The motivation-hygiene theory divides work motivation into two factors: motivation and hygiene (Herzberg et al., 1959). Among them, hygiene factors are related to a person’s basic survival needs, such as the reward system, wages, and interpersonal relationships. However, these factors cannot motivate or bring satisfaction to employees, they can only prevent dissatisfaction and maintain the original work efficiency and psychological state of employees. Motivation factors involve growth needs, which refer to internal factors such as the work itself, such as recognition of work completion, sense of accomplishment, responsibility, progress, and the work itself. These factors represent higher-level needs and are an important way to improve job satisfaction and thus enhance job motivation (Herzberg, 1971; Tietjen and Myers, 1998). If motivation factors can be satisfied, employees will be satisfied and improve their work enthusiasm and efficiency. If motivation factors are not met, they will not create negative emotions in employees like hygiene factors. For example, a person who leaves a teaching job they love because they do not earn enough to cover living expenses (hygiene factor), or a person who leaves a teaching job that pays well but they do not like teaching every day (motivation factor). Looking at this example, the ideal situation would be to hire someone who intrinsically enjoys teaching and is paid well enough not to worry about the salary. “The modern strategic human resources management (HRM) practices, argue that school leaders should intentionally design a supportive employee experience for teacher support as an organizational talent management strategy” (Tran and Smith, 2020). A major cause of teacher turnover is insufficient employer support for teachers’ needs, i.e., administrative support has been identified as the most important factor for teacher retention (Ingersoll, 2001; Horng, 2009), so we focus on administrative support strategies.

What factors affect the turnover intention of rural teachers? The motivation-hygiene theory states that teachers, when choosing a teaching career, are influenced by hygiene factors related to needs at lower levels in the hierarchy of needs - i.e. money, working conditions, relationships or company policies, etc., as well as motivation factors related to needs at higher levels - promotion, achievement, responsibility, etc. On the whole, rural teachers’ turnover intention is affected by institutional factors including teachers’ salaries, working conditions, policy formulation and implementation, and individual characteristics including gender, age, experience, and ability (Hanushek and Rivkin, 2010; Ping and Yao, 2019).

According to Herzberg’s motivation-hygiene theory, hygiene factors represent lower levels of needs, and when these factors are lower than the teacher’s acceptable level, teachers will become dissatisfied with their jobs and even leave their jobs (Sanjeev and Surya, 2016). Based on the survey of rural teachers in western China, this paper focuses on the analysis of three important hygiene factors that rural teachers are most concerned about, namely rural incentive policies, personnel policies, and teacher pay.

The reason for choosing these three hygiene factors is that the Chinese government and schools have reformed these three aspects of hygiene factors for rural teachers, and we pay attention to the impact of these policy interventions.

Motivation factors can meet the personal psychological and growth needs of employees and are an important way to improve employee job satisfaction and reduce turnover intentions. Among them, promotion, work itself, and achievement are the key motivating factors (Sanjeev and Surya, 2016; Hur, 2018).

Influenced by personal preferences, individual characteristics, etc. there are individual differences in teachers’ work motivation, job satisfaction, and turnover intention. Gender, age, marital status, and teaching subjects may all affect teachers’ turnover intention (Guarino et al., 2006; Shirol, 2014; Geiger and Pivovarova, 2018; Suárez and Wright, 2019). Studies have shown that female teachers have a stronger willingness to move than male teachers at the beginning of their careers, but when teachers are in the middle and late stages of their careers, the turnover tendencies of male teachers and female teachers gradually tend to be the same, and the turnover intention of teachers shows a “U”-shaped trend with the increase of teaching age (Ingersoll, 2002). Scholar Du Ping studied the relationship between the salary of rural primary and secondary school teachers and turnover intention and found that marital status has a significant impact on teachers’ turnover intention, among which male teachers and unmarried teachers have the higher turnover intention (Ping and Yao, 2019). Due to the general shortage of rural art teachers and English teachers, this study focuses on the moderating effects of these two disciplines on the influence of incentive hygiene factors on teachers’ turnover intention. Therefore, the following assumptions are made, see Figure 1:

The present study was a part of a larger research project on attracting and retaining rural teachers conducted by the first author. The research population targeted in-service teachers working at rural schools in Western China. Multi-step stratified sampling was used for reaching the participants from May to July 2018. First, we choose three areas 14 poverty-stricken areas in Western China based on the poverty area list. Second, we choose one typical county from each poverty-stricken area. Third, we choose rural schools located in nearby counties, remote towns, and the remoter town. At last, we send out 1,000 questionnaires to all teachers in the sampling schools with the help of a local county bureau officer and got 973 valid questionnaires.

