The Effect of Emotional Intelligence, Self-Efficacy, Subjective Well-Being and Resilience on Student Teachers’ Perceived Practicum Stress: A Malaysian Case Study

Stress is inevitable in the world of teaching and practicum training and therefore, student teachers naturally incur a certain level of stress due to the demands for them to use various knowledge and skills in real school and classroom environment. Hence, practicum stress needs to be addressed accordingly. The central focus of this study is using a partial least square-structural equation modeling to explore the inter-relationships among the student teachers’ personal resources to mitigate practicum stress. A sample of 200 student teachers selected by purposive sampling from teacher education institutions in Sabah, Malaysia was used in this study. This study collected data via survey methods using a questionnaire developed from several existing scales. Findings showed that emotional intelligence, self-efficacy, and subjective well-being were able to explain resilience with good predictive accuracy and relevance but poorly for practicum stress. These findings were suggestive of the need to include additional constructs to explain perceived practicum stress better in future exploratory research.


Introduction
Recent literature shows that in many world countries, the teaching profession is a stressful job and teacher stress is in fact, a global phenomenon (Skaalvik & Skaalvik, 2016). Alemu, Teshome, Kebede and Regassa (2014) stated that teaching is considered as one of the most stressful profession where 33 to 37% of teachers suffer from extreme stress due to the intrinsic characteristics of the profession. Stress is a complex issue among teachers and well documented in literature (Klassen & Durksen, 2014;Malik & Ajmal, 2010). In fact, Danyluk (2013) mentioned that the teaching practicum also raises concern among the student teachers as it has been identified as a source of stress and anxiety too. Studies also show that student teachers also experienced stress during their teaching practicum (Malek & Ajmal, 2010). Therefore, the teacher education programme must ensure that teachers are prepared to deal with a stressful working situation. Abebe and HaileMariam (2011) defined stress as any factor that act internally or externally causing it harder to adapt, thus greater efforts are needed from the individual to maintain equilibrium in himself and of the external environment. There are two main types of stressor: personal and situational. Personal stressors include the thinking, beliefs and internal feeling which make it hard for the individual to function whereas situational stressors include specific incidence such as a teacher witnessing a student fight. The existence of stressors everywhere and yet the different ways of how individuals react to stress imply that responses to stress is not universal as one person can cope better with internal stress than another person (Xin et al., 2017).
Past studies defined stress as the difference and incongruence between demand in a situation and the coping strategies (Okeke, Adu, Drake & Duku, 2014;Okeke, Shumba, Rembe, & Sotuku, 2015). Therefore, resources to cope with stress are needed to balance the situation that causes symptoms of stress during a teaching practicum. As a teacher, stress is inherent in their daily life that it is thus, necessary for them to know how to cope with stress (Kaur, 2011;Maphosa & Shumba, 2010;Richards 2012).
The coping resources refer to the coping behaviours used by the individual to cope with a situation that is stressful, driven by the intention to reduce the impact of stress and enabling the teacher to become adaptive to stress (Mapfumo, Chitsiko & Chireshe, 2012). Personal resources which include emotional intelligence, self-efficacy, and subjective wellbeing are able to promote resilience and deal with perceived practicum stress. Past studies have proven the existence of interrelationships among emotional intelligence, self-efficacy, subjective well-being and resilience (Abebe & HaileMariam, 2011;Hong, 2012;Koydemir & Schlutz, 2012;Lane & Wilson, 2011;Maher et al., 2012;Mapfumo, et al., 2012;Windle, Bennett, & Noyes, 2011). Alhebaishi (2019) stated that there is also a great need to consider the teacher emotion because it can either inhibit or enhance the teacher's ability to teach students effectively. Emotional intelligence refers to the social intelligence which enable one to identify his and other's emotion. It enables the differentiation of these emotions so that suitable decision can be made to think and act accordingly (Sharfras Navar & Vijayakumar, 2018). Emotional intelligence is perceived as the ability to recognize emotions in an adaptive manner, express emotions, control emotions and use these emotions (Moradi & Ardahaey, 2011;Nordin, 2012;Shafiq & Akram Rana, 2016). Therefore, in dealing with a stressful situation, emotional intelligence can play a critical role to mitigate the situation and ensure that teachers maintain their abilities to teach students accordingly.
