The secret of reducing turnover intention: Evidence from Indonesia

Abstract Turnover still occurs in various organizations, including both profit and non-profit. Accordingly, a study of turnover intention (TI) to detect and anticipate its prevalence among employees is urgently needed. This study explored the impact of psychological capital (PsyCap) and emotional intelligence (EI) on TI mediated by organizational commitment (OC). A survey-operated causal design with a Likert-scale questionnaire was employed to gather data from 439 honorary private school teachers in Indonesia. Structural equation modeling analysis showed that EI and OC negatively affected TI, PsyCap positively affected OC, and PsyCap negatively affected TI through OC. However, PsyCap did not significantly affect TI, EI did not significantly affect OC, and EI did not significantly affect TI mediated by OC, which is inconsistent with previous studies. These inconsistencies create a research gap that requires scientific clarification and confirmation in future research. Practitioners can use this study’s results to reduce TI through the PsyCap, EI, and OC perspectives.


ABOUT THE AUTHORS
Furtasan Ali Yusuf has a doctorate in Education Management from Jakarta State University and is an associate professor at Bina Bangsa University, Serang, Banten. He actively writes articles in various national and international journals in HRM and education management. His published works include Human Resource Management (2020). W. Widodo has a doctorate in Education Management from Jakarta State University and is a senior lecturer at the Department of Social Sciences Education, Indraprasta University PGRI, Jakarta. He actively writes articles in various national newspapers and national and international journals in educational psychology, education management, and HRM. His published works include Popular and Practical Research Methodology (2019) and Foundation of Modern Education (2020).

PUBLIC INTEREST STATEMENT
Turnover intention negatively impacts organizations, especially when facing changes and challenges in society and industry. This study provides an alternative solution to reduce turnover intention through psychological capital (PsyCap), emotional intelligence (EI), and organizational commitment (OC). The results prove that EI and OC affect turnover intention, PsyCap influences OC, and PsyCap affects turnover intention mediated by OC. However, PsyCap does not affect turnover intention, EI does not influence OC, and EI does not affect turnover intention mediated by organizational commitment. This evidence is crucial for future researchers and practitioners.

