Our study explored the latent profiles of teacher burnout and the question of whether these profiles could be differentiated based on their relations with individual resource factors, such as psychological capital, professional identity, and positive coping. The results showed that there were six consistent latent profiles of teacher burnout with a meaningful difference in the whole study population at T1 and T2 and that teachers in one profile would transfer to another profile with a latent transition probability over time. It was demonstrated that the burnout dynamics and the temporal sequence of burnout symptoms varied across teachers, but that the same profiles were generally maintained on the whole. In addition, psychological capital and professional identity were important influencing factors in reducing the occurrence of teacher burnout and in increasing transition probability toward burnout symptom alleviation over time, while positive coping played an important role in reducing the occurrence of teacher ineffectiveness.
5.1. Profiles of Burnout—The Differential Roles of Individual Resource Factors
The present study using the LPA approach confirmed the first hypothesis that the same profiles existed at varying times throughout the whole study population and that the “Low/no burnout” profile accounted for the largest proportion of the sample (see
Table 2). Specifically, two-wave longitudinal tests found that there were six well-distinguished stable latent burnout profiles, namely, “Low/no burnout”, “Highly ineffective”, “Ineffective instigated”, “Exhaustion instigated”, “EE&DP dominated burnout” and “Burnout”. The result of six latent profiles was similar to the five profiles developed by Leiter and Maslach (2016) [
16]. The difference was that the present study distinguished the “Highly ineffective” profile, although the proportion of this profile was very small. This profile was similar to but different from the “Ineffective instigated” profile. As seen from the results of this study (
Table 3), when EE and DP were low, the symptom of highly reduced personal accomplishment was easily alleviated and there was a high latent transition probability to the “Low/no burnout” profile. However, the symptoms of “Ineffective instigated” were relatively difficult to alleviate and had a higher latent probability of transferring to the “Burnout” or “Highly ineffective” profiles. This result suggested that the median split method may obscure some important subtypes. Although these subtypes appeared to be similar, they may lend themselves to different intervention strategies. The ineffective profile (including both “Highly ineffective” and “Ineffective instigated”) reflects a psychological relationship with work, and it suggests that work-life experience was not at the same level of self-actualization. This profile was more prevalent among teachers in both tests (>20%), and it suggested that ineffectiveness (reduced PA) was actually a far more common experience among teachers; even if not yet well understood, this ineffective profile deserves more research attention [
8,
16,
51].
Although the three symptoms of burnout have been widely accepted, their independence and dependence on each other have been debated [
50,
52,
53]. In the present study, the ineffective profile (including “Highly ineffective” and “Ineffective instigated”) and the exhaustion profile (“Exhaustion instigated”) were identified, with the exception of a profile named “Depersonalization instigated” (high DP, low EE and reduced PA). Higher DP appeared only when EE was high (“EE&DP dominated burnout”) or when all three dimensions were high (“Burnout”), not in isolation. Leiter and Maslach (2016) [
16] claimed to have discovered this profile, but it has not been found by other studies [
17,
54], including the profile study of teacher burnout [
8]. In studies using the median split method, this profile existed, but in a very small proportion [
12]. In this study, the mean value of this dimension was obviously lower than that of the other two dimensions, meaning that most teachers did not think that they treated students negatively and coldly. Whether this manifests differently in different occupations is unclear. However, depersonalization plays an important role in the development of teacher burnout [
11], and the question of how to better identify depersonalization among teachers is an issue of concern. The most direct effect of depersonalization was the disruption of interpersonal relationships [
55,
56], and poor interpersonal relationships may also increase burnout [
57,
58]. Thus, an alternative way of identifying early symptoms of depersonalization may be to include interpersonal assessments in teacher evaluations.
This study adds to what is known about the impact of individual resource factors on the development process of teacher burnout. The results of this study concurred with the COR theory [
29,
30] and the JD-R model [
31,
32], suggesting that individual resources can reduce burnout symptoms. Previous studies have revealed the direct or moderating effects of psychological capital, professional identity, and positive coping on burnout [
6,
9,
10,
41,
59]; however, the present study contributed to the burnout literature by showing that psychological capital and professional identity are important influencing factors in reducing the occurrence of teacher burnout and increasing the transition probability of each profile toward burnout symptom alleviation over time, while positive coping plays an important role in reducing the occurrence of teacher ineffectiveness.
