TEACHERS’ TURNOVER INTENTIONS: ROLE OF HRM PRACTICES IN PUBLIC SECONDARY SCHOOLS IN KENYA

ABSTRACT Employees play a key role in the achievement of organizations’ objectives and thus every organization endeavors to ensure employee retention. However, as the organization endeavors to do this, they are faced with the challenge of employees’ turnover intentions. Although it has been suggested that Human Resource Management (HRM) practices may affect employees’ turnover intentions, the extent of this effect has been inconsistent in previous studies. Moreover, literature on HRM practices and turnover intentions in Kenyan context is limited hence prompting this research. The study thus sought to investigate the effect of compensation, career management, performance appraisal and training on turnover intentions of teachers in public secondary schools in Murang’a county Kenya. The study had its basis on Social exchange, Organization equilibrium and Expectancy confirmatory theories. The target population was 3752 Teachers Service Commission (TSC) teachers in 304 public secondary schools in Murang’a county Kenya. A two multi stage random sampling method was applied to select a sample size of 40 schools and 400 teachers. A self-administered semi-structured questionnaire and interview guide were data collection tools. The study used descriptive statistics and logistic regression and content analysis for data analysis. The study findings indicated that compensation and training significantly influenced turnover intentions while career development and performance appraisal insignificantly influenced public secondary schools’ teachers’ turnover intentions in Murang’a county Kenya. Policy implications of study findings have been discussed.


PUBLIC INTEREST STATEMENT
Teachers play an important role in the society by ensuring a learner acquires knowledge and discipline. To ensure this role is performed, retention of teachers in the classrooms is paramount. Teachers' retention is influenced by turnover intentions, manifested through intention to stay or leave the current job. Intention to leave is a challenge since it interferes with the learning and teaching process because a teacher with an intention to leave has poor morale and thus not settled in his or her work. Factors contributing to turnover intentions include: pay, training, promotion, performance appraisal, and work conditions among others. These factors are implemented through organization's Human Resource Management (HRM) practices. Thus, this study sought to investigate the effect of four HRM practices namely: compensation, career development, performance appraisal and training on teachers' turnover intentions. The findings indicated that compensation and training had significant effect while career development and performance appraisal had insignificant effect in influencing teachers' turnover intentions. Policy implications has been discussed. size of 40 schools and 400 teachers. A self-administered semi-structured questionnaire and interview guide were data collection tools. The study used descriptive statistics and logistic regression and content analysis for data analysis. The study findings indicated that compensation and training significantly influenced turnover intentions while career development and performance appraisal insignificantly influenced public secondary schools' teachers' turnover intentions in Murang'a county Kenya. Policy implications of study findings have been discussed.

Subjects: Management Education; Human Resource Management; Organizational Studies
KEYWORDS: HRM Practices; public secondary schools; turnover intentions; teachers; logistic regression

