A scoping review of research on the turnover intention of military personnel in Taiwan

Turnover intention can be defined as ‘employees' awareness or thoughts about leaving the job’ (Park and Min 2020, p. 2). Meta-analysis has shown that intention is a critical causal antecedent of actual behaviour (Hagger et al. 2023). Hence, studying turnover intention is of paramount importance, as it may become too late to prevent the consequences of individuals leaving an organisation once they have made their decision. The leaving of employees can result in increased personnel, recruitment and training costs, while also indirectly impacting team dynamics (Ongori 2007). The weighted median cost of turnover was estimated to be $71,613—approximately 1.45 times the annual median household income in the United States in 2010—and was associated with several adverse effects on remaining employees, including understaffing, increased workload and elevated stress levels (Patterson et al. 2010; DeNavas-Walt et al. 2011). Yarbrough et al. (2017) also reported that staff turnover has a huge financial impact on an organisation. The cost of each lost staff member is estimated to be between $44,380 and $63,400, and each organisation may face financial losses of up to $6,020,000 per year, including recruitment, training costs and reduced productivity during the adaptation period of new personnel. Furthermore, the departure of employees is likely to lead to lost knowledge and intellectual assets and, hence, negatively influences the competitiveness of an organisation (Alshanbri et al. 2015).

Scholars typically categorise turnover into two main types: broad and narrow. The broad definition encompasses all forms of labour mobility, such as changes in geographical location, occupation and industry within an organisation, as outlined by Huang (2016). In contrast, the narrow definition focuses on the exit mode of turnover, highlighting the transition of employees from within an organisation to outside it, as described by Bolt et al. (2022). The current review will focus on the narrow definition of turnover.

In the current study, we leveraged Werner and DeSimone's (2011) model of employee behaviour to analyse and summarise the existing literature on the turnover intention of military personnel in Taiwan (Figure 1). This model was chosen in that it was developed specifically for understanding how environments, employees' psychological processes and behaviours are related to one another. Furthermore, this model has been empirically tested in the context of the workplace (Long et al. 2012; Lin and Yoo 2013). According to the model, employees' turnover intention (Box E) serves as an immediate antecedent of employee behaviour (Box F), which is consistent with conceptual frameworks and findings of many meta-analytic studies (Hagger et al. 2023). Individuals' turnover intention is directly determined by individuals' psychological processes (Box D), which encompass the unobservable and malleable factors – such as motivation, affect, attitude and cognition – that drive behaviour (Deci et al. 2017; Garrick and Chan 2017; Elfenbein 2023; Kammeyer-Mueller et al. 2024). These psychological processes are assumed to be shaped by factors in the work environment (Box B), such as supervision, reward structure, job design and co-workers, and by personal characteristics (Box C), which represent relatively stable and less malleable factors, such as demographics and personality traits (Kang et al. 2023; Wang et al. 2024).

Fig. 1:

A model of employee behaviour.

While both psychological processes and personal characteristics fall under the broader category of intraindividual factors, they differ in observability and malleability. Personal characteristics – such as gender and race – are typically more observable and less modifiable within a given context, whereas psychological processes are less visible and more susceptible to change. Given that individuals enter organisations with inherent or long-developed characteristics, our model of employee behaviour posits that both factors in the work environment and these relatively stable, observable personal characteristics serve as activators of individuals' in-situ psychological processes. This conceptualisation aligns with other well-established theoretical frameworks (e.g., Pintrich and Zusho 2007; Pekrun et al. 2023) and is supported by empirical findings (Amarasena et al. 2015; Iman and Lestari 2019; Hatinoğlu and Ergün 2020). The aforementioned factors will be impacted by factors outside organisations (Box A), such as economic conditions, labour market conditions, or laws and regulations. The model further stipulates that factors in the work environment and personal characteristics will only indirectly influence turnover intention through the pathways of individuals' psychological processes. According to this model, any differences in turnover intention between military personnel can be traced back to factors in the work environment. This conceptualisation emphasises the importance of environments and prompts scholars and stakeholders to focus on solutions that can change the structure of an organisation. It should be noted that, although the model may appear to suggest a linear causal pathway from work environment factors and personal characteristics to turnover intention, it in fact emphasises a feedback loop from individuals' behaviour back to their turnover intentions, psychological processes and aspects of the work environment. This design highlights the role of human agency – the capacity of individuals to reflect on and influence their own thoughts and surroundings. This assumption aligns with several established theoretical frameworks that emphasise reciprocal determinism and agentic processes (Bandura 2001; Pintrich and Zusho 2007; Pekrun et al. 2023).

A country's security and long-term development rely on the overall combat effectiveness of the armed forces. With the military pension reform and the low social status of military careers, the Taiwanese government is facing severe challenges in retaining military personnel. As reviewed in previous sections, turnover has a huge impact on an organisation's competitiveness. Furthermore, turnover intention is closely related to the actual behaviour of leaving a job. Therefore, there is a need to review the existing literature to identify factors that are associated with the turnover intention of military personnel in Taiwan. The current review was guided by the following questions:

RQ1: What are the types of evidence available on the turnover intention of military personnel in Taiwan?