In addition, The Research project team also conducted a field visit and in-depth interviews with the relevant persons, such as school principals, teachers, the Bureau Officer of county Education, and students.

Among all valid questionnaires, there were 545 (56%) female teachers, 428 (44%) male teachers, and 788 (81%) minority teachers. The age of teachers is mainly concentrated 34 years old and below and 34 ~ 45 years old. Among them, 315 teachers (32.4%) are 34 years old and below, and 380 teachers are between 34 and 45 years old (39.1%). Married teachers accounted for the majority of 841 (86.4%), and unmarried teachers accounted for 132 (13.6%). In addition, there are 402 (41.3%) teachers in schools around the villages, 324 (33.3%) and 247 (25.4%) teachers in the remoter and remotest villages, respectively. There are 664 (68.2%) teachers whose hometowns are places of residence, and 309 (31.8%) whose hometowns aren’t places of residence, as shown in Table 1.

The 18 measurement items in this study were summarized from the existing literature and partially adjusted according to the current research. The projects were based on validated instruments previously (Smerek and Peterson, 2007; Lundberg et al., 2009). All items used a 5-point Likert scale (1 = strongly disagree; 5 = strongly agree). Hygiene factors are divided into three parts: rural incentive policy, personnel policy, and teacher pay. According to policies formulated by most western counties, rural incentive policies are measured according to the rural incentive policies formulated by most western counties (Jiang and Mingze, 2019; Liang, 2021). The three personnel policies are from Anzia and Moe and have been modified according to the actual situation of rural teachers in China (Anzia and Moe, 2013). Finally, the salary evaluation adopts three items of Ryan and adjusts it according to the situation of rural teachers (Ryan et al., 2017).

Motivation factors are divided into three parts: advancement, the work itself, and achievement. All motivation factors are drawn from domestic and foreign literature and adjusted according to the current research context (Sanjeev and Surya, 2016; Qingye, 2018; Zhihui and Weihong, 2018). Teachers leaving their current school or quitting the teaching profession completely represent two aspects of turnover intention (Lindqvist et al., 2014). In addition, moderator variables are selected from individual variables, such as gender (1 = male, 0 = female), marital status (1 = married; 0 = unmarried), art teacher (1 = yes, 0 = no), English teacher (1 = yes, 0 = no), age group (1=under 35 years old,0=over 35 years old). Minority, school location, workload, and hometown were selected as control variables for this study (Zhichun and Jiang, 2007; Perryman and Calvert, 2019).

In this study, higher-order confirmatory factor analysis (CFA) was used to test the reliability and validity of the scale, respectively. The results showed that 16 of the 18 measurement indicators had factor loadings greater than 0.70, and only two indicators had slightly lower factor loading, 0.678, and 0.679, respectively. The t-value shows that each index is significantly correlated with its respective dimensions (p < 0.001), which indicates that all the indexes selected in this study have a satisfactory level of reliability. In addition, the overall Cronbach α of the questionnaire is 0.831, only the Cronbach α of the work itself is 0.678, and the rest of the items all exceed the critical value of 0.70, indicating that the internal consistency of each dimension is good. The multiple correlation coefficients (R²) of all measures ranged from 0.334 to 0.899, indicating that the percentages of these measures that could be explained by predictors ranged from 33.4 to 89.9% (see Table 2). In addition, the Kaiser-Meyer-Olkin value of the questionnaire design in this study was 0.877 > 0.6, and the value of p of the significant probability was 0.000 < 0.01. The sample validity was high, which was analyzed by the validity test.

To examine the influence of motivation-hygiene factors on the turnover intention of teachers in western rural areas, a correlation test was carried out. Table 3 shows the correlation between the various dimensions of motivation-hygiene factors and turnover intention. According to the results of the Kendall harmony coefficient, the dimensions of hygiene factors were significantly negatively correlated with turnover intention, and the dimensions of motivation factors were also significantly negatively correlated with turnover intention. And among the hygiene factors, personnel policy has the strongest correlation with turnover intention, r_5 = −0.322; among motivation factors, the work itself has the strongest correlation with turnover intention, r_2 = −0.412.