The teacher education program prepares teachers to teach students in the classroom and deal with the numerous and varied responsibility of the teaching profession in the school setting. Hence, the student teachers must develop a strong sense of belief in their own capabilities to carry out the tasks demanded from them. Savas, Bozgeyik and Eser (2014) explained that efficacy relates to the necessary knowledge, skills and attitudes needed by the individual to behave in an expected manner. According to Maddux (2011), self-efficacy is the belief that one has the ability to coordinate both abilities and skills to achieve an intended goal in a certain domain or situation. An individual believes he can lead in a situation when he feels that he is in control of his environment, and this belief can overcome the challenging demands by using an adaptive action.
Teacher self-efficacy, according to Moulding, Stewart and Dunmeyer (2014) is a motivational construct to indicate the perception and belief of the teacher to perform specific teaching tasks. The concept of self-efficacy stems from the cognitive social theory that stresses on the significance of social experience and the need for observational learning in the process of developing an individual's personality (Mahler, Groschedl & Harms, 2018). According to this theory, the individual's choice in a certain situation depends on his personal observation. The observed behavior that is kept in his memory will shape his cognitive process and social behaviours in future events. According to Qureshi (2015), the interaction of cognitive (personal factors), behavioral factors and environmental factors shape the individual's action. Knowledge, expectation and attitude are cognitive factors whereas environmental factors relate to social norms, community access and influence on other people (Mahler et al., 2018).
Perceived self-efficacy can also be considered as the control of personal actions that is able to mitigate stress (Schwarzer & Warner, 2013). In other word, efficacy reflects an optimistic perception of one's ability to overcome stress and ensure that he is able to cope successfully with a stressful situation. External events such as practicum stress influences one's behavior through cognitive process, which in turn influences how external events are perceived and used by the individual in future endeavors. It also governs the effect of the external events to the person whereby he is able to regulate his environment that influences his behavior. The thoughts, expectations and behavior of a person in the future depends on the acquired experience from this behavior.
Apart from that, teachers who are happy are also shown to be more capable of dealing with their multitude responsibilities and heavy workload. The efforts to deal with stress are also related to the meaning of life to an individual. Satisfaction with life is a measure of well-being that determines the satisfaction that the individual feels about various domains in his life (Yusuf & Khan, 2018). Satisfaction with life is a cognitive constituent in the subjective well-being. Erdogan, Bauer, Truxillo and Mansfield (2012) explained that life satisfaction is a global assessment on the quality of life of the individual in general. When a person always develops a positive feeling but less on negative feelings, he becomes greatly satisfied with his life and this determines his subjective well-being (Eryilmaz, 2012).
These explanations on the resources to cope with stress implied that a person depends on internal resources as well as external resources to deal with a stressful situation. There have been studies investigating the interrelationship of these personal resources among themselves and how they were able to develop one's resilience in dealing with stress.
A study by Vesely, Saklofske and Leschied (2013) suggested that teachers employ emotional intelligence to help them in stressful teaching situation whereby high emotional intelligence is positively correlated to greater teachers' selfefficacy. According to Brackett, Rivers, Reyes, and Salovey (2012), those who can regulate emotions is able to predict emotional fatigue and build personal relationship with students, as well as employ a handful of constructive emotions including the sense of achieving something and having satisfaction with the profession of teaching.
Past studies have also shown that emotional intelligence and subjective well-being are interrelated (Austin, Saklofske & Mastoral, 2010;Razia, 2016;Soave, 2014). In a study by Razia (2016) on 120 student teachers in Aligarh, India, they found that emotional intelligence and subjective well-being were positively and significantly related. Austin et al. (2010) in their study among university students in Canada found that emotional intelligence and subjective well-being were positively correlated.
Further to that, individuals with higher emotional intelligence showed greater resilient capacity that makes the adaptation to change easier in a stressful situation, and they do not consider it as a threat, but rather, as a challenge (Schneider, Lyons & Khazon, 2013). Armstrong, Galligan and Critchley (2011) explained that emotional intelligence relates with resilience because a behaviour based on emotional intelligence in a stressful situation is adaptive. Armstrong et al. (2011) had investigated the relationship between resilience and emotional intelligence among 414 respondents aged between 24 and 58 years old and found that the relationship between resilience and emotional intelligence was positive as individuals in the higher resilience group also have higher emotional intelligence scores.
Earlier literature on emotional intelligence showed that individuals with a high score of emotional intelligence can deal adaptively to stressors and life's adversities more easily (Zeidner, Matthews & Roberts, 2012). Mayuran (2013) who studied the effect of emotional intelligence on stress management among respondents in schools and banks in Jaffna, confirmed that the relationship between emotional intelligence and stress management was moderate and positive. Punia, Balda and Poonam (2016), in their study in Haryana, India found that teachers with high level of emotional intelligence suffered from a low level of job stress. Thus, emotional intelligence can be related to self-efficacy, subjective well-being, resilience and perceived practicum stress.
