Personality and commitment as predictors of turnover intentions among Greek employees in the lodging industry

Employee turnover rates in the tourism industry are globally considerably high. Research on the topic has focused mostly on environmental and situational factors, with little attention given to employees’ different characteristic. In the current research the effect of the Big Five personality traits and commitment on turnover intentions of lodging employees was examined. The effect of personality on commitment was examined, as well. Utilizing structural equation modelling (SEM), emotional stability and organizational commitment were found to be predictors of turnover intentional behaviour. An additional impact of occupation commitment-related variables on the prediction of organizational commitment was observed. Finally, conscientiousness was found to be the best predictor of organizational commitment. The implications of these results for future research are discussed, as well.


Introduction
Tourism is considered the world's fastest growing industry and it is among the highest earners of foreign currency (Boella & Goss-Turner, 2019). In 2019, the travel and tourism sector grew at a rate of 3.5% (World Travel & Tourism Council [WTTC], 2020). Tourism depends a lot on the human factor, as it is a labour-intensive industry (Durbarry & Sinclair, 2003;Boella & Goss-Turner, 2019;González-Santacruz, Sánchez-Cañizares, & López-Guzmán, 2014). The travel and tourism sector accounts for 10.3% of global GDP and 330 million jobs, which can be translated in 10.4% of total employment in 2019. Globally 10% of the employment is provided by the tourism industry (WTTC, 2020). However, the hospitality industry has a reputation for high levels of labour turnover worldwide, ranging from 60% to 300% (Boella & Goss-Turner, 2019;Riley, 2018;Kim, 2014;Tesone, 2008;Hinkin & Tracey, 2000;Nadiri & Tanova, 2010;Park & Min, 2020), holding one of the highest position in employee turnover among private-sector industries (US Department of Labor, 2015).
Turnover in hospitality affects profitability, competitive advantage, quality and level of service provided to customers, as well as customers' engagement and brand's image (Kang, Busser, & Choi, 2018;Dusek, Ruppel, Yurova, & Clarke, 2014). When employee turnover occurs, the company must spend time and resources to fill the vacancy. This results to in both direct and indirect costs, that can hardly be measured in detail (Karsan, 2007;Faldetta, Fasone, & Provenzano, 2013) and often exceed 100% of the annual salary for the vacated positions (Bryant & Allen, 2013). Separation, replacement, and training costs can reach 1.5 to 2.5 times the annual salary for each employee that leaves the organization (Rust, Stewart, Miller & Pielack, 1996, p. 63).
High employee turnover rates are an important issue for the Greek lodging industry, as well (Chalkiti & Sigala, 2010). The lodging sector represents 45% of Greece's total tourism income (Institute SETE [INSETE], 2015). The tourism sector represents 20.8% (around 23 billion) of Greek GDP and 21.7% of all employment (around 846,200 jobs) in Greece (WTTC, 2020). Despite the crucial impact of tourism on the Greek economy, the research focus on employee turnover issues is surprisingly low (Chalkiti & Sigala, 2010), and no official records regarding employee turnover in the Greek hospitality industry are available (Sigala, 2006).
There is thorough literature on employee turnover, but researchers and academics have been focused mostly on environmental and situational factors (Zimmerman, 2008). However, according to Zimmerman (2008), instead of trying to retain high-performing employees after they are hired, it might be worth taking into consideration their propensity to quit, regardless of the working conditions or circumstances.
According to Mercer (1988, p. 38) "the most direct, cost-effective way to reduce turnover is probably to use valid, reliable employment selection tests that pinpoint which applicant fits the model of the longterm, effective employee". Cho, Johanson and Guchait (2009) in their study of management practices on hospitality firms (hotel and restaurants with more than 100 employees) in the US, revealed that even though pre-employment tests are rarely used in the industry, when they do so they contribute to lower the turnover rates of non-managerial employees. This implies that using pre-employment tests can indeed decrease turnover rates. However, tourism and especially the hospitality sector has built a reputation of poor and opportunist staff selection methods lacking even widely established practices (Anastassova & Purcell, 1995;Hoque, 1999;McGunnigle & Jameson, 2000;Armstrong & Taylor, 2014).
Employees joining an organization already have some previous work experience but also feelings and attitudes towards responsibilities, tasks and the working conditions under which they are asked to work Therefore, the second purpose of the study is to examine the role of personality on commitment and in turn the role of commitment on turnover intention in the lodging context.
After reviewing the related literature through Web of Science database to the authors' knowledge, no research has been found in the accommodation sector focusing on exploring employees' turnover intentional behaviour on a dispositional base. Should a dispositional feature of turnover and commitment be confirmed, then evaluating personality during the selection procedure can become valuable for managers, who will be supported to hire people that have less vocation to express turnover behaviour. Additionally, the results of the current study can contribute to the growing body of literature regarding personality, commitment, and turnover intentions of employees in the hospitality context, as research on the topic is very limited.