The mean values of each dimension of incentive factors, hygienic factors, and turnover intention are between 2.290–3.338, among which the mean value of advancement is the highest, and the mean value of work itself is the lowest. The standard deviations of incentive factors, hygienic factors, and turnover intention are between 0.981–1.112, among which turnover intention has the highest standard deviation and advancement has the lowest standard deviation.

According to the research hypothesis, this study constructed a structural equation model of motivation factors, hygiene factors, and teachers’ turnover intention. To evaluate the structural equation model and obtain the path coefficient and its significance level, a preliminary structural model fit evaluation was performed on 973 samples. The test results show that χ^²/df = 7.73, RMSEA = 0.08, GFI = 0.91, AGFI = 0.88, and CFI = 0.88. Except for the good fit of RMSEA and GFI, the fit of other indicators is less than 0.9 but still meets the Baumgartner recommendation above 0.8, which may be due to the large sample size (Baumgartner and Homburg, 1996; Jackson et al., 2009). Next, this study will add control variables to the initial model, and use the Bollen-Stine method to correct the fitness of all models (Bollen and Stine, 1992; Fisher and King, 2010), proving that the low fitness of the model is not caused by the poor fitness of the model itself.

To investigate whether there are significant differences in the turnover intention of rural teachers in Western China with different ethnic minorities, school locations, workload, and hometowns. This study conducted a one-way analysis of variance on 973 samples. There are significant differences in teachers’ turnover intention. Therefore, ethnic minorities’ school location workload and hometown were introduced into the structural equation model as control variables. After the introduction of control variables, the model fit index χ^²/df = 6.96, which still did not meet the standard requirement of 5.0, but was lower than the χ^²/df of the initial model. Second, while GFI = 0.89, AGFI = 0.86, CFI = 0.84, RMSEA = 0.078. Although GFI, AGFI, and CFI all decreased slightly, according to existing research, in addition to motivation and hygiene factors that may have an impact on turnover intention, it is undeniable that the individual characteristics of teachers do have a certain impact on the dependent variable turnover intention.

According to the Bollen-Stine Bootstrap report, the model has a good fitting effect in 2,000 Bootstraps. The poor fit is 0 times, and Bollen-Stine bootstrap = 0.00 < 0.05, which proves that the model is well fitted. Also, the chi-square before correction is 1234.90, while the chi-square after correction is 205.34, which is a big improvement. The modified fit indices are shown in Table 4. It can be found that each fit index has reached the ideal standard value, which is a very obvious improvement compared with fit indexes before the correction. The χ^²/df before the modification is 6.82, and it is not only smaller than the ideal standard value of 5.0 but also smaller than the idealized standard value of 3.0. The corrected RMSEA is not only less than the standard of 0.1 but even close to 0. GFI and AGF are less than or equal to 0.9 before the correction, and are not only greater than the standard of 0.9 but also gradually approached 1 after the correction. The CFI is even more equal to 1 after correction. This shows that the model has a very good fit.

The standardized path coefficients among the variables are shown in Figure 2. The motivation factors consisting of promotion, work itself, and achievement and the hygiene factors consisting of rural incentive policies, personnel policies, and remuneration all have significant negative effects on the turnover intention of rural teachers in western China. The standardized coefficients are −0.26 and − 0.12, respectively. Both H1 and H2 are supported by the test results, and the influence of motivation factors on turnover intention is greater than that of hygiene factors, as shown in Table 5. After adding control variables, ethnic minorities, school location, and workload all have significant effects on turnover intention, and their standardized coefficients are −0.08, 0.15, and 0.15, respectively. Among them, hometown has no significant impact on turnover intention. It can be seen that it is difficult to retain rural teachers only by relying on abundant monetary incentives, and non-monetary incentives such as the realization of self-worth, and the satisfaction of emotional needs, can affect rural teachers’ work behavior and decision to leave.

According to the results of multi-group analysis, it is investigated whether H3-H6 is established, and then the moderating effect of individual factors on the turnover intention of rural teachers in western my country is analyzed. First, this study uses the chi-square test to compare different models to test the path differences between different groups, that is, the strength of the relationship between structures, such as the moderating effect of male and female, married and unmarried, different teaching subjects groups and different ages on turnover intention. The results show that, except for H3a and H3b, 6 of the 8 hypotheses have been verified, and the structural model coefficient of gender = 0.29 > 0.05, indicating that gender does not significantly moderate the relationship between motivation-hygiene factors and turnover intention effect.