Introduction
Turnover is the enemy of organizations as it can seriously disrupt their life dynamics. Hence, the phenomenon of turnover as a form of voluntary termination of employment at the initiative and desire of employees has long attracted the attention of practitioners, researchers, and academics (Hom et al., 2017). Previous studies have indicated that turnover negatively affects employees, groups, organizations, and business continuity. At the individual level, TI potentially reduces productivity (Park et al., 2013) and employee performance (Lin & Huang, 2021). At the group level, TI can affect team quality and performance (Gupta, 2019). At the organizational level, turnover negatively influences organizational performance (Lai & Chen, 2012), efficiency (Rana & Abbasi, 2013), and effectiveness (Ahmed et al., 2016). Finally, turnover is particularly detrimental to a company in the business context because it is associated with enormous expenses (Hall & Smith, 2009).
Turnover refers to when an employee leaves the organization of his/her own free will or due to the decision of the organizational authority and is replaced by a new employee (Wei, 2015). Turnover can occur in two forms, which are based on the underlying motives: that which occurs at the will of employees and due to the organization's decisions to terminate an employee is called voluntary and involuntary turnover, respectively (Long et al., 2012;Robbins & Judge, 2019;Wei, 2015). Turnover intention (TI) basically refers to employees' tendency to try to find another job after they leave their current one (e.g., Chen et al., 2021;Ghadi, 2017;Wombacher & Felfe, 2017). Thus, TI is characterized by the intention and desire to leave the organization, the possibility of leaving the current job, the desire to find another job, and the opportunity to work elsewhere Suyono et al., 2021).
Symptoms of TI also occur in Indonesia among honorary teachers who work in private schools on a part-time basis and are paid based on the number of teaching hours. As a result, their monthly income accumulation is far below the regional (provincial) minimum wage standards. Under these conditions, it is unlikely that they will be able to live appropriately. However, teachers have addressed this issue in various ways. A small number of them willingly accept low wages; however, most are forced to accept the situation. For the latter, the poor working conditions create pressure that motivates them to look for other, more decent jobs. In this context, they are in a situation of voluntary TI; that is, their desire to resign and stop working at private schools is based on their own desires and decisions. This situation can interfere with the implementation of the learning press in schools, which in turn can have implications for students' academic achievement and the quality of school output. On a broader scale, these conditions can interfere with achieving national educational goals with respect to attracting and retaining quality and highly competitive human resources. Thus, a critical question related is "What factors trigger TI among honorary teachers?" In reality, TI can be triggered by many factors, including psychological capital (PsyCap), emotional intelligence (EI), and organizational commitment (OC). Studies by Rivaldi and Sadeli (2020) and Suganda (2022) in Indonesia show that PsyCap influences TI, and others by Giao et al. (2020) and Sharma and Tiwari (2022) have demonstrated that EI is related to TI. Recent research results from several countries also prove that TI is affected by OC (e.g., Faliza et al., 2023;Solikhah & Andriani, 2023;Zhu et al., 2022). However, other studies have reported contrasting (inconsistent) results. For example, Sutrisno and Widhianto's (2021) investigation indicated that EI does not affect TI. Hakim et al. (2022) results also revealed that OC has no significant relationship with TI. Other studies have also found that in addition to affecting TI, OC is also influenced by PsyCap (Pant & Parveen, 2022;Ribeiro et al., 2021) and EI (Alismail et al., 2022;Dhungana & Kautish, 2020). This suggests that OC may mediate the relationship between PsyCap, EI, and TI. However, previous studies have reported contradictory results. For example, Pariat et al. (2017) demonstrated that PsyCap did not significantly affect OC. Aghdasi et al. (2011) also indicated that EI does not significantly influence OC. The inconsistent research results have created a research gap that requires scientific clarification. Based on this urgency, this study focuses on how PsyCap, EI, and OC affect TI and attempts to confirm that PsyCap and EI affect TI through OC.

PsyCap and TI
PsyCap is currently receiving attention from scholars for many reasons. Based on studies in several countries, PsyCap positively contributes to a person's life and career. For instance, Imran and Shahnawaz (2020) claimed that PsyCap significantly affects performance, including academic performance (Carmona-Halty et al., 2019). Other studies' results indicate that PsyCap affects organizational citizenship behavior (OCB) (e.g., Chamisa et al., 2020;Waters et al., 2020;Yildiz, 2018). PsyCap is also related to innovative behavior (Slåtten et al., 2020;Wang et al., 2021) and career outcomes (Baluku et al., 2021). PsyCap explores beyond human and social capital theory and answers the following questions: "What do we know?", "Who do we know?", and "Who are we?" (Luthans & Youssef-Morgan, 2017). PsyCap is the primary capital for determining and building awareness, effort, and work efficiency (Çavus & G€okçen, 2015). It reflects favorable conditions and individual development via four characteristics: self-efficacy, optimism, hope, and resilience. Self-efficacy is the belief in one's ability to complete tasks. Optimism relates to the careful assessment and consideration of specific situations to position oneself well. Hope is the belief in the determination of ways to achieve goals. Finally, resilience is the ability to face difficulties or setbacks, challenges, and constantly changing environments (Luthans & Youssef, 2004;Luthans & Youssef-Morgan, 2017;Tang et al., 2019). In high conditions, these four indicators can potentially reduce teachers' TI. Karatepe and Avci (2017) revealed that PsyCap affects TI, mediated by work engagement. Likewise, several other studies concluded that PsyCap influences TI (e.g., Chen et al., 2021;Da et al., 2020;Dhiman & Arora, 2018;Saraswati, 2019;Wen, 2020;Yim et al., 2017;Çelik, 2018). Accordingly, we formulate the first hypothesis (H): H 1 : PsyCap negatively affects TI.