Literature review and hypothesis development Personality and turnover
There are different models examining personality. The Five Factor Model (FFM) is one of the leading models that evaluates personality and has been widely examined in numerous countries and languages (Armstrong & Taylor, 2014;Tews, Stafford, & Tracey, 2011;Wu & Mursid, 2019). In the work context, the Five Factor Model has been used in different studies to examine the predictive validity of personality at work (Barrick, Mount, & Judge, 2001). However, according to Salgado (2002), little systematic research has been focused on investigating whether counterproductive behaviours such as turnover could be predicted by estimating the Big Five personality traits. Extraversion, emotional stability, agreeableness, conscientiousness, and intellect are the elements of the Big Five Factor Model, that can explain personality differences among individuals (Costa & McCrae, 1992). These traits are very important for the tourism industry, as well (Kusluvan, 2010), but their connection to turnover has been limited (Ariyabuddhiphongs & Marican, 2015).
Extraversion reflects individual traits such as being sociable, gregarious, assertive, talkative, and active (Kim, Shin, & Umbreit, 2007;Tews, Stafford, & Zhu, 2009). It also refers to the extent to which people are warm, friendly, and dominant in social situations (Zhao, Seibert, & Lumpkin, 2010). As extraversion is characterised by positive feelings and experiences, it is considered a positive affect (Watson & Clark, 1997). Positive emotions and better networking as results of extraversion traits imply that people with high extraversion will be less likely to quit (Zimmerman, 2008;Wanberg, Kanfer, & Banas, 2000). Given that the lodging industry requires advanced social skills (Kim et al., 2007), and positivity is foregrounded in hospitality employees (Gatling, Kang, & Kim, 2016), extraversion is expected to play an important role for employees and be negatively related with turnover intentions.
Emotional Stability refers to people that are usually calm, relaxed, generally free from worry (Tews et al., 2009), even-tempered, able to face stressful situations without becoming upset (Rothmann & Coetzer, 2003), coping easily with negative emotions (Michel, Clark, & Jaramillo, 2011;Choi & Lee, 2014), and hardy (Zhao et al., 2010). People with a lack of emotional stability (i.e., neuroticism) are sensitive to negative feedback and easily become discouraged by small failures, feel easily worried, hopeless, or panicked when they face difficult situations (Zhao et al., 2010). Dunn, Mount, Barrick, & Ones (1995) found that emotional stability is the second most important characteristic that affects the fit to work of candidates. Given that hospitality is a very stressful and demanding sector, where the needs of clients should be fulfilled around the clock daily (Kim et al., 2007) positively, emotional stability is expected to play an important role in the lodging industry and have a strong negative correlation with organisation turnover intentions.
Agreeableness is a dimension that evaluates an individual's attitude and behaviour toward people (Zhao et al., 2010). Agreeableness refers to the warmth, friendliness, kindness, empathy in social interactions of individuals (Kim et al., 2007). People high in agreeableness tend to be altruistic, generous, trusting, and cooperative (Tews et al., 2009). Additionally, people high in agreeableness prefer social occupations due to the frequent interpersonal interactions (Zhao et al., 2010). On the other hand, individuals low in agreeableness seem to quit their job in an unexpected manner (Zimmerman, 2008), and are ignorant of workplace rules displaying deviant behaviours as they disagree easily with others (Durak & Saritepeci, 2019). Agreeableness is considered a valuable predictor in jobs that require teamwork and in the customer service sector (Rothmann & Coetzer, 2003). Thus, individuals high in agreeableness are less likely to quit as they tend to create better interpersonal relationships in the workplace which may help them having positive working experiences (Zimmerman, 2008). According to Kim et al. (2007), though, research results on this dimension vary regarding the relationship with work performance. Hotel employees high in agreeableness seem to be the best to easily identify guests' needs in accommodation and in this way strong positive guest relationships can be created, ensuring repeat guests and more profitability. Hiring employees high in agreeableness in the lodging industry can be an asset for organisations, as guests pay much attention to personalised services (Kim et al., 2007). Thus, a strong negative relationship is reasonable to be expected between agreeableness and turnover intentions.
Conscientiousness reflects an individual's dependability and volition. Dependability refers to how careful, organised, thorough, and responsible is someone and volition refers to how hardworking, persevering and achievement-oriented is someone (Kim et al., 2007). Individuals high in conscientiousness are trustworthy, committed (Durak & Saritepeci, 2019), punctual, and well-organised (Tews et al., 2009). However, there is a negative side of high conscientiousness, as it may lead individuals to annoying fastidiousness, compulsive neatness, and workaholic behaviour (Rothmann & Coetzer, 2003). A strong relationship between job performance and conscientiousness was found in various studies (Rothmann & Coetzer, 2003;Kim et al., 2007) and is reported to be the most important predictor among the five dimensions of FFM in the workplace. Given that the hospitality industry is a very demanding sector where clients' needs should be fulfilled all day long in a professional and dependable manner, conscientiousness is expected to play an important negative role on turnover intentions of employees.
Intellect represents how imaginative, cultured, curious, broad-minded, artistically sensitive (Barrick & Mount, 1991;Durak & Saritepeci, 2019), and flexible (Michel, Clark, & Jaramillo, 2011) people are. These people tend to be original, unconventional, independent (Tews et al., 2009) and they have a breadth of interests (Fang et al., 2015). In the workplace people high in intellect face difficulties as opportunities to develop themselves and not as cases of negative emotions (Durak & Saritepeci, 2019). Research on this dimension reveals contradictory results regarding its relationship with job performance, something that maybe has to do with the different requirements of each job (Rothmann & Coetzer, 2003). Regarding the hospitality industry, it is expected that intellect will be positively correlated with turnover intentions as this industry needs dedicating and obeying people to offer the "perfect" hospitality product to the "demanding" consumers of tourism products. Based on these approaches, the following hypotheses are formulated: Hypothesis 1a. There is a negative relationship between four of the dimensions of the Big Five Factor Model (extraversion, agreeableness, conscientiousness, and emotional stability) and intention to leave organisation. Hypothesis 1b. There is a positive relationship between the dimension of Intellect/Imagination of the Big Five Factor Model and intention to leave organisation.

