait and monotrait, HTMT ratio test, employed to corroborate the study constructs for discriminant validity. Fornell-Larcker criterion estimates the square root of AVE for each construct; the construct AVE of square root needs to be higher than the other constructs (Hair et al., 2019). The HTMT ratio essentially is less than 0.90 to provide the sign for discriminant validity for study constructs (Henseler et al., 2015). Cross-loading table establishing the discriminant validity is provided in the Annexure Table 6.

Path analysis

Afterwards, model measurement accomplished later the recognition of model validity and reliabilities. In this stage, the influence of the ABS, CMT, and FTB on the RVA is assessed. The adjusted r2 value for the three response variables (i.e., ABS, CMT, and FTO) on the individual RVA elucidates that 45.2% of change in revenge attitude. The Q2 value for the part of the model is 0.274 indicating a medium predictive relevance (Chin, 2010). Moreover, the adjusted r2 value of RVA and FTO explains 39.1% of the variance in the EKH. The Q2 value for the part of the model is 0.195 indicating a medium predictive relevance (Chin, 2010). Additionally, the adjusted r2 value of FTO and RVA explains 4.1% of the variance in the PKH. The Q2 value for the part of the model is 0.013, indicating a small predictive relevance (Chin, 2010). Moreover, the effect of FTO and RVA adjusted r2 value explains 12.1% of the variance in the EKH. The Q2 value for the part of the model is 0.083, indicating a medium predictive relevance (Chin, 2010).

Study standardized path values, t-values, and significance level are illustrated in Table 4. The path coefficient between ABS and RAV (β = 0.253, p = 0.001) indicates a significant and positive effect of the ABS on the RAV. This result forms statistical support to accept the H1a. The path value for the CMT and RVA (β = 0.436, p = 0.000) shows the CMT’s effect on the RVA as positive and significant and offers statistical provision to agree with the H1b. The path value for the FTO and RVA (β =  − 0.043, p = 0.077) displays the FTO’s influence on the RVA as negative but insignificant and offers no statistical endowment to agree with the H1c. The path coefficient for the RVA and EKH (β = 0.477, p = 0.000), depicting the effect of RVA on the EKH as positive and significant, provides statistical evidence to support the H2a. The path coefficient for the RVA and PKH (β = 0.098, p = 0.030) depicts the effect of RVA on the PKH as positive and significant; it provides the statistical evidence to support the H2b. Moreover, the path coefficient for the RVA on the RKH (β = 0.196, p = 0.000) depicts the effect of RVA on the RKH as positive and significant; it provides the statistical evidence to support the H2c. The path value for the FTO on the EKH (β =  − 0.246, p = 0.000) shows that the FOB’s impact on the EKH comes significant and negative, thus providing statistical support for the H3a. The path coefficient for the FTO and PKH (β =  − 0.109, p = 0.039) depicts the effect of FTO on the PKH as negative but significant; it provides the statistical evidence to support the H3b. Moreover, the path coefficient for the FTO on the RKH (β = 0.035, p = 0.090), depicting the effect of FTO on the RKH as positive and insignificant, provides no statistical evidence to support the H3c. Path coefficients are shown in Table 3.

Table 3 Hypothesis testing
Full size table

Moderating effect of gender and work experience

Multiple group analyses were executed to appraises the group based on the sample characteristics of experience and gender. A non-parametric test of MGA was utilized to estimate the variances between the model paths based on experiences and gender. Tables 4 and 5 show the path values for two groups and the differences within the groups with the p-values as recommended by Henseler et al. (2015). The PMGA signifies the p-values realized by the multiple group analysis of PLS-SEM as the degree of significance of the variance among the groups in the study (Henseler et al., 2015).

Table 4 Multiple group comparison based on experience
Full size table
Table 5 Multiple group comparison based on gender
Full size table

Effects on experience within the groups

The outcomes of the two groups were created on the experience of the study sample. The groups based on experience signifies difference exists for the relationship between CMT and RVA and between FTO and RVA. Moreover, significant differences exist for the path between FTO and PKH, FTO, and RKH. The variance of experience does not impact the relationship between other paths of the study.

Effect of gender within the groups

The consequences of two groups based on the gender of study subjects confirmed that between the groups based on the gender significant difference exists for the relationship between the ABS and RVA, and between the FTO on PKH. The difference in gender does not affect the relationship between other paths of the study.