Introduction
Education is fundamental aspect in the development of a country. Teachers play an important role in education by aiding the learning process and fostering discipline which in the end improves the quality of life, therefore benefiting both the individual and the society as a whole. However, these benefits are usually reduced through teachers' turnover since according to Allensworth et al. (2009) teacher turnover compromises learning process by diverting resources from the classroom.
Teacher turnover, which is a global challenge and a phenomenon that is complex has been frequently suggested to be either caused by problem with workforce planning, or as an indicator of lack of teacher morale and motivation (Waititu, 2013). It has been suggested that employees leave their current jobs in situations where they are not satisfied with their jobs or employers, or where, chances of being promoted are minimal, and when better alternative job opportunities are available (Kaur et al., 2013;Manthi et al., 2018). Verhees (2012) posits that turnover intentions as opposed to real turnover continue to feature in many studies as a predictor of turnover in organization because of the ease to measure turnover intentions as opposed to real turnover which may be difficult due to lack of data on employees of who have actually left and also because of its capability of predicting actual turnover (Bhat, 2014). Park and Kim (2009) assert that when employees are making decisions about their current jobs turnover intentions constitute the final cognitive step whereby an employee actively opts to quit and starts looking for employment elsewhere. They further stated that employee intentions to quit the current job starts with mere thoughts and then statements by the employee that he/she desires leave the employer. Curry and Hill (2012) argue that turnover intentions just like actual turnover manifests itself in two ways; first, an employee (teacher) intention to leave the employer within the same profession and second, an employee (teacher) intention to change the profession altogether.
Just like actual turnover intention (intention to leave) has negative effects. Chughtai (2013) asserts that when employees have thoughts of quitting it affects their loyalty to the organization, and it may finally lead to the actual quitting of the valuable employee. Also according to Tarmizi (2008) intention to leave also affects other employees as it reduces their commitment within the organization. Krishman and Singh (2010) assert that teachers' turnover intention (intention to leave) leads to reduced motivation hence hindering achievement of schools goals. Albaqami (2016) posits that employees with high levels of intention to leave have poor morale, perform their duties poorly and negatively affect service delivery efforts.
It is therefore important to enhance teachers' retention by reducing intention to leave so as to improve teaching and learning processes in schools. One suggestion that has been put forward in order to reduce intention to leave is proper implementation of Human Resource Management (HRM) practices. Armstrong and Stephen (2014) postulate that it is important for organization to be consistent in implementing these practices since it provides a proof that the organization is concerned about those that contribute to achievement of its goals. It is important to note that individuals attitudes tend to be less affected by the way the practices are implemented but by how on a daily basis managers implement them. Thus, how HRM practices are operationalized tends to affect employees attitude towards the organization and consequently their turnover intentions.
Berry and Morris (2008) note that, employers are more concerned when most productive individuals reason for intending to quit is because of factors that the business can influence e.g., inadequate pay, unfavorable working conditions, and unsupportive supervisors and challenging working environment. Since HRM practices are within the control of organizations' managers, it is necessary to identify which HRM practices will significantly influence employee turnover intentions and strengthen them. This study considered the following HRM practices compensation, performance appraisal, career development and training. The choice of these HRM Practices is because of having been suggested to possess the potential to motivate employees alongside offering them recognition, and improved self-worth (Döckel et al., 2006;Kamau et al., 2020a;Mensah, 2014).
Previous studies on teachers' turnover in Kenya have indicated a worrying situation of high intent to leave among teachers. Study by Chepkemboi et al. (2013) found that majority of teachers indicated that given an alternative employment they would leave Teachers Service Commission (TSC). Another study by Njung'e (2015) among teachers in Gatanga Sub County of Murang'a County found that 65.7% of teachers had expressed intention to leave teaching. Nationally, it had been reported that 129,600 out of the 288,000 teachers had intention to leave the teaching profession in 2015 (Oduor, 2015). According to Republic of Kenya (2016), out of the 47 counties in Kenya, Murang'a County ranks third in terms of the number of teachers in public secondary schools. However, data from TSC Murang'a county office indicates the county had a shortage of 1291 teachers in 2018. Besides this shortage an average of 200 teachers leave the teaching service annually on voluntary basis. This translates to about 5% voluntary turnover rate. This shows need to address the problem of turnover intentions so as to improve retention of teachers In response to this, the government of Kenya through TSC in order to make teaching attractive and also to ensure teachers' retention has developed various policies such as teacher performance appraisal and development (Teachers Service Commission, 2016), policy on appointment and deployment of institutional (Teachers Service Commission, 2017), performance recognition, reward and sanction framework for the teaching service (Teachers Service Commission, 2018b), policy on employee wellbeing (Teachers Service Commission, 2018c), and career progression guidelines for teachers (Teachers Service Commission, 2018a). Besides these policies the TSC has also delocalized some of the HR functions to the counties thus bringing them closer to teachers (Manthi et al., 2018). However, despite the government intervention through various policies the problem of teacher turnover intentions (intention to leave) still persists (Kamau et al., 2020a). For example, a recent study by Ekabu (2019) revealed 64.2% of the respondents intended to leave teaching in secondary schools in Meru County.
Although there have been previous studies based on these variables, the findings of some of the studies were varied, for instance study by Nawaz and Pangil (2016) found Career development relating positively and significantly with turnover intentions while Mensah (2014) found Career development relating negatively and insignificantly with intention to stay. Acquah (2015) study on the hotel industry in Ghana revealed that rewards significantly affected turnover intentions more than the other variables that were studied. However, other early researchers had found contrary results such as study by Kim (2005) and another by Chew and Chan (2008) found that compensation is not significant in retention of employees. Since these studies' finding were not similar, this showed the need for a replica study in order to provide empirical evidence of the role of HRM practices in turnover intentions decision in the Kenyan context.

Literature review
Social Exchange Theory (SET) stipulates that parties that are in a situation that they are reciprocally depended on each other become indebted towards each other due to a number of mutual actions (Kular et al., 2008). SET is built on the principle that in organizations, relationships advance with time when parties involved can be trusted are loyal and are mutually committed to the stipulated rules (Crospranzano & Mitchell, 2005). By abiding to these rules, each party feels obligated to each other.
From the SET perspective, employees tend to examine the quality of their relationship with their organization and when they find the presence of quality relationship, they get motivated (Kamau et al., 2020b). Social relationship are based on the sense of reciprocity between the parties involved such that individuals have expectations from the organization which may include fair and adequate rewards, career opportunities, job security and emotional support (Eisenberger et al., 1986). In reference to SET, Cheche et al. (2017) posit that human beings socially tend to conduct themselves in ways that will enable them maximally attaining their personal interests. This means that if the social set up is beneficial employees will reciprocate positively if otherwise negatively. Hannah and Iverson (2002) posit that employees consider HRM practices as a sign of concern towards them by the organization, which leads them to reciprocate to the establishment by displaying good behaviors and attitudes. This view is also held by Skinner (2004) that the type of strategies and policies that an organization choose and the way they implement them is an assurance to the employees that the organization can be trusted in its operations. This implies that HRM practices can be used to foster positive attitudes and behavior among employees such as increased intention to stay. The theory is therefore suitable in explaining the relationship between HRM practices and employees' turnover intentions.
Organization equilibrium theory was propagated by March and Simon (1958). This theory constitutes the earliest structured turnover intention theory. The theory stresses the need to balance between employees' and organization's contributions and inducements. The equilibrium theory postulates that when an institution rewards an employee with income that is equal to or more than the employee's contribution to the institution, the individual will not think about leaving the institution (March & Simon, 1958). This theory implies that an employee would stick with his/her organization when he/she is satisfied with the work and pay and also when the employee perceives that the alternative job options are few.
Expectancy Confirmatory Theory (ECT) was developed by Vroom (1964), and it postulates that individuals have expectations for an event and if those expectations are not met it leads to dissatisfaction. The concept behind ECT is that before any event, one is usually expectant. If what is being expected is realized, satisfaction is gained however if one does not realize what was expected then one becomes unsatisfied (Jiang & Klein, 2009). Vroom (1964) argued that both the expectations before an event and subsequent evaluation after the event combine to affect satisfaction with the event. Ngo-Henha (2017) suggest that based on this theory, management's effort to evaluate and meet the employees' expectations can be considered as a strategy to retain and ultimately control intentions to leave. Snell and Bohlander (2010) posit that compensation consists of different kinds of payment and benefits that employee get after performance of their jobs. Employee compensation can either be direct or indirect. Salaries, commission and bonuses represent direct compensation while various fringe benefits and non-financial compensation such as medical insurance, meals, overtime and pension plans which employee receives are indirect compensation. Therefore when both types of compensation are provided to the employees it represents total reward (Armstrong & Stephen, 2014). Comprehensive compensation accompanied by effective disbursement system can be used to retain employees and at the same time inculcate positive employees' behavior (Khan, 2010). It has also been suggested that benefits may help attract and retain employees. Rusbult and Farrel (1983) posited that employee's intention to leave is increased when benefits are inadequate.