Commitment and turnover
Within highly competitive environments, organisations nowadays acknowledge the crucial need for committed staff (Albrecht, Bakker, Gruman, Macey, & Saks, 2015). Meyer (2009) defined commitment as an "internal force that binds an individual to a target (social or non-social) and/or to a course of action or relevance to that target" (p. 39).
There are two main groups of foci: the internal that includes commitment to organizational goals or an organization's members and external foci that include commitment to factors that are outside of the organization such as occupation, customers and suppliers (Siders, George, & Dharwadkar, 2001). The most frequently studied topic in the commitment literature is organizational commitment (Vance et al., 2020) and, therefore, empirical evidence regarding other foci like occupational commitment, is scarce.
There is rich literature suggesting that there is a strong negative relationship between an employee's intention to turnover and organisational commitment in the hospitality context (Park & Min, 2020;Yang, 2008;Cho et al., 2009). No data about the Greek hospitality industry have been found.
Occupational commitment was also found to be correlated with organizational turnover in a way that people high in occupational commitment are less likely to quit from the organisation (Blau & Lunz, 1998;Lee, Carswell, & Allen, 2000;Yousaf, Sanders & Abbas, 2015;Chang, 1999). According to Morrow (1983), occupational commitment is a more stable type of work commitment than organisational commitment. Occupational commitment is also perceived to have a direct relation to the individuals' personality and not always influenced by the contextual organisation (Freund & Carmeli, 2003). People who feel committed to their occupation are approaching organisation as a place that allows them to apply their occupational skills (McAulay, Zeitz, & Blau, 2006). Consequently, they experience higher organizational commitment, as a bidirectional beneficial relationship is established. Lee et al. (2000) claimed that occupational commitment, in comparison with organisational commitment, is a stronger predictor of organisational turnover intentions. Therefore, they proposed that measuring occupational and organisational commitment simultaneously may improve our ability to predict organisational turnover. Similar results were found by other studies (Tsoumbris & Xenikou, 2010;Yousaf et al., 2015). Yousaf et al. (2015) found that people who are committed to their organisation express low intention to leave and this relationship is getting stronger for those being at the same time highly occupationally committed.
In their meta-analysis, Lee et al. (2000) also found that the relationship between occupational commitment and intention to leave the organisation is fully mediated by intention to leave the occupation. A full mediating role of intention to leave the occupation was found by Chang, Chi and Miao (2007), as well. This means that the total effect of occupational commitment on intention to leave the organisation is explained through intention to leave the occupation. Career stage was also found to moderate the relationship between organisational commitment and intention to leave organisation the organisation (Cohen, 1993). This highlights the importance of measuring the occupational commitment-related variables as part of turnover models, as these variables are highly neglected in the literature (Lee et al., 2000).
However, within the context of the hospitality sector, there is no evidence on the effect of occupational commitment variables on turnover. Therefore, it seems valuable to examine the role of occupationalrelated variables on organisational turnover intentions (i.e. occupational commitment and intention to leave the occupation) concurrently with organisational commitment. In this way a better prediction of turnover intentions may be achieved.
The present study adopts the Three-Component Model (TCM) of Allen and Meyer (1991). This model is the leading model to the conceptualisation and measurement of employee commitment (Vance et al., 2020). Although TCM has been criticised by various researchers, it remains the leading model for commitment, which has been widely examined in different settings during the last years, and it has been tested thoroughly, theoretically, and methodologically (Allen, 2016). This model has been extended in different work-related foci beyond the organisation. Besides, meta-analytic evidence suggests that there is negative correlation between the TCM dimensions and turnover intention (Allen, 2016).
Among the three components of commitment, only affective commitment -both in organisation and in occupation -was measured in this study. There is evidence that affective commitment has the strongest and most favourable relationship with organisation-relevant and employee-relevant outcomes. Therefore, is supposed to be the most beneficial type of commitment for organisations (Meyer & Allen, 1997). Affective organizational commitment "refers to the employee's emotional attachment to identification with and involvement in the organization" (Meyer & Allen, 1991, p. 67). In addition, affective organisational commitment has the strongest negative correlation with turnover (Meyer, Stanley, Herscovitch, & Topolnytsky, 2002;Vandenberghe, Bentein, & Panaccio, 2014;Joung, Goh, Huffman, Yuan, & Surles, 2015) and is the dominant measure in commitment studies (Μeyer et al., 2002).
Accordingly, affective occupational commitment refers to an employee's emotional identification with his/her occupation and work goals (Lee et al., 2000;Vandenberg & Scarpello, 1994;Yousaf, Sanders, & Shipton, 2013). Thus, employees that are affectively committed to their occupation feel positive emotions about their occupation.
Several studies conducted since 2008 have shown the existence of high turnover rates that occur in the tourism industry (Boella & Goss-Turner, 2019;Riley, 2018;Park & Min, 2020;Chalkiti & Sigala, 2010). Additional studies covering the same period have also shown the predictive features of commitment variables in terms of organisational turnover intention (Cho et al., 2009;Zopiatis et al., 2014;Park & Min, 2020). Thus, it seems crucial to investigate the role of commitment on intentional turnover behaviour within the lodging industry and more specifically in Greece. This will provide further implications for academics and especially for hospitality managers, who can use this approach in their hiring process.
Based on these approaches, the following hypotheses have been formed.

Hypothesis 2a:
There is a negative relationship between affective organisational commitment and intention to leave the organisation. Hypothesis 2b: There is a negative relationship between affective occupational commitment and intention to leave the organisation. Hypothesis 2c: The relation between occupational commitment and intention to leave the organisation is mediated by intention to leave the occupation. Hypothesis 2d: The relationship between organisational commitment and intention to leave the organisation is getting stronger for higher levels of occupational commitment.