Another personal resource used in dealing with stress is self-efficacy. The relationship of self-efficacy with life satisfaction was shown in Cakar's (2012) study of 405 young adults in Turkey. His findings showed that self-efficacy is a predictor of life satisfaction which represents the subjective well-being of the respondents. Santos, Magramo, Oguan and Paat (2014) involved 969 college students in the Philippines and found that their general self-efficacy and subjective well-being was positively related. Strobel, Tumasjan and Sporrle (2011) who assessed the relationship between self-efficacy and subjective well-being showed a positive correlation.
Past studies also indicated that individuals with greater self-efficacy withstand in their job longer compared to lower efficacious people (Hong, 2012;Narayanan & Cheang, 2016). Hong (2012) who compared two groups of teachers over a period of five years of teaching noted that teachers who continued in the teaching profession showed positive responses relating to resilience and self-efficacy compared to those who resigned. Narayanan and Cheang (2016), in their study had examined the influence of perceived social support and self-efficacy towards resilience among 377 first year students in public universities in Malaysia. Findings showed that those with self-efficacy reduced the possibility of being in the low resilience category and increased their chances to be in the moderate and higher resilience categories.
Self-efficacy is a proactive attitude and serves as a personal resource factor to control the problem of burnout among teachers (Skaalvik & Skaalvik, 2010). Natovova and Chylova (2014) explained that self-efficacy balances the demands from the imposing situations of the environment in a process of a cognitive stress appraisal that identifies stress as harmful, threatening or challenging. A low level of self-efficacy among teachers will likely lead to their perception of having more problems faced in their work and becoming a challenge towards their efforts to deal with stress arising from the perception. On the other hand, a highly efficacious teacher is more likely to use classroom management strategy and a more successful instructional style to address the demanding situation at work (Abu-Tineh, Khasawneh & Khalaileh, 2011).
Subjective well-being reflects the satisfaction of the teacher regarding his or her general life, including the teaching profession. However, there are very few studies that shows the effect of subjective well-being on resilience as most studies regard subjective well-being as the final outcome and not the means of achieving something. Earlier studies are more focused on determining the effect of resilience on subjective well-being (Franklin, 2013;Mahmood & Abdul Ghaffar, 2014). However, in this study, subjective well-being is regarded as a personal resource to achieve resistance against stress.
Studies on subjective well-being and stress are also more focused on exploring the effect of stress on subjective wellbeing (Coyle, 2010;Hou & Liu, 2016;Kuykendall & Tay, 2015). However, Vazi, et al. (2013) found that among the primary school teachers in East Cape, South Africa, their subjective and psychological well-being were able to predict stress significantly.
Resilience is also a resource to combat stress. According to the American Psychological Association (2014), resilience is the process of adapting when one faces hardships, trauma, tragedies, threats or significant stressors. However, this definition only highlighted its importance but does not describes its complex nature (Southwick, Douglas-Palumberi & Pietrzak, 2014). According to Pietrzak and Southwick (2011), resilience cannot be used with a binary approach, that is, determining that there is or there is no resilience because in reality, it exists at a continuum with different stages along the various domains in life. Southwick, Bonnano, Masten, Panter-Brick and Yehuda (2014) stated that an individual can adapt well with stress at the workplace and in the academic environment but might experience adapting failure in his personal life or in his relationship with others. Therefore, resilience changes with time as a function of development and it is the result of the interaction of the individual with his surrounding (Kim- Cohen & Turkewitz, 2012). Studies such as Chou, Chao, Yang, Yeh and Lee (2011) and Ng and Hurry (2011) stated that adolescents with low personal resilience are less capable of dealing and coping with stress while a good personal resilience helps individuals to recover from a stressful event (Kwok, Wong & Lee, 2014).