EI and TI
Like PsyCap, empirical evidence has found that EI positively impacts individuals' and organizations' lives. Silva and Coelho (2019) demonstrated that EI is related to creativity. Additionally, EI was found to influence leadership change and effectiveness (Issah, 2018;Lone & Lone, 2018). Another study indicated that EI increased OCB, task performance, job satisfaction and organizational commitment (Alsughayir, 2021;Miao et al., 2018). At the organizational level, EI drives projects to be more successful (Doan et al., 2020), and in the educational context, EI affects teaching effectiveness (Shahid et al., 2015), teachers' professional competence (Widodo et al., 2022), and academic performance (MacCann et al., 2020;Sánchez-Álvarez et al., 2020). EI is the capacity to understand, feel, and actualize emotional sensitivity as a source of energy, connection, information, and influence on humans to act (Cooper & Sawaf, 1997). According to Mayer and Salovey (1997), EI relates to understanding, appreciating, and expressing emotions appropriately to evoke feelings that facilitate the development of (intellectual) thinking and emotions. Bradberry and Greaves (2009) described EI as ability, talent, careful judgment, and good sense management in dealing with other individuals. Conceptually, EI includes both intrapersonal and interpersonal aspects. Intrapersonal intelligence is an individual's internal intellectual capacity to understand themselves and build awareness, inspiration, and self-regulation. Meanwhile, interpersonal intelligence is an external intellectual capacity that individuals use to understand and establish relationships with others, such as sympathy and empathy (Suleman et al., 2020). According to Wijekoon et al. (2017), individuals with higher EI tend to understand, regulate, and control emotions well in themselves and others. They also tend to better understand the causes of their and others' behavior. For example, they understand why people behave in certain ways and how they can regulate their own behavior and that of others, which can help direct the growth and success of a person and those around them (Mahanta & Goswami, 2020). This suggests that EI can help individuals become proficient in differentiating constructive and destructive emotions so that they can express emotions that lead to better growth, creativity, efficiency, and effectiveness. EI comprises five indicators: self-awareness, self-regulation, motivation, empathy, and relationship management. Self-awareness refers to the ability to perceive and use emotional preferences in making decisions and to provide realistic assessments of rational abilities and self-confidence. Self-regulation is related to managing emotions that can facilitate the proper completion of tasks, being aware of delaying gratification for a moment to pursue goals, and recovering immediately from emotional stress. Motivation refers to using the deepest preferences to move and guide individuals toward the desired goal by taking initiative and continuing to fight. Empathy is the strong desire to feel what others feel based on their perspective. Finally, relationship management is related to handling emotions and fostering good relationships (Goleman, 1998). In high conditions, these indicators can potentially reduce TI among teachers. Researchers have also claimed that EI significantly affects TI (e.g., Bartk, 2019;Fernando & Gamage, 2019;Hanib & Wan Abdullah, 2020;Hosain, 2019;Mir, 2018;Soleimani et al., 2017;Yiğit, 2018;Zeidan, 2020;Zulmi et al., 2020). Therefore, we propose the second hypothesis: H 2 : EI negatively affects TI.

OC and TI
TI can also be influenced by OC. Several prior studies in various organizations and countries have found that OC impacts TI (e.g., Güllü et al., 2020;Nurtati et al., 2020;Nurtjahjono et al., 2023;S. Zhou et al., 2020;Sabarrudin & Djamil, 2023;Scales & Brown, 2020). This finding indicates that OC is a vital predictor of TI. Conceptually, commitment refers to the strength of employees in identifying with themselves, participating actively, and making the best efforts to benefit the organization (Newstrom, 2017;Noe et al., 2019). Thus, OC reflects employees' strong desire to be part of the organization, readiness to make greater efforts to achieve organizational goals, and acceptance of organizational values (Doan et al., 2020;Riana, 2021), which is reflected in daily work activities (Beardwell & Thompson, 2017). In addition, OC can be described as an attitude that demonstrates employees' loyalty to an organization (Hadian, 2017). OC comprises three components as indicators: affective, continuance, and normative commitment. Affective commitment refers to employees' emotional attachment to identifying with themselves and being involved in the organization. Continuance commitment relates to feelings of loss if employees leave an organization. The third is normative commitment, which reflects employees' feelings of fulfilling the various normative rules of the organization (Meyer & Allen, 1991). At a high level, these three components may potentially reduce teacher TI. Hence, we postulate the following: H 3 : OC negatively affects TI.