Commitment and personality
Although personality and affective commitment of employees are connected with turnover, there has been limited research on investigating the dispositional basis of organisational commitment (Erdheim et al., 2006;Zettler et al., 2011;Spagnoli & Caetano, 2012;Choi et al., 2015;Guay et al., 2016). Empirical research had focused mostly on situational and experiential antecedents regarding organisational commitment.
The person-centred approach regarding commitment only lately began to draw the attention of researchers (Bergman & Jean, 2016, Meyer, Stanley, & Vandenberg, 2013. Choi et al. (2015) through their meta-analysis stated that, apart from work environments, organisational commitment is also influenced by personality traits of individuals. This might be explained because perceptions of work environments are affected by individuals' idiosyncratic traits. However, the meta-analytic results of Choi et al. (2015) were based on a small number of studies due to a lack of literature on the topic and, therefore, further research on the topic is needed.
Some additional studies using different frameworks confirmed the utility of dispositions in predicting organisational commitment, even when an individual's traits are evaluated before any work experience take place (Rubenstein, Zhang, Ma, Morrison, & Jorgensen, 2019;Judge, Higgins, Thoresen, & Barrick, 1999;Indarti, Fernandes, & Hakim, 2017). However, in the context of hospitality employees, only Silva (2006) analysed the connection between personality and commitment. This study has limited implications, though, as only correlation between personality and organizational commitment was examined and, therefore, no information on the way that personality affects commitment was provided. In addition, commitment was treated as a singular trait and not according to Meyer's Model classification of three distinct components.
The main feature of extraversion is positive emotionality, which encourages people to perceive their work environments more positively and makes them recall more easily positive information about their organisation (Watson & Clark, 1997). These people are more motivated to stay in their organisations (Meyer & Allen, 1991). Due to the social aspect of extraversion, people who are high in this dimension can more easily create social relationships in their work environment, which consequently can make them feel more connected with the organisation (Costa & McCrae, 1992;Michel et al., 2011;Choi & Lee, 2014). Given the fact that social skills are important in hospitality, it is reasonable to expect that extraversion will be positively correlated with affective organisational commitment in the lodging industry. This positive connection between the two variables has already been examined and confirmed by Erdheim et al. (2006), Zettler et al. (2011), Spagnoli and Caetano (2012), as well as Choi et al. (2015). The interpersonal components of agreeableness help individuals to create a pleasant work environment and strengthen the emotional attachment to the organisation (Ilies, Fulmer, Spitzmuller, & Johnson, 2009). Individuals high in agreeableness usually develop positive relations with colleagues and integrate easily into new environments (Michel et al., 2011;Choi & Lee, 2014). In addition, people high in agreeableness can easily identify themselves with the organisation and they can be more patient with unfair experiences within the workplace (Choi et al., 2015). A positive correlation between affective occupational commitment and agreeableness has been observed by Choi et al. (2015), whereas a negative correlation has been observed by Zettler et al. (2011). Considering that affective organisational commitment represents a positive emotional attachment to organisation, and that coping with demanding customers' needs is inextricably linked with the hospitality industry, it is reasonable to expect a positive relationship between agreeableness and affective organisational commitment.
Conscientiousness makes employees feel responsible and have an obligation to stay with the establishment (Michel et al., 2011;Choi & Lee, 2014). A strong relationship between conscientiousness and job involvement has been observed (Rothmann & Coetzer, 2003;Kim et al., 2007). Additionally, people high in conscientiousness are more likely to become affectively committed to the organisation (Farrukh, Ying, & Mansori, 2017). A positive correlation between conscientiousness and affective organisational commitment has been confirmed also by Choi et al. (2015), Erdheim et al. (2006) and Farrukh et al. (2017). Taking into consideration that in the hospitality industry responsibility and professionalism play an important role in fulfilling guest expectations, it is reasonable to expect a positive relationship between conscientiousness and affective organisational commitment.
People high in emotional stability present a more stable personality and tend to focus on the positive side of the situations they face within their work environment (Durak & Saritepeci, 2019). These people easily create positive relationships, too. As affective organisational commitment is referred to as a positive attachment to the organisation, a positive correlation can be expected. Previous researchers on the topic have confirmed this positive correlation (Erdheim et al., 2006;Choi et al., 2015). Given that creating positive relationships with customers is important in the hospitality sector, we can assume that individuals high in emotional stability are more likely to be affectively committed to their organisation, as well.
People high in the dimension of intellect are open to new experiences and always consider other job opportunities (Michel et al., 2011;Choi & Lee, 2014). Individuals high in intellect tend to unconventionality and doubting authority (Durak & Saritepeci, 2019). Intellectual individuals are primarily interested in new experiences and independence and, consequently, feel less attached to an organisation and its values. Previous evidence of Choi et al. (2015) supported a positive correlation of intellect with affective commitment, whereas Spagnoli and Caetano (2012) found a negative correlation in their study. Considering that hospitality requires dedication and stability, we assume that people high in intellect will be not affectively committed to the organisation.
Based on the above discussion, the following hypotheses are formed:

Hypothesis 3a:
There is a positive relationship between the four dimensions of the Big Five Factor Model (extraversion, agreeableness, conscientiousness, and emotional stability) and affective organisational commitment. Hypothesis 3b: There is a negative relationship between the dimension of intellect/imagination and affective organisational commitment. Table 1 summarises previous studies conducted in regard to the aforementioned relationships that have focused on the hospitality sector.

Data collection and analyses
The detailed review of relevant literature enabled the development of a self-administered quantitative questionnaire with the aim of addressing the aforementioned hypotheses. The population for this study comprised Greek accommodation-sector employees. Online survey distribution was selected due to the lack of a reliable point of reference about the population of Greek hospitality workers. At the same time this method offers easy accessibility, has low cost, is speedy, the participation is voluntary and there is a lack of time limitation (Schonlau, Ronald, & Elliott, 2002;Riva, Teruzzi, & Anolli, 2003;Tasci & Knutson, 2003). Thus, non-probability sampling was chosen as the preferred method of finding respondents to the online survey for this research. More specifically, the currents' research target population was employees in different accommodation businesses in different locations in Greece. Participants came from different department of accommodation businesses such as sales and reservations, food service, kitchen, housekeeping, management, accounting etc. Finally, participants consisted of frontline workers, supervisors and managers. It is important to mention that no particular subgroup among the target population has been excluded. Therefore, the resulting sample of the present survey should be considered reliable enough.  (Ariyabuddhiphongs & Marican, 2015) Negative relationship between personality (measures as set) and turnover intentions among hotel employees in Thailand Commitment-turnover intentions Predicting hotel managers' turnover cognitions (Carbery et al., 2003) Affective organisational commitment negatively predicts hotel managers' intention to leave organisation in Ireland Employees intent to leave: A comparison of determinants of intent to leave versus intent to stay (Cho et al., 2009) Affective organisational commitment negatively predicts turnover intentions among employees in restaurant and hotels in U.S. A study on factors affecting turnover intention of hotel employees (Lee et al., 2012) Organisational commitment negatively predicts turnover intentions among hotel employees in Taiwan. Job involvement, commitment, satisfaction and turnover (Zopiatis et al., 2014) Affective organizational commitment negatively predicts turnover intentions among hotel employees in Cyprus Turnover intention in the hospitality industry: A metaanalysis (Park & Min, 2020) Organisational commitment is the strongest negative antecedent of turnover intention in hospitality Personality and commitment Effects of disposition on hospitality employee job satisfaction (Silva, 2006) Positive relationship between organisational commitment and the personality traits of extraversion, emotional stability, conscientiousness. Conscientiousness had the strongest correlation with organisational commitment For the pilot study, the questionnaire was sent to 20 employees of 4 different hotels by using Facebook Messenger. After making minor changes to the questionnaire, the online survey was posted on Greek Facebook group pages for hospitality workers. The online questionnaire was available for seven weeks between November 2019 and January 2020. The response rate cannot be calculated as there is no data available about the number of post views. The questionnaire took about 15 minutes to complete. An introductory text explained the purpose of the study and assured respondents that their answers would be anonymous and confidential and to be used only for research purposes. Data collection was in accordance with the guidelines of the Helsinki's Declaration for ethics in medical research (World Medical Association, 2001). The Google Forms tool was used for data collection.
A total of 800 responses were received, 665 of which were included in the study, resulting in a final effective response rate of around 83%. Of the 800 responses, 135 responses were excluded due to respondents' failure to correctly answer the control questions that were included in the questionnaire to avoid careless responses. The final sample size of 665 participants is considered adequate, taking into considerations the SEM and PLS requirements, which will be analysed in the following section (Roscoe, 1975;Siddiqui, 2013;Anthoine, Moret, Regnault, Sébille, & Hardouin, 2014;Hair, Sarstedt, et al., 2017).