The estimation results of the multi-group analysis show that H4a/H4b H5a/H5b and H6a/H6b are supported and the results are significant, as shown in Table 6, and the standardized and unstandardized estimates of the paths between motivation factors, hygiene factors, and turnover intention are given, respectively. The results of the moderating effect show that there are significant differences in the selected moderating variable paths between the variables of the married and unmarried groups, English teachers and non-English teachers, art teachers and non-art teachers, and teachers over 35 years old and teachers under 35 years old.

First, the hygiene factor of unmarried teachers has a significantly stronger influence on turnover intention than married teachers (β = −0.31 and − 0.102, p < 0.05), and the influence of unmarried teachers’ motivation factors on turnover intention is also significantly higher than that of married teachers. Marriage teachers (β = −0.426 and − 0.236, p < 0.05). Compared with married teachers, unmarried rural teacher’s pay more attention to institutional environmental factors, and are eager to improve their teaching subjective initiative, attach great importance to personal development, and better devote themselves to rural education construction and development.

Secondly, English teachers paid more attention to hygiene factors than non-English teachers (β = −0.33 and − 0.105, p < 0.05); art teachers’ motivation factors and hygiene factors had significantly stronger effects on turnover intention than non-art teachers, but art teachers placed more importance on motivating factors (β = −0.562 and − 0.302, p < 0.01). The results of the study may be related to the higher opportunity cost of English teachers teaching outside the school. At the same time, the characteristics of art education and art teachers should be fully taken into account. The higher investment in the early education cost of art teachers leads art teachers to think that motivation factors are more effective to keep them in their current teaching positions.

Finally, the motivation factors and hygiene factors of teachers over the age of 35 have a stronger impact on turnover intention than those of teachers under the age of 35, and the impact of motivation factors of teachers over the age of 35 on turnover intention is significantly stronger than that of teachers under the age of 35 (β = −0.457 and − 0.225, p < 0.05). Likewise, the effect of hygiene factors on turnover intention was stronger for teachers over 35 years old than for teachers 35 and under (β = −0.131 and − 0.098, p < 0.05). Teachers of different age groups pay more attention to the satisfaction of motivating factors. If rural schools pay more attention to the motivating factors that teacher’s value, rural teachers are more willing to stay and continue teaching.

Drawing on motivation-hygiene theory, this study theoretically and empirically tests a model of hygiene and Motivation factor that explains teachers’ turnover intention and examines the moderation effect of gender, marital status, and teaching subject, respectively. Three major findings from this study are as follows:

First, Hygiene factors, composed of compensation, incentives, and personnel policy are the significant predictor which restrains teachers’ turnover intention. Particularly, personnel policy is the most influential factor in building hygiene factor. This finding implies that rural teachers in China indeed respond to the three key hygiene factors, and their turnover intention can be decreased by raising compensation level (Shen, 1997; Stinebrickner, 1998; Podgursky et al., 2004), attractive rural incentives (McEwan, 1999) and reasonable personnel policy (Loeb et al., 2009; Anzia and Moe, 2013).

Second, Motivator factors composed of achievement, work itself, and advancement, are negatively related to rural teachers’ turnover intention. Besides, the impact of motivational factors on turnover intention is greater than that of hygiene factors. Among the three kinds of Motivator factors, Advancement contributes the most as a factor. Thus, rural schools providing internal growth opportunities such as training, and teaching instruction, would increase teachers’ teaching skills and competence, which results in lower teacher turnover (McGilton et al., 2013). In addition, schools providing teacher respect, recognition, and appreciation may also reduce rural teachers’ turnover intention (Hogan et al., 2013).

Third, marital status, teaching subject, and age have a significant moderation effect on the relationship between motivation/hygiene factors and turnover intention, whereas gender has no significant moderation effect. Both the hygiene and motivator factor are significantly stronger among unmarried teachers than married teachers. English teachers concern more on hygiene factors, which affect their turnover intention stronger than non-English teachers. For art teachers, both the effect of hygiene and motivation factors on turnover intention is significantly stronger than for non-art teachers. The motivation factors and hygiene factors of teachers over 35 years old have a stronger impact on turnover intention than those of teachers under 35 years old.