There are no studies in the past that have examined compounding effect of emotional intelligence, self-efficacy and subjective well-being on resilience and practicum stress. Based on past studies, it was shown that these personal resources are inter-related with one another, implying that some of these resources might act as a mediator in the research framework. Further to that, there are also a great lacking in studies exploring these constructs based on student teachers undertaking teaching practicum in the Malaysian teacher education institution. Thus, the findings from past studies might show some variance due to cultural difference as these studies were based on Western culture. Hence, the aims of this study are to investigate the inter-relationships of several personal resources, namely, emotional intelligence, self-efficacy, subjective well-being and resilience with perceived practicum stress. The main objectives of this study are to: 1. Determine the relationships of emotional intelligence with self-efficacy, subjective well-being, resilience and practicum stress; 2. Determine the relationships of self-efficacy with subjective well-being, resilience and practicum stress; 3. Determine the relationships of subjective well-being with resilience and practicum stress; 4. Determine the relationship of resilience with practicum stress; 5. Determine the mediating effect of self-efficacy on the relationship between emotional intelligence and resilience; 6. Determine the mediating effects of subjective well-being on the relationships between self-efficacy and emotional intelligence with resilience; 7. Determine the mediating effects of resilience on the relationships between self-efficacy, emotional intelligence and subjective well-being with practicum stress; 8. Develop a relational model between emotional intelligence, self-efficacy, subjective well-being, resilience and practicum stress with a PLS-SEM approach.
The investigation of the inter-relationships of the personal resources (emotional intelligence, self-efficacy, subjective well-being and resilience) with perceived practicum stress is based on the research framework shown in Figure 1. The population of this study is identified as student teachers from four teacher education institutions in Sabah, Malaysia who have underwent a recent practicum during the fifth semester. Purposive sampling was used to select the respondents from the student teachers. Purposive sampling is categorized as a non-probability sampling to obtain a representative of a larger group as close as possible (Teddlie & Yu, 2007). The respondents consisted of 70.5% females and 29.5% males with 96.5% of them who are 25 years old and a few of them who are more than 25 years old. A greater percentage of the student teachers took Teaching English as a Second Language (TESL) (64.5%) while the remaining took Chinese Language (9.5%) and Mathematics (26.0%).

Research Design
This case study is a correlational and quantitative study using questionnaires to collect information from the respondents. The sample size was determined based on the sample size determination for a PLS-SEM approach. Wong (2013) stated that the determination of a sample size needs to consider the background of the model, data distribution characteristics, the psychometric characteristic of the variables and the magnitude of their relationships. Hair, Ringle and Sarstedt (2013) proposed that a sample size should consider a significant level at 5%, a statistical power of 80% and a minimum coefficient determination (R 2 ) of 0.25. In order to determine an accurate sample size with good reliability, the G*Power 3.1.9.2 software was used. Based on the G*Power 3.1.9.2 calculator, the adequate sample size determined was 138. However, since the objective of this study is to investigate the inter-relationships of factors with possibly low quality of indicators, increasing the sample size is recommended (Wong, 2013). Therefore, the sample size for this study was decided at 200.

Measures
The main instrument in this research is a questionnaire developed from several scales to measure each of the study variables. The Teacher Sense of Efficacy Scale (TSES) adapted from Tschannen-Moran and Woolfolk-Hoy (2001) measures teachers' self-efficacy. This scale consists of 12 items using a 9-point Likert scale response. The Trait Emotional Intelligence Questionnaire (TEIQue) which was initially developed by Petrides (2009) measures emotional intelligence. The shorter version comprising of 30 items and responded using a 7-point Likert scale was used to measure specifically for global trait emotional intelligence (Cooper & Petrides, 2010). The Satisfaction with Life Scale (SWLS) developed by Diener, Emmons, Larsen and Griffin (1985) is used to measure subjective well-being. This scale has a 7-point Likert scale with five items. The measurement of resilience is based on the Wagnild and Young (1993) resilience scale consisting of 25 items and a response with a 7-point Likert scale. The 10-item Perceived Stress Scale (PSS-10) which was adapted from Cohen, Kamarck and Mermelstein (1983) is used to measure the perceived practicum stress of the student teachers in this based on a 5-point Likert scale, with "0"as the lowest and "4"as the highest.
The questionnaire was translated in Malay language using back translation method (Brislin, 1970) by language experts from a teacher education institution in Kota Kinabalu, Sabah. A panel of five experts comprising of three lecturers from the Teacher Education Institution and two lecturers from Universiti Malaysia Sabah validated the contents of the questionnaire after it was translated.

Procedures
Upon obtaining the permission to carry out data collection from the Education Policy Planning and Research Division (EPRD) of the Education Ministry in Malaysia and from the Teacher Education Institutes in Sabah, the questionnaires were distributed. Student teachers selected for this study were briefed on the purpose of the research and how to answer the questionnaire. A consent form was signed by those who indicated their willingness to participate in the survey. A period of three weeks from the distribution of the questionnaire was taken to ensure that all of the questionnaires were filled in completely and returned.