PsyCap and OC
OC has a unique position. Apart from being an exogenous variable, it is also an endogenous variable. For example, in addition to affecting TI, OC is significantly influenced by PsyCap (e.g., Chugh, 2020;Huynh & Hua, 2020;Nangoy & Hamsal, 2018;Nguyen & Ngo, 2020;Sürücü et al., 2020;Tang et al., 2019), indicating that PsyCap is a crucial antecedent of OC. As an illustration, teachers with high self-efficacy-reflected in their belief in their ability to complete specific taskswill be actively involved in various school-specific activities such as teaching. Likewise, teachers with higher optimism tend to consider school conditions as a reason to survive and continue teaching at school. This confirms that PsyCap affects teachers' OC. Thus, we propose the following: H 4 : PsyCap positively affects OC.

EI and OC
The research results in several countries indicate that as an endogenous variable, OC is also affected by EI (e.g., Alsughayir, 2021;Baker et al., 2019;Johar et al., 2020;Mahanta & Goswami, 2020; Ramli & Novariani, 2020; Rizki et al., 2019). In practice, teachers with high motivation possess a strong will to try hard to achieve the desired goals and are actively involved in their professional tasks. Likewise, teachers with high self-awareness in utilizing emotional preferences in making decisions and providing assessments tend to be more realistic in accepting values and realizing school organizational goals. This supports the following hypothesis: H 5 : EI positively affects OC.

Mediating role of OC
Research that explicitly investigates the mediating effect of OC on the causal relationship between PsyCap, EI, and TI still needs to be conducted. However, from the results of previous research in several countries and contexts, several important findings indicate that PsyCap and EI affect TI through OC. For example, Nguyen and Ngo (2020), Chugh (2020), Ribeiro et al. (2021), and Pant and Parveen (2022) demonstrated that PsyCap significantly affects OC. Ramli and Novariani (2020), Dhungana and Kautish (2020), Alsughayir (2021), and Alismail et al. (2022) showed that EI significantly impacts OC. Other studies have found that OC is related to TI (e.g., Faliza et al., 2023;Güllü et al., 2020;Scales & Brown, 2020;Solikhah & Andriani, 2023;Zhu et al., 2022). From these studies, it appears that OC simultaneously acts as both an exogenous and an endogenous variable so that it can be positioned as a mediator variable. Accordingly, we propose the following hypotheses:

Participants
The target population for this research was private school honorary teachers in Indonesia, while the accessible population as a sample frame was private school honorary teachers in the provinces of West Java, Central Java, DKI Jakarta, Banten, Riau Island, East Kalimantan, and East Nusa Tenggara, which are spread across five islands. Finding valid data on the number of honorary private schoolteachers in Indonesia is difficult because honorary teachers can work in several schools. Consequently, it is difficult to record the exact number of honorary teachers. As data on the number of honorary teachers were not available as a basis for determining the sample, this study used an accidental sampling technique, which relies on honorary teachers' voluntary willingness to fill out a complete questionnaire during the research without receiving any compensation (Widodo, 2019). A total of 455 questionnaires were collected using this technique. Of these, 439 questionnaires were completed and analyzed as the sample (participants) for this study. As presented in Table 1, the majority were female (64.01%) and aged 26-35 years (40.77%). In addition, most participants had a bachelor's degree (87.24%) and were married (70.62%). Finally, most had≤5 years (40.55%) of teaching experience.

Data analysis
This study uses a Likert-scale questionnaire, a sample of 439 teachers, and a causal design. Therefore structural equation modeling (SEM) was considered an appropriate data analysis technique for testing the hypotheses and research models. Additionally we employ descriptive analysis to describe the condition of the indicators in each construct (variable) and correlational analysis to determine the relationship between indicators in all constructs (variables). SEM analysis was performed using LisRel 8.80, whereas descriptive and correlational analyses were performed using SPSS version 26.