Discussion

Development of the RVA

Results of the study postulate that ABS significantly influence the RVA; the effect size is small but significant (Cohen, 1988). The results offer support to accept the H1a. Our study finding agrees with the postulation made by Liu et al. (2018) that the ABS promotes deviant behaviours as employees cannot tackle the ABS actively. The employee adopts the passive reaction of revenge as a last parsimonious approach to deals with the ABS. Subsequently, the study analysis suggests that the CMT significantly affect the RVA and offer statistical sustenance to accept the H1b. Our study result coincides with Shao and Skarlicki (2014) findings that the CMT instigates the negative emotions of revenge to come forward as a coping strategy to deals with the CMT, while organizational management is not taking action to protect the employees. The CMT also harms service delivery, and FLEs shows disengagement at the workplace (Heizmann & Olsson, 2015).

However, the FTO insignificantly influence the RVA and offer no support to accept the H1c. FTO promotes affective commitment and instigates pro-social behaviors among the FLEs (Eisenberger et al., 2001). Our study findings suggest that FTO reduces the RVA and promotes positive psychology among the FLEs, but the results are insignificant.

Effects of RVA on knowledge hiding behaviors

Study subsequent hypotheses were proposing the influence of RVA on the three types of knowledge hiding. Our study confirms that the RVA significantly influences evasive knowledge hiding and offers statistical support to accept the H2a. Our study outcome coincides with the result suggested by Riaz et al. (2019) that the personal attitude instigates the knowledge hiding with the purpose. Withholding the vital knowledge at the FLEs end is triggered by fear, anxiety, and loss of prestige and reduces work performance (Heizmann & Olsson, 2015).

Next, our study outcome settles that the RVA significantly influence the knowledge hiding by playing dumb and suggest significant backing to accept the H2b. The outcome accords with the result proposed by Khalid et al. (2018) that the negative emotions initiate the knowledge hiding while playing dumb and escape the situation. Vengeance attitude instigates the concealment of vital information while moving the buck to another or displaying an attitude of having no idea about the information requested (Heizmann & Olsson, 2015).

The succeeding result suggests that the RVA significantly influences rational knowledge hiding and support the acceptance of H2c. Our study result agrees with the outcome proposed by Riaz et al. (2019) that activates the knowledge hiding with justified reasons. FLEs have valid reasons to evade the knowledge, and revenge attitude triggered the knowledge hidings based on the positional capacity to avoid the requested knowledge (Connelly & Zweig, 2015). The dual reason of revenge and justified purpose makes it acceptable to hide knowledge at the workplace.

Effects of FTO on knowledge hiding behaviors

Furthermore, the study evaluates the impact of felt reasonability on the three types of knowledge hiding behaviors. Our study confirms that the FTO significantly impacts the evasive knowledge hiding and reduces the intention knowledge hiding at the workplace to dodge the workplace peers. It offers support to accept the H3a. Our study findings coincide with the outcome proposed by Connelly and Zweig (2015) that the pro-social behaviors support to reduce the antisocial workplace behaviors like knowledge hiding. The positive psychology of taking an acceptable level of responsibility induce the employees to curtail the behaviors that may harm the organizational performance (Nga & Feldman, 2015).

Next, our study suggests that the felt responsibility significantly reduces the knowledge hiding by playing dumb at the employee end and accepting the H3b. Our study finding agrees with the results postulated by Kern and Grandey (2009) that reciprocity harnesses positive behavior and employees reduce the negative behaviors of knowledge hiding. Playing dumb to hide knowledge at the workplace is counterproductive workplace behaviors (Nga & Feldman, 2015). Only innate personal feeling of duty induces to curtail the knowledge hiding behaviors like playing dumb to hide the knowledge (Eisenberger et al., 2001).

Subsequently, our study outcome advocates that the felt responsibility insignificantly impacts the rational knowledge hiding and suggests not to accept the H3c. Our study finding agrees with the results that Connelly et al. (2012) postulated that employees might hide knowledge with positive intentions, and justified reason promotes knowledge hiding. Felt reasonability induces justified knowledge hiding based on the employee hierarchal position. A clearly defined organizational structure offers guidelines to employees to hide knowledge or not fully disclose requested information (Serenko & Bontis, 2016). However, the organizational culture and established work norms also influence the knowledge sharing environment in the organization (Moon & Lee, 2014).

Moderating effect of gender and work experience

The study also proposes to reconnoiter the moderating effect of FLEs experience and gender on the relationship between the study paths of the model. The results show a significant variance among the respondents’ experience between high and low experience FLEs for the CMT for the RVA and FTO and RVA. The FLEs having low experience can takes the customer to mistreat and not develop the revenge attitude. However, FLEs with a higher level of work experience develop a revenge attitude from the customer mistreatment. Young FLEs with low work experience significantly manage personal revenge attitude with the felt obligation, which is not valid for the high experienced FLEs.