Data from the questionnaire was analyzed with descriptively using IBM Statistics SPSS 23.0 and inferentially with SmartPLS3.0 software. PLS-SEM approach determines the direct and indirect relationships of the research variables. There are two main stages of assessment done which are: the measurement models assessment and the structural model assessment.
The measurement model assessment is carried out to determine the reliability and validity of the indicators and their latent constructs (Hair et al., 2013). The assessment at this stage includes the measure of indicator reliability, construct reliability and validity, convergent validity and discriminant validity (Wong, 2013). Indicator reliability is determined using outer loading. Kock (2015) stated that the acceptable outer loading should be 0.708 and above. Construct reliability is represented by composite reliability while internal consistency is checked using Cronbach's Alpha. Both measures must exceed 0.70 to indicate construct reliability (Nunnally & Bernstein, 1994). Construct validity is measured by average variance extracted (AVE) with a threshold of 0.50 (Hair et al., 2013). Convergent validity is met when AVE is 0.50 and more with all outer loadings at least 0.40 and above (Henseler et al., 2009). Indicators with outer loading less than 0.4 are deleted while those with outer loading between 0.4 and 0.70 are considered for retention subjective to the gaining of construct reliability and validity (Hair et al., 2013). Fornell-Larcker criterion and Heterotrait-Monotrait (HTMT) ratio are used to determine discriminant validity. The Fornell-Larcker criterion explains that the square root of AVE must be greater than the correlations among the latent variables (Wong, 2013). HTMT ratio indicated a threshold of 0.850 and below to show discriminant validity (Hair et al., 2013).
Once the measurement model assessment has been done to ensure reliability and validity of the indicators and constructs, the structural model assessment is carried out. The direct and indirect relationships of the constructs were examined to determine their path coefficient, T statistics and p value, thus showing the relevancy, significance and direction of the relationships (Kock, 2015). The predictive accuracy and relevancy of the structural model were also assessed using the R 2 and Q 2 values (Wong, 2013). The effect sizes, f 2 of the exogenous construct on the endogenous construct were also determined (Hair et al., 2013). The assessment of the structural model shows how well the personal resources explain practicum stress and the inter-relationships of these resources to mitigate practicum stress.

The Measurement Model Assessment
The reliability and validity of the measurement models used in this study were determined using factor analysis. Table  1 summarizes the findings that showed the indicator reliability for the five measurement models. The result shows that the teacher self-efficacy and subjective well-being showed good indicator reliability with no item deleted. For emotional intelligence, six items were removed, while for resilience and practicum stress, two items each were removed. The composite reliability and Cronbach's Alpha provide the construct reliability of the measurement models. Table 2 presents the research findings. All values of Cronbach's Alpha and composite reliability need to exceed 0.70 to indicate construct reliability (Hair, et al., 2013), which was attained by each of the measurement models. Convergent validity is based on the average variance extracted (AVE) and the outer loading of each construct. Table 3 presents the convergent validity of the latent constructs. The results show an additional six more indicators were removed from the emotional intelligence measurement scale resulting in the removal of 12 indicators to reach an AVE value from 0.45 to 0.50 which is acceptable (Wong, 2013). Two indicators were removed from subjective well-being while another one more indicator was removed from practicum stress measurement scale resulting in a remaining seven indicators. After the removals of the indicators, the convergent validity of the measurement models was satisfied.  