Compensation and employee turnover intentions
Study by Ekabu (2019) among secondary school teachers in Meru Kenya established remuneration significantly correlated with turnover intentions. The findings were similar with Zahra et al. (2013) study among faculty members in Pakistan universities. Regression analysis showed that remuneration contributed negatively towards turnover intentions of the faculty members. Weldeyohannes (2016) study also among teachers in Tigray concluded that there was high dissatisfaction among teachers in relation to the total rewards that they receive. These three studies used different methods of analysis from the current study which used logistic regression to interrogate the relationship between the variables. Kwenin et al. (2013) study collected data from 142 employees from a telecommunication firm in Ghana. The results showed that increased retention can be experienced when there are adequate reward systems in organizations. A study by Acquah (2015) on the hotel industry in Ghana revealed that rewards have more significant effect on turnover intentions than the other variables that were studied. These two studies were in a similar context but different sectors yet the findings were similar. The current study therefore attempted to establish whether similar results will be registered in another context. Waititu (2013) study among 136 teachers in public secondary schools in Limuru, Kenya established that 45.51% were dissatisfied with the salaries they earn. Further, the study indicated that 53.36% of respondents felt that salary which they received was not equivalent to the work they performed, hence leading to them quitting. However, this study by Waititu (2013) focused on actual turnover as opposed to turnover intentions and applied descriptive analysis and so the relationship between compensation and turnover intentions was not interrogated. Other researches have indicated compensation and turnover intentions having a negative relationship (Batt et al., 2002;Ovadje, 2009;Rhoades et al., 2001) From this discourse, it is evident that majority of empirical studies indicate that compensation and turnover intention relationship is significant (Grace & Khalsa, 2003;Griffeth et al., 2000;Joarder & Sharif, 2011). However, some studies such as study by Kim (2005) and Chew and Chan (2008) indicated contrary results whereby compensation was not significant in employee retention. The fact that other studies have given contrary results in relation to these variables this study explored this relationship in a different context.

Career development and employee turnover intentions
According to Osibanjo et al. (2014) career development in an organization program individuals' mindset to focus on their future career prospects within and without the organization. This suggests that if an organization does not embrace HRM practices that promote career development it risks losing the much-needed employees as these will look for opportunities in organizations where their career concerns are considered. Xu and Thomas (2011) assert that besides career development placing employees in position compatible with their individual career needs and interests, also helps to retain and motivate employees. Manthi et al. (2018) study among tutors in teachers college established that career development and turnover intention had a negative but significant relationship. A study conducted by Nawaz and Pangil (2016), among faculty members in Pakistan's private universities found career concerns and turnover intention relating positively and significantly. This implied that if individuals care more about their careers, their likelihood of leaving their current organization would be high. The two studies had similar results however, these studies focused on respondents who were working in colleges unlike the current study that focused on teachers in secondary schools. (2016) did a study in a private owned Schools and Colleges at Karachi that involved 120 respondents. The study findings established that the career growth and turnover intentions relate negatively but significant. However a study carried by Nurita et al. (2013) that involved 457 employees from different sectors in Malaysia found career management relating positively but insignificantly with turnover intention. The two studies had different outcomes, used different analysis method and were also conducted in a different context from the current study. It is on the basis of this review that this study was out to determine how career development relates to teachers' turnover intentions in Kenya.