Measurement
The questionnaire consists of five sections and aims to cover the research questions. At the beginning of each section, detailed instructions were provided about properly responding to the questions. Participants were also asked to respond in honestly and always in relation to how they feel, think and react in the workplace.
The first section contains 16 closed-type questions regarding participants' working status, position, current working conditions, working experience in the tourism industry and more specifically in the accommodation sector, and the duration of the longest period they have worked for the same employer. The second section contains the 50-item version of Goldberg's (1999) International Personality Item Pool (IPIP) that evaluates the personality dimensions of the well-known Big Five Factor Model. The 50item version of IPIP is available in the Greek language and its validity has already been tested among the Greek population (Ypofanti et al., 2015). Besides, IPIP is cost-free, anyone can have access to all of its items, scoring keys are available on the internet, its items can be translated into other languages, and can be administered worldwide without any need of permission of anyone (Goldberg, et al., 2006).
Respondents were asked to rate this instrument's statements on a five-point scale (1=Very inaccurate, 2=Moderately Inaccurate, 3=Neither Inaccurate nor Accurate, 4=Moderately Accurate, 5=Very accurate). Each of the five dimensions was measured by 10 items. After reversing some questions (according to keys), the final score of each participant in each of five dimensions is measured by the sum of 10 items answers. The closest the participants' final score is to value '50' ('10'), the higher (less) is characterised by the evaluated personality trait.
In the third section affective organisational and occupational commitment were measured using the affective commitment scale developed by Meyer and Allen (1991). Affective occupational commitment scale consists of 6 items. This scale was scored by means of a seven-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). The affective organisational commitment scale consists of 6 items too, also scored by means of a seven-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). After reversing some questions, the final scale score is measured by the average score of related items answers. The closest the participants' final score is to value '7' ('1'), the higher (less) is characterised by the evaluated personality trait.
Actual turnover is challenging to measure (Lingard, 2003), because once employees leave, either the employees are difficult to find or their response rates to surveys are usually low (Johnsrud & Rosser, 2002). However, there is literature supporting that turnover intention is the best immediate predictor of actual turnover behaviour (Bluedorn, 1982;Balogun & Agesin, 2014;Zopiatis et al., 2014). Thus, in the questionnaire's fourth section, intention to leave the organisation and the occupation was measured by providing six items (three items for each variable). Turnover intentions have been measured reliably by tools of three to five items in many studies (Jang & George, 2012;Zopiatis et al., 2014). More specifically, to evaluate participants' intention to leave the organisation, they were asked how frequently they thought about getting out of the occupation, how likely it was that they would explore other career opportunities, and how likely it was that they would leave their present occupation within the next year.
Respectively, intention to leave the organisation was evaluated by asking participants how frequently they thought about leaving their current employer, how likely it was that they would search for a job in another organisation and how likely it was that they would actually leave the organisation within the next year. Responses were made on appropriately labelled 7-point scales and were averaged across items to yield composite intention scores according to the instructions of Meyer, Allen, and Smith (1993). The closest the final turnover score is to value '7' ('1'), the more this participant is characterised by high (low) intention to leave the organisation or occupation. The same back-translation method was used as with organisational and occupational commitment for identifying the intention to leave the organisation and occupation, as well.
Finally, the fifth section contains questions about demographic information including age, gender, marital and educational status.

Analysis and results
The SmartPLS 3 software (Ringle, Wende, & Becker, 2015) was used for analysis. Partial Least Squares-Structural Equations Modeling (PLS-SEM) is a widely accepted technique that is preferred when theory is underdeveloped, and prediction and explanation of endogenous constructs is the focus of the research. In addition, PLS does not require distributions assumptions as it is nonparametric and is more flexible in handling more complicated models. (Hair et al., 2014;Hair, et al., 2016;Matthews, Hair & Matthews, 2018;Hair, Sarstedt, et al., 2017). Although one of PLS advantages is that is working properly in small samples sizes (N<100), it is effective in analysing data of large sample too (Hair, Hollingsworth, et al., 2017). Thus, as this research focuses on predictions, data is non-parametric and the sample is quite large (N=665), PLS-SEM is the appropriate choice for analysis.
The average employment within the tourist industry was 7 seasons (IQR 4-12 seasons), whereas particularly in the accommodation sector was 6 seasons (IQR 3-11 seasons). Respondents' average longest period working for the same lodging employer was 3 years (IQR 2-6 years). Descriptive statistics of measurement scales including mean and standard deviation are shown in Table 3.  (Fornell & Larcker, 1981). The average variance extracted ranges between 34% and 82%, and so for the constructs of Agree, Consc, Exta and Intell is below the recommended level of 0.50. A small increase on AVE was observed when deleting loadings lower than 0.40 but AVE still remained lower than 0.50 which is the minimum value required. An AVE lower than 0.50 though can be accepted in case that composite reliability is greater than 0.60 as according to Fornell and Larcker (1981) "on the basis of pn (composite reliability) alone, the researcher may conclude that the convergent validity of the construct is adequate, even though more than 50% of the variance is due to error" (p. 46). As the composite reliability of all nine constructs is above the recommended level of 0.60, the internal reliability of the measurement items is acceptable. The results for reliability and validity are presented in Table 4. Discriminant validity was evaluated according to the recommended guidelines (Hair et al., 2016). Using the Fornell-Larcker criterion, the square root of AVE values were compared with the inter-construct correlation (Henseler et al., 2015;Hair et al., 2016). As it is shown in Table 5 all constructs had the square root of AVE values higher than the correlations among constructs. Discriminant validity has also been assessed by the HTMT approach (Table 6) and all values where lower than 0.90 (Hair et al., 2016;Henseler et al., 2015), Thus discriminant validity of the model has been confirmed. Lastly, collinearity of variables has been examined. Every item's variance inflation factor (VIF) was under 3,3, implying that multicollinearity was not an issue.