Felt obligation also significantly varies the playing dumb and rational knowledge hiding among the low and high experience FLEs. The insignificant differences exist between the high and low experience police FLEs for other study’s paths.

Moreover, for the effect of gender on the relationship between ABS and RVA, and FTO for PKH displays variances of gender have significant effects. For male FLEs, the abusive supervision significantly initiates the revenge attitude than the female counterparts. Nevertheless, gender plays a significant role in reducing the incidents of playing dumb knowledge hiding with the felt obligation among the female than the male FLEs. However, insignificant variances exist for the other paths of the study.

Conclusion

This research aims to explore the effects of ABS, and CMT on the RVA, and how the FTO reduces the effect of RVA on the knowledge hiding behaviors. ABS and CMT have established a negative effect on the performance of the services sector employees. The study found that ABS and CMT develop the RVA that leads to the knowledge hiding behaviors among the service industry employees. However, the FTO regulates the possible knowledge of hiding behaviors. The study results support the claim that the ABS and CMT positively significantly influence the RVA among the FLEs. The influence of the FTO on the EKH was medium and significant; FLEs FTO reduces the EKH and PKH. Although, FLEs FTO insignificantly controls the RKH.

Moreover, FLEs experience varies for the CMT for the revenge attitude, and the FTO for the revenge attitude. However, the experience makes no difference for other paths of the model. The effect of gender makes significant differences for the ABS on the RVA, and FTO for PKH. Study outcomes bid noteworthy policy and practice directions. CMT strongly influences the RVA among the service worker than the ABS received by the employees in the service sector (Tepper et al., 2009). CMT is perceived as negative more than the ABS. That shows the FLEs understanding of the ABS, as ABS is negative but less harmful for negative behaviors than CMT for the FLEs. Moreover, ABS remains the way of working in many sectors like PSO in developing countries. The training and have a strict code of service technique allow the FLEs to hand mistreatment from the customer (Khalid et al., 2018).

Furthermore, customer needs to provide with the instruction as well to request the services from FLEs. Another practice path is to improve the FTO to enhance work performance and reduce negative work behaviors. Felt obligation improved the job holders to deal with the ABS and CMT and take a right suggested measures to get rid of developing revenge attitude and concentrate on the performance of the job.

Reducing knowledge hiding enables us to achieve efficiency and effectiveness at the workplace. However, it is essential to explore the predictors of the ABS and CMT to control incidents of negative emotions among FLEs and create trustworthiness and enjoyable workplace (Riaz et al., 2019). Supervisor roles require management training to get work done from subordinates FLEs with understanding effectively and better emotional management at the workplace. Moreover, the customer needs to tailor and be educated to behave with the FLEs. The document and strict rules enable reducing CMT events at the workplace by not following the standard role and procedures (Shao & Skarlicki, 2014). Customer mistreatment is managed well by adopting a protocol to deal with customers, and management support empowers the FLEs to deal with disturbing customers (Yeh, 2015). Moreover, knowledge hiding incidents reduce by offering incentives to perform well and taking collective responsibility of the supervisor and subordinates FLEs to cultivate the open sharing culture based on mutual respect and trust.

The current study have its strength; the current work is associated with three limitations as well. The current work aims to reconnoiter the backgrounds of the knowledge hiding in the service organization among the FLEs grounded on abusive supervision and customer mistreatment. However, the motivation of the knowledge hiding varies. Therefore, more factors also need to add to determine the influence on the knowledge hiding behaviors. The factors of organizational culture, service delivery nature, industry and work system are factors at the macro-level inducing the knowledge hiding behaviours, and disposition, attention to detail, and locus of control can bring more thoughtful of the knowledge hiding at a personal level (Tepper et al., 2009). Future research must take into account these macro- and micro-level factors to evaluate the knowledge hiding behaviors. These micro-level factors may include personality traits, personal inclination, and attitude towards the job. However, the macro-factors that need attention from the scholars are organizational climate, work structure, workplace politics, and general workplace employee engagement. The second limitation related to the current work is past knowledge hiding at a personal level. Future work needs to exclude the past behaviors of knowledge hiding towards the target or the data collected longitudinally to explore the knowledge hiding incidents previously happen with the organization. Future studies can take the aspect of service sabotage from the knowledge hiding of FLEs and backend employees. Furthermore, customer mistreatment also needs to explore from the point of view of all kinds of employees, influencing service delivery. The last limitation associated with the study is to use the sample from police employees. FLEs in the public sector face pressure from supervisors and their respective customers. So, the study has limited generalization. Therefore, future research must explore the study model with a sample taken from business marketing firms and take the sample from different geographic locations to enhance the study generalization.