First, this study enriches teacher turnover intention literature by exploring the effect of institutional factors in the context of the Chinese western rural labor market. Previous studies of teachers’ turnover intention are based on a set of factors including personal attributes, school conditions, and family support, respectively, (Yanling et al., 2017). However, institutional factors especially rural teacher incentives have not been thoroughly explored. Our study sorted the institutional factors from a theory perspective and examined their influences on rural teachers’ turnover in western China.

Second, by adopting the hygiene-motivation theory, this research empirically verifies three types of hygiene factors and motivator factors with distinct effects on turnover intention. The previous study mainly tested this theory on employees of the private sector, while only a few directly examine the campus teacher. Our research identifies different weighted effects of Hygiene and Motivator factors on turnover intention and extends the current turnover intention literature to the context of the rural teacher labor market.

Third, this study extends the Hygiene-motivator theory by examining the moderation of personal factors between teacher motivation and turnover intention in the Chinese context. A limited study is concerned with the moderation effect of individual factors. Our research verifies the effect of hygiene and motivator factors on turnover intention differed between married teachers and unmarried teachers, which supports Ping and Yao (2019) conclusion. Besides, a distinct effect also exists between teachers of different ages, English teachers, non-English teachers, art teachers, and non-art teachers in rural China.

Our findings may have important implications for relevant education practitioners to better manage the rural teachers’ turnover problems in less developed countries.

First, policymakers should emphasize the influence of hygiene factors on teachers’ turnover decisions, especially design a reasonable and flexible personnel policy for rural teachers. For example, the performance evaluation plan should aim to help rural teachers to improve their teaching skills and ability not just assess them. Also, the county government should make a more flexible teacher exchange plan to help teachers to transfer their job between rural schools and urban schools. Besides, our results have verified that the various rural incentive policy enacted in Chinese has the effect to reduce teachers’ turnover intention. Hence, the county government should continue to strengthen the monetary and non-monetary incentives for rural teachers, such as increasing the standard of subsidy for rural, more chances for rural teachers to have promotions, and so on.

Meanwhile, it is worthwhile noting that motivator factors play a more important role in restraining teachers’ turnover intention, especially the advancement factor. This finding implies rural schools should focus on rural teachers’ internal growth opportunities, such as training, and teaching instruction, to increase teachers’ teaching skills and competence. Besides, recognizing employees’ effort and contribution is also an effective and less expensive approach to attracting rural teachers for school managers.

At last, both the hygiene and motivator factor are significantly stronger among unmarried teachers than married teachers. This result implies that policymakers should pay more attention to unmarried teachers’ hygiene and motivation needs, including career development, recognition, and incentive. Besides, English/art teacher’s pay more attention to hygiene factors than non-English/art teachers, this may be because the salaries of English teachers in rural areas are relatively insufficient compared with civil servants (Soodmand Afshar and Doosti, 2016). If English /art teachers choose to work in the off-school tutoring institution, they can get more compensation. Therefore, the opportunity cost for English teachers to teach in other schools will increase accordingly as well as for art teachers. So policymakers should design art and English teacher’s salary above the market level. Finally, the results of age group adjustment show that, for rural teachers of different age groups, improving hygiene factors and motivating factors can significantly reduce their turnover intention, and the effect of motivation factors is much greater than hygienic factors. In addition, the same motivation and hygienic factors have a greater effect on teachers over the age of 35. It can be seen that the various reform policies implemented for rural teachers have a better retention effect on teachers over the age of 35.

This study contains limitations that can turn into suggestions and opportunities for further research. First, the present study only examined the individual perception of hygiene and motivator factors and turnover intention. These variables can also be estimated from a school perspective, so future research can apply a school-level analysis to investigate the effect of hygiene and motivation factors on a school’s turnover rate. Second, despite our best effort to incorporate the three kinds of hygiene-motivator factors into the model according to our investigation, we cannot exclude the possibility of having other important factors that may also contribute to the hygiene-motivator construct. For example, our model did not consider the effect of the working condition of the school, which also play a key role in turnover intention (Boyd et al., 2009; Bullough et al., 2014). Third, this study is not comprehensive as some mediating variables were not included in the research model, for example, job satisfaction, and self-efficacy. At last, the generalizability of our results is limited. As we only collected data from western China, the sample may not be representative enough of eastern and middle China.