Structural Model and hypothesis testing
The goodness of the model is determined by the strength of each structural path determined by R 2 value for dependent variable, the value of R 2 should be equal to or over 0.1. Using bootstrapping (5000 bootstrap samples) the results in Table 6 show that all R 2 are above 0.1. Thus, predictive capability of the structure model is established. Moreover, the predictive relevance of the endogenous constructs is examined by the Q 2 value, which should be greater than 0. Using blindfolding the results on the same table (Table 7), show that prediction of the constructs is significant. Model fit was assessed by the SRMR value which is 0.064 and it confirms the model fit as the SRMR value is lower than the 0.10 threshold (Hair et al., 2016).
Path coefficient, total effect, direct effect, indirect effect and significance of the hypothesised relationships have been assessed to test the postulated hypothesis. The results revealed that hypothesis H1a has been partially confirmed. Among the four traits (Agree, Con, Emot and Extra), only emotional stability (Emot) had a significant impact on intention to leave the organisation (β=-0.095, t=3.307, p=0.001). Hypothesis H1b examined whether intellect has a positive impact on intention to leave the organisation (IntLOrg). Results revealed that even though the relationship is positive (β=-0.020, t=0.521, p=0.602), it is not significant, and thus H1b was not supported (Table 7).
The results for hypothesis H2a are presented in Table 7, whereas results of H2b are presented in Table  8.

Mediation analysis (hypothesis H2c)
Mediation analysis was performed to assess the mediation role of intention to leave occupation (IntLOcc) on the relationship between occupational commitment (OcCom) and intention to leave organisation (IntLOrg). The results (Table 8) revealed that the total effect of OcCom on IntLOrg was significant (H2b: β=-0.096, t=2.551, p=0.011). In the presence of IntLOc the impact of OcCom on IntLOrg remains significant (β=0.155, t=4.182, p=≤0.001). The indirect effect of OcCom on IntLOrg was found significant (β=-0.251, t=9.325, p=≤0.001), too. This shows that the relationship between OcCom and IntLOrg is partially mediated by IntLOc and, thus, H2c Hypothesis has been supported, which in other words means that high levels of occupational commitment leads to lower levels of intention to leave the occupation, which in turn leads to lower levels of intention to leave the organisation. Thus, some of the effects of occupational commitment on intention to leave organisation is explained through intention to leave the occupation.

Moderation effect (hypothesis H2d)
One of the study's goals was to examine the moderating effect of occupational commitment on the relationship between organisational commitment and intention to leave the organisation. Results revealed that OcCom actually is a moderator on the relationship between OrgCom and IntLOrg (β=-0.067, t=2.573, p=0.010). Regarding the size of the moderation effect, the interaction term has a negative effect on IntLOrg (-0,067), whereas the simple effect of OrgCom on IntLOrg is (-0,592). These results suggest that the relationship between OrgCom and IntLOrg is (-0.592) in an average level of occupational commitment. For higher levels of OcCom (OcCom increased by one standard deviation point), the relationship between OrgCom and IntLOrg increases by the size of interaction term (-0.592-0.067=-0.659). On the other hand, for lower levels of OcCom (OcCom decreased by one standard deviation point), the relationship between OrgCom and IntLOrg becomes (-0.592+0.067=)-0.525 ( Figure 10). Therefore, hypothesis H2d has been confirmed.

Figure 1. Moderating effect.
Note: The middle slope represents the mean value of OccCom, while the upper one +1 SD and the lower one -1 SD from the mean value.