Performance appraisal and employee turnover intentions
The reason behind performance appraisal is to ensure that work standards are set, evaluating whether those standards are realized by employees and ensuring employees get feedback, which can motivate them leading to better performance (Dessler, 2008). According to Elliott (2015), performance appraisal process assists both teachers and an administrator in determining the teachers' performance and it also assists to identify areas for future development. Ingvarson (2012) asserts that performance appraisal system can affect teachers' attitudes, behaviors and performance and in the end affecting students' learning outcomes. This shows that the way performance appraisal is implemented can result to either good or bad organization's outcomes. Tuytens and Devos (2010) assert that analysis of teachers' perceptions regarding the performance appraisal is necessary. Therefore, there is need for education managers to evaluate the attitudes of the teachers towards the appraisal system since the attitudes have effects on its implementation and effectiveness. In Kenya, Kenya National Union of Teachers (KNUT) has been in conflict with TSC concerning the teachers' appraisal system. The union describes the exercise as full of paperwork and a waste of time for teachers and threatened to call for a strike (Oduor, 2018).
Flores (2011) did a study to establish teachers' perception on performance appraisal in Portugal. The study found that teachers were skeptical and uncertain about the appraisal system. Specifically, teachers were skeptical on the role of performance appraisal on their professional development. This negative perception towards the appraisal system can have effect on its implementation and teachers morale and thus need to identify ways of making the process generally acceptable. However, this study was in a different context, only focused on the role of performance appraisal and applied different method of analysis from the current study. Bekele et al. (2014) did a study among auditors in Ethiopia whereby a turnover intention was one of the work outcomes. The study's findings demonstrated that when performance appraisal was negatively perceived by employees, it significantly leads to thoughts of leaving the current job. However this study was conducted in one firm so could not be generalized. Kibe (2013) did a study among teachers in selected schools whereby one of the indicators for commitment was turnover intentions. The study results indicated that performance appraisal and commitment related positively and significantly. However, this study did not use turnover intention as a variable but as an indicator of commitment.
Long and Perumal (2014) study among 80 supervisors and security officers of a firm in Singapore revealed that performance management correlates with intention to leave at 0.61 while the regression analysis showed that performance management had a significant high relation to intent to leave compared to other variables (employee relation, career development, compensation and benefits and training). The study used correlation analysis while the current study used logistic regression. This research was also done in one firm so could not be generalized and also was in a different industry and context. There was need to find out whether the same will be realized in a different industry and context and with different method of analysis.
From the foregoing, it is evident that performance appraisal significantly influences turnover intentions. However, there are cases where the effect was insignificant and besides this, there is dearth of literature on effect of performance appraisal on turnover intentions in Kenya and thus this prompted this study.

Training and employee turnover intentions
Training is usually a deliberate move by an organization to influence positive organization's outcomes. The positive outcome expected out of training explains why organizations spend a lot of money to train its employees. A study by Kesen (2016) among workers in service firms in Turkey used questionnaires to collect data. The finding revealed that training had negative correlations with turnover intention. Thus, the study concluded that using training can reduce turnover in organizations. This study took place in a different context and therefore need to do another study in another context. However, a study by Joarder and Sharif (2011) in the private universities revealed that perception of faculty training influences turnover negatively but not significantly. This implied that although the relationship between training and turnover was negative, and its influence was rather weak.
Verhees (2012) did a study on employees' turnover intentions in a Dutch bank. Distribution of 250 questionnaires to the employees was done; out of this 100 were filled and returned. However, only 79, which represented response rate of 31.6% were found suitable for further analysis. The study's analysis was by Pearson's correlation and hierarchical multiple regression analyses. The findings indicated that training and turnover intentions relate negatively. However, these results could not be generalized since a single bank was involved in the study and also had a low response rate.
Aslami (2013) study targeted teachers in public secondary schools in Kabul, Afghanistan. It focused on what contributes to teachers leaving or intending to leave the profession. The study used interview, questionnaires and observations to collect data whereby 71 teachers participated. This study found that for new teachers, high attrition was due to of the absence of effective and continuous programs. The findings further showed that unpreparedness for the reality of the classroom, led to 56% of former teachers leaving their profession. This implies that lack of induction program can lead to frustrations among new teachers resulting to intention to leave. However, this study applied only descriptive statistics and so did not bring out the relationship between the study variables.
There are however some researchers who had a different viewpoint on training and turnover relationship (Acquah, 2015). There are cases where studies indicated employee training and turnover intentions, having no significant relationship (Batt et al., 2002;Lincoln & Kalleberg, 2006;Way, 2002) whereas other researches indicated that inadequate training actually increases turnover intentions (Manthi et al., 2018). This therefore shows need to further investigate training and employee turnover intentions relationship and this prompted this study. Hence, based on this review, the study had these hypotheses: H0 1 : There is no significant effect of compensation on turnover intentions of teachers in public secondary schools in Murang'a county Kenya.
H0 2 : There is no significant effect of career development on turnover intentions of teachers in public secondary schools in Murang'a county Kenya.
H0 3 : There is no significant effect of performance appraisal on turnover intention of teachers in public secondary schools in Murang'a county Kenya.
H0 4 : There is no significant effect of training on turnover intentions of teachers in public secondary schools in Murang'a county Kenya.