Table 4 presents the Fornell-Larcker criterion while Table 5 presents the HTMT ratio to show that the measurement models have discriminant validity. According to Hair et al. (2013), discriminant validity based on Fornell-Larcker criterion is determined based on the larger square root of the AVE compared to the correlation of latent constructs. Based on HTMT ratio of correlation, the threshold for discriminant validity should not exceed 0.850 (Ab Hamid, Sami & Mohmad Sidek, 2017). Both tables indicated acceptable discriminant validity of the measurement models. The result in Table 6 shows the inter-relationships of the latent constructs based on the path coefficient and T statistics values. A significant relationship is indicated when T statistics is more than 1.96 based on a one-tailed test of significance for direct relationships with a probability, p value of less than 0.05 (Hair et al., 2013;Kock, 2015;Wong, 2013). The result shows that three relationships among the constructs were not significant. Emotional intelligence is not significantly related to subjective well-being (β = 0.041, T = 0.751, p > 0.05) and practicum stress (β = -0.101, T = 1.208, p > 0.05), while self-efficacy is not significantly related to practicum stress (β = 0.075, T = 0.684, p > 0.05). All other relationships were significant. Emotional intelligence is significantly related to self-efficacy (β = 0.349, T = 4.955, p < 0.05) and resilience (β = 0.188, T = 2.757, p < 0.05) while self-efficacy is significantly related to subjective wellbeing (β = 0.657, T = 14.468, p < 0.05) and resilience (β = 0.459, T = 6.331, p < 0.05). Subjective well-being has a significant relationship with resilience (β = 0.203, T = 3.633, p < 0.05) and practicum stress (β = 0.272, T = 3.080, p < 0.05). Resilience and practicum stress are also significantly related (β = 0.190, T = 1.977, p < 0.05). The path coefficient values showed that self-efficacy (β = 0.459) is the most relevant predictor of resilience and subjective well-being (β = 0.657) while subjective well-being (β = 0.272) is the most relevant predictor of practicum stress. The indirect relationships of the research variables are considering mediators that affect the relationships of the exogenous construct on the endogenous construct. The assessment of these indirect relationships uses the measures of T statistics of more than 1.64 on a two-tailed test of significance to show significance of mediation (Hair et al., 2013). Mediation shows an effect when the paths leading from the exogenous construct to the mediator and the endogenous construct as well as the path leading to the exogenous construct direct to the endogenous construct are significant.
The first assessment on mediation is on the construct of self-efficacy that mediates the relationship of emotional intelligence and resilience. Figure 1 depicts the portion of relationship within the research model to show the mediation between these constructs. The result shows that the three paths: emotional intelligence to self-efficacy (T = 5.330, p<0.05), self-efficacy to resilience (T = 2.867, p<0.05) and emotional intelligence to resilience (T = 3.994, p<0.05) were all significant. Thus, this study concludes that self-efficacy mediates the relationship between emotional intelligence and resilience. Figure 1. The Mediation of Self-Efficacy on the Relationship between Emotional Intelligence and Resilience Figure 2 shows the portion of the research model highlighting the relationship of self-efficacy and resilience through subjective well-being. The result shows that relationships of self-efficacy with subjective well-being (T = 14.455, p<0.05), subjective well-being with resilience (T = 3.236, p<0.05) and self-efficacy with resilience (T = 9.277, p<0.05) were all significant. Therefore, subjective well-being mediates the relationship between self-efficacy and resilience.

Subjective
Well-Being

Self-Efficacy Resilience
T = 9.277, p<0.05 Figure 3 shows the mediation effect of subjective well-being on the emotional intelligence and resilience relationship. The result shows that the relationship of emotional intelligence with subjective well-being (T = 0.769, p>0.05) was not significant while the relationships of subjective well-being with resilience (T = 3.236, p<0.05) and emotional intelligence with resilience (T = 2.935, p<0.05) were significant. Wong (2013) stated that the relationships among the three constructs: the exogenous, mediator and endogenous must be all significant to show a mediating effect. Therefore, subjective well-being does not mediate the relationship between emotional intelligence and resilience. Figure 4 shows the mediation effect of resilience on the self-efficacy, emotional intelligence and subjective well-being relationships with practicum stress. The result showed that the direct path from resilience to practicum stress has a T statistic of 1.935 and a p value of more than 0.05. Two other paths were found insignificant: the indirect effect of emotional intelligence to practicum stress (T = 0.56, p > 0.05), and the indirect effect of self-efficacy to practicum stress (T = 1.79, p > 0.05). The relationships of subjective well-being with resilience (T = 3.425, p < 0.05) and practicum stress (T = 3.739, p < 0.05) were significant.