Discussion
The current study's results revealed that of the five traits of FFM, only emotional stability has a significant negative effect on the intention to the leave the organisation, thus partially confirming Hypothesis H1a and failing to confirm Hypothesis H1b. Present results will be compared with the findings of previous studies by Zimmerman (2008), Salgado (2002), Timmerman (2006) as well as Ariyabuddiphongs and Marican (2015).
Present results are partially similar to Zimmerman's (2008) meta-analysis, who found strong negative correlation between four of the five dimensions of FFM (i.e., extraversion, conscientiousness and emotional stability, agreeableness) and the intention to leave the organization. Remarkably, among those three dimensions of FFM, emotional stability was the best predictor, followed by conscientiousness and extraversion. Salgado (2002) in his meta-analysis found that emotional stability, conscientiousness, and agreeableness are determinants of actual turnover, whereas Timmerman (2006) in a study conducted among customers service representative in US found strong correlations only between extraversion and openness to experience (intellect) and actual turnover.
Little previous research has examined the impact of FFM traits on intention to leave the organisation generally and more specifically, in the hospitality industry. To the author's best knowledge only Ariyabuddiphongs and Marican (2015) have examined the relationship between the Big Five personality traits and turnover intentions. In their study using regression analysis, they have examined personality as one set and they found that personality negatively predicts turnover intentions among hotel employees in Thailand.
To cover this gap, the present study examined the relationship between personality and turnover intention among lodging employees and found that among the big five traits, only emotional stability can predict intention to leave the organisation, in a way that employees high in emotional stability are less likely to quit. The aforementioned results are important for staff selection in the accommodation sector, as they imply that personality can contribute to predicting turnover intentions only in terms of emotional stability.
The small effect size of emotional stability should not be neglected mainly due to two reasons. First, in the context of psychological research it is more or less foreseeable (for a discussion see Cumming, 2012, andFunder &Ozer, 2019) and, second, underestimation of results of small effect sizes might lead to misleading results and information, which otherwise would be useful in the improvement of theory and practice (Funder & Ozer, 2019;Ahadi & Diener, 1989;De Boeck & Jeon, 2018).
These results seem reasonable as in the hospitality industry employees are often asked to cope with stressful situations, to positively face the challenges, and stay calm even in the most demanding situations. Therefore, if employees are not emotionally stable, they will face difficulties to stay in a hospitality organisation and they will be more likely to leave regardless of the working conditions. However, further research on the topic in the hospitality context is needed, as empirical evidence is limited.
Moreover, the results indicate that the work attitude of organisational commitment has a significant negative effect on the intention to leave the organisation. More specifically, it was revealed that organisational commitment predicts intentional turnover behaviour in a way that people high on organisational commitment are less likely to have turnover intentions. These results are in accordance with previous research in the hospitality context (Cho et al., 2009;Yang, 2010;Zopiatis et al., 2014;Park & Min, 2020) and add to the existing hospitality literature especially in terms of the Greek literature where little research on the topic has been conducted, so far. The effect size of organisational commitment on intention to leave organisation is quite large (f²=0,585), which implies that organisational commitment has a large effect on turnover intentional behaviour of employees in the context of tourism and more specifically in the accommodation sector.
Thus, the present findings from Greece are in line with previous research from other countries. This provides added assurance to hospitality managers to base their staffing decisions by assessing commitment of job candidates for limiting turnover.
Furthermore, occupational commitment was found to negatively affect the intention to leave the organisation in a way that people with high occupational commitment have less intention to leave organisation. These results are in accordance with previous research from other sectors (Lee et al., 2000;Chang et al., 2007) and add to the hospitality literature as there was no previous research on the topic in the hospitality context. The effect size of occupational commitment (f²=0.025) is reasonably smaller than that of organisational commitment, but still of notable importance.
These results seem reasonable considering that people who are committed to their occupation are more motivated to stay in the organisation and apply their professional skills within such environment. Therefore, assessing occupational commitment as a selection criterion by hiring managers might be useful in the unstable hospitality context, but further research on the topic is needed to confirm our results in other countries, as well, before we can use these results safely.
In addition, this study's results revealed a mediating role of the intention to leave the occupation on the relationship between occupational commitment and the intention to leave the organisation, supporting Hypothesis H2c. More specifically, a partial mediation was revealed which means that some of the effect of occupational commitment on intention to leave the organisation can be explained through intention to leave the occupation. More specifically, a low intention to leave the occupation leads to higher occupational commitment, which in turn leads to lower levels of intention to leave the organisation.
Studies have so far neglected to examine occupation commitment-related variables and, therefore, the research is very limited on the topic (Lee et al., 2000). Interestingly, the present results are different from previous researchers who found full mediation in their studies (Lee et al., 2000;Chang et al., 2007). Even though the present study found partial mediation, the effect size of occupational commitment on intention to leave the organisation is smaller when intention to leave occupation is not considered in the equation. (f²=0,023). However, in the presence of intention to leave occupation the effect size of occupational commitment on intention to leave organisation is is much greater (f²=0,165).
These results indicate that intention to leave the occupation is a much better predictor for intention to leave the organisation. This means that intentions of employees who are committed to their occupation cannot be safely predicted if their intention to leave the occupation is not measured simultaneously. An occupationally committed hospitality employee might not intend to leave an organisation if they feel that the organisation gives them the opportunity to apply their professional skills, else they might continue within the same occupation by changing their employer for better opportunities. Therefore, it cannot be safely predicted whether an employee will quit only by measuring occupational commitment, but also by considering intention to leave the occupation, as well. These results add to the hospitality literature as no other research on the topic exists.
Regarding Hypothesis H2d it was found that occupational commitment can affect the strength of the relationship (moderating role) between organisational commitment and the intention to leave the organisation. The results indicate that individuals who are committed to organisation are less willing to quit and this relationship is getting stronger for those who are highly committed to their occupation. According to Cohen's (1988) classification of effect size, the effect size of the moderator (f²=0.018) was not significant. However, as has already been discussed, in the moderation literature the average value of f² in the case of moderators is 0.009 (Aguinis, Beaty, Boik, & Pierce, 2005), which lead us to adopt less strict thresholds like the ones of Kenny's (2016) in accordance with the recommendations of Hair, Sarstedt, et al. (2017). Thus, the effect size of the moderator can be considered as a medium, considering that the size of the effect should reflect the research aims and the evidence of the current literature.
These results indicate that the more a hospitality employee is occupationally committed, the less he/she intends to quit. These findings are in line with the results of similar research projects (Lee et al., 2000;Yousaf et al., 2015) and add to the hospitality literature as no evidence on this issue exists within the hospitality industry.
The above results imply that the work attitude of commitment (either in terms of organisational commitment-related variables or in terms of occupational commitment-related variables) highly contributes to predict the intention to leave the organisation in the hospitality context.
Hypothesis H3a was confirmed as a strong positive relationship between four of the five traits (i.e., extraversion, agreeableness, emotional stability, conscientiousness) and organisational commitment was revealed. The impact of intellect on organisational commitment, however, despite being negative, was insignificant and thus failed to support Hypothesis H3b. These results are similar to previous yet limited studies. More specifically, the present results are in accordance with Erdheim et al. (2006), Zettler et al. (2011), Spagnoli and Caetano (2012), Silva (2006), and Choi et al. (2015) with respect to extraversion. Regarding agreeableness, the present results are in line with Choi et al. (2015), but in contrast with Zettler et al. (2011). Concerning conscientiousness, the present results are similar to those of Choi et al. (2015), Erdheim et al. (2006), Silva (2006), and Farrukh et al. (2017). Regarding emotional stability, this study found the same results as Erdheim et al. (2006), Silva (2006), and Choi et al. (2015). Finally, as far as intellect is concerned the present results are in contrast with both Choi et al. (2015) and Spagnoli and Caetano (2012), as there was no significant relationship between intellect and organisational commitment.
The effect sizes of personality dimensions on organisational commitment are quite small, which is not uncommon according to the analysis provided regarding Hypothesis H1. More specifically, among the personality traits conscientiousness had the highest effect size (f²=0.027) on organisational commitment, although it is still considered small. The effect size of agreeableness (f²=0.012) and emotional stability (f²=0.014) on organisational commitment can be considered very small. However, extraversion (f²=0.006) and intellect (f²=0.001) can be considered as having no effect as they are smaller even than the lowest threshold that was proposed for moderators (f²=0.009). Consequently, when the research aim is to predict organisational commitment, conscientiousness can be considered the most valuable predictor, followed by agreeableness and emotional stability. Silva (2006) first reported a positive correlation between conscientiousness, extraversion and emotional stability, which was also confirmed by the present study. The latter also revealed that conscientiousness is a predictor of organisational commitment. This study's results add to the limited literature regarding the dispositional feature of commitment especially in the hospitality context.

Conclusion
In sum, this study examined the dispositional feature of turnover in the Greek accommodation sector. For this purpose, personality, commitment and turnover intentions among 665 lodging employees were examined. After analysing the research data, emotional stability was found to be a predictor of organisational turnover intentions. In addition, conscientiousness was found to predict organisational commitment. Organisational commitment and intention to leave the occupation were found to be strong predictors of organisational turnover intentions, as well.
These findings provide useful input to hiring managers in the hospitality sector by helping them to tackle the high staff turnover rates in this particular sector both in Greece and globally. The present study also adds to the limited literature and broadens its scope as previous research on turnover models is restricted to organisational and situational factors. Additionally, this study provides primary data concerning turnover, commitment, and personality about Greek hospitality employees, which can be compared with future surveys in other countries and cultures.