RQ2: What are the main findings on the turnover intention of military personnel in Taiwan?

The current study aimed to determine the scope of existing literature and map the types of evidence available regarding the turnover intention of military personnel in Taiwan. Given this objective, a scoping review was deemed more appropriate than a systematic review, as it better aligns with the goal of synthesising and organising a broad and diverse body of research (Munn et al. 2018). The current review followed the methodological framework for scoping reviews outlined by Mak and Thomas (2022). Studies considered for this review were primarily related to the turnover intention of military personnel in Taiwan. The decision to focus the review on Taiwan reflects the call for more ‘situationally specific and culturally bound’ research in psychology (Jacquelynne and Allan 2020, p. 2). To identify relevant studies, we initially searched Scopus [1] to examine the availability of English-language literature. This search yielded only eight studies, none of which directly addressed the turnover intention of military personnel. These results suggest that relevant research may be published in languages other than English. Since Mandarin Chinese is Taiwan's official language, we extended our search to the Airiti Library, the largest electronic database for Chinese-language academic resources. The Boolean search terms used were: ([ALL] = ((“turnover intention” OR “intent* to stay” OR “intent* to leave” OR “intent* to remain” OR “intent* to serve” OR “intent* to resign”)) AND [ALL] = ((“military” OR “army” OR “navy” OR “air force” OR “soldier*” OR “sergeant*” OR “troop*”))). The country/region filter was set to ‘Taiwan’. The entire research team conducted the search in November 2023, which initially identified 346 potentially relevant studies. Based on the pre-established inclusion and exclusion criteria, the team screened titles and abstracts, followed by full-text reviews to determine eligibility. Cohen's kappa coefficients for interrater reliability were .78 for abstract screening and .44 for full-text screening, indicating moderate to substantial agreement (Cohen 1960).

Eligibility criteria were established a priori in accordance with scoping review best practices (Levac et al. 2010). To be included, studies had to meet all six of the following conditions: (a) be published in peer-reviewed journals; (b) be based on a sample of Taiwanese military personnel; (c) address the topic of turnover intention; (d) employ quantitative research methods; (e) include a methods section and report empirical findings; (f) be written in Chinese. Applying these criteria, 317 studies were excluded after abstract screening. Of the remaining 29 studies, 12 were excluded after full-text review. Ultimately, 17 studies met all inclusion criteria. The PRISMA 2020 flow diagram (Figure 2) summarises the study selection process and enhances transparency in reporting (Tricco et al. 2018).

Fig. 2:

PRISMA 2020 flow diagram.

An initial coding schema was developed that included the following elements: author, year, sample size, data source, research design, analytical approach, causal antecedent (X), mediator (M), moderator (W), outcome variable (Y) [2] and research focus. ‘Research focus’ was categorised according to the dimensions outlined in Werner and DeSimone's (2011) model. The included studies were first divided equally among members of the research team. Each member then extracted the relevant information from their assigned articles. To ensure consistency and accuracy, the first author reviewed all coded data. Any discrepancies were resolved through discussion among the team during the coding process.

Figure 3 presents the number of records on the turnover intention of military personnel in Taiwan per year. Studies on this topic date from 2003. Since 2013, there have been an increasing number of related works published. The number peaked in 2019 and fluctuated afterward. The sample size ranged from 176 to 994 participants. These 17 included studies encompassed a total of 6,256 members of military units in Taiwan. Self-reported survey is the main data source (n = 17; 100%). Furthermore, most of the studies (n = 16; 94%) employed a non-experimental correlational and predictive research design and were based on cross-sectional data, aiming to examine the relations between variables of interest and turnover intention. Of the included studies, eight (47%) studies explored mediation mechanisms, while seven (41%) studies examined moderation mechanisms. Regarding research focus, 12 studies (71%) examined factors in the work environment, 13 (74%) studies examined individuals' personal characteristics, 10 studies (59%) examined individuals' psychological processes, and no studies examined factors in the external environment. Tables 1 and 2 summarise the included studies.

Fig. 3:

Trends in studies on the turnover intention of military personnel in Taiwan.

A growing body of literature has identified a wide range of work environment factors associated with military personnel's turnover intention. These factors include social, supervisory and organisational support; internal and external communication; alignment between organisational culture and members' values; workplace incivility; perceived usefulness of organisational policies; onboarding and socialisation processes; task assignment practices; work–family conflict; autonomy and equity; and uncertainty (e.g., Day and Huang 2011; Hsu and Tsai 2015; Chang et al. 2017; Liu and Chen 2018; Lee et al. 2023). Several studies have demonstrated that such environmental factors predict turnover intention above and beyond the influence of personal characteristics and psychological processes (e.g., Lo and Chou 2006; Day and Huang 2011; Lee et al. 2021). As shown in Table 3, a smaller subset of studies examined multiple work environment factors within the same analytical models. These studies suggest that perceived support from supervisors and organisations, organisational culture and both person–organisation fit and person–job fit play particularly salient roles in shaping turnover intention—often more so than factors such as the physical work environment, peer collaboration, or general uncertainty (e.g., Chang et al. 2017; Liu and Chen 2018; Lee et al. 2023).