Methodology
The research designs applied were descriptive and explanatory designs. The use of descriptive design in this study facilitated vivid description of the characteristics of the participants and nature of their responses to the research questions (Kothari, 2004). Explanatory design helped to explain how the predictor and the response variables relate (Firebaugh, 2008). In this study, the dependent variable turnover intention was considered a dichotomous binary variable and was indicated by intention to leave or stay. The binary logistic regression was therefore used since it is suitable where one wants to show the presence or absence of a particular attribute of interest (Kamau et al., 2020b;Rodriguez, 2007).
The study targeted 3752 teachers from 304 public secondary schools from Murang'a County. The study thus used multi stage random sampling method to select the 400 teachers from 40 schools in the county. The researchers used the Yamane (1967) formula to obtain a representative sample size of teachers from the county. n =N ÷ 1 + N (e 2 )(3752) ÷ 1 + 3752 (0.05 2 ) = 400.
Source: Yamane (1967) Where n = number of teachers sampled, N = total number of teachers, e = level of precision. Table 1 shows distribution of the sample size in the eight sub counties in Murang'a County, Kenya. Table 1 indicates that out of 3752 teachers in 304 schools, 400 teachers were sampled from 40 schools from the eight sub counties in Murang'a County. This sample size for both the teachers and schools were considered since they fell within the acceptable sample size of 5-20% of a population which is recommended for studies that requires generalization of the findings (Creswell, 2003;Sekaran, 2003). Mugenda and Mugenda (2003) also assert that 10% sample size is sufficient.
Besides the 400 sampled teachers who filled the questionnaires, 15 principals were randomly chosen for an interview from the 40 schools that were sampled for the study. The sample size of 15 principals was chosen because according to Crouch and Mckenzie (2006) for collection of qualitative data, a sample size of 12-15 is suitable for interviews involving homogeneous respondents. This argument is also supported by Creswell (2012) who suggests that for an in-depth interview a sample size of 15-25 is appropriate. It was necessary to interview the 15 principals in order to validate and supplement the data from teachers. The questionnaire was developed from review of previous studies. For instance compensation scale was adapted from Guest (2002), career development from Delery and Doty (1996). Training was adapted from Boselie et al. (2001). Turnover intention scale was adopted from Mensah (2014).
To ensure the instruments achieve the objectives with accuracy (internal validity) and that generalizing will be beyond the sample (external validity) and that the instrument has both face and content validity, experts' opinions were sought. Construct validity on the other hand was ensured by operationalizing variables related to constructs reviewed from literature (Nzuki, 2014). Reliability of the instrument was ensured by deriving and adopting from the literature those items that other researchers have tested their reliability (Muathe, 2010). Cronbach's alpha correlation that measures internal consistency was also used to determine reliability where items with 0.7 and above Cronbach's alpha were deemed suitable for the study.
During data collection, the researchers first contacted the schools' principals for consent and explanation for study purpose and intention to drop the questionnaires to the participants. The questionnaires were given to the respondents who were randomly selected who were given two weeks to fill the questionnaires. The questionnaires were proportionately distributed to the  To determine percentages, measure of central tendency (mean) and dispersion (standard deviation) for the study variables, descriptive analysis was used. On the other hand binary logistic  regression analysis helped to determine how the predictor variables affect the dependent variable (turnover intentions). To test the model fit, log likelihood and the associated Chi-square statistics was used (Muathe, 2010). The Wald statistics was used to test contribution by each explanatory variable. If the odd ratio value generated by Wald statistics was equal to or greater than one it indicated positive contribution to the model. Estimation and test of hypotheses was done at 95% confidence interval. To reject the null hypothesis, the p-value had to be below 0.05, otherwise it was not rejected.
To analyze qualitative data collected content analysis was used.

Descriptive analysis
A pre-study was carried out to test whether research instruments were reliable. Cronbach alpha correlation was used where items with Cronbach's alpha of 0.7 and above were deemed reliable. Table 2 shows the outcome of the reliability test. As shown in Table 2, all the variables had α greater than 0.7 while the overall reliability was 0.812 which confirmed that scale used in questionnaire construction was dependable.
Out of the four hundred (400) teachers who were sampled, two hundred and ninety-eight (298) responded to the questionnaires that were administered. In addition, 15 principals participated in the interview, resulting to a 75.4% response rate hence it was considered satisfactory to conduct data analysis and interpret the results (Mugenda & Mugenda, 2003). Tables 3 and 4 present the characteristics of the respondents. Table 3 show that gender disparity was small among the teachers who responded to the questionnaires since the males were 54% while the females were 46%. Majority of the teachers were between 21 and 50 years and worked with TSC for more than 5 years (70.5%). On level of education, majority of the respondents (68.1%) had bachelor degree. This imply that majority of secondary schools teacher have higher qualifications since only 14.4% had diplomas. The findings also show that out of the eight sub counties Murang'a south had the highest response rate (18.1) followed by Gatanga (15.1) and Kangema (14.8).