Practical and theoretical implications
The present research has important implications for both human resource managers in the accommodation sector and academic research in terms of turnover models.
Hospitality-related research has shown that selection procedures in hospitality industry are quite poor and even though the use of structured tests, like personality and attitudes tests, has widely been acknowledged, the use of structured tests is very limited (Armstrong & Taylor, 2014). Especially in terms of turnover, personality has rarely been assessed. Therefore, the present study results encourage managers to consider personality during staff selection.
Hospitality managers -especially those in charge of hiring new employees -can benefit from the present findings by considering the following points. First of all, HR managers will be better supported in discussing staffing needs with the respective department heads of a hotel. Emphasizing on personality and commitment propensities will help in preparing more accurate job profiles. Second, HR managers can apply personality tests in the selection procedure for screening job candidates in a more efficient way. More particularly, the use of standardised psychometric tests will help in better identifying the more suitable candidates.
According to our results, candidates high in emotional stability and conscientiousness should be preferred in the hospitality industry as they are expected to express lower intention to leave the organisation. In this way, recruiters may contribute to decreasing high turnover rates, which represent a pressing problem for the industry. Furthermore, as employee turnover is accompanied by huge direct and indirect costs, hiring on turnover intention criteria can lead to a reduction of these damaging effects, as well. Moreover, by administrating personality tests that assess all five traits, recruiters will gain additional information about other traits that have practical implications for other work-related outcomes. Is should be stressed, however, that as recommended by Hughes and Batey (2017), personality tests should not be used alone as a selection tool, but they should always be combined with other selection methods. Therefore, people that fulfil all the other established hiring criteria and are at the same time high in emotional stability and conscientiousness, should be preferred.
Managers being aware of commitment and turnover tendency of an employ may be alerted earlier and apply more personalised practices to prevent turnover behaviour. At the same time managers will accept that some of the quits are out of their control. More specifically some quits are related to employee ad se, and therefore even the best management practices cannot prevent employee turnover in some cases. Managers are also encouraged to assess candidates' occupational commitment-related variables during the selection procedure. Employees that have intentions to leave occupation are prone to leave the organisation as well. This might be explained by the fact that it is difficult for an employee to find a different job in the same organisation. The connection of candidates with their occupation may be valuable while making managerial decisions in an organisational level. If managers are able to detect that a valuable employee has intentions to leave their occupation, then they may try to find a sustainable solution for both parts, in order to avoid losing a key employee at all.
The findings of the present study highlight the usefulness of psychometrics in the staff selection process of hospitality businesses with the aim of reducing the sector's high turnover rates. However, the limited financial resources of small accommodation businesses -which are the majority among this sectorthe lack of independent human resource departments and the special education and training that is required to evaluate personality, should not be ignored. Co-operation with qualified consultants in the selection process might be a solution, however, it would be best if HRM professionals themselves will master these techniques and be able to include them in their day-to-day work. In addition, educational programs on human resource management in tourism should be enriched with staff selection courses, so that future HRM professionals will be able to handle these issues in the proper manner and to be able to make the most of the available literature, as most of the times research papers seem to have only theoretical implications but no actionable advice. Finally, tourism-related government agencies should put more emphasis on employment issues by developing training programs for HRM professionals. These programs should highlight the importance of using structured selection tools, like personality tests, during the hiring process.
Through this study, it was intended to broaden our horizons regarding turnover models in the hospitality context focusing on individuals' differences. This study confirmed the contribution of personality to turnover models. Personality was also connected with the work attitude of commitment contributing to the organisational behaviour theory. The connection between different foci of commitment provided further data on the theory of work attitudes, as well. The present study provided primary data regarding personality, commitment, and intention to leave the organisation within the Greek accommodation sector. Therefore, our study highly contributed in the growing body of literature in the hospitality context and more specifically in the context of hospitality organisational behaviour.

Limitations and future research
A limitation of the present research is that it is based on a non-probability sample, as respondents were recruited through Facebook and, thus, the results cannot be safely generalised. Even though recent literature states that samples through Facebook are acceptable and present equally representative samples as traditional sampling methods, there is only limited literature on this topic. Self-selection bias is might also be a limitation of online surveys, as in any given online community some individuals tend to participate more frequently in online surveys than others and this might lead to systematic bias (Sekaran & Bougie, 2016).
Also, the majority of the respondents were employees of large upscale four-and five-star hotels, whereas the large majority of tourist accommodation in Greece comprises quite smaller businesses. Nevertheless, the present sample can be considered quite representative of Greek reality, as employees of different accommodation businesses, different positions and different locations in Greece have participated in the present survey. Thus, this study provides valid evidence for readers interested in further investigating turnover issues in the hospitality industry.
A limitation of the present study is that turnover intention and not actual turnover has been measured. Even though there is evidence that actual turnover can be measured effectively by intention to leave the organisation, it should not become a surrogate of turnover (Tett & Meyer, 1993). The fact that previous research found a strong correlation of different personality traits of FFM with actual turnover, reveals the need of further focus on the topic of measuring actual turnover in the hospitality industry. Therefore, actual turnover should be measured, as well.
Personality in this study was measured through broad traits, whereas there is evidence that in some cases, more narrow traits become more useful in the organisational field. As the focus of the present research was not in a specific job, but on different jobs in the broader accommodation sector, the use of broad traits seems appropriate. More narrow traits would be more appropriate in the case of focusing on a particular job's specific dimensions (Spector, 2012). In addition, as claimed previously personality can be predicted in general terms and, thus, using more broad traits seems to be more effective. However, future research on specific jobs in lodging industry is highly encouraged.
Finally, a limitation of the study is that all variables have been measured with self-reports and, therefore, the absence of faking cannot be guaranteed. For minimizing this problem, participants received instructions to honestly answer how they see themselves and not how they wish to be in the future, highlighting that their responses will be kept in absolute confidence. However, future research on the topic with different research methods (e.g. observer ratio) for assessing personality might be valuable, as well.
The current study's results will support academics to further examine personality and commitment (both organisational and occupational) in terms of turnover models in hospitality, as the related literature is very limited. Considering that the present study covered the Greek accommodation sector, future studies in other countries and other hospitality sectors would help to generalise these findings. This may enable researchers to find differences that might exist in personality and commitment features among different cultures in the hospitality industry.