Results
As presented in Table 2 Overall, the mean value obtained was higher than the SD value. Therefore, the overall representation of the data is good. In addition, the correlation analysis results between the indicators revealed a significant relationship (p < .01). This condition shows that all the indicators have a reciprocal relationship. However, this relationship does not indicate the currency of multicollinearity symptoms because all the correlation coefficient values are <.8. Table 3, the results of the confirmatory factor analysis show that all factor loading values (λ) are greater than .30, while the t value is greater than the critical value (1.65). This indicates that all indicators are valid (Costello & Osborne, 2005), so it can measure all research constructs (variables). In addition, it obtained a construct reliability (CR) value greater than .70 and an average variance extracted (AVE) value greater than .50. It shows excellent internal consistency (reliability) and acceptable convergent validity (Hair et al., 2018).

As displayed in
The output of LisRel 8.80, summarized in Table 4, shows the goodness of fit (GOF) index results. Of the 11 measurement criteria, nine indices met the requirements (good fit), while the others did not (poor fit), namely Chi-Square and Sig. probability. In many cases, Chi-Square tends to be sensitive to large sample sizes of more than 200 (Hair et al., 2018), as in this study involving 439 teachers. Therefore, in this study, the Chi-Square index did not meet these requirements (poor fit). However, overall, these results can be considered fit because the majority (nine) of the criteria tested fulfill the requirements. Table 5 and Figures 1 and 2, the SEM results indicate that four hypotheses are significant (supported) with a t value > t table at α = .05and .01; they are H 2 , H 3 , H 4 , and H 6 . Therefore, EI has a negative direct effect on TI (γ = −.21, p = .05), OC has a negative direct effect on TI (β = −.39, p = .01), PsyCap has a positive direct effect on OC (γ= .67, p = .01), and PsyCap has a negative indirect effect on TI mediated by OC (β = −.26, p = .01). This suggests that improving EI and OC can reduce TI. In addition, high PsyCap potentially increases OC and mitigates TI supported by OC. However, the other three hypotheses (H 1 , H 5 , and H 7 ) are non-significant (unsupported) with a t value > t table at α = .05.Hence, PsyCap does not negatively directly affect TI (γ = −.15, p = .05), EI does not positively directly affect OC (γ = −.04, p = .05), and EI does not indirectly affect TI mediated by OC (β = .02, p = .05). This shows that adequate PsyCap cannot be relied on to overcome TI among teachers. Further, high EI cannot increase teachers' OC, nor can it even mitigate TI when obtaining OC support.