Specifically, teachers with high psychological capital or high professional identity were more likely to be in the “Low/no burnout” profile and less likely to be in the “Ineffective instigated”, “Exhaustion instigated”, “EE&DP dominated burnout” and “Burnout” profiles. Additionally, during the development of individual burnout profiles, teachers had a significantly higher probability of transferring to “Low/no burnout” and had a significantly lower probability of transferring to “Burnout”. In other words, psychological capital and professional identity could not only reduce the occurrence of various burnout profiles, but also help teachers reduce burnout symptoms and avoid worse situations when burnout occurs. This effect was not obviously associated with different burnout profiles or burnout symptoms. Compared with psychological capital and professional identity, the role of positive coping in reducing the occurrence of various burnout profiles was limited; however, positive coping played an important role in improving the efficacy of teachers. Previous studies have not emphasized this point [
6,
41]. Of course, most previous studies have analysed the relationship between coping and teacher burnout in cross-sectional data, while no study has analysed the role of positive coping in the development of teacher burnout by using longitudinal data. As mentioned above, depersonalization is a form of dysfunctional coping that reduces the accomplishment of teachers [
11]. Could positive coping improve teachers’ accomplishment? In an intervention study on coping strategies and socioemotional competence, participating teachers demonstrated an increase in positive coping strategies and a significant increase in personal accomplishment [
41]. More longitudinal studies are needed to analyse the impact of positive coping on teacher effectiveness.However, there were two unexpected findings: professional identity was associated with the “Highly ineffective” profile compared to the “Low/no burnout” profile and was associated with the transition from the “Low/no burnout” profile to the “Highly ineffective” profile compared with staying in the low/no burnout profile. One possibility was that there was a high proportion of male teachers among the “Highly ineffective” profile (34%, compared with 18.9% in the total sample, 17.4% in “Low/no burnout” profile, and 22.7% in “Ineffective instigated” profile). Male teachers were more likely than female teachers to be in the “Highly ineffective” profile (OR = 2.44,
p = 0.030) and were more likely to transfer from the “Low/no burnout” profile to the “Highly ineffective” profile (OR = 2.82,
p = 0.048). When gender effects were present, higher professional identity did not increase the probability of teachers being in the “Highly ineffective” profile compared the “Low/ no burnout” profile (OR = 1.34,
p = 0.232), but it still increased the probability of transition from the “Low/no burnout” profile to the “Highly ineffective” profile (OR = 5.10,
p = 0.013). The second possibility was that the professional identity of teachers among the “Highly ineffective” profile was too high, even higher than teachers among the “Low/no burnout” profile (
Table 4). According to the environment perspective of job burnout, burnout occurs when individual goals and expectations are not successfully translated into actual value [
55,
60]. However, whether there are differences in burnout developmental trends between male and female teachers among the “Highly ineffective” profile and whether there is a covariant relationship between such trends and professional identity needs to be explored in future longitudinal studies with more time points and a larger sample.
5.2. Implications for Interventions
These profiles and the roles of individual resource factors could also have implications for interventions. As burnout interventions and rehabilitation are not always effective in treating severe chronic burnout [
61,
62], more emphasis needs to be placed on taking proactive measures to prevent burnout, for example, by distinguishing different burnout profiles [
16,
17] and improving individual resources [
63,
64].
For example, “Burnout” teachers exhibited low scores on all three resource factors, and an increase in each factor would help alleviate their burnout symptoms. “Ineffective instigated” teachers had high scores on professional identity but low scores on psychological capital and positive coping, so improving psychological capital and positive coping strategies among this group of teachers would be helpful. For “Exhaustion instigated” teachers, their relatively high positive coping scores prevented them from transferring to “Highly ineffective” and “Ineffective instigated” but had no obvious effect on the alleviation of exhaustion. They may need psychological capital intervention training or an increase in their recognition of the value of their teaching. The proportion of teachers who fell into the profiles of “Highly ineffective” and “EE&DP dominated burnout” was very small, and it was more important to distinguish the two profiles. The former group needs an increase in positive coping strategies, while the latter group needs an increase in psychological capital and professional identity. Besides, burnout is a very common issue among teachers, even in the recent context of COVID-19 [
65,
66]. We believed that our study may provide some suggestions for intervention for teachers suffering from burnout during the pandemic.
There have already been some intervention programs for teachers or other groups that have focused on these three resource factors [
41,
67,
68,
69]. We can learn from these intervention programs to solve the problems of the five profiles of teacher burnout (except the “Low/no burnout” profile). However, we suggest that it is desirable to distinguish different burnout profiles before initiating targeted interventions, as burnout interventions are not always effective for different samples [
61,
62]. In addition, the reduced PA dimension has been more neglected in previous burnout research [
16], while in the profile study of burnout, researchers proposed to pay attention to this dimension [
16], which may be a more decisive symptom of burnout [
17]. Furthermore, the results of this study indicate that no matter how individual burnout profiles transfer over time, there are six relatively stable profiles of teacher burnout across the population for which corresponding intervention measures can be formulated (except the “Low/no burnout” profile). This approach has at least two advantages: one advantage is that doing so improves the effectiveness of intervention and constantly refines the intervention based on the effect of intervention; the other advantage is that doing so facilitates group intervention in one or more schools, saving educational costs and benefiting more teachers.
5.3. Limitations and Implications for Future Research
There were also some limitations in this study. First, our data consisted of primary school teachers, a female-dominated group; consequently, we cannot be sure that the same latent profiles of burnout and their transferring probabilities emerge in other groups of professionals. Therefore, the results of this study are applicable only to primary school teachers, and there is a need to replicate the profiles in the context of other professionals and to compare those groups with teachers in future studies. In addition, the study sample was from a district with strong educational level, whether the results of this study are applicable to other districts with relatively weak educational levels needs more comparative studies in the future. Second, the time-lags between measurements were not theoretically determined. Although a three-year time lag has been included in previous burnout studies [
18], it is unknown whether this length of time best describes the burnout process among teachers. However, the temporal aspects of burnout are unclear, although some studies have suggested that burnout symptoms accumulate and develop over a long period [
70]. In addition, there is some question as to whether the temporal aspects of burnout are related to occupation. This may be a question that needs to be answered once research on burnout development has been sufficiently enriched. Third, there are many factors affecting teacher burnout, including individual factors and situational factors (e.g., job-related factors, lifestyle and economic status, etc). The extension of the investigation of various individual or situational factors of teachers and the analysis of the main effect and coaction mechanism of these factors are necessary for future research. For example, this might include other risk factors or protective factors, and their possible interactions, that are in line with JD-R theory [
31,
32].