Figure 4. The Mediation of Resilience on the Relationships between Subjective Well-Being, Emotional Intelligence and Subjective Well-Being with Practicum Stress
As shown in Table 7, the R 2 and Q 2 explain the predicting accuracy and relevancy respectively of the structural model. The result shows that the predicting accuracy and relevancy of self-efficacy is small (R 2 = 12.2%; Q 2 = 0.085). The sole predictor of self-efficacy which is emotional intelligence is not enough to predict self-efficacy. The predicting accuracy of subjective well-being is moderate while its predicting relevancy is large (R 2 = 45.3%; Q 2 = 0.342). Self-efficacy and emotional intelligence contributed 45.3% of explanation to the variance in subjective well-being. Resilience also has a moderate predicting accuracy and a slightly large predicting relevancy (R 2 = 49.3%; Q 2 = 0.293). This prediction was provided by the combination of self-efficacy, emotional intelligence and subjective well-being. Lastly, practicum stress

Resilience Subjective
Well-Being has a weak predicting accuracy and relevancy (R 2 = 19.5%; Q 2 = 0.108). The combination of self-efficacy, emotional intelligence, subjective well-being and resilience predicts practicum stress weakly.  Table 8 shows the effect size in the structural model. Self-efficacy has one exogenous predictor, emotional intelligence with a moderate predicting accuracy effect size (f 2 = 0.139). Subjective well-being has two exogenous predictors, selfefficacy and emotional intelligence. Self-efficacy has a large predicting accuracy (f 2 = 0.694). In comparison, emotional intelligence has very small predicting accuracy (f 2 = 0.003). Resilience is predicted by three exogenous latent constructs, self-efficacy, emotional intelligence and subjective well-being. Self-efficacy has a moderate predicting accuracy (f 2 = 0.215) but in comparison to emotional intelligence and subjective well-being, self-efficacy has larger size effects of predicting accuracy. Emotional intelligence (f 2 = 0.061) and subjective well-being (f 2 = 0.044) have smaller predicting accuracy. Practicum stress is predicted by self-efficacy, emotional intelligence, subjective well-being and resilience. The predicting accuracy's effect size of self-efficacy (f 2 = 0.003), emotional intelligence (f 2 = 0.010), subjective well-being (f 2 = 0.048) and resilience (f 2 = 0.023) are all small. In comparison, subjective well-being predicts practicum stress with the largest predicting accuracy effect size than other exogenous predictors.  (4) 0.023

Discussion
The ability of the student teachers to identify and use emotion is important as they deal with the daily task of teaching during practicum. It has been shown in this study that emotional intelligence was able to explain 12.2% of variance in self-efficacy with a moderate predicting accuracy effect size and a small predicting relevance effect size. In addition, emotional intelligence has a significant relationship with self-efficacy. Previous study such as Gurol, Ozercan, and Yalcin (2010) also showed similar findings. Vesely et al. (2013) stated that teacher self-efficacy is a competence under emotional intelligence. The intellectual and emotional resources are needed by the teacher to deal with students' behaviour, work satisfaction and heavy workload (Chang & Davis, 2009;Vesely et al., 2013). Thus, this study concludes that student teachers need to enhance their emotional intelligence as this could help them cope with their environment.
In this study, it is shown that emotional intelligence has an insignificant relationship with subjective well-being. This contradicted with findings of Razia (2016) who noted a moderate relationship between the two constructs. The ability to control and manage emotions is not certain to make one satisfied with his life. It maybe that controlling and managing the problem helps to accept the situations that causes the dissatisfaction but does not solve the issues. Therefore, even if someone is able to accept a situation with matured emotion, yet, he might still not feel satisfied and therefore, this does not drive subjective well-being.
In the structural model of this study, subjected well-being is predicted by self-efficacy and emotional intelligence. However, self-efficacy dominates the prediction of subjective well-being with a 45.3% predicting accuracy. When a person feels confident of his capability to perform his roles and duties in the classroom, this pushes the person towards self-satisfaction. Barker and Martin (2009), showed similar findings that self-efficacy is significantly related to subjective well-being. The current finding and the supporting evidence from past studies suggest that a teacher who believes in his capability to manage a classroom is a happy and satisfied person.
The structural model was able to provide a moderate predicting accuracy for resilience. The three exogenous predictors, self-efficacy, emotional intelligence and subjective well-being explain 49.3% of variance in resilience. The effect size of predicting accuracy was moderate due to self-efficacy but smaller due to self-efficacy and subjective wellbeing. The result of this study also supported studies' findings in the past such as Sosa and Gomez (2012) and Hong (2012), indicating the positive effect of self-efficacy on resilience.
Emotional intelligence was also found to be significantly related to resilience. Schneider et al. (2013) explained that emotional intelligence can enhance the psychological resilience of a person. Studies by Dhamodharan and Ravikumar (2014) as well as Trapp (2010) also showed the positive relationship between emotional intelligence and resilience. Emotional intelligence mediates the effect of personality on stress (Austin, et al., 2010;Saklofske, Austin, Mastoras, Beaton, & Osborne, 2012). Therefore, Downey, Johnston, Hanser, Birney and Stough (2010) explained that by nurturing the emotional ability, this can increase coping resources personally.