The findings in
The results in Table 4 indicate that male principals were the majority (9) while in terms of age majority of the principals were in the age bracket 45-54 years. This implies that for one to be principal it requires many year of experience. Those who had worked as principals for more than 10 years were 7 almost half the number interviewed. In regard to level of education majority had bachelors and masters degrees 46.6% and 40%, respectively. In terms of job groups majority of the principals were in job group N while 63.37 had stayed in the same job group for more than 6 years. According to TSC career progression guidelines (Teachers Service Commission, 2018a) one qualifies to apply for promotion interview in order to move to a new job group after three years in the current job group. These results indicate that there is slow progression among the principals from one job group to another. This research was out to confirm the effect of HRM Practices on turnover intentions and thus respondents were required to state their degree of disagreement or otherwise on various aspects of 4 HRM Practice: compensation, career development, performance appraisal and training. The Likert scale was based on a scale of 1-5 whereby 1 being the lowest level indicated strongly disagreed, while the highest 5 represented strongly agreed. The responses are summarised in Table 5   Table 5 shows that the participants disagreed that the employer had done enough to enhance better compensation for their various responsibilities. This was indicated by an aggregate score  Most respondents commented on compensation that "compensation is not satisfactory therefore need to review" Also many respondents commented that "The new compensation scheme is not fair in that it mostly favors those in administration". These findings show that most of the study participants were not satisfied with the compensation they received which could be the reasons for increased intention to leave among the public secondary schools TSC teachers in Murang'a county Kenya.
Has hardship allowance contributed to your continued stay with your employer? Majority of the respondents mentioned that "performance appraisal is full of paper work and does not raise level of performance." However there are those who indicated that "Performance appraisal was a good initiative by the employer but it needs improvement and should be done once a year. The results implied that performance appraisal used by TSC for teachers elicited different reactions from teachers, that is, there are those who felt it is okay but others felt it is not okay. There is need to promote general acceptance Training What do you think need to be done by your employer on training of employees?
Respondents were of the opinion that "the employer should carry out frequent training programs". According to respondents there were no adequate training programs conducted by the employer to enhance the capacity of the teachers therefore there was need for improvement on training.
Turnover Intention Main reason for not leaving the current job Majority of the respondents indicated: l like my job' This implies that majority of those who have intention to stay love their teaching job and so they are satisfied.
Reason for teachers leaving teaching in public secondary schools Major reason was: Greener pastures. This indicated that better terms in other jobs made teachers to leave. of 2.17. A standard deviation of 1.02 was an indication that the respondents had less varied views concerning compensation. The findings agree with Waititu (2013) study among teachers in public secondary schools in Limuru, Kenya that established teachers were dissatisfied with the salaries they earn. This findings also agree with findings from interview with principals that indicated teachers particularly those not in administration were not satisfied with compensation being offered to them leading to high intention to leave. There is therefore a need for teachers' employer to continuously improve compensation for teachers who have no administrative roles and also to improve the benefits in order to increase intention to stay. According to Cho et al. (2006), when non-managerial employees are provided with incentive plans, they will unlikely quit their jobs. During the interview with the principals, it was also established that teachers were not happy with the medical insurance being offered by the employer thus triggering intention to leave. This agrees with Khan (2010) assertion that employees' intention to leave is increased when benefits are inadequate and suggested that an organization can offer a package that has moderate salaries accompanied by large amounts of benefits with a goal of retaining employees.
In relation to career development, the respondents were neutral that various activities and programs by the employer and supervisors enhanced furtherance of their careers as confirmed by mean score of 2.69 and a 1.15 standard deviation. However, respondents disagreed that the employer promotional policies are fair and that they are better career opportunities with the employer as compared to others represented by a mean score of 2.24 and 2.35, respectively. This agreed with the findings from the interview with the principals that the employer's promotion policy is not fair. Majority of the principals stated "The promotion policy is not fair at it only emphasises on general academic performance but not on value addition" Performance appraisal had also a mean score of 2.69 and a 1.09 standard deviation. This implied that teachers were neutral on various aspects of the appraisal system. However, the respondents disagreed on three aspects which were: they are satisfied with the way the performance appraisal system is used to evaluate and rate performance, the supervisor takes the performance appraisal review discussion seriously and that current performance appraisal is fair and unbiased. This means score were 2.05, 2.33 and 1.04, respectively. Interview with the principals also revealed that many teachers were negative about the performance appraisal and they saw it as a bother to them and adding no value to them but wasting time needed for teaching and learning process. This implies that teachers' morale has been affected by the performance appraisal process.
The participants were also neutral on various issues concerning training represented by a mean score of 2.84 and a standard deviation of 1.18. However, the respondents disagreed with the statement that there are better training opportunities in the organisation as compared to others with a mean score of 2.44. Principals during the interview also indicated that they are not able to adequately support training of teachers due to inadequate funds. One principal said: "The employer does nothing as far as training of teachers is concerned. When teachers compare what is done in other government ministries on training of employees, they think about leaving". It was also revealed during the interview with the principals that although induction training is conducted for new teachers each school does it in their own way. In general, interview with principals indicated that the HRM practices implemented by the employer were not popular with teachers. Majority of the principals also indicated they have inadequate Human Resource (HR) skills so were not able to effectively carryout some of the HRM functions.
Turnover intentions, which was the response variable in this study, were considered as a binary variable. This section of the analysis therefore presents the results on turnover intentions. Respondents were asked whether they will leave if they get a better job. The findings are presented in Table 6.
As shown in Table 6, the teachers who had the intention to leave were 82.9% (247). However, 17.1% (51) of them intend to stay. The findings provide a clear indication that teachers in public secondary schools who were the participant in the study that intend to leave are more than those that want to stay. There is therefore a need to address high intention to leave by teachers to avoid compromising quality of education. Ngo-Henha (2017) suggests that based on ECT, management effort to evaluate and meet the employees' expectations can be considered as a strategy to retain and ultimately control turnover intentions.