Although, in theory, personal characteristics are positioned as distal antecedents of turnover intention, several studies have been interested in exploring the direct effects of personal characteristics. Overall, many personal characteristics were found to be significantly and directly associated with turnover intention, including gender, age, rank, marital status, education, years of service, occupational specialties, salary, residence, location of deployment, organisational identification, grit, power distance and family influence (e.g., Hsu and Chiang 2019; Lee et al. 2021; Chang and Hsiung 2023; Lee et al. 2023). However, discrepant results have been reported regarding the direct effects of personal characteristics on turnover intention. For instance, while Hsu and Chiang (2019) and Lee et al. (2023) found significant effects of gender on turnover intention, many studies had different conclusions (e.g., Lo and Chou 2006; Day and Huang 2011; Wu 2014; Hsu and Tsai 2015). In a similar vein, Wu (2014) found significant effects of marital status on turnover intention, many studies found otherwise (e.g., Day and Huang 2011; Tsai and Lee 2016; Cheng 2019).

Although individuals' psychological processes have been examined less frequently, all factors of interest were found to be significantly associated with turnover intention. These factors included job stress, achievement motivation, organisational commitment, job satisfaction, surface acting, job burnout, well-being and organisational frustration (e.g., Chen and Huang 2019; Li 2019; Chang and Hsiung 2023; Lee et al. 2023). Different from the findings of personal characteristics, many research studies examined the same construct in different contexts and obtained similar results. For instance, both Wu (2014) and Li (2019) found a significant and positive association between job satisfaction and turnover intention. Both Wang and Huang (2018) and Chang and Hsiung (2023) found a significant and negative association between job burnout and turnover intention.

In addition to direct effects, a few studies were interested in examining the indirect relation with turnover intention. In terms of work environment factors as mediators, Chang et al. (2017) and Liu and Chen (2018) found that the perceived practicality of policies and organisational socialisation were indirectly related to turnover intention through the pathways of perceived organisational support and person–organisation fit, respectively. In terms of personal characteristics as mediators, Lee et al. (2021) found that social responsibility was indirectly related to turnover intention through the pathway of organisational identification. In terms of individuals' psychological processes as mediators, Hsu and Tsai (2015), Chang and Hsiung (2023) and Lee et al. (2023) found that co-worker incivility, work–family conflict and person–job misfit were indirectly related to turnover intention through the pathways of surface acting, job burnout and organisational frustration, respectively. Finally, Chen and Huang (2019) reported that job stress was indirectly related to turnover intention through the pathways of well-being and organisational commitment.

Some studies have considered the role of moderators in relation to turnover intention. Lo and Chou (2006) found that the relation between perceived supervisor support and turnover intention was moderated by achievement motivation. Li (2019) also found that the relation between perceived risk and turnover intention was moderated by organisational identification and the relation between job satisfaction and turnover intention was moderated by transformative leadership. On the contrary, Shih and Tseng (2003) found that the relationship between job stress and turnover intention was not moderated by social support acquisition. Chang and Hsiung (2023) found that the relation between job burnout and turnover intention was not moderated by grit.

Instead of focusing on turnover intention, other studies examined the moderation mechanisms involving variables of interest within the nomological network of turnover intention. Liu and Chen (2018) found that the relationship between organisational socialisation and person–organisation fit was moderated by leadership style. Chang and Hsiung (2023) also found that the relation between work-family conflict and job burnout was moderated by grit. Lee et al. (2023) reported that the relation between person-job misfit and organisational frustration was moderated by authoritarian leadership. On the contrary, Chang et al. (2017) found that the relation between the perceived practicality of policies and perceived organisational support did not differ by military personnel's rank. Figure 4 summarises the findings of the existing literature.

Fig. 4:

Hypothesised model and empirical findings [3].

Here are four suggestions for improving the analytical approaches in research on military personnel's turnover intention. First, the sample size for each group, mean, standard deviation and bivariate correlation need to be provided. Second, multiple comparisons need to be controlled for type I error rate. Third, coefficients of covariates need to be reported. Fourth, if interaction terms are not statistically significant, coefficients of the baseline model need to be reported.

There are at least two limitations associated with the current review. First, the current review is situated in Taiwan. Therefore, the findings and implications may not be generalisable to military personnel in other countries. Second, the current review only included quantitative studies published in peer-reviewed journals and excluded grey literature. Accordingly, the conclusions may not reflect all the relevant works on the turnover intention of military personnel. Despite these limitations, the current review provides important insights into what is known about the turnover intention of military personnel in Taiwan and points out what are the gaps and need to be addressed in future research.