In this study, the relationship between subjective well-being and resilience is significant. This supports past studies like Utsey, Hook, Fischer and Belvet (2008) and Windle et al. (2010). Soave (2014) stated that a person who perceives his life as happy and satisfying will be able to reduce stress due to his higher resilience. The relationship between emotional intelligence with resilience is mediated by self-efficacy and subjective well-being. Therefore, this suggests that emotional intelligence can improve self-efficacy and subjective well-being and in turn, drive more resilience.
Subjective well-being as a significant mediator to the relationship between self-efficacy and resilience implies that when a person is satisfied with his life, this can further reinforce his belief in his self-capabilities and, thus becoming more resilient. Subjective well-being relates to the cognitive perception about life satisfaction. Therefore, subjective well-being drives a positive spirit in self and also in his self-esteem. Increasing one's subjective well-being leads to greater self-efficacy and assures a resilient person against adversities. As a mediator, self-efficacy is important to ensure resilience though subjective well-being. Thus, the student teacher needs to be happy besides having a strong belief in his ability to become resilient.
In this study, subjective well-being was found as the best predictor of practicum stress compared to other predictors. A satisfied person or one who has high subjective well-being is always thinking positively, happy and less affected by stress. Subjective well-being can create a good feeling, enhances resilience and thus, reducing the impact of stress (Soave, 2014). Therefore, subjective well-being has a significant influence on stress.
Resilience was also shown to have a significant relationship with perceived stress. This is because resilience has a threshold in an individual to feel the effect of stress and among those who are resilient, this threshold is high, that the stressful environment is perceived to be relatively mild compared to the less resilient person. It becomes an important aspect to teachers who are always overloaded with work in their teaching profession (Soave, 2014). They can be facing a great amount of work, but their resilience helps them to see these things as challenges and not something to be too stressed about.
Another interesting finding discovered from this study is the insignificant relationship between self-efficacy and practicum stress. Studies like Klassen and Durksen (2014) showed that stress can affect the teachers' self-efficacy. Selfefficacy influences the tolerance on stress whereby a high self-efficacy leads to a reduced job stress. This study however, did not show that self-efficacy reduced the perceived practicum stress. This means that although a teacher may show high confidence in his capabilities to do his work, but he might still be sensitive to the effect of stress from his work environment. This study identifies a moderate level of self-efficacy for classroom management but in most studies, it has been found that a major cause for stress among teachers is classroom management (Paker, 2011;Reupert & Woodcook, 2010). Hence, teachers may be efficacious on class management but may still be subjected to feeling stress handling the situation in the classroom.
It is acknowledged from findings in this study that personal resources like self-efficacy, emotional intelligence and subjective well-being contribute significantly to predict resilience but not practicum stress. Therefore, this further strengthens the common notion that stress is a complex concept that cannot be explained from a personal resource aspect only. This study had focused mainly on personal resources which is not enough to predict practicum stress. The proposed structural model was correct in assuming the covariance among the constructs and in particular, highlighting the role of subjective well-being to explain perceived stress which self-efficacy and emotional intelligence failed to do. It also highlights the role of self-efficacy to drive subjective well-being and resilience. Although, findings from this study failed to provide a solid explanation to predict practicum stress, but the pattern of prediction from each variable were identified.
There were limitations of this study which could have contributed to the extent of findings of this study. The study used a sample of student teachers who have undergone practicum training in their fifth semester. The duration of the practicum might not be enough for the teachers to really feel the stress of teaching in real situation. Furthermore, these teachers are normally sent in batches to schools and the support from their friends might have helped them to deal with stress. The experiences that these student teachers get from their practicum training are also influence by school factor. Some schools might be supportive of the student teachers while in other schools, the student teachers may encounter an unsupportive workplace. Thus, this could have contributed to the insignificant relationships of emotional intelligence with subjective well-being and practicum stress, and self-efficacy with perceived stress.
The structural model in this study has a low predicting accuracy, at 19.5% for explaining practicum stress. The low level implies that there may be other factors to explain practicum stress; other that the coping resources examined in this study. Besides personal resources, there are also social resources to consider. This study has shown that the structural model is valid but incomplete. It has shown the importance of self-efficacy, emotional intelligence and subjective well-being to drive resilience and reduce the impact of practicum stress. The scope of the current study was limited to explore the effect of personal resources to provide resilience and reduce stress. Hence, further studies should include other variables into the model such as social resources (peer support, supervisor support) and school factors (workload given) to further enrich it and therefore, gives a more accurate prediction of practicum stress.