Hypotheses testing
A binary logistic regression was applied to test how the predictor variables affect the response variable. However, logistic regression assumptions had to be met through the performance of diagnostic tests prior to performance of hypotheses testing. According to Kirigiti and Ahmed (2018), these tests are a prerequisite so as make the model reliable. The tests of multicollinearity and influential values were done, and the results indicated all the basic assumptions were met. This made it possible to carry out the analysis through logistic regression. The results for logistic regression are presented in Table 7.
From Table 7, Hosmer and Lemeshow chi-square test of 6.294 with 8 degree of freedom and p value of 0.614 is not significant implying that there is good fitting of the data to the model while the overall model prediction was 84.2%. The Nagelkerke R Square 0.246 implies that HRM practices accounts for 24.6% of the variation in turnover intentions. This implies that other determinants other than HRM practices affect teachers' turnover intentions in public secondary schools. The binary logistic regression model findings are also presented whereby the result constitutes B, Wald's X 2 , odds ratio (exp B) and the significance value (P-value).
The effect of compensation on public secondary schools' teachers' turnover intentions in Murang'a county was found to be significant (B = 0.8040, Wald X 2 = 7.676, exp B = 2.234, P < 0.05). The contribution of compensation to the model is significant as the Wald's X 2 is greater than zero while the combined odd ratio is greater than one. The null hypothesis was therefore rejected. The implication of the result is that compensation practices significantly influenced turnover intention of public secondary school teachers in Murang'a County Kenya. The findings implied that contribution of training to intention to stay is positive and that when holding other predictors constant, for a unit positive change in compensation it is expected a 0.8040 positive change in the log odds or 123% increase in odds of intention to stay. Therefore, compensation is a critical component in leading to intention to stay of teachers.
This outcome match those of Khan (2010) who established that all-inclusive compensation together with effective disbursement system can be used to retain employees and at the same time inculcate positive employees' behavior. The findings also agree with those of Manthi et al. (2018) and Ekabu (2019), however they contradict the findings by Way (2002), Batt et al. (2002), and Lincoln and Kalleberg (2006).
The career development's effect on public secondary schools' teachers' turnover intentions in Murang'a county was however found to be insignificant (B = −0.2460, Wald's X 2 = 0.612, expB = 0.782, P > 0.05). Though the Wald's X 2 was greater than zero the combined odd ratio was less than one indicating a negative contribution to the model. The null hypothesis was thus confirmed which was an indication that the public secondary schools' teachers' turnover intention in Murang'a county is not significantly affected by career development. The fact that the coefficient is negative implies that career development and intention to stay have an inverse or negative relationship. These findings agree with those of Nurita et al. (2013) and Mensah (2014) which established an insignificant relationship between career development and turnover intention. However, they are inconsistent with the findings by Manthi et al. (2018), which established significant relationship between career development and turnover intentions.
The insignificant relationship between career development practices and intention to stay was however unanticipated because usually, provision of career development opportunities is an appreciated human resource activity that is generally expected to enhance employees job satisfaction, self-esteem as well as morale (Mensah, 2014). A possible justification for the insignificant relationship between career development practices and intention to stay could be the fact that the career development activities may not be a continuing activity among public secondary schools teachers. According to Tan (2008), career development ought to be a continuous development process that is a part of organization's policy and structure. The other factor that could explain the insignificant relationship between career development and turnover intentions of teachers could be the fact that the teachers' employer the TSC reviewed teachers' career development practice not long ago and so teachers are still not be acquainted with the terms of the new policy as indicated by the response to the open-ended question in Table 8.
The effect of performance appraisal on turnover intentions of public secondary schools' teachers in Murang'a county was also found to be insignificant since the p value was greater than 0.05 (B = 0.1670, Wald's X 2 = 0.29. expB = 1.182, P > 0.05). The null hypothesis was therefore not rejected. However, the contribution of performance appraisal to the model was not positive since Wald's X 2 was not zero while the combined odd ratio was greater than one. This confirmed the prediction that turnover intentions of teachers in the county are not significantly influenced by performance appraisal. The findings agree with those of Long and Perumal (2014) who established an insignificant association between performance management and intention to leave. However, the findings disagree with Elliott (2015) assertion that performance appraisal process enhances teacher attachment to the institution.
The insignificant influence of performance appraisal on turnover intentions obtained in this study can also be justified by a number of factors. First, the performance appraisal program may have failed to attain the set targets. This is line with Tuytens and Devos (2010) assertion that performance appraisal is aimed at showing whether employees have achieved the set targets and also identify professional development gaps. If this is not achieved, employees may fail to attach any importance to the program. Secondly, the insignificant relationship could be explained by the manner in which the performance appraisal was implemented. There is therefore a need to consider the attitudes of the teachers towards the appraisal system since poor attitude may adversely affect its implementation as well as effectiveness. If teachers are skeptical about the purpose of a performance appraisal program, there is a high likelihood that they will perceive it negatively (Flores, 2011). This will in turn affect its implementation and effectiveness. One principal said that "Teachers don't own it; this is because at its introduction teachers as stakeholders were not involved, so they feel it is imposed on them".
Training had a significant effect on public secondary schools' teachers' turnover intentions in Murang'a county (B = 1.1080, Wald's X 2 = 12.779 expB = 3.028, P < 0.05). The contribution of training to the model is significant as the Wald's X 2 is greater than zero and combined odd ratio is greater than one thus positive contribution to the model. The null hypothesis was therefore rejected which implied that training activities significantly influenced turnover intentions of public secondary schools' teachers in Murang'a County, Kenya. Accordingly, holding other predictors constant, for a unit increase in training among secondary schools teachers we expect increase in intention to stay by 1.1080 log odds or 203% odds. The results agree with those of Kesen (2016) which confirmed a significant and positive effect of training on the intention to stay of employees. They also agree with findings of Manthi et al. (2018) which established a significant and negative effect of training on tutors' turnover intentions. However, these findings contradict the findings by Batt et al. (2002) and Lincoln and Kalleberg (2006) where the researchers indicated that training actually decreases intentions to stay.
It is thus necessary for every entity to enhance training of its human resources as this increases the intention to stay. According to Long et al. (2012), besides improving employee's capabilities, training also enables employees to achieve organizational goals and objectives. Further, Pare and Tremblay (2007) argued that reduced employees' turnover is experienced in organizations where existing employees receive training to better their competence.