Discussion
This study proved that EI negatively affects TI and confirmed that EI can reduce TI. Thus, improving EI can mitigate potential TI. This evidence aligns with previous scholars' finding that EI significantly affects TI (e.g., Bartk, 2019;Fernando & Gamage, 2019;Hanib & Wan Abdullah, 2020;Hosain, 2019;Zeidan, 2020;Zulmi et al., 2020). In addition, the research results provide empirical evidence that OC negatively affects TI, confirming that OC's existence potentially reduces TI; therefore, increasing OC among teachers can reduce TI. This finding is similar to those of several prior studies that concluded that OC influences TI (e.g., Güllü et al., 2020;Nurtati et al., 2020;S. Zhou et al., 2020;Scales & Brown, 2020). This study also revealed that PscyCap positively   affects OC. These findings indicate the empirical fact that a person's PsyCap can increase OC. Thus, improving the actual condition of PsyCap teachers could trigger an increase in OC. This is consistent with the results of Huynh and Hua (2020), Sürücü et al. (2020), and Nguyen and Ngo (2020), who found that PsyCap significantly affects OC.
In addition, this study found a significant mediating effect of OC on PsyCap's relationship with TI, which shows the strategic position of OC in the causal relationship between PsyCap and TI. From this perspective, OC can mediate the effect of PsyCap on TI; therefore, its existence needs to be considered in reducing TI via PsyCap. This evidence is consistent with that of previous studies that concluded that PsyCap impacts OC (e.g., J. Zhou et al., 2018;Nguyen & Ngo, 2020;Pant & Parveen, 2022;Ribeiro et al., 2021) and OC with TI (e.g., Faliza et al., 2023;Ong et al., 2019;Solikhah & Andriani, 2023;Terason, 2018;Zhu et al., 2022).
This study also reveals other empirical facts inconsistent with those of several previous studies. For example, EI does not affect OC, PsyCap does not influence TI, and EI does not indirectly affect TI mediated by OC. The relationship between EI and OC indicates that teachers' EI does not affect OC. It is difficult to understand whether teachers' problems related to their income from schools as compensation for teaching, which is very limited and far from sufficient to fulfill a decent life, is not considered. However, if it can be cited as an excuse, then the absence of EI's influence on OC is understandable. Therefore, it makes sense that these findings are inconsistent with those of Mahanta and Goswami (2020), Ramli and Novariani (2020) and Alsughayir (2021). However, this is similar to Aghdasi et al. (2011), who found that EI had no significant relationship with OC. These findings further deepen the research gap regarding the relationship between EI and OC, which requires scientific clarification as soon as possible.
Regarding the link between PsyCap and TI, PsyCap cannot reduce TI. As in the case of EI, this is related to the minimal teacher salary. When teachers are economically depressed, no matter how strong PsyCap is, it cannot reduce TI. Therefore, it can also be understood why this finding contradicts those of previous studies, such as Wen (2020), Da et al. (2020), and Chen et al. (2021), that determined that PsyCap influences TI. The influence of EI on TI mediated by OC seems to be due to the absence of EI effects on OC, which subsequently reduces the chances of EI's influence on turnover through OC. Hence, the findings are inconsistent with the results of previous studies and create a research gap deeply more that requires scientific clarification and confirmation through future research. However, the inconsistency in the research results does not negate the contribution of PsyCap, EI, and OC. These three variables either directly or indirectly contribute to TI. Accordingly, these three variables need to be improved and enhanced to contribute more to reducing TI. although the theoretical model, along with the hypotheses of this study, are not fully confirmed, the empirical evidence generated through hypothesis testing makes a scientific contribution, especially linked to the causal relationship between PsyCap, EI, OC, and TI. In addition, the empirical evidence produced by this study provides practical implications for the delivery of education in private schools to mitigate the intensity of turnover among teachers. School management can take advantage of the potential of PsyCap, EI, and teacher OC, with mediating dynamics to reduce teacher TI. Other organizations, such as business organizations, can also use this study's results as part of a strategy to anticipate employee turnover.

Conclusion
This study focuses on PsyCap, EI, and OC affecting TI and attempts to confirm that PsyCap and EI affect TI through OC. The results showed that EI and OC have a negative direct effect on TI, PsyCap has a positive direct effect on OC, and PsyCap has a negative indirect effect on TI mediated by OC. In addition, PsyCap has no negative direct effect on TI, EI has no positive direct effect on OC, and EI has no negative indirect effect on TI mediated by OC, which is inconsistent with the results of prior studies. These inconsistencies have generated a research gap deeply that requires scientific clarification and confirmation in future research. Practitioners can take advantage of these studies' results to reduce TI through PsyCap, EI, and OC perspectives.

Limitations and future research
Although this study has been conducted carefully based on scientific procedures and methods, it has several limitations. For example, it only focuses private schools with a limited unit of analysis for Indonesian honorary teachers in the provinces of West Java, Central Java, DKI Jakarta, Banten, Riau Island, East Kalimantan, and East Nusa Tenggara. Therefore, further research should be expanded to include other provinces to broaden the generalization area. This study also does not use all theoretical dimensions or indicators in the literature; hence, different dimensions or indicators are needed for future research. Finally, this study used only one source (teachers). Accordingly, further research should consider other data sources, such as colleagues (teachers), so that they are more promising for objectivity.

Funding
The authors received no external funding for this research. Faculty, Universitas Indraprasta PGRI, Jakarta, Indonesia.

Disclosure statement
No potential conflict of interest was reported by the author(s).