Qualitative data analysis
The questionnaire used to collect data had also semi-structured questions. The semi-structured questions were thus analyzed and presented into their particular categories as summarized in Table 8.

Conclusion
HRM practices continues to play a significant role in employees' turnover intentions decisions. However, its influence seem to be different from one context to the other. In reference to the outcome of this research, the following conclusions were made. First, there is clear evidence that compensation significantly affects pubic secondary schools' teachers' intention to stay in Kenya. Thus, improvement in compensation practices such as adequate payment of salaries and other benefits comparable to the market, provision of better medical covers as well as satisfactorily reviewing compensation frequently will result in significant increase of public secondary schools' teachers' intention to stay.
The study also concluded that training significantly affects intention to stay of public secondary schools' teachers in Kenya. Accordingly, improvement in teacher training activities such as carrying out a systematic analysis, determining the needs for training programs, conducting induction training to prepare new teachers to acquire skills necessary for their jobs and occasionally evaluating the effectiveness of employee training programmes will result in significant increase in intention to stay of public secondary school teachers. Further, this research concluded that career development and performance appraisal had insignificant role in teachers' turnover intentions in Kenya. However, such insignificant role was anticipated since these HRM practices are likely to affect significantly employees' behaviors and attitudes.

Recommendations
This study recommends to teachers' employer and other organizations that the focus should not only be on the actual implementation of HRM practices but also on employees perception of HRM practices so as to incorporate their employees' expectations so as to realize lower intention to leave. The TSC and other employers with regard to this should implement a participatory approach system before and during implementation of HRM practices and policies so that the employer will have feedback from their employees and where expectations are not met appropriate remedies can be done or consensus reached. This will also help the employees (teachers) to own the policies and practices and thus enhancing implementation and achievement of the intended goals.
Since the findings indicated compensation significantly affect turnover intention among teachers, the study also recommends TSC to review compensation for teachers who are not administrators since studies have shown that compensation of non-managerial employees play a crucial role in determining employees' turnover intentions. It is also recommended the TSC policy on performance recognition, rewards and sanctions on the teaching profession, should not only consider general performance but also value addition on performance as a basis for reward in order to motivate teachers who are in schools that do not get higher grades. The study also recommends current medical scheme should be reviewed to ensure better medical services to teachers.
From the findings, induction of new teachers is currently being conducted in an unsystematic manner, which has contributed to some teachers joining TSC not well prepared on the job expectations, practices and policies that affect them. It is therefore recommended that TSC should come up with policy guidelines on induction training of its teachers, which will ensure uniformity in training and thus prepare all the new teachers with the work ahead. Proper induction will help to reduce intention to leave particularly among the new teachers.
The Principals being the immediate Human Resource Managers in the schools need to have HRM skills in order to discharge the HRM functions effectively. However, from the findings most of the Principals indicated they are not well equipped with the skills. It is therefore recommended that TSC in collaboration with Kenya Secondary Schools Heads Association (KESSHA) and Kenya Education Management Institute (KEMI) should take the initiative to equip the school Principals with relevant HR skills through organizing refresher training to principals. This will assist the Principals to implement the HRM practices effectively thus reducing intention to leave among teachers. Additionally, from the findings the schools' management has been directly involved in sponsoring the teachers for training, but this has not been done as per the teachers' expectations. Thus, it is recommended that additional funds should be allocated to schools to enable the management to adequately offer training opportunities to teachers. This is necessary because from the findings training of teachers significantly influences turnover intention.

Limitations and considerations for further research
From the logistic regression results, HRM practices accounted for a low variation in turnover intention. This study could consider four HRM Practices. This implies that there are other practices responsible for the variance in turnover intentions which were not considered in this study. Therefore, it is recommended future research should seek to determine other variables that influence change in turnover intentions of teachers in public secondary schools. Other studies should also focus on other sectors, both public and private, as this study predominantly focused on the education sector. The study was conducted in one of the 47 counties in Kenya and thus the findings may not be generalized to the whole country. Thus, this calls for future research to widen the scope of the study by including other counties. Since this was a cross sectional study other studies can consider a longitudinal study in order to counter